Home

View/Open

1. var lib ffi Library libapi init bool AL destroy bool apply_settings bool J set background img bool UShortArray set background subtraction bool bool set camera par bool ushort uint32 set deadtime bool ushort set deadtime correction bool bool set live mode bool bool get img position bool UShortArray uint32 get live image bool UShortArray get snap bool J prepare snap bool average image bool DoubleArray save image to disk bool uint32 uint32 string string 4 Module functions var sdk module exports Connect to the camera sdk init function init logger info Connecting checking connection to camera return lib init F3 Disconnect from the camera sdk destroy function destroy return lib destroy I Enables or disables dead time correction The argument enable is a boolean which decides wether to enable true or disable false the correction algorithm sdk setDeadTimeCorrection function setDeadTimeCorrection enable var success true success success amp amp lib set_deadtime_correction enable success success amp amp lib apply_settings var statusObj success success if success I var verb enable enable disable var errMsg Could not verb dead time corre
2. 4 5 6 7 15 16 17 18 19 20 21 26 27 28 29 30 31 Figure 5 Download page This page contains only one form that allows you to download measurements from the database Fill in the from and to dates and click Download data The data is a standard comma separated values file CSV with a header 12 B Specification The following pages contain the software specification It is discussed in more detail in section 2 45 Software specification The purpose of this document is to specify a software system the program for controlling the SPC2 Photon Imager the sensor Software functionality The purpose of the sensor is to measure light by imaging the light field using photon counters The raw output from the sensor is an image with 32 x 32 pixel resolution Each of these pixels represent the photons counted by one counter during the exposure I e the sensor consists of 1024 individual counters The main purpose of the program is to control the sensor The program must be able to send commands to and receive information from the sensor The following specifications are prioritized into three categories Priority A Priority B and Priority C Functionality listed under Priority A is minimum functionality that must be implemented for the program to be functional Priority B lists functions that are non essential but important for a good user experience Functionality listed under Priority C are requests from
3. This sum of frames is referred to as an image Integration time indirectly specifies how many frames we want to sum in each image The number of frames is set to the value that comes closest to the specified time The integration time is in milliseconds 11 Each measurement takes a number of images and returns the average of those images The number of images captured each time is specified in the input field Number of images Dead time is the time from the camera detects a photon until it is ready for detection again Dead time correction is an algorithm that tries to com pensate for this unproductive time It can be turned on and off with the checkbox Calbration factor is a factor that is multiplied with the calculated irradiance before it is saved in the database Use it if you need to calibrate the camera Capture interval is the amount of time between each measurement in auto matic mode The interval is in seconds Note Although unlikely in practice it is possible to set parameters such that each measurement takes longer than the interval between measurements Avoid this 2 3 2 Lens specifications In this section you can record the specifics about the lens used with the cam era The information is stored in the database with a timestamp such that it is possible to go back and see which lens was in use for the measurements 2 4 The Download page Download data December 2014 Mo Tu We Th Fr Sa Su a 2 3
4. and light colors 24 hour irradiance in Trondheim 15 16 December 2014 Irradiance u mol m s Local time Fig 3 24 hour time series Sunrise in Trondheim 16 December 2014 Irradiance p mol m s Local time Fig 4 Sunrise detail for the 24 hour time series toward yellow signify high dark counting rates Because of the long exposure time some of the pixels have saturated MPD guarantees that the dark counting rate will be within certain bounds Specifically they guarantee that half of the pixels will have a dark counting rate below 4000 cps counts per second Also 75 of the array should have dark counting rates below 25000 cps For the test image shown in figure 6 729 pixels have a DCR value under 4000 cps and 841 pixels have a DCR value less than 25000 cps In percentages that is 71 19 and 82 13 respectively The lowest dark counting rate in the array was 1955 cps The maximum is unknown because the pixels with the Sunset in Trondheim 16 December 2014 Irradiance ji mol m s 15 30 Local time Fig 5 Sunset detail for the 24 hour time series 10 i 5 4 1 5 10 15 20 25 30 Fig 6 The Dark Counting Rate DCR or noise for all the pixels in the camera 8 w N wn highest counts had saturated The mean of the DCR values was 13815 and the standard deviation was 22039 Figure 7 shows the dark counting rates in a histogram where the y axis represents the number o
5. with units pW m The irradiance value is a function of the direction This is called the cosine effect see section 2 5 Irradiance is often specified as either downwelling irra diance E4 or upwelling irradiance F Downwelling irradiance is caused by light coming from above such as sunlight hitting the ocean surface Upwelling is reflected light coming from below such as sunlight reflected off the seafloor in shallow waters The sum of irradiance from all directions in one point is called scalar irradiance Eo Scalar irradiance most closely resembles what is seen by marine organisms Sakshaug et al 2009 and is therefore the preferred unit of measurement for these purposes The spectral composition of light quantifies how much energy there is in each waveband on the visible spectrum Figure 2 shows an example of how the spectral composi tion of sunlight varies with the sun s altitude The sun s altitude is defined as the angle between the horizon and the center of the sun s disk Relative scale o ep oO fi f o K 1 300 400 500 600 700 Wavelength nm Fig 2 Spectral composition of sunlight at different solar altitudes Taken from Sakshaug et al 2009 On the Marine Night campaign the focus was to investi gate wether there is discernible differences in the available light throughout the day during the polar night Hence irradiance was the most important characteristic Irradi ance was meas
6. The camera is not built for outdoor use Thus it needed to be protected from the weather Protection was fashioned out of plastic bags It was rudimentary but sufficient for the weather conditions during the campaign 3 6 IMO irradiance sensor The IMO irradiance sensor is developed by In situ Marine Optics located in Perth Australia It is a prototype with model name ISSU and serial no 2 It has been developed to address the problem of measuring very low levels of irradiance such as is seen during the Arctic polar night The IMO sensor measured photons as digital counts from which E can be calculated It also measured temperature in C In contrast to the SPC2 camera the IMO sensor was purpose built for marine use It is built for underwater use and is equipped with a battery pack and a logging computer Thus it is fully autonomous and able to take measurements in a much wider range of conditions 3 7 Limitations The field of view of the camera was estimated to be 20 Thus the camera will only capture part of the lightfield For most of the time the measurements were done in diffused light Diffused light is the opposite of direct light meaning that it is reflected or transmitted through 9 http www insitumarineoptics com contact html a medium such as clouds This is the available light on an overcast day In these conditions it was assumed that the camera captured a representative value of the total irrad
7. a conversion factor for the sensor must be calculated Finding the conversion factor would most likely be achieved by measuring different light sources with known irradiance values and calculating the conversion factor based on those measurements 0 00025 I 0 00020 fi 0 00015 fi Epar 0 00010 fi 0 00005 fi 0 00000 fi T T T Tue 17 00 Wed 03 00 Wed 13 00 Day and time CET Fig 8 Light conditions in Ny Alesund measured with the IMO sensor Measurements taken between Tuesday 21 January and Thursday 23 January 5 2 Software The software has only basic functionality In a full featured system it should be able to e Set the camera parameters e Take test images and display them to the user e Adjust the parameters online according to preset goals e Show a live update of the acquired data e Run in both headless mode for autonomous logging and with a graphical user interface 5 8 Autonomous unit The prototype is limited in its usefulness by the fact that it needs to be connected to a computer to run A functional system should be able to operate autonomously Thus it would need both a computer to run the software and a power supply For the computer it should be as small as possible and preferably running a low resource demanding operating system That would imply porting the program for the camera to the new operating system also The power demand o
8. at the beginning of the mission and left running without user input 15 until the platform it is deployed on is recovered Lastly if the sensor needs to interface with the platform s systems special interface code might be called for 3 2 Power system The camera is powered by a 5V 2 4 power source resulting in a 12 W power requirement Power for the controlling computer is also required and is typically around 60 W In an indoors environment for relatively short periods of time power supply reliability is usually not a problem For the sensor s future intended use where it is expected to reliably measure the available light for months on end reliable power can be an issue Especially if the sensor is deployed in a location where maintenance is hard to come by the power supply must always be available A constant power supply can be obtained with a battery The battery must be sized to provide the sensor with power for longer than the worst case outage within reason A further development of the sensor would be to investigate the possibilities of an outage and calculate the probable worst case scenario A battery package could then be designed to power the sensor while the battery itself is charged as long as there is available power 16 References J Berge A S B tnes G Johnsen S M Blackwell and M A Moline Biolu minescence in the high arctic during the polar night Marine Biology 159 1 231 237 2012
9. believed that most all life came to a stop during winter Recent observations however has shown that this is not the case Berge et al 2009 2 2 The physics of visible light The following subsection is based on Young and Freedman 2012 YS og pe 750 a v 450 w 495 Fig 1 Illustration of the visible spectrum Visible light is electromagnetic radiation with a wave length A in the range 400 nm to 700nm Radiation with either shorter or longer wavelengths are invisible to the human eye The color of the visible light is a function of its wavelength see figure 1 Visible light ranges from violet with a wavelength around 400nm through blue green yellow and orange to red which has a wavelength in the range 620 nm to 700 nm The wavelengths of visible light also coincides with the wavelengths of Photosynthetically Available Radiation PAR which is discussed in section 2 4 In the above description light is assumed to behave like a wave with a given wavelength and frequency In some situations such as when describing colors this model works very well However light can also be modeled as discrete particles called photons Modern physics actually sees light as something which behaves both like a wave and like a particle This is called the wave particle duality The manifestation of light either as particles or as waves depend on the mode of measurement it is not possible to obse
10. but in JSON format It is used for plotting data in the web application apiRouter get plotData from to function req res var from new Date req params from var to new Date req params to if from to I var msg Date format passed to plotData is not valid logger warn msg res status 400 send msg else var data dataAccess getData from to function err results if err return logger error err var plotData results forEach function result I var time result time getTime value result quantumScale plotData push time time value value H logger info Sending plotdata res send plotData H 4 Sets parameters for the camera and stores the parameters in the database This route expects the parameters as a JSON object included in the request The object should have the properties integrationTime and nImages integrationTime is the exposure time of the camera in milliseconds and nImages is the number of images captured in each snap The actual measurement stored in the database is the average of these nImages apiRouter post setParameters jsonParser function req res I var args req body if largs I return res sendStatus 400 var integrationTime parseFloat args integrationTime 10 var nImages parseInt args nImages 10 if integrationTime nImages var errMsg Params to setParameters are not valid numbers logger er
11. factor for the measurements Each measurement is multiplied with the calibration factor before it is stored in the database The calibration factor is stored with each measurement apiRouter post setCalibrationFactor jsonParser function req res var postData req body if postData postData calibrationFactor logger error Invalid data sent to setCalibrationFactor logger debug postData postData return res sendStatus 400 dataAccess updateSettings calibrationFactor postData calibrationFactor function callback err if err logger error err res sendStatus 500 logger info Set calibration factor res sendStatus 200 H 4 Enables the dead time correction algorithm After a sensor has detected a photon it needs to reset before it is ready for detection again During this reset no detections will be made This time is called the dead time Dead time correction is an algorithm that tries to correct the errors in the measurements due to the dead time apiRouter post enableDeadTimeCorrection function req res I var statusObj sdk setDeadTimeCorrection true dataAccess updateSettings deadTimeCorrection true function err if err logger error Could not update dead time correction in settings err res sendStatus 500 res status status0bj success 200 500 send status0bj H H Disables the dead time correction apiRouter post di
12. for the server 2 3 1 Requirements s sal pr ey Saws hea Dk A Be od 2 3 2 Candidate C hava 4G elas Ae a 2 0 0 Candidate Java Spring Le saja ded pr d ee 2 3 4 Candidate node js and express 200 SONCISION e sl se GBPS SP ed GK pda tee as 2 4 Architecture of the server application 2 4 1 Separation of server and web application 20 Rob ststorag Lad sg se aS Pe A Pers SAG 2 5 1 Fault prevention Ak Ke 20 S28 OE BE GA EG 2 5 2 Fault tolerance suse rss dl Jo Red ndancy qg ses bros kerk a SE bekle 2 5 4 Redundancy implementation in the server application 2 6 User acceptance test sara beh a eed es st DOL DAR ST LSA ee hae BS 3 Future adaptations 3 1 Underwater operations ras Son Sage Gr Sie ER ve ka S wc OWE SY TON seat fre bare s r ge bugs s Paper 1 Development of a Prototype Sensor for Measuring ii ili iv 12 Irradiance in the Polar Night 18 Paper 2 Testing and Verification of a Low Light Irradiance Sensor 25 A User manual 32 B Specification 45 C Code documentation 50 vi 1 Introduction 1 1 The need for a new sensor Until recently it was a widely held belief that biological processes in the Arctic slows down or stops during the polar night In Berge et al 2009 the opposite is shown to be true Using acoustic measurements the process known as Diel Vertical Migration DVM is shown to persist through the Arctic winter DVM repres
13. from toUTCString and to toUTCString var model getMeasurementModel model find time gte from 1t to callback Defines the interface to this module dataAccess is an alias for module exports which is the object exported from a node js module dataAccess connect connect dataAccess disconnect disconnect dataAccess getMeasurementModel getMeasurementModel dataAccess getSettingsModel getSettingsModel dataAccess getCameraSpecsModel getCameraSpecsModel dataAccess saveMeasurement saveMeasurement dataAccess updateSettings updateSettings dataAccess getSettings getSettings dataAccess getCameraSpecs getCameraSpecs dataAccess getData getData dataAccess saveCameraSpecs saveCameraSpecs data processor js data processor js Purpose This module is a collection of functions that manipulate data in some way and where the function are too large to include in the other modules without violating the single responsibility principle Conversions and calculations on the data typically belongs here var winston require winston var logger new winston Logger transports new winston transports Console level debug colorize true H var config require confert process cwd config var tmp require tmp fs require fs async require async os require os var processor module exports function sum coll return coll reduce ad
14. get settings err res sendStatus 500 else logger debug Getting settings object res status 200 send result H H Returns the stored camera specifications from the database These are the specifications stored with saveCameraSpecs apiRouter get getCameraSpecs function req res dataAccess getCameraSpecs function gotSpecs err result if err I logger error Could not get camera specs err res sendStatus 500 else I logger debug Getting camera specs res status 200 send result PS 3 Startup Creates the server instance and starts listening for requests var listenOn config port 8080 var server app listen listenOn function Attempt to connect to the camera var connected sdk init if connected logger error Could not connect to camera Exiting process exit 1 The following functions are called in series because the database must be connected before the other functions can be called async series function connectToDatabase callback dataAccess connect config dbHostAddresses config databaseName callback This creates the autoCapture closure The reason for using a closure is to encapsulate the intervalDbject which is needed for stopping a running timer function createAutoCaptureClosure callback autoCapture function I var interval0bject return start function dataAccess getSett
15. have only one reason to change Extending this from classes to modules each of the modules in the server application should have only one responsibility Thus the module data access handles all commu nication with the databases data processor collects functions which pro cesses data in some way e g calculating the number of photons per second from the raw data The spc2 sdk and c api modules handles communica tion with the camera on different levels of abstraction The c api module is written in C and communicates directly with the camera s library of func tions It exists to abstract away C pointers from the node js application The spc2 sdk module exists to translate the C function calls into javascript The server module contains the top level logic for the entire application It defines the Application Programmer Interface API that enables other application to communicate with the SPC2 camera The API is built as a REST API which means that all communication with the API is done with standard HTTP request methods such as GET and POST It is built as an Express application Express js 2014 and handles the routing of requests from the clients to the other modules An important feature of the architecture is that there is no cross communication between modules on the same level of abstraction This is intentional and assures that each model can be used as a black box abstraction Thus each module can be extended and maintained
16. independently of the others As an example if in the future there is a desire to change the database system to the latest and greatest database one need only rewrite the data access module It does not depend on any of the other modules and can be replaced by another assuming that the new module provides the same functions as the existing 2 4 1 Separation of server and web application The server application and the web application have been developed as two separate modules This decision increases the complexity as the two could have been merged and created as one web application The reason for the separation is future use For this project it is assumed that the camera will be used to create time series of light measurements and remote access through a web application is a reasonable way to communicate with the sensor In the future it is very likely that the camera will be part of an underwater sensor platform either moored or mounted on an underwater vehicle In this case it may still be desirable to use the server client architecture but not with a web application client The server code can then easily be reused or modified for new use without concern for the web application part 2 5 Robust storage This section will present some theory on the robust storage before detailing the way data storage is implemented in the server application The theory in this section is based on Burns and Wellings 2009 Two approaches
17. on separate hard drives should be a minimum requirement Run ning on separate machines is better and on separate machines in separate locations would be ideal Strive to separate the running databases as much as possible Never run all the instances from one hard drive 1 2 3 Starting mongod A MongoDB database instance is started with the command mongod which stands for mongo daemon It takes a number of command line options some of which are detailed here The rest can be found on mongod s help page accessible with the command mongod help Options are specified on the command line with two dashes preceding it such as mongod option port lt port number gt specifies the port that the process will listen on The default port is 27017 and should be used if the instances are running on separate machines If the database processes are all running on one ma chine they must listen on separate port numbers Starting on 27017 and incrementing for each one is a good choice dbpath lt path gt specifies the path to the database files This depends on the host machine but data db is often used replSet lt set name gt specifies the name of the replica set that the process belongs to It can really be anything as long as all the members belong to the same replica set An example could be spc2 repl set logpath lt path gt saves the database logs to the file at lt path gt instead of printing it to the console This should alwa
18. the end users that most likely will not be implemented in this development cycle due to time constraints Priority C functions will not be attempted until A and B are complete Capturing images The sensor captures images of the light field in frames which is defined as an exposure of 20 74 1s The parameter integration time decides how many frames should be summed up in the output image from the sensor Dead time is the time it takes to restore the photon detector after detection of a photon During this time no detection takes place This effect introduces a non linearity in the measurements which can be reduced by a setting known as dead time correction This correction is most effective at low to moderate light intensities Priority A e The program must enable the user to set and change integration time and dead time correction e Focal length field of view and aperture all depend on the lens used and are not controlled by the program The program must include a text field where info about the lens can be entered This info will be saved in the database with the captured data e Shutter speed is controlled indirectly by the integration time e The program must set default values for all parameters and the pro gram should be able to reset the parameters to the default values Examples of default values could be integration time 0 58 and dead time correction on Actual values will be decided after testing the sensor in its ope
19. the conversion is suspect Other factors that might have affected the result but unlikely in a major way are reflectance of the light from nearby walls and background light Finally the sensor might be in error However other tests such as the 24 hour irradiance measurements have proved to be reasonable Thus it seems prudent to rule out the possible sources of error before assuming that the measurements are in the wrong Clearly other experiments must be performed to assess the correctness of the sensor This is a task for future studies 6 4 Comparison with the Walz Diving PAM The Diving PAM measured values in the range 3 pmol m s to 6ymol m s while the SPC2 measured 8 pmol m s to 9umol m s Thus the results differed with a factor of about 2 The Diving PAM has a much lower sensitivity which probably accounts for some of the error Also the SPC2 has never been calibrated against a known light source meaning that its absolute values must be treated with some suspicion The values were within the same order of magnitude which implies that the SPC2 s measurements are within reasonable limits Further calibration and a more rigorous procedure is needed to confirm this more accurately 6 5 Images of the light field An image of the light field indoors is shown in figure 9 The image was captured in a room with windows in the morning when the sun was up The exposure time is 15 ms The resolution of the image i
20. to drop to 40 compared to a clear day In terms of irradiance 40 corresponds to roughly 500 pmol m s Clouds can also increase the radiation if they are bright white and are not covering the sun Light will then be reflected off them and return to the earth surface The increase in radiation can be as much as 10 Lastly reflection is an important factor in the attenuation of light below the sea surface Glassy water i e water with no waves reflects the most light especially for low solar elevations This sea state is normal in ice filled Arctic waters A rougher sea state causes lower reflection The proportion of the downwelling irradiance that is reflected back into the atmosphere is known as the albedo Albedo ranges from 0 to 1 0 being total absorption and 1 being total reflection While the average albedo of the whole earth is 0 31 values for ice and snow is much higher and can be as high as 0 97 for ice covered with fresh snow The average albedo for the Arctic Ocean during summer is about 0 46 The reflecting properties of ice causes a destabilizing positive feedback in the climate change system In cold periods more ice forms causing more light to be reflected and further lowering the ocean temperature In the course of warm periods the ice melts which accelerates the heating of the ocean 3 METHOD The sensor used for this project is an SPC2 Photon Imager developed by Micro Photon Devices The SPC2 Phot
21. to the complexity of most larger software systems e A test can only prove the presence of faults not their absence 10 e It may be impossible to test under realistic circumstances e Errors may not manifest themselves before the system goes operational In spite of all testing and verification hardware components will fail Fault prevention is therefore insufficient if the system must continue operating without the aid of maintenance personnel 2 5 2 Fault tolerance In many ways fault tolerance is the counter example to fault avoidance While avoidance focuses on not experiencing faults in the first place fault tolerance assumes that faults will be present and implements strategies to continue operating despite this fact It is important that fault tolerant sys tems caters for unforeseen faults as well as the anticipated ones Fault tolerance can be implemented on three levels of complexity Full fault tolerance The system continues to operate in the presence of faults albeit for a limited time period It experiences no significant loss of functionality or performance This state of operation is not possible to sustain indefinitely Graceful degradation Also called fail soft The system continues to oper ate accepting a partial degradation of functionality or performance Fail safe The system saves its state and comes to a halt 2 5 3 Redundancy Redundancy is a common way to implement fault tolerance Protective re d
22. var processor require modules data processor data processor js var autoCapture var logger new winston Logger transports new winston transports Console level debug colorize true H HTTP access control CORS The following middleware adds the Access Control Allow Origin to every re sponse The purpose of this is to allow clients to make requests to this domain which is different from the domain the request was made from For more information see this article on MDN app use function req res next res header Access Control Allow Origin res header Access Control Allow Headers Origin X Requested With Content Type Accept next 4 Application Programming Interface The rest of the file sets up the routes for the API and defines the functions to be called for requests to those routes Common for all of the routes is that they respond with 200 on success and either 400 or 500 for errors on the client or server side respectively Use the apiRouter for all calls to addresses starting with api app use api apiRouter Checks for a connection to the camera Responds with 200 if successful and 500 if a connection does not exist and a new one can not be made apiRouter get connected function req res I var connected sdk init res sendStatus connected 200 500 4 Disconnects from the camera Note that this route responds to POST requests while co
23. 69 72 L nne L 2014 Development of a prototype sensor for measuring irradiance in the polar night Sakshaug E Johnsen G H and Kovacs K M 2009 Ecosystem Barents Sea Tapir Academic Press A User manual The following pages contain the user manual meant for the end users of the system It details installation of the server and the web application and the user interface 32 User manual for the SPC2 software package Lars L nne larslonn stud ntnu no 14 December 2014 1 Installation 1 1 Connecting the camera to the server 1 1 1 Inputs The SPC2 camera has five inputs Only two of them will be used for this application One is the power which has no label and the other is the USB Figure 1 shows the inputs on the camera 1 1 2 Connections Connect the USB cable to the camera and to the server Connect the power cable to the camera The camera can take up to 10 seconds to start up after the power has been connected 1 2 Databases The server application needs at least one database to save measurements settings and metadata in The database must always be available so failure preventions must be considered when implementing the database solution The chosen database system is MongoDB It was chosen because it is an open source document driven database which works particularly well with node js which is the platform that the server application is written in It Figure 1 SPC2 inputs also support
24. A e The program must enable the user to access data stored in the database The program must offer the user to download the entire database either as a database file or as a dump of the database e The program should display the captured data through a web inter face in the form of a plot The units on the displayed data should be umol m s Priority B e The program should also make parts of the collected data available such as data captured within a specified timeframe This timeframe is specified by the user The data can then be downloaded as a structured plain text file such as CSV Remote access Priority A e The program must enable the user remote access to the sensor This means that the user must be able to access the captured data over a network either the internet or a local network e Read access must be provided remotely i e the user must be able to inspect and download captured data Priority B e Controlling the sensor e g by setting parameters may be provided remotely Security Priority A e The program must be secured against unauthorized access e The program must ensure that the user is authenticated before the user is allowed to download data or set sensor parameters Priority B e Access to view the captured data may be allowed without authentica tion Hardware functionality Power supply Priority B e Review the available power supply e Plan a robust and failsafe system for continuo
25. CEDURES There are three elements of the camera that must be calibrated the sensor s active area diameter noise and the photon detection efficiency The information on calibration comes from communications with the manufacturers of the camera Micro Photon Devices Active area The active area of the sensor is perfectly determined by the manufacturing process It needs no further calibration The active area diameter is 20 um Noise Noise is called Dark Counting Rate DCR by MPD It is the photon count registered by the camera in a completely dark environment The procedure to calculate it is as follows 1 Put the camera in a completely dark environment and acquire a frame with a long exposure time MPD recommended 10 s 2 Divide the count in each pixel with the exposure time to get the DCR for each pixel Use the data as is or sum it to get DCR for the whole pixel array 3 Check that the measured DCR is compatible with MPD s measurements which are provided with the camera Photon detection efficiency Measuring the photon detec tion efficiency is a simple concept but difficult in practice The difficulties lie in that the procedure calls for calibrated monochromatic source and a method for determining the exact power the camera sees at the given wavelength MPD has performed tests of photon detection efficiency for every wavelength from 400nm to 1000nm with a 50nm step The results are included with the camera and
26. ISSN 0025 3162 doi 10 1007 800227 011 1798 0 URL http dx doi org 10 1007 s00227 011 1798 0 Jorgen Berge Finlo Cottier Kim S Last Oystein Varpe Eva Leu Janne Sor eide Ketil Eiane Stig Falk Petersen Kate Willis Henrik Nygard Daniel Vogedes Colin Griffiths Geir Johnsen Dag Lorentzen and Andrew S Brierley Diel vertical migration of Arctic zooplankton during the polar night Biology Letters 5 1 69 72 2009 Alan Burns and Andy Wellings Real Time Systems and Programming Lan guages Ada Real Time Java and C Real Time POSIX Addison Wesley Educational Publishers Inc USA 4th edition 2009 ISBN 0321417453 9780321417459 Hakan Erdogmus On the effectiveness of test first approach to programming In Proceedings of the IEEE Transactions on Software Engineering 31 1 2005 Express js Express node js web application framework December 2014 URL http expressjs com Robert C Martin The single responsibility principle The Principles Pat terns and Practices of Agile Software Development pages 149 154 2002 MongoDB MongoDB December 2014 URL http www mongodb org Mongoose js Mongoose odm v3 8 20 December 2014 URL http mongoosejs com 17 Development of a Prototype Sensor for Measuring Irradiance in the Polar Night Lars L nne Norwegian University of Science and Technology Trondheim Norway e mail larslonn stud ntnu no Abstract This report describes the process of developing a p
27. NTNU Trondheim Norwegian University of Science and Technology Development of an Irradiance Sensor Based on a Photon Counting Camera Lars L nne Master of Science in Engineering and ICT Submission date December 2014 Supervisor Asgeir Johan Sgrensen IMT Norwegian University of Science and Technology Department of Marine Technology NTNU Trondheim Norwegian University of Science and Technology Department of Marine Technology MASTER THESIS IN MARINE CYBERNETICS FALL 2014 FOR STUD TECH Lars L nne Development of a Prototype Sensor for Measuring Irradiance in the Polar Night Work description In January 2014 the Marine Night project conducted a research campaign in Ny lesund As part of this campaign a new prototype sensor was tested The sensor measures minuscule amounts of light such as are found in the Arctic polar night This light might be important for life in the Arctic and researchers need tools such as this sensor to measure it The initial test of the sensor was promising Now for the researchers to be able to fully use this new sensor there exists a need for a more complete software package than what was developed for the previous campaign The aim of this thesis is to develop a software package that will enable researchers to use this sensor for measuring light in the Arctic polar night This work is a continuation of the candidate s previous work This master thesis should consist of two scientific
28. arch Council of Norway The thesis is edited as a collection of papers with a r sum in front The first paper is titled Development of a Prototype Sensor for Measuring Irradiance in the Polar Night and the second paper is titled Testing and Verification of a Low Light Irradiance Sensor ill Acknowledgements I would like to thank my supervisor Professor Asgeir J Sorensen and my co supervisor Professor Geir Johnsen for their contribution to this thesis Asgeir has been an excellent guide in this project and is always a source of great advice Geir has always been enthusiastic about the project and has provided a number of good ideas along the way They have been a joy to work with I would also like to thank my colleagues at Kantega in Trondheim While working there I have learned a great deal about software development and web development in particular That knowledge has proved invaluable for this project A special thanks goes to Inga and Sigrid who makes it all worthwhile Contents Abstract Sammendrag Preface Acknowledgements Contents 1 Introduction 1 1 The need for a new sensor 0 0 00484 1 2 Development of the new sensor 1 3 Structure of the thesis o oo a a 2 Software system 2 1 Development concepts 4 2 oa ee a a a 2 1 1 Test driven development 2 2 Software specification aa dads Pont oes a a FPS ican Gt ei she Ain gs 2 3 Selecting a language and framework
29. are also shown in figure 2 The dotted lines show the PAR band i e 400nm to 700nm The camera counts photons up to 1000 nm If the spectral distribution of the measured light is known figure 2 can be used directly to calibrate the sensor The raw measurements would be saved by the sensor and Photon detection efficiency of the SPC2 Efficiency in Frequency in nm Fig 2 Photon detection efficiency adjusted later using a weighted average with the weights taken from the spectral distribution of the incoming light A more realistic scenario is that the spectral distribution is unknown In the case of the polar night all available light is indirect The indirect light is mostly blue Sakshaug et al 2009 The SPC2 is also most sensitive in the blue band making the polar night an ideal environment for this sensor The users will have to make some assumptions about the actual spectral distribution and calibrate the measurements according to these assumptions 5 TEST SETUP 5 1 24 hour time series This test aimed to capture the light dynamics in a 24 hour period The test was performed in Trondheim Norway The camera was situated in a window inclined at roughly 30 from the horizontal with the lens pointing to the sky The test was performed in an urban area so some light pollution should be expected but the camera was positioned to avoid pollution as much as possible To protect the camera s buffers from numeri
30. asure the outside ambient light The weather was clear The light sensor on the Diving PAM is much smaller than the lens on the SPC2 The SPC2 is believed to be able 2 Heinz Walz GMBH Eichenring 6 91090 Effeltrich Germany 3 From the company s website www walz com to capture a larger light field than the Diving PAM Also the light sensor on the Diving PAM is cosine corrected meaning that a diffuser is placed over the light sensor to account for light coming in at different angles These two differences will probably account for some deviations in the results 5 5 Capturing the light field as an image Lastly the sensor was tested as a more conventional camera to see how well it performed when capturing images of the light field This experiment used the Capture test image function of the web application which captures an image of the light field converts it to the PNG format and displays it to the user Two images were captured in normal daylight conditions indoors In the first image only the ambient light was present The second image was captured with the same settings but with an artificial light source pointed at the camera to see if it is possible to discern light sources on the images 6 RESULTS 6 1 24 hour time series Figure 3 shows the result of measuring the irradiance in Trondheim for 24 hours on 15 and 16 December 2014 The graph shows a couple of interesting facts First irradiance values when t
31. atabase Saves are done asynchronously and the provided callback is called when the save is complete function saveMeasurement measurement callback var Measurement getMeasurementModel var m new Measurement measurement m save callback F Similar to saveMeasurement but updates instead of saving a new object function updateSettings newSettings callback var Settings getSettingsModel Settings update set newSettings upsert means to insert if a record does not exist and update otherwise upsert true callback Identical to saveMeasurement but for camera specs function saveCameraSpecs specs callback var CameraSpecs getCameraSpecsModel var cs new CameraSpecs specs cs save callback Reads the settings from the database function getSettings callback logger debug called getSettings getSettingsModel findOne callback Reads the camera specs from the database Each set of specs are saved by itself with a timestamp This function gets the last one function getCameraSpecs callback getCameraSpecsModel find sort time Limit 1 exec callback Get measurements taken between from and to All parameters must be specified function getData from to callback if from to callback return logger error data access getData called with invalid arguments logger info Getting measurements between
32. ated in the autoCapture closure apiRouter post startAutoCapture function req res logger info Starting autocapture autoCapture start res sendStatus 200 H Stops automatic capturing apiRouter post stopAutoCapture function req res logger info Stopping autocapture autoCapture stop res sendStatus 200 4 The user can store metadata about the camera in the database for later refer ence This route accepts the meta data as a JSON object and saves it to the database The JSON object may include the properties focalLength aperture and description apiRouter post saveCameraSpecs jsonParser function req res var postData req body if postData logger error Invalid data or no data posted to saveCameraSpecs return res sendStatus 400 dataAccess saveCameraSpecs postData function err if err logger error Could not save camera specs err res sendStatus 500 else logger info New camera specs saved res sendStatus 200 H H Returns the stored settings from the database Settings include the number of integrated frames the number of images captured in each snap seconds between each capture in automatic mode the calibration factor and a boolean telling us if dead time correction is enabled or not apiRouter get getSettings function req res dataAccess getSettings function gotSettings err result if err logger error Could not
33. b application must run on https only 2 Web interface The web interface is the main interface for end users Through it users can examine the latest measurements set capture parameters and download data from the camera This section describes each part of the web interface in detail and explains how to use it 2 1 The main page pmol m s Time in GMT from to Draw graph Figure 2 Main page This is the main page of the application By default it shows a graph of the measurements taken in the past six hours Time in GMT is on the x axis while light measurements on the quantum scale is on the y axis By moving the mouse pointer over the graph it is possible to read the exact measurement at that time instant Below the graph are the graph controls By choosing the from and to date and clicking the Draw graph button you can show measurements for a specific time range Beware that all the data must be read from the database so if you choose a large time range it might take some time before the graph is shown again 2 2 The Control page Automatic capturing In automatic capture mode the sensor captures an image processes it and saves the data to the database in regular intervals The interval is controlled from the Settings page The buttons below start and stop automatic capture Start automatic capture Stop automatic capture Capture test image Capture a test image by setting the integration time and pres
34. base section To run the server appli cation the node js platform must be installed See the node js webpage for instructions Also before proceding the camera should be connected to the server 1 3 2 The config file The server s configuration file is called config yaml and should exist in the same directory as server js server js is the main file of the server application The database specific entries in the configuration file are dealt with in the sec tion on databases Specifically the variables dbHostAddresses and databaseName should exist and have its correct values The config file may include a variable called port which specifies which port the server should listen for connections on If port is not specified the server defaults to listening on port 8080 The config also has an entry named sensorArea which should be 3 14e 10 This is the surface area of the sensor in the camera and is used by the server application for calculating the irradiance values 1 3 3 Installing dependencies The server application depends on a number of external modules to function The dependencies are all listed in the file package json and can be installed using npm the node package manager npm is installed with the node platform and should be available on your system Run npm install to install all dependencies 1 3 4 External dependencies The server application depends on http www graphicsmagick org for con verting the t
35. cal overflow the camera operates in normal mode In this mode the camera captures frames with constant exposure time of 20 74 ys Multiple frames are integrated or summed to achieve a longer exposure time For this experiment the exposure time was set to 10ms which corresponds to 482 frames per image 482 frames does not add up to exactly 10ms but rather 9 997 ms because the exposure time can only be set in discrete steps No calibration was done prior to this experiment so the calibration factor was set to 1 Each data point is the average of 5 images 5 2 Dark counting rate When the camera is placed in a completely dark environ ment it will get counts from the photon counters These counts are false positives and contribute to the noise in the measurements The noise is referred to as the Dark Counting Rate DCR To measure the DCR it was necessary to capture images in a completely dark environment This was achieved by capturing images in a dark room To further minimize stray light the lens was removed from the camera and the sensor was covered with a cap The camera was then positioned face down on a table Measuring the dark counting rate requires that the images are acquired with a long exposure time Because the camera operates in normal mode the exposure time has to be a multiple of 20 74 us and the number of captured frames is limited to 65535 Thus the maximum exposure time is limited to 1s The exposure t
36. change integration time and dead time correction Yes Save information about camera lens Yes Reset parameters to default values No Store values for all parameters Yes Capture images in automatic mode Yes Store data in a reliable and secure manner Yes Each image or equivalent data must be stored Yes Store result in quantum scale No Parameters must be stored with the data Yes Calibration factor must be stored with the data Yes User access to stored data Yes Download entire database No Display captured data as plot in web interface Yes Remote access to the sensor both read and write access Yes Secured against unauthorized access No Table 1 Results from the acceptance test Overall the user was satisfied with the product There were a couple of points that needed to be addressed before delivery The functionality to reset parameters to their default values was dropped The function was not yet implemented but the user did not think that the reset functionality had much value Parameters will have to be ad justed for every experiment and it would be very difficult to find good defaults In addition to the values that were already stored in the database the user would like to also record temperature The camera does not have a temperature sensor and the logging of temperature were not in the original specifications It was decided that temperature logging would not be implemented in this version When examining t
37. ction logger error errMsg statusObj error errMsg return status0bj I Sets the camera s capture parameters the number of integrated frames and the number of images per snap sdk setCameraParameters function setCameraParameters nIntegratedFrames nImages var success lib set camera par nIntegratedFrames nImages if success logger error Could not set camera parameters success lib apply_settings if success logger error Could not apply camera settings return success success Captures an image according to the parameters set with sdk setCameraParameters The returned image is an array of values each value representing one photon detector in the camera The values are the averages of all images captured in the snap sdk captureImage function captureImage I var time new Date prepareAndGetSnap var image new DoubleArray 1024 var gotImage lib average_image image if gotImage return logger error Could not get average image of snap var arr new Array 1024 for var i 0 i lt 1024 i I arr i imagelil return data arr time time ja This function also captures an image like sdk captureImage but its purpose is a little different Instead of capturing multiple images and returning the average data this function captures just one image and returns the image data as a tiff file It also accepts parameters as inp
38. ctions Writing a server program from scratch is time consuming and unnecessary Frameworks exist that will handle the low level details allowing developers to focus on the parts that are unique to their implementation The choice of language should have a good framework that will make development easy and efficient To enable the client to communicate with the server and indirectly with the sensor the server must expose some kind of API Application Programming Interface This server will expose a REST API which uses an internet media type often JSON to transfer data through standard HTTP methods such as GET POST PUT and DELETE The client will then make AJAX calls to query the server and control the sensor AJAX is an acronym for Asynchronous Javascript XML and is the standard technique to make asynchronous calls to an API on the client side Nowadays JSON is often favored over XML as the data format language of choice The choice of programming language and framework should facilitate the creation of this API The API must be secured against unauthorized use and attacks For this application basic access authentication will suffice The choice of language and framework should enable the developer to easily implement this authen tication scheme Shttps developer mozilla org en US docs Web HTTP Basic access authentication 2 3 2 Candidate C The server has to make calls to a C library and this is most easily done f
39. d function add x y I return x y Module functions Calculates the number of photons per second measured in an image The image is an array where each element represents one photon detector in the camera processor getPhotonsPerSecond function photonsPerSecond numIntegFrames image if numIntegFrames lt 0 I logger error Called photonsPerSecond with invalid numIntegFrames return 0 In normal mode the exposure time is always a multiple of 20 74 microseconds var exposureTime numIntegFrames 20 74e 6 var detectedPhotons sum image return detectedPhotons exposureTime 3 Converts the photons per second count to the quantum scale micromol photons per square meter per second Requires that we know the sensor area which is specified in the config file processor convertToQuantumScale function quantumScale photonsPerSecond calibrationFactor var avogadro 6 022141e 23 var area config sensorArea 0 if area 0 logger warning Sensor area is not defined in the config file Returning 0 for the quantum scale return 0 return photonsPerSecond 1e6 avogadro area calibrationFactor 3 Builds a standard CSV comma separated values file with as the separator and a descriptive header results is an array of measurement objects from the database Note Measurements are not sorted processor createCSV function createCSV results callback var h
40. e This is the translation of the SDK from C to javascript This module uses the module ffi which stands for foreign function interface It lets us create javascript functions from the C functions by defining them as parameters to the ffi Library function External modules var ffi require ffi ref and ref array simplifies handling of C pointers and arrays from javascript by using its predefined types var ref require ref var ArrayType require ref array var winston require winston var logger new winston Logger transports new winston transports Console level debug colorize true H tmp is a module for creating temporary files var tmp require tmp gm is a module for working with images It requires GraphicsMagick to be installed var gm require gm Type definitions Here we define the C types and C array types that we re going to need in the library function calls var ushort ref types ushort var double ref types double var DoubleArray ArrayType double var UShortArray ArrayType ushort Creating the javascript functions The call to ffi Library creates a javascript object which contains javascript functions for all the C library functions that were specified in the parameters After this call the functions can be called just like any other javascript function Each library function is defined with its name return type and the types of its arguments
41. e livsformene trenger en ny type lysm ler siden m lerene som er tilgjengelig i dag ikke er f lsomme nok for denne m rke tiden p ret I forbindelse med et forskningstokt i Ny lesund i januar 2014 ble en proto type av denne nye sensoren testet Toktet ble utf rt som en del av forsk ningsprosjektet Marine Night Resultatene var lovende og arbeidet begynte dermed for utvikle et mer komplett system Dette nye systemet skal kunne m le de arktiske lysforholdene gjennom lange perioder av ret Utviklingen av dette sensorsystemet og de f rste testene er beskrevet i denne oppgaven Dette prosjektet endte med et system som tilfredsstiller de prim re kravene fra forskerne Det skal testes p det neste toktet i regi av Marine Night i Ny lesund i januar 2015 En brukermanual har blitt skrevet for dette toktet og er vedlagt som en del av oppgaven Dokumentasjon av vesentlige deler av koden er ogs vedlagt og skal v re til hjelp for framtidige utviklere av systemet il Preface This master thesis is written during the fall semester of 2014 as the final part of the master program Engineering and ICT at the department of Marine Technology at the Norwegian University of Science and Technology NTNU in Trondheim This thesis is written in close cooperation with researchers from the Marine Night research project Marine Night is part of the larger Mare Incognitum research project registered as project number 226417 with The Rese
42. e of a point source of light Light source marked with an arrow 7 CONCLUSION The experiments that have been described in this paper were meant to test the new sensor system and confirm that the measurements it provides are reasonable Most of the experiments were successful in this regard The 24 hour time series demonstrated that the sensor is able to measure the dynamics of the light field It also established that the sensor has a saturation limit which can cause measurement problems when the difference in irradiance is large For measuring light in the polar night this should not pose much of a problem since the irradiance levels are relatively low throughout the day For other uses this issue will have to be addressed either by manual intervention or by extending the sensor to set the exposure time autonomously The dark counting rate experiment shows that the DCR values are very reasonable and well within the guaranteed levels from the manufacturer Additionally the experiment increases the confidence in the new sensor system in that it captures and delivers the correct images Measuring the irradiance from a candle was not success ful No conclusions will be drawn from this experiment because there are too many uncertainties which are dis cussed in section 6 3 Comparison with the Walz Diving PAM showed that the sensor s output falls in a reasonable range This indicates that the sensor s measurements needs f
43. eader time numIntegratedFrames numImages photonsPerSecond quantumScale calibrationFactor os EOL tmp file function tmpFileCreated err path if err return logger err err logger debug Created temp file path var stream fs createWriteStream path flags w encoding utf8 H stream write header Process each measurement object in the results There is no guarantee that the results will be processed in order ref the async module async each results addResultToFile function err if err logger error createCSV Could not write to stream err stream end H stream on finish function callback null path H Called for each measurement object in the results array Assembles a CSV line and writes it to the file function addResultToFile result cb delete result data var s result time toISOString result numIntegratedFrames result numImages result photonsPerSecond result quantumScale result calibrationFactor os EOL stream write s cb H Deletes unlinks a file Typically used to delete a file created with the createCSV function when the user has downloaded it processor deleteFile function deleteFile path fs unlink path function err if err return logger error Could not unlink file err logger debug Unlinked file path H spc2 sdk js spc2 sdk js Purpos
44. eak at all This was a snowy day with a low cloud cover Micro Photon Devices claim that the sensor has a very high noise immunity as described in section 3 1 This is verified by figure 5 There is little if any trace of electronic noise Figure 5 shows the values in Epar The output from the program is in photons s A rough calibration has been done based on the values from the IMO sensor This calibration procedure is explained in section 3 8 4 06 04 5 8 60 04 4 8 20 04 4 2 8e 04 4 240 04 4 20004 4 1 686 04 4 1 2e 04 4 8 0e 05 4 4 0e 05 4 0 0e 00 12 00 at gven date January 2014 Fig 5 Light conditions in Ny Alesund measured with the SPC2 camera On clear days such as 19 January there is a clear peak at noon This corresponds well with photographs that were taken at the same time This is shown in figure 6 The background photograph is taken with a fisheye lens just outside the Marine Lab The blue line on the plot shows exactly when the photo was taken Fig 6 Light conditions at midday Background photo Geir Johnsen The graph also corresponds well with the conditions at night Figure 7 shows the irradiance values at night under a clear sky The low light levels can be verified both on the photograph and on the plot a Fig 7 Light conditions on a clear night Background photo Geir Johnsen The plot also shows multiple isolated spikes The spikes are severa
45. ensor Lars L nne Norwegian University of Science and Technology Trondheim Norway e mail larslonn stud ntnu no Abstract A new sensor has been designed and developed for measuring the low levels of irradiance in the Arctic polar night As a part of the development a number of experiments have been carried out to ensure that the sensor performs as expected The experiments test the sensor s ability to capture the dynamics of the changing light and its measurement noise The sensor s correctness is also tested by comparing the output to another sensor The tests are mostly positive indicating that the sensor works as expected 1 INTRODUCTION Researchers working at the University Center on Sval bard UNIS have discovered that life forms in the Arctic might be dependent on the light in the polar night Berge et al 2009 The level of illumination during the polar night is often too low for the human eye to register Also commercially available light meters are not sensitive enough to measure these microscopic light values The need to measure these low light levels has led to the testing and development of a new irradiance sensor which is presented in this paper The new sensor is based on a product from Micro Photon Devices MPD called the Single Photon Counter Cam era SPC2 The SPC2 is capable of single photon detec tion with low dark counting rates and capturing frame rates up to 48000kHz The camera was
46. ents a huge movement of biomass and is of great importance to the Arctic marine ecosystem Further Berge et al 2009 argues that DVM in the Arctic polar night is controlled by variations in solar and lunar light intensity Light intensities in the polar night are miniscule and well below the intensities where the human eye is effective To study the effects of the light in the polar night the intensity of the light must be reliably measured Such low intensities calls for a sensor which is much more sensitive than what is generally available in the market today thus it must be custom made A camera called the Single Photon Counting Camera SPC2 made by Micro Photon Devices was chosen as a new light measurement instrument It is a precision instrument capable of single photon detection with an efficiency from 15 to over 40 in the 400nm to 700nm band The sensitivity of this camera will enable researchers to quantify the light levels of the Arctic polar night and further investigate the biological processes during winter in the high Arctic 1 2 Development of the new sensor The SPC2 provides the possibility of measuring tiny intensities of light but it is not delivered with suitable software for this purpose The company provides a Software Developer Kit SDK and a test application which is able to capture images and try out the camera This project needed to capture time series of light measurements for this purpose the test app
47. eriod of time Thus the software is expected to be reasonably robust and the storage facilities must be able to safely record the data The data must be protected against such eventualities as power outages and hardware failure Together with the users of the sensor a specification of the sensor functionality was drawn up before development began The specifications are discussed in detail in the r sum that accompanies this paper One point that is contrasting to the last iteration of the sensor is that it must have remote control capabilities The users expect to monitor and control the sensor from any other location over the internet Ultimately the most important expectation is that the sensor stores correct measurements and that the users are confident in its continued safe operation 3 THEORY 3 1 PAR measurements The sensor measures light as a count of the incoming photons in a given time frame The count is then con verted to pmol m s which is a standard way to quantify Photosynthetically Active Radiation or PAR All photons in the PAR band carry enough energy to start photosyn thesis It is the number of photons that count not the total energy Thus it is convenient to measure irradiance on the quantum scale when talking about photosynthesis Sakshaug et al 2009 PAR can also be expressed in energy units W m This is preferred when doing energy calculations such as mea suring solar energy flux in the pola
48. est images from TIFF to PNG format It must be installed on the machine that the server application is running on On an Ubuntu Linux server it is installed like this sudo apt get install graphicsmagick If you are running a different system consult the GraphicsMagick webpage 1 3 5 Compiling the C API The last step before starting the server is compiling the C API This job is done with the tool scons If it is not installed on your machine install it now through your package manager SConstruct files are provided for both Mac OS X and Linux Compile the C API by calling scons like this scons f SConstruct linux Change linux to osx if you are running on a Mac 1 3 6 Starting the server The server application includes a startup script called run server sh It is written in bash and will run on all Mac OS X and Linux systems It will not run on Windows except through Cygwin or something similar The server is run using a program called supervisor which restarts the server in case of failures If this program is not installed it can be installed with npm npm install g supervisor Then run the script to start the server run server sh 1 3 7 Running The server should now be running It will try to connect to the camera and the database and log any errors to the console Usual problems at this stage is not having the camera connected or not having a running database process The server should now be up and run
49. f pixels and the x axis represents the counts per second Clearly most of the pixels have DCR values which are less than 10000 cps 6 3 Irradiance of a candle Figure 8 shows 10 irradiance measurements of a burning candle at a distance of approximately 1 ft The red line shows the mean value which is 8 1283 umol m s 1 foot candle is approximately 10 1x 1 klx is approximately 14 18 pmol m s see section 5 3 Converting the measured irradiance of the candlelight to foot candles gives a value which is about 50 times too large before taking into account the photon detection efficiency Count N 3 4 Counts per second x10 Fig 7 Histogram showing the dark counting rates for all the pixels in the camera Irradiance measurements of a candle at 1 ft o Irradiance u mol m s 4 5 6 ta Measurement number Fig 8 Irradiance measurements of a candle The red line shows the mean of the dataset Note that the y axis does not start at zero There are multiple possible sources for this error First the experiment is very simple It is not known for certain how much light is emitted by the candle it was only assumed to be 1 foot candle at a distance of 1 ft The conversion factor in Sakshaug et al 2009 is valid in sunlight In this experiment it has been assumed that the conversion factor holds which it might not Converting between pmol m sand lx is dependent on the spectral distribution and without knowing it
50. f the whole system must then be calculated to come up with a power supply that is reasonably small and lightweight but still can run the camera for an acceptable length of time 6 CONCLUSION This paper has described the development of a prototype irradiance sensor using the SPC2 Photon Imager which measures the photon count of the ambient light field Development of the sensor included both development of the necessary software deployment of the sensor in the Arctic environment during the polar night and a comparison of the measurements from this sensor with those of another prototype The results from the prototype sensor were promising The measurements clearly showed differences in the ambient light levels throughout the day These measurements will be of value to researchers studying life in the high Arctic during the polar night REFERENCES Berge J Cottier F Last K S Varpe O Leu E Soreide J Eiane K Falk Petersen S Willis K Nygard H Vogedes D Griffiths C Johnsen G Lorentzen D and Brierley A S 2009 Diel vertical migration of Arctic zooplankton during the polar night Biology Letters 5 1 69 72 Sakshaug E Johnsen G and Kovacs K M 2009 Ecosystem Barents Sea Tapir Academic Press Trond heim Young H D and Freedman R A 2012 University Physics With Modern Physics Volume 1 CHS 1 20 Pearson Education Testing and Verification of a Low Light Irradiance S
51. for the new functionality 3 Write production code New code is now added to the code base Only the minimal code needed to make the test pass is added at this stage 4 Re run tests All tests are now run again to make sure that the new test passes If it does not the developer goes back to the previous step This cycle is repeated until all tests pass 5 Refactoring As the code base grows larger refactoring becomes in creasingly important Duplication must be removed code can be moved to where it more logically belongs and code can be rewritten to increase readability or maintainability At this point all functionality is covered by the functional tests After refactoring all tests can be run again to make sure that the refactoring has not introduced any new bugs According to a study from the National Research Council of Canada Er dogmus 2005 following the TDD technique leads developers to write more tests be more productive and it raises the minimum quality of the finished software It also increases confidence in the correctness of the program which is especially important for a long running process such as this 2 2 Software specification A software specification is a detailed list of the customer s requirements for the software system The specification is a crucial part of the development process because the specification document lists the requirements for the system Both functional and non functional requiremen
52. ging 3 8 Calibration The values from the IMO sensor was used to calculate a rough calibration constant for the SPC2 camera By matching the peaks in the data set from the IMO sensor with the same peak from the camera and assuming that the data from the camera was representative for the entire light field 1 a rough calibration constant was calculated The data from the IMO sensor was filtered using a Butter worth filter This reduced most of the dark current so the data was easier to analyze Disregarding the outliers which are likely anthropogenic in nature the maximum of the dataset was 1 8232 x 1074 This occurred on 21 January 12 26 47 CET The corresponding time was found on the graph of the data from the camera The photon count at that time was 1 9298 x 107 Thus the calibration factor was 1 8232 x 1074 Using this factor the values from the camera were con verted from photons s to pmol photons m s 4 EXPERIMENTAL RESULTS 4 1 SPC2 photon imager The measurements done with the SPC2 camera is shown in figure 5 The horizontal axis shows the time each tick 10 See discussion in section 3 7 mark is placed at noon on the given date The vertical axis shows Epar values the combined value of E for all wavelengths in the 400 nm to 700 nm band On most days there is a peak in the measurements around noon The peak is lower or higher depending on the weather conditions Some days like 20 January hardly has a p
53. god being the process that runs an instance of the database system One of the processes is named the primary and receives all write commands from the client The client in this case being the SPC2 server application Every other mongod process is called a secondary Every secondary process ensures that they have the same data set as the primary by applying the primary s operations To maintain the set of database processes each member of the replica set sends heartbeats to all other members every two seconds If one member is not heard from in 10 seconds it is assumed offline Then the remaining members elect a new primary among themselves In the server application the addresses of database processes is entered into the config file On startup the application through the mongoose module Mongoose js 2014 will connect to the replica set and provide seamless static redundancy The application assumes that the mongod processes are configured and belongs to the same replica set This process is detailed in the user manual in appendix A 2 6 User acceptance test On 24 November 2014 an acceptance test was performed with the user The software was presented along with a table of the priority A requirements from the specification The results are shown in table 1 I The user in this context is Prof Geir Johnsen Department of Biology NTNU 12 Functionality Accepted Set and
54. gration time and dead time correction are set here Changing settings Automatic capturing will use the parameters that were in effect when it started Changing parameters after It has started will not affect the running process To make automatic capture use new settings go back to the Control page and restart the capture process Integration time 15 Number of images 10 Dead time correction Calibration factor 1 Capture interval 60 Submit Set lens specifications The details about the lens used with the sensor can be saved with this form The data Is stored in the database with a timestamp Focal length 8mm Aperture na Description Made in Japan by Tamron Submit Figure 4 Settings page The settings page allows you to set the capture parameters for the camera and save information about the lens used with the camera 2 3 1 Capture parameters The SPC2 camera works is very different from a traditional camera and this is reflected in the parameters The camera actually counts the number of photons it detects during its exposure time However if the user is allowed to set the exposure time directly the photon counters might overflow and the results would be undefined To prevent this the camera has a constant exposure time of 20 74 microseconds The data captured during this time is referred to as a frame To enable longer exposures than 20 74 microseconds the camera sums mul tiple frames
55. he quantum scale output the numbers seemed to be too low The fault would need to be found and corrected Downloading the entire database was not implemented but the user decided to drop it The database files can be downloaded over ssh if needed Security was not implemented yet thus it was not accepted 13 2 6 1 Delivery The corrections from the acceptance test were made and the software system was delivered and accepted on 15 December 2014 14 3 Future adaptations 3 1 Underwater operations When the sensor has been thoroughly tested on land the next step will be to further develop it for use underwater Accurate measurements of the underwater light environment is crucial for further research on marine life in the polar night In Berge et al 2012 the activities of bioluminescent zooplankton in the Arctic is described The zooplankton is shown to follow a diel vertical migration pattern that does not follow any detectable changes in illumination This suggests that the zooplankton either react to light that is too low for current sensor solutions to register or that the levels of light are so low that other mechanisms must be driving these vertical migrations A better understanding of the available light in the underwater environment could help drive this line of research forward The SPC2 is capable of high speed captures at up to 48000 kHz This could be utilized to measure light dynamics e g of the bioluminescent z
56. he sun is below the horizon is approximately constant The values are in the range Opmol m s to 0 5 umol m s Typical irradiance values at night is between 0 and 1 pmol m s which proves that the values from the sensor are perfectly reasonable The sky was clear on the night of the 15th which corresponds well with the constant measurements Some small disturbances can be seen they are believed to be caused by light pollution from the urban environment On 16 December the sun rose at 09 58 local time A detail of figure 3 is shown in figure 4 which shows the period just before and after sunrise The graph is seen to rise sharply just after 08 30 which agrees well with the timing of the sunrise As the sun comes over the horizon the sensor seems to have saturated This was confirmed by checking the original measurement data All counters had saturated Figure 5 shows a detail for the sunset It is almost a mirror image of the sunrise The local sunset was at 14 32 and the irradiance can be seen to sharply drop some time after that The saturation of the sensor hides the dynamics between 14 30 and 15 00 from view After 15 00 the irradiance is seen to drop steady to a constant value below 1 as expected 6 2 Dark counting rate The camera s sensors are laid out in an array with di mensions 32 x 32 A visualization of the dark counting rate for the entire array is shown in figure 6 Dark colors correspond to low dark counting rates
57. iance because the irradiance would be roughly equal in every direction On days with a point source of light such as the moon the measured values would be drastically different depending on the orientation of the camera A camera pointing at the moon would measure much higher values than if it was pointing elsewhere The camera is sensitive to temperature changes The cam era was tested indoors in close to constant ambient light During this time the temperature increased noticeably Also the photon count kept increasing although the am bient light was more or less constant During this test the sensor was programmed to capture images every second During the measurements the imager was placed outside in temperatures between 16 C to 3 C Thus heating of the camera was not a problem If the camera is to be used in warmer temperatures especially indoors the temperature issue would have to be addressed The camera is not made for underwater use An underwa ter housing must include battery power and a computer running the necessary software making it completely au tonomous For now measurements can only be made in air The software for the camera is of very limited capability The only thing it does at the moment is to capture images every minute and save the average image and the sum ina database Full featured software should be able to capture and display test images set parameters and configure different types of log
58. ibrary and which includes the header created in step 2 2 3 4 Candidate node js and express From the node js webpage https casablanca codeplex com Shttp spring io http nodejs org Node js is a platform built on Google Chrome s JavaScript run time for easily building fast scalable network applications Node js uses an event driven non blocking I O model that makes it lightweight and efficient perfect for data intensive real time applications that run across distributed devices 0 Express is an addon to node js It is a minimal and flexible Node js web application framework that provides a robust set of features for web and mobile applications Together node js and express makes the job of creating a REST API and securing it with basic authentication very easy Node js is made for building network applications and includes many time saving modules such as http http is a module that supports many features of the protocol which have been tra ditionally difficult to use For calling code from the C library node js has an addon called node ffi the Node js Foreign Function Interface It enables node js developers to load and call dynamic libraries using pure JavaScript node ffi also handles the conversion of types between JavaScript and C 2 3 5 Conclusion The server will be developed with node js and Express Node js appears to offer the easiest and most time saving solution being p
59. ime was set to 48216 frames which is amp 1s Multiple images were captured with these settings and formed the basis for the DCR calculations 5 3 Measuring the irradiance of a candle The camera was positioned 1 foot away from a burning candle facing the light The camera and the candle were placed in a dark room Multiple irradiance measurements were taken of the candle s flame The measurements were then converted to the unit foot candle to verify that the sensor s output was reasonable This experiment was performed as a rough verification of the sensor s output values In Sakshaug et al 2009 a crude conversion between kilolux and pmol is given as 1klx 14 18 umol m s Since 1 foot candle is approximately 10 1x a candle can be used as a basic verification of the sensor s output 5 4 Comparing the SPC2 to the Walz Diving PAM To investigate if the SPC2 were measuring reasonable values it was compared to another light meter The comparison was done with a Diving PAM underwater fluorometer manufactured by Walz The Diving PAM is a worldwide unique instrument for study ing in situ photosynthesis of underwater plants including sea grasses macroalgae and zooxan thellae in corals The Diving PAM has a lower limit of reliable measure ments at about 1 umol m s The experiment was performed on 19 December 2014 in the afternoon between 14 30 and 15 00 local time The two sensors were set to me
60. iments that were performed to determine the new sensor s correct operation The tests were simple but illustrates that the sensor is giving output in the correct range and that it is capable of capturing the dynamics of changing light conditions The appendices include the user manual software specification and the code documentation The user manual is intended for the end users of the soft ware while the code documentation is intended for future developers and maintainers of the software modules There is some repetition in the texts in the appendices especially in the user manual and the code documentation This is intentional so that the manuals can be read independently of the thesis A zip archive accompanying this thesis includes the source code for the soft ware system 3Professor Geir Johnsen Department of Biology NTNU 2 Software system This section describes the software package that was developed as part of this thesis The software package has been developed using the test driven development technique Subsection 2 1 discusses test driven development and the test first concept Subsection 2 2 describes the software specifica tion and discusses the its importance Subsection 2 4 presents the software architecture and details the ideas behind it especially the decision to sepa rate the server and web applications into two independent modules Finally subsection 2 5 discusses some theory on robust storage and dem
61. in Ny Alesund from 13 to 28 January 2014 Ny Alesund is the world s northernmost permanent settlement and with its position at 78 55 N 11 56 E it is perfectly situated for studies of the polar night The purpose of the campaign was twofold to carry out the research activities of the Marine Night project and to educate students taking part in the course AB 834 Underwater Robotics in the Arctic Polar Night at the University Centre in Svalbard The author was enrolled in said course The paper is organized as follows section 2 presents the theory and characteristics of light and its importance for life in the Arctic section 3 presents the prototype sensor and the software that was developed for it and section 4 presents the results from the Marine Night campaign I http www mare incognitum no 2 THEORY 2 1 The polar night Due to the inclination of the earth the Arctic regions experience polar night during the winter season The polar night is a period of the year in which the sun never rises above the horizon For a few hours at midday there is some available light due to scattering of the sunlight in the atmosphere Sakshaug et al 2009 but for most of the day the environment is seemingly completely dark In Ny Alesund the sun sets on 25 October and does not rise again until 17 February This has a major effect on life in the Arctic Primary production halts during the winter months and until recently it was
62. ings function err settings if err return logger error Could not start auto capture err The timer function uses the sdk to capture an image and saves the data to the database The camera parameters are saved along with the measurement intervalObject setInterval function var measurement sdk captureImage measurement numIntegratedFrames settings numIntegratedFrames measurement numImages settings numImages measurement photonsPerSecond processor getPhotonsPerSecond settings numIntegratedFrames measurement data measurement quantumScale processor convertToQuantumScale measurement photonsPerSecond settings calibrationFactor measurement calibrationFactor settings calibrationFactor dataAccess saveMeasurement measurement function err m I if err return logger error err else return logger debug Measurement saved at m time H settings secondsBetweenCaptures 1000 Ji stop function I clearInterval interval0bject HO callback null I function setStoredParameters callback Gets settings from database and apply them to camera If any errors occurs here the application assumes that either the camera is not connected or the database is not running Either way there is no point for the application to start up so it just exits dataAccess getSettings gotSettings function gotSettings err settings if err logger error Could not ge
63. ion 1 2 5 Creating the settings collection For the server application to run the settings and the cameraspecs collec tions must exist in the database In the scripts directory there is a script called populate db js that will create a settings object and save it to the database Open the script and edit the line that says conn new Mongo spc2 server 27017 Change spc2 server to the hostname of your machine and change 27017 to the port your mongod process is running on Next run the script with the following command mongo lt hostname gt 27017 scripts populate db js Change lt hostname gt and 27017 if your database is running on a different host and or port The settings and cameraspecs collections should now exist in the database 1 2 6 The server application config file When the databases are up and running the addresses must be entered into the server application config file so that the server can connect to them Open the config file and update it so it includes the following lines dbHostAddresses lt addressl portl1 gt lt address2 port2 gt databaseName spc2db Each member of the replica set must be listed on its own line under dbHostAddresses Make sure that the file only includes one instance of the dbHostAddresses list and the databaseName variable 1 3 Server 1 3 1 Prerequisites The server is a node js application with MongoDB databases Setting up the databases is described in the data
64. is the amount of time that passes between the end of one acquisition and the start of the next The Application Programming Interface API also has a subroutine that enables dead time correction which can further reduce the effect of dead time The SPC2 can be operated in two modes normal and ad vanced Each pixel contains an 8 bit counter which counts the number of photons detected during the exposure time In normal mode the exposure time is fixed to 20 74 ps ensuring that the counter will not overflow Longer expo sures are possible by integrating multiple frames Because of the noise immunity this integration does not increase the noise In advanced mode the user can set the exposure time for better control of the camera Longer exposures makes counter overflow possible and this must be checked for and handled by the user Other parameters include number of acquired frames and number of frames to integrate on each acquisition 3 When the camera is commanded to acquire images it returns a set of images The number of images in this set is the number of acquired frames which must be set beforehand In normal mode the number of integrated frames is set to decide the exposure time 3 2 Application Programming Interface API The SPC2 Photon Imager includes an API written in the C language An API is an interface between two systems in this case the SPC2 camera and the user s application It is used for writing programs t
65. l orders of magnitude larger than the average measurements These spikes most likely correspond with artificial light sources such as headlamps on vehicles and the floodlights of the research vessel R V Helmer Hanssen which docked at the pier on several occasions during the campaign There is some missing data roughly between 03 00 and noon on 19 January This was due to a computer restart following system updates It was discovered in the morning and the program was restarted The missing data is seen as a straight line in the plot between tick marks 18 and 19 4 2 IMO sensor The irradiance measurements from the IMO sensor are shown in figure 8 These measurements are clearly more noisy than the corresponding values from the SPC2 cam era However there are clear peaks around noon for both Tuesday 21 January and Wednesday 22 January This corresponds well with the measurements from the camera both showing a lower peak on the 22nd The IMO sensor came with calibration instructions to convert the raw measurements into Fpar which is shown on the plot 5 FURTHER WORK The SPC2 camera as an irradiance sensor is a prototype In its current form it lacks many features Detailed in this section are suggestions for further work that would make the camera a functional sensor for measuring irradiance in the marine environment 5 1 Calibration The conversion from photon count to irradiance is a rough estimate For more refined results
66. lication fell short Thus it was necessary to build a custom made software package The new light sensor was developed for researchers in marine biology asso ciated with the research program Marine Night Together with Professor http www micro photon devices com http www marinenight mare incognitum no Geir Johnsen a detailed specification of the software system was drawn up This specification is included in Appendix B 1 3 Structure of the thesis This thesis is presented as a collection of two papers with a r sum in front The purpose of the r sum is to collect and present background theory and concepts in one place leaving the papers to concentrate more on the results Section 2 describes the developed software system The software architec ture is presented and explained along with some methods on software de velopment Also described is the theory and implementation of robust data storage Section 3 presents some possibilities for further development of the sensor This section contains ideas and plans for future use especially concerning use of the sensor together with underwater platforms Paper 1 is called Development of a Prototype Sensor for Measuring Irradiance in the Polar Night and describes the first testing and deployment of the sensor during the January 2014 Marine Night field campaign Paper 2 is called Testing and Verification of a Low Light Irradiance Sensor It details several exper
67. model Measurement Measurement Defines the settings schema Settings include the camera parameters the capture interval and the calibration factor var Settings mongoose Schema numIntegratedFrames type Number min 1 numImages type Number min 1 secondsBetweenCaptures Number calibrationFactor type Number min 0 deadTimeCorrection Boolean H mongoose model Settings Settings Schema for saving metadata about the camera var CameraSpecs mongoose Schema time Date focalLength String aperture String description String H mongoose model CameraSpecs CameraSpecs if callback amp amp typeof callback function callback null H Disconnects from the database function disconnect mongoose connection close Concatenates multiple addresses into one string Used for connecting with mongoose function concatAddresses addresses databaseName var pattern mongodb S var arr addresses reduce function acc address var matches pattern exec address return acc concat matches 2 ID return mongodb arr join databaseName Standard get functions function getMeasurementModel return mongoose models Measurement function getSettingsModel I return mongoose models Settings function getCameraSpecsModel return mongoose models CameraSpecs Saves a new measurement to the d
68. n the tropics due to the sun being up all night Sakshaug et al 2009 2 6 Attenuation of light When radiation from the sun hits the atmosphere it is attenuated by the gases it travels through Oxygen water and carbon dioxide all cause attenuation of the irradiance At higher altitudes because of the cosine effect the light must travel a longer distance through the atmospheric gases In fact at latitudes lt 30 the distance is about twice as long as for zenith radiation i e when the sun is directly overhead Hence attenuation is stronger for light hitting the Arctic regions than for any lower latitude The attenuation at different wavelengths can also be seen from figure 2 At low solar altitudes the direct radiation in the blue band is strongly attenuated due to scattering in the atmosphere Thus direct radiation at low angles is mostly in the red band As the angles approach zero the available light will appear blue again because direct radiation decreases much faster than indirect radiation The indirect radiation in mostly in the blue band Another severe source of irradiance attenuation is clouds Depending on the type of clouds and their thickness the irradiance can be attenuated by as little as 20 and as much as 70 During the Arctic summer fog is often encountered over open water in calm weather Also the cloud cover associated with a passing low pressure system can cause the daily average surface irradiance
69. ning and listening for HTTP requests on the configured port or 8080 by default 1 4 Web application The web application is a couple of simple pages for controlling and monitoring the SPC2 camera through the server This section assumes that the server is running and that you have the server address 1 4 1 Setup The web application requires minimal setup First it needs to know is the address to the server This information must be entered in the file js config js which is a standard javascript file Change the host and port to the address and port number of your server Do not change the api part A sample config js is shown below var host http 129 241 143 245 port 8080 var config apiUrl host port api Second the web application s dependencies must be installed This is done in two steps because it uses both npm and bower to handle its dependencies Change to the top directory where the file package json is located Run the command npm install This should install all the dependencies for node js Then change to the resources directory and run the command bower install This should install all dependencies for the web application itself 1 4 2 Hosting The web application is also a node js application and can be started with the command NODE_ENV production node app js Important It uses basic authentication to authenticate the user If security is a concern the we
70. nnected responds to GET The reason for this is reflected in the name of the routes also connected is meant as a status request as in the question is the camera connected disconnect on the other hand assumes that the camera is connected and that the user issues the command disconnect the camera apiRouter post disconnect function req res I var disconnected sdk destroy res sendStatus disconnected 200 500 H Gets data from the database between the dates given in the route name from and to must be on a form that is recognized by Date parse The recommended format is the ISO 8601 format because it is clear and unambiguous On success the data is written to a CSV file and offered to the user as a download apiRouter get getData from to function req res var from new Date req params from var to new Date req params to if from to I var msg Date format passed to getData is not valid logger warn msg res status 400 send msg else var data dataAccess getData from to processData function processData err results if err return logger error err processor createCSV results downloadFile function downloadFile err path if err return logger error err var filename data csv res download path filename function err processor deleteFile path PD F Calls to this route returns the same data as calls to getData
71. o control the camera and includes multiple functions for connecting setting parameters and acquiring images to name a few The API was used for writing the custom software that was used on the campaign A few of the most important functions are detailed below SPC2_Constr is used for getting a handle to the camera A handle is an object which is a unique identifier to the camera and it is used to communicate with it The handle must be acquired before any other functions can be called SPC2_Set_Camera_Par sets the camera parameters ex posure the number of frames per acquisition and the number of integrated frames This function must be called before acquiring any images SPC2_Snap acquires a set of images as specified in the parameters The acquired images are stored in the working memory of the camera and must be retrieved by one of the other functions SPC2_Average_Img or SPC2_Save_Img_Disk SPC2_Average_Img returns one image which is an average of the previously acquired images 3 3 Software developed for the project A software package from now on referred to as the program was developed during the campaign to enable the SPC2 to measure the ambient light in the polar night Some interesting light events were expected on 18 and 19 January To ensure that the sensor would be operational in time for these events a minimal set of specifications were developed by the end users The sensor had to be able to e Measu
72. on Imager is a camera capable of imaging the light field by counting the number of incoming photons during the acquisition period In the Marine Night campaign the photon imager has been used to measure the light field in the Arctic polar night The photon imager is shown in figure 3 http www micro photon devices com Contact Fig 3 The SPC2 Photon Imager 3 1 Capabilities of the photon imager The information in this section is taken from the SPC2 User Manual The photon imager captures images of the light field in much the same way as a normal camera does Light enters through the lens and is measured by the sensor on the imager s circuit board The lens is mounted on a standard c mount which enables the camera to easily utilize many different lenses The lens used for these measurements was a Tamron 8mm with 1 4 maximum aperture The field of view of the camera with this lens was estimated to be about 20 The light detecting sensor in the SPC2 is very different from the sensors found in conventional cameras It consists of an array of 32 x 32 pixels Each pixel is made up from a single photon avalanche diode detector with integrated electronics for digital processing Because of this fully digital acquisition of the light signal the camera has a high noise immunity thus giving a very clean output signal The SPC2 is capable of acquiring images at up to 49 000 frames s with a negligible dead time Dead time
73. onstrates how this is implemented in the server application 2 1 Development concepts The software was developed following the guidelines of test driven develop ment This development technique increases confidence in the correctness of the program and has also been found to increase productivity Erdogmus 2005 2 1 1 Test driven development Test Driven Development TDD is a technique based on writing functional tests for a piece of production code before writing the code itself A typical TDD workflow is as follows 1 Writing the test The first step in TDD is to read and understand the requirements and write a functional test The test is often written in a unit testing framework such as JUnit if one is developing Java applications or Mocha for the node js platform The requirements must be described in a clear concise manner with as little ambiguity as possible This is the job of the specification which will be discussed further in subsection 2 2 Writing the test first forces the developer to really understand the requirements beforehand 2 Run all tests This step should conclude with all previous tests suc ceeding and the new test failing This confirms that the new test is adding new information to the system in the sense that it requires http junit org http mochajs org new code to pass It also confirms for the developer that the testing platform works and that the new test is actually looking
74. ooplankton described in Berge et al 2012 Continuous measurements at high speed could give a level of detail in the light measurements that is not possible with slower or less sensitive sensors For the sensor to be usable in underwater missions a number of modifications and further developments will have to be done Most importantly the sensor needs an underwater housing The camera is not built for underwater use thus a custom housing will need to be built The housing must have room for the camera itself a computer for controlling the camera a logging unit for storing the collected data and a power supply The sensor might be de ployed on a number of different platforms such as Autonomous Underwater Vehicles AUV Remotely Operated Vehicles ROV or stationary platforms such as a moored sensor station The housing will need to take into account the restraints of the platform such as weight allowances geometric shape and possibly hydrodynamic efficiency in the case of the AUV The sensor s software will have to be further developed for underwater use At this point the software is designed for prolonged use in a safe environment indoors The focus has been on reliability and ease of use For an underwater operation the specification will change in a major way Reliability will be important perhaps more so The sensor must continue working even when failures are present The sensor must be autonomous such that it can be started
75. ose var db var winston require winston var logger new winston Logger transports new winston transports Console level debug colorize true H Module functions var dataAccess module exports Connects to the database s Addresses is a list of addresses to running mongod instances The use of multiple databases assures redundancy in case some of them fail function connect addresses databaseName callback Concatenates the addresses to one URI that mongoose accepts var uri concatAddresses addresses databaseName var options Sets keepalive to true to make sure that the database connection doesn t time out server socketOptions keepAlive 1 replset socketOptions keepAlive 1 3 mongoose connect uri options db mongoose connection db on error logger error Asynchronously sets up the database schemas once the connection has been made db once open function defineModels I The schema for a measurement data holds the original data from the camera and time holds the timestamp from when the measurement was made The rest of the fields are saved with the measurement to make sure the results are reproducible var Measurement mongoose Schema data type Number min 0 time Date numIntegratedFrames Number numImages Number photonsPerSecond Number quantumScale Number calibrationFactor Number H mongoose
76. papers with a r sum The first paper should look at the results of deploying the sensor during the 2014 campaign and the implications of those results The second paper should detail testing and verification of the sensor Scope of work R sum e Discuss the need for a new sensor and its importance to ongoing Arctic research Describe the architecture and design of the software package Paper I Review the theoretical background for light its characteristics Review the need for irradiance measurements Describe the experiment carried out during the 2014 campaign Present the results from the campaign Discuss the results and their importance for further work Paper 2 Describe the new sensor structure and implementation Describe experiments for testing the correctness of the new sensor Discuss the results of the experiments and provide suggestions for further work CN NTNU Fakultet for marin teknikk Norges teknisk naturvitenskapelige universitet Institutt for marin kybernetikk The report shall be written in English and edited as an article collection with a r sum in front in the format of a report It is supposed that Department of Marine Technology NTNU can use the results freely in its research work unless otherwise agreed upon by referring to the student s work The thesis should be submitted by 08 February 2015 Co supervisor Professor Geir Johnsen Professor Asgeir J Sorensen Supervi
77. r Simple test cases are performed to verify that the sensor performs as expected and that it registers reasonable irradiance values Section 2 lists the expected functionality of the sensor and what the end users expect it to be capable of Section 3 presents some theory on the different ways to measure illuminance and irradiance and presents the foot candle which is used in one of the experiments Section 5 describes the setup of the different tests and briefly discusses their limitations and their strong points Section 4 discusses the different calibration options that the sensor offers and why they may be useful Section 6 presents the results of the tests Section 7 presents the conclusion of the tests and finally section 8 lists possibilities for how the sensor s capabilities can be extended in the future 2 EXPECTATIONS The SPC2 was initially tested in January 2014 during the polar night in Ny Alesund Svalbard Ny Alesund is located at 78 55 30 N 11 50 20 E and is well within the polar night region The initial tests were successful in the sense that the camera was able to capture the little light that was present at this time of the year with high fidelity Also the results were comparable with those measured by other sensors taken at the same location The new sensor is expected to perform as well as last time It is also expected to record time series of the light conditions in Ny Alesund for an extended p
78. r regions Converting between pmol m s and W m is difficult un less the spectral distribution is known Photons carry different amounts of energy for different wavelengths and this must be accounted for in the conversion 3 2 The lux scale Another scale for measuring illuminance is the luz scale This scale was designed for measuring illuminance indoors and for architecture The scale is therefore adjusted for the spectral sensitivity of the human eye The human eye responds well to light in the green band while plants absorb light mostly in the red and blue bands of the spectrum In fact algae and higher plants hardly respond to green light at all In Sakshaug et al 2009 there is an average conversion factor from klx to pmol m s in the range 14 18 This conversion factor must be used with care as the spectral composition can change a lot especially in water For sunlight however the spectral distribution varies little as long as the elevation of the sun is greater than 10 8 8 The foot candle The foot candle abbreviated fc is a unit of illuminance It is the amount of light one foot away from a point source of one candle It is equal to one lumen per square foot The fe unit is intuitive and makes possible a rough procedure to calibrate a light meter One foot candle is roughly equal to 10lux and 1klx is roughly equal to 14ymol m s to 18 pmol m s see Sakshaug et al 2009 appendix A 4 CALIBRATION PRO
79. rational environment e The program must store the values for all parameters i e the param eters must not change or be reset by any other means than user inter action e The program must be able to capture images in automatic mode Auto matic mode means that the program captures an image every t seconds where t is specified by the user Priority B e Enable the user to capture test images A test image is an image captured at the command of the user The program may then show the image to the user along with data connected to the image such as the irradiance in quanta or energy mode Storing data Priority A e The program must store all data from the sensor in a reliable and secure manner Each captured image or equivalent data must be stored e The stored data should be calibrated to the quantum scale i e it should be stored as pmol m s e The parameters of the sensor must be stored with the data for future recalculation and confirmation e A calibration factor if calculated should be stored with the captured data This way the original measurements raw counts are always available e The storing of data must be robust meaning that captured data from the sensor must be stored even if there exists some problem with the database or the program Priority B e Implement alarms to warn the user about storage problems e Also store data calibrated to the energy scale i e pW m Access to stored data Priority
80. re the ambient light once per minute e Store the measurements safely The program acquires an image approximately once every minute Each returned image is the average of 100 images each of which is the integrated result of 5 frames This average is then stored in a database In addition the program calculates the photon count per second and stores that in the same database The photon count is calculated by summing the values of each pixel in the image which represents the photon count detected by 3 In normal mode 4 Only in advanced mode 5 The users in this context are Prof Geir Johnsen Dep of Biology NTNU and Prof J rgen Berge Dep of Arctic and Marine Biology University of Troms the representative pixel on the sensor This sum is then divided by the acquisition time to get the photon count per second 3 4 Database The database system used in the program is a SQLite database From the web page SQLite is a software library that implements a self contained serverless zero configuration transactional SQL database engine SQLite is the most widely deployed SQL database engine in the world The source code for SQLite is in the public domain The above description highlights why SQLite was chosen for this project It is self contained meaning that the program can run on any machine without having to set up a database server That the source code is in the public domain means that anyone is free to use the sof
81. rom C C Thus one candidate programming language for the server is C Writing a REST API in C is not common The language is more often used for applications where either hardware access or speed of computation are top priority Thus there are not many server frameworks written in C One exception is Microsoft s C REST SDK codename Casablanca 2 3 3 Candidate Java Spring Spring is a framework for Java EE From the spring webpage Spring helps development teams everywhere build simple portable fast and flexible JVM based systems and applications Java is in the author s opinion easier to use than C Spring enables rapid development of a REST API through the use of Java annotations and the Model View Controller software architectural pattern Basic authentication is also available through the use of the BasicAuthenticationFilter class To call the C library from the SPC2 SDK Java makes use of JNI the Java Native Library Using the JNI with a precompiled library requires three steps 1 Write a declaration in Java for the library function you want to call This declaration must use the native keyword to specify that the code is written in some native language in this case C 2 Create a C header file for use by the native code This is done auto matically with the javah tool 3 Implement the native method in C This requires the developer to write another C program which is a wrapper around the l
82. ror errMsg return res status 400 send error errMsg The camera operates in normal mode where one exposure is 20 74 microseconds var exposureTime 20 74e 6 var nIntegratedFrames Math round integrationTime exposureTime async parallel function setCameraParameters callback var status sdk setCameraParameters nIntegratedFrames nImages callback null status function saveParameters callback var settings settings numIntegratedFrames nIntegratedFrames settings numImages nImages dataAccess updateSettings settings callback function after err results if err return logger error err var status results 0 res status status success 200 500 send status H J Sets the interval between each measurement when in automatic capture mode The POST request must include a JSON object with the property captureInterval The interval is given in seconds apiRouter post setCaptureInterval jsonParser function req res var postData req body if postData postData captureInterval logger error Invalid data sent to setCaptureInterval logger debug postData postData return res sendStatus 400 dataAccess updateSettings secondsBetweenCaptures postData captureInterval function callback err if err logger error err res sendStatus 500 logger info Updated capture interval res sendStatus 200 H H Sets the calibration
83. rototype sensor for measuring the low levels of irradiance in the Arctic polar night A highly sensitive photon imager capable of single photon detection was used for this purpose The special purpose software developed for the prototype is also described The prototype was deployed at 78 55 N during the Marine Night research campaign Experimental results are presented and discussed along with results from another prototype for comparison Keywords Prototype sensor Polar night Irradiance 1 INTRODUCTION The purpose of this project was to develop a prototype for measuring the low levels of light in the polar night This project report details the different aspects of this project such as description of the prototype hardware and development of the required software Also included are experimental results and suggestions for further work This project was done as part of the Marine Night project Marine Night aims to increase our understand ing about the marine biological processes during the polar night and the light environment is a crucial part of this Recently discovered marine biological processes Berge et al 2009 are among the motivating factors for this re search project One of the hypotheses is that the organisms are able to react to light changes in the polar night which seem insignificant to the human eye Experiments with the prototype light sensor were carried out during the Marine Night field campaign
84. rve both modes at the same time In the particle mode light is characterized by the energy of each particle and the particle flux Photons have no inherent color but when they reach the eye the energy of the photon is translated by the brain into colors Brightness is associated with the photon flux a higher flux corresponds to a brighter source of light The wave model and the particle model is connected according to Planck s law e he A e is the energy h 6 62 x 1073 Js is Planck s constant and c is the speed of light in a vacuum From Planck s law it is seen that the energy is inversely proportional to the wavelength From this it can be deduced that a photon in the blue band of the spectrum has an inherent energy that is roughly 1 6 times greater than a photon in the red band Photons in the visible band all carry enough energy to bring about photosynthesis and it is the number of photons that matter not the total energy Therefore the particle model is well suited when studying the effect of light on biological life Sakshaug et al 2009 The photon counter described in section 3 measures the incoming light as photons 2 3 The characteristics of light Light can be characterized by its intensity spectral char acteristics polarization and direction The intensity of light is measured as irradiance Irradiance is measured either on a quantum scale with units pmol photons m s or on an energy scale
85. s 32 x 32 pixels where each pixel s value is based on the photon count of a single photon detector It is nearly impossible to recognize any thing in the image although contrasts show up reasonably clearly An object with a high contrast to its background should be possible to recognize Fig 9 Image of the light field Figure 10 shows an image of the same room with a point source of light pointed towards the camera The point source is a flashlight The camera is not saturated as the image is still gray saturated pixels are completely white The contrasts from figure 9 is gone presumably because of the new light source It is nearly impossible to discern anything in the image except for some darker patches in the corners Most importantly it is hard to identify the light source in the image In figure 10 the light source is pointed out with the arrow The light from the flashlight causes increased counts in almost all pixels causing the entire image to become more indistinct When using the sensor to capture images of the night sky or the polar night environment users have expressed that it would be advantageous to be able to identify point light sources such as the moon On the basis of these tests it seems like it would not be possible to identify point sources from the images The identification would have to be done some other way e g by simultaneously taking images with a traditional camera Fig 10 Imag
86. s the notion of replica sets which is an easy way of implementing redundancy This section describes how to set up redundant databases and how to tell the server application which databases to use 1 2 1 Getting MongoDB MongoDB can be downloaded from the download page Choose the correct download for your system and follow the installation instructions For many systems such as Mac OS X and many flavors of Linux MongoDB can also be installed through a package manager See the MongoDB web page for details Once the installation is finished you can proceed with the next section 1 2 2 Replica sets MongoDB supports redundancy through what it calls replica sets This means that the application connects to two or more instances of the database at the same time One of the databases is designated as the primary and the others are secondary The data is first saved to the primary database and then replicated on the secondaries In case of a failure such as a power outage or a hardware failure one of the secondaries is elected to be the new primary and the system can continue in its normal mode of operation Important Replica sets should always have an odd number of members to ensure that elections proceed smoothly Thus 3 members should be sufficient to endure most problems The MongoDB manual recommends that replica set members in production environments maintain as much separation as possible Storing the different databases
87. sableDeadTimeCorrection function req res var statusObj sdk setDeadTimeCorrection false dataAccess updateSettings deadTimeCorrection false function err if err logger error Could not update dead time correction in settings err res sendStatus 500 res status status0bj success 200 500 send status0bj H H Captures a single image and returns a JSON object containing the image data as a base64 encoded string The image data is in the imageBase64 property apiRouter get getTestImage function req res I Expects a query parameter intTime which specifies the integration time in milliseconds var intTime req query intTime 1 if intTime intTime NaN intTime lt 0 logger error getTestImage invalid parameter intTime intTime res sendStatus 400 var nintFrames time2Frames intTime sdk captureTestImage nIntFrames function path fs readFile path function err data if err I logger error getTestImage Could not read image file err return res sendStatus 500 var base64Data data toString base64 res status 200 send imageBase64 base64Data H H H H Starts automatic capturing Automatic capturing means that the camera cap tures a set of images at set intervals The average of these images are stored in the database as a measurement along with the capture parameters Most of the functionality of this route is encapsul
88. sing Capture test Image The test image can be used to find the correct Integration time for automatic capture Integration time Be aware that the integration time set here is not saved It is only used for capturing the test image To set a permanent integration time use the form on the Settings page 15 Capture test image Figure 3 Control page The Control page enables you to control the camera operations There are two options controlling Automatic capturing and capturing test images Automatic capturing is used for capturing time series of light measurements It has two buttons which starts and stops the automatic capturing mode In this mode the camera captures a series of images at given time intervals The average of these images is stored in the database along with the capture parameters that were active when the image was aquired Also stored are the total photons per second that was detected and the irradiance value on the quantum scale micromol photons per square meter per second The time interval is set on the Settings page Capture test image is for testing different integration times Integration time is analogous to exposure on a traditional camera Enter the integration time in milliseconds and press the button to capture an image After the image has been captured and sent from the server it will appear below the button 10 2 3 The Settings page Set capture parameters The sensor parameters inte
89. sor Abstract This thesis describes the development and testing of a new irradiance sensor Light from the sun affects all life on earth and recent studies give reason to believe that some life might even react to the microscopic levels of light found in the Arctic polar night Commercial light meters are not sensitive enough to register the light in these dark periods of the year thus the need for a new sensor has emerged During a field campaign in Ny Alesund in January 2014 conducted as part of the Marine Night Research project a prototype of the sensor was first tested The results were promising and work began to develop a more complete software system that is able to measure the light environment of the Arctic for extended periods of time The development of this sensor system and initial tests are described in this thesis This part of the project ended with a working system that is to be tested on the next field campaign of the Marine Night project in January 2015 A manual has been written for the users and is included as part of this thesis Also included is the code documentation intended for the future developers of the system Sammendrag Denne masteroppgaven beskriver utviklingen og testingen av en ny type irra dianssensor Sollyset pavirker alt liv pa jorden og nyere forskningsresultater tyder pa at noen livsformer kan veere avhengig av de mikroskopiske lysnivaene i den arktiske vinteren Forskerene som studerer diss
90. t settings from database err process exit 1 10 var status sdk setCameraParameters settings numIntegratedFrames settings numImages if status success logger error Could not set camera parameters process exit 1 status sdk setDeadTimeCorrection settings deadTimeCorrection if status success logger error Could not set dead time correction process exit 1 logger info Camera parameters have been set callback null 3 function initDone err logger info Initialization done H var host server address address var port server address port logger info Server listening at http s s host port H Returns the number of integrated frames that comes closest to the given integra tion time integrationTime is the time in milliseconds function time2Frames integrationTime var var var var micro 1e 6 milli 1e 3 exposureTime 20 74 micro timeInSeconds integrationTime milli frames Math round timeInSeconds exposureTime return frames 11 data access js data access js Purpose This module includes functions that read and write to the database It acts as an abstraction layer between the server application and the database hiding the database details from view MongoDB The server uses MongoDB as its database Mongoose is a node js module for interfacing with MongoDB var mongoose require mongo
91. t the calibration can take into account the photon detection efficiency at different wavelengths 8 8 Alarms The specification of the sensor lists alarms as a priority B task This is not yet implemented due to time constraints The sensor should alert the user of any trouble it encoun ters during operation Examples of problems could be hard drive failure causing one of the databases to shut down It could also be other sources of hardware failure such as problems connecting to the camera or problems with the power supply The alarm system should also warn the user in the event of software failures such as failure to save measurements A more sophisticated alarm system could also monitor the output of the sensor and issue warnings when the output seems suspect Problematic output could be very high or low measurements or a significant increase in noise which could be an indication of hardware prob lems The alarms would have to be issued to the user through the user interface e g on a list on the main page of the web application The interface would need some way of dismissing the alarm when the cause has been fixed REFERENCES Berge J Cottier F Last K S Varpe O Leu E Soreide J Eiane K Falk Petersen S Willis K Nygard H Vogedes D Griffiths C Johnsen G Lorentzen D and Brierley A S 2009 Diel vertical migration of Arctic zooplankton during the polar night Biology Letters 5 1
92. tested during the January 2014 campaign of the Marine Night research project in Ny Alesund Svalbard and showed promising results Most importantly the camera was able to capture the light dynamics of the polar night with very little measurement noise See L nne 2014 for details Together with the camera the new sensor also consists of two pieces of newly developed software The sensor is built around the client server model A server application runs on a computer that is connected to the camera and the user interacts with the sensor through a client application which runs in a web browser The server is responsible for communicating with the camera saving measurements from the camera in a robust and safe way and enabling clients to take measurements and retrieve data The client is responsible for presenting the user with an interface that offers sufficient control but at the same time is easy to use The overall structure of the new sensor system is shown in figure 1 Testing and verifying the sensor is part of the overall system testing Before the sensor can be used in the field users must have confidence in the equipment and be relatively sure that it will perform as expected 1 Micro Photon Devices Via Stradivari 4 I 39100 Bolzano Italy Web application Intemet Sensor server Fig 1 The structure of the sensor system Camera This paper describes the initial testing and verification of the senso
93. that can help designers improve the reliability of their sys tems are fault prevention and fault tolerance Prevention attempts to elimi nate sources of faults Tolerance enables the system to continue functioning in the presence of faults 2 5 1 Fault prevention There exists two stages of fault prevention fault avoidance and fault removal Avoidance tries to limit the introduction of potentially faulty components during the construction of the system For hardware avoidance includes the following strategies e Use the most reliable components e Use refined techniques for interconnecting components and assembling subsystems e Package the hardware to eliminate interference It is virtually impossible to write fault free code Even bugs that are very unlikely to occur can come to the surface in a long running program To improve the quality of software developers can e rigorously specify the requirements e use proven design methodologies e use analysis tools to verify properties of the software e use languages that facilitate modular structure and abstractions No matter how careful the developers are faults will inevitably be present in the system after its construction Hence the second stage of prevention is fault removal Fault removal consists of procedures for finding and removing bugs A ubiquitous example of fault removal is system testing in all its forms Unfortunately testing can almost never be exhaustive due
94. ts should be doc umented Functional requirements specifies how the program should work Non functional requirements specify other requirements to the software such as uptime requirements for a server It is vital that both the customer and the developer fully understands and adheres to the specification Any changes to the requirements either functional or non functional should first be listed in the specification A detailed specification was developed together with Prof Geir Johnsen The final specification is included in Appendix B The document is divided first into Software functionality and Hardware functionality then into subgroups under these headings Each subgroup such as Capturing images and Storing data is further divided into priority groups Priority A are requirements for a functional system These have top priority Priority B tasks are not absolutely required but functionality that the customer would like to see in the system Priority C tasks are tasks meant for future development and have lowest priority The specifications were consciously made with too many tasks for this de velopment cycle Thus it also lists future adaptations of the system that are not realized at this point 2 3 Selecting a language and framework for the server 2 3 1 Requirements The server must communicate with the SPC2 sensor Communication and control of the sensor is done through a C library Hence the server must be able to call C fun
95. tware for any purpose i e no licence is needed Finally SQLite is used in many large projects by well known companies which implies that it is a well tested and stable software product Name Time Max aperture Value Focal length Latitude Longitude Direction Fig 4 The database schema for the program Figure 4 shows the database schema that was used for the program The database consists of four tables e Measurement is the central table It stores the time and value of each measurement The time is local time the value is in photons s This table has a many to one relationship to both Position and Lens e Position records the position where the measurement was taken Latitude and Longitude are the coordi nates and Direction is the direction the camera was facing when the measurements were taken e The table Lens stores information about the lens used for the measurements It is identified by its name maximum aperture and focal length e Image contains the image returned from the camera It is stored as binary data and can be retrieved in a double array of length 1024 Image has a one to one relationship to Measurement 6 http www sqlite org T http www sqlite org copyright html 8 http www sqlite org famous html 3 5 Experimental setup The SPC2 camera was set up outside the Marine Lab in Ny Alesund The Marine Lab is situated just outside the settlement and next to the pier It has a
96. undancy is adding extra elements to the system enabling it to detect and recover from faults While redundancy increases the fault tolerance of the system it also increases the complexity The aim then must be to minimize redundancy while at the same time maximizing reliability One specific implementation of redundant fault tolerance is static redun dancy Here faulty components are masked out or hidden from the system As an example in Triple Modular Redundancy three identical components work in parallel and a voting circuit masks out an eventual faulty compo nent This can be extended with more components and is in the general case called N Modular Redundancy 11 Dynamic redundancy is code running inside a component which tests the component and indicates the system if the component is faulty It provides no redundancy in itself in the sense that another component must take over for the faulty one 2 5 4 Redundancy implementation in the server application To achieve robust storage in the server application it makes use of a feature in MongoDB called replication The documentation MongoDB 2014 states A replica set in MongoDB is a group of mongod processes that maintain the same data set Replica sets provide redundancy and high availability and are the basis for all production deployments Hence MongoDB has static redundancy built in albeit under a different name Replication works with a set of mongod processes mon
97. ured with the SPC2 photon imager and the IMO sensor see sections 4 1 and 4 2 2 4 The biological importance of light The information in this subsection is taken from Sakshaug et al 2009 Light from the sun is a prerequisite for all life on earth This light is electromagnetic radiation with a wavelength from 10 x 10714m to 2000 km The human eye can register light waves in the band 400nm to 700 nm Photosynthetically Available Radiation PAR is defined as the total radiation available in this band PAR is expressed either in nmol m s i e the flux of photons per area per time or as irradiance W m 2 5 Light in the Arctic At high altitudes the rays from the sun strike the earth at a much more oblique angle than closer to the equator The result is that the available light is spread out over a larger area lowering the intensity This lowering of intensity is proportional to the cosine of the angle of the rays to the surface thus this is often referred to as the cosine effect Summer in the Arctic is characterized by midnight sun and winter by the polar night During the polar night the sun never rises above the horizon and during the summer the sun never sets The midday midsummer elevation of the sun is strongly dependent on latitude It is 23 45 for the North Pole 34 for Ny lesund and 43 for Troms However during the summer months the 24 h integrated irradiance can be almost as great in these latitudes as i
98. urposely built for developing network applications Also the node ffi module appears much easier to use than the Java Native Interface which will save much development time Java and Spring was not chosen mostly because of its complexity While it is perfectly feasible to write this program with Spring node js seems to present a much cleaner easier solution without much of the overhead For a simple project such as this Spring seems to complex C was not chosen for three reasons First it appears to be much too complex for this simple project Second the lack of frameworks seem to indicate that C is not much used for this kind of development Third the developer does not have much experience with C and using it for this l0http expressjs com project would require some time studying the language itself That time is better spent on developing the actual program 2 4 Architecture of the server application The architecture of the server application is visualized in figure 1 Server module server js SPC2 SDK module spc2 sdk js CAPI api c libapi SPC2 library libSPC2 SPC2 camera Data access module data access js Data processor module data processor js Figure 1 The architecture of the server application The design of the applications modules follows the Single Responsibility Prin ciple a term coined by Robert C Martin Martin 2002 The principle states that A class should
99. urther calibration but that it is absolutely fit for the purpose Finally the images of the light field goes some way in demonstrating the camera s effectiveness in visually describing the incoming light With some training it might be possible to discern points of interest on these images The resolution is so low however that all such identifications must be treated as uncertain 8 FURTHER WORK 8 1 Automatic adjustment of exposure time As was seen in the 24 hour time series experiment there is a need for adjusting the exposure time when the light levels change In the interest of minimal manual adjustment it would be preferable if the sensor adjusted the exposure time automatically to avoid saturating the photon coun ters when light levels rise above a critical level Each element in the captured image is a 16 bit counter meaning that the maximum value it can represent is 65535 The automatic adjuster algorithm could be implemented as a regulator with the aim of keeping the mean value of the measurements well below this limit but also above the dark counting rate By capturing a test image the mean of the counters can be calculated and the exposure time set accordingly 8 2 Calibration against a known light source For the sensor to be able to reliably measure irradiance in absolute values it needs to be calibrated against a known source of light The spectral signature of the light source must also be known such tha
100. us power supply to the sensor Priority C e Implement the planned power supply system Future adaptation Underwater operations Priority C e Review systems and methods for adapting the sensor to underwater operations e Plan design a system for adapting the sensor to underwater operations e g on landers and mobile platforms such as AUV ROV C Code documentation This appendix includes the documented code for the server application The documentation was autogenerated with docco an open source documenta tion generator The code documentation is primarily intended for future developers and maintainers who want to understand and extend the code The code is also included in the accompanying zip archive http jashkenas github io docco 50 server js server js var express require express var app express0 var basicAuth require basic auth var async require async var bodyParser require body parser var confert require confert var config confert config var winston require winston var apiRouter express Router var jsonParser bodyParser json var fs require fs sdk dataAccess and processor are local modules that handle the connection to the camera access to the database and intermediate processing of data respectively var sdk require modules spc2 sdk spc2 sdk js var dataAccess require modules data access data access js
101. ut It is used for capturing test images to test the parameters before enabling automatic capture mode sdk captureTestImage function captureTestImage nIntegratedFrames callback if nIntegratedFrames logger error captureTestImage missing parameters return sdk setCameraParameters nIntegratedFrames 1 prepareAndGetSnap tmp file function fileCreated err path if err return logger error captureTestImage Could not create tmp file var startImg 1 endImg 1 fileFormat tiff no compression var imageSaved lib save image to disk startImg endImg path fileFormat if imageSaved logger error captureTestImage Could not save test image to disk callback else logger info captureTestImage Test image saved var pngPath path png gm path write pngPath function imageConverted err if err logger error Could not convert image to png err return callback logger info captureTestImage Image converted callback pngPath H H Helper function Used in both sdk captureImage and sdk captureTestImage function prepareAndGetSnap var prepared lib prepare_snap if prepared return logger error Could not prepare camera for snap var gotSnap lib get_snap if gotSnap return logger error Could not get snap
102. veranda facing northwards towards Kings Bay This is where the camera was deployed with the lens facing out towards the bay When doing light measurements light pollution is a problem Light pollution is measured light from artificial sources such as street lights or headlamps on cars or boats To get reliable measurements light pollution must be reduced as much as possible At the Marine Lab this was accomplished by covering all the windows with black plas tic The street lights around the lab had also been switched off for the duration of the campaign Some light pollution from cars and especially the research vessel R V Helmer Hanssen was unavoidable The times of these events were noted so they could be disregarded in further study of the data material In addition to these precautions the camera was positioned against the north wall of the Marine Lab facing towards the bay This way the Marine Lab shielded the camera from any artificial light coming from the settlement of Ny Alesund The camera was mounted on a tripod with its power and data communication cable leading from the camera through an open window to a computer The computer was running the software described in section 3 3 At this point the camera needed no attention from an operator It was left to take measurements from 18 26 January 2014 In addition to the camera another prototype irradiance sensor was running This sensor is described in section 3 6
103. ys be enabled If you want the process to append to the logfile instead of overwriting it use logappend An example of how to start a mongod instance mongod port 27018 dbpath data db replSet spc2 repl set0 logpath var log mongod log This would start a mongod instance listening on port 27018 saving data at data db belonging to the replica set spc2 repl set0 and saving its logs at var log mongod log 1 2 4 Adding members to the replica set Open a mongo shell on the machine that runs the primary process This can be any of the processes A mongo shell is opened by issuing the command mongo port lt port number gt lt port number gt is the port number that the process on this machine listens on E g if the process was started with mongod port 27018 you would start the shell with the command mongo port 27018 Once the shell has started run the command rs initiate This will initiate the replica set and add this process as the primary When it has finished issue the following command for each of the other members in the replica set rs add lt member address gt lt member address gt is the address of the other process that was started When all processes has been added to the replica set run rs status to confirm that everything is set up correctly All members in the set should be listed in the members property If they are not or you encounter any other errors consult the documentat

View/Open

Contents

Download Pdf Manuals

Related Search

Related Contents