ME 411 LAB 4: REMOTE SENSING LAND SURFACE
Contents
1. Use Macroclass ID is checked increase the Size to 500 the size of the classification preview in pixel unit choose the button on the right hand side of the Classification preview box then click on any area of your map A small preview of the classification output will appear and you can compare it to the type of land cover evident in your base raster layer 10 ME41 1 Winter 2015 Lab 4 Classification algorithm Select classification algorithm Threshold Spectral Angle Mapping 0 0000 2 Use Macroclass ID assiiicaton preview F Do E BatFGERAZS la Wala M Ei Size 500 2 Redo U RT_LC80460282014315LGN00_811 TIF RT_LC80460282014315LGN00_810 TIF RT_LC80460282014315LGN00_89 TIF RT_LC80460282014315LGN00_B7 TIF RT_LC80460282014315LGN00_82 TIF RT_LC80460282014315LGN00_81 TIF ao Tce ene oe SCP Classification Layers 7 If any of the signatures appear inaccurate too similar to Breer land class or not miar enough to its own land class you can delete them by highlighting them in Signature list and using the remove icon S Cla MC info CID C info Color EJA E E E E er wor 8 When you are satisfied that you have produced a good classification by visually comparing the preview to the rgb image classification of the entire image can be completed Otherwise you should remove spectral signatures and or add new spectral signatures by creating other ROIs2. 1 From the SCP toolbar select the Preprocessing icon and then the Landsat tab BL JE 2 Select directory where you saved the Landsat bands and metafile For the output directory save the converted bands to a new folder 3 Check the box next to Apply DOS1 atmospheric correction and leave checked Create Virtual Raster which we will use to create a red blue green color composite 4 Perform conversion This may take up to several minutes When it is complete the converted bands and the virtual raster landsat vrt are loaded into the Layers panel ME41 1 7 Winter 2015 Lab 4 To create an rgb color composite double click on landsat and navigate to the Style tab Set Band 5 as the red bed Band 4 as the green band and Band 3 as the blue band This false color combination is best for viewing vegetation because healthy vegetation reflects near infrared wavelengths Make sure Contrast enhancement is selected as Stretch to MinMax and then press Load in the right hand box to redefine the min and max values of the bands you selected in the previous step Your map should look like similar to the one to the right Select the Band Set icon from the SCP toolbar Select bands 2 7 do not select any bands outside of this range and Add rasters to set The SCP will average the value of spectral signatures in each band to create the land cover classification so we do not w
3. http www kyngchaos com software python c Note Some of the steps contained in this tutorial require a significant amount of space on your hard drive Before beginning you should make sure you have at least 30 GB or more of free space 2 The Landsat8 images can be downloaded from http earthexplorer usgs gov a Enter Portland OR as the Address Place clicking the result will populate the coordinates for you b Goto the Data Sets tab Expand Landsat Archive and check L8 OLI TIRS Good review http pubs usgs gov fs 2013 3060 pdf fs2013 3060 pdf c Press Results and find the image from November 11 2014 You can preview the image and its attributes by clicking on the image icon d Choose the download icon and download the Entity ID LC80460282014315LGNOO Level 1 GeoTIFF Data Product you will need to epee anne create an account and login before you can do this E Row 28 it s easy quick and free jaa e Once the package has completed downloading move it to a file on your computer that is easy to access You may need to download an external Archive Utility to unzip the file 3 Now you re ready to start mapping GIS Data Processing 1 GIS Basics a In order to get a good introduction to Geographic Information Systems GIS what they can do and how they work read these pages i Introducing GIS Vector Data Raster Data and Coordinate Reference Systems All can
4. 14380 In b 15 ME41 1 Winter 2015 Lab 4 4 Choose where to save the result i e landsurfacetemp tif and press Run After a few seconds the land surface temperature raster in kelvin will be loaded in QGIS In order to make the image more visually meaningful you must specify the style of the image 5 Double click on the Result layer this should bring up the Properties gt Style window As Render type choose Singleband pseudocolor Under the Generate new color map box keep Spectral as the color scheme but check to Invert it the mode to use is Equal Interval with 3 Classes The Min and Max values can be estimated by looking at the Raster Histogram found under the Histogram tab on the sidebar Finally back in the Style tab press the button Classify Press OK to apply the results and get back to the main window 6 Congratulations you now have a working image of land surface temperature over Portland on November 11 2014 eee QGIS 2 4 0 Chugiak Final Tutorial ee eS Ue PP PHP LR Awe Bo Se BE ayo ee gt BY a Mier Gye z Tisa ee E Cetin lt lt band set gt gt T EB eo Processing Toolbox eames gt dp GDAL OGR 32 geoalgorithms Bi 220 000000 PF Reclassified Grid I errormatrix3 tif JF classification3 tif Hl 0 Unclassified Water Vegetation Builtup Bare soil Snow Class_temp_group C ROLshp 8 g
5. Color interpolation Exact kd BU Specral w Invert raster in the Layers panel to bring up the Dee aaa ae Properties gt Style window This is also where a ooa Mex 4 2 200000 2 tiam Classify you can change the colors and labels of the F E Men menor classes EE oe 20 980 gt Extent Accuracy Ofa Estimate taster Current Actual slower cip Restore Default Style Save As Default Load Style Save Style Help Apply Cancel a 14 ME41 1 4 Winter 2015 Lab 4 Select where to save the emissivity raster by clicking the button under the Reclassified Grid heading Press Run and after a few moments the reclassified grid i e emissivity will be added into QGIS A F E pBAa3r or n a gt K m X lt lt band set gt gt ka v PF Reclassitied Grid Trvrororororryvy Hl 0 92995 0 97987 IF errormatrix3 tit Unclassified Water Vegetation Builtup Bare soil Snow Class_temp_group ROI shp a pF landsat P RT_LC80460282014315LGN00_B11 TIF P RT_LC80460282014315LGN00_B10 TIF PF RT_LC80460282014315LGN00_B9 TIF P RT_LC80460282014315LGN00_B7 TIF P RT_LC80460282014315LGN00_B6 TIF D RT_LC80460282014315LGN00_B5 TIF B RT_LC80460282014315LGN00_B4 TIF E RT_LC80460282014315LGN00_B3 TIF D RT_LC80460282014315LGN00_B2 TIF E RT_LC80460282014315LGN00_B1 TIF ss SCP Classification BarF gh QGIS 2 4 0 Chugiak Final Tutorial w EA g Fd 2 98 2
6. END OF IN LAB WORK HAVE YOUR LAB TA CHECK THAT YOU HAVE PERFORMED THE CLASSIFICATION PREVIEW 9 Inthe Classification output dock press the Perform classification button and save the output ex classification tif The classification will take several minutes to complete so wait until it has finished executing and has been loaded into the Layers panel It should look similar to the below image 11 ME411 Winter 2015 Lab 4 Note Some inconsistencies will remain in your result For example take a look at the lower right hand corner where there is some cloud cover in the original image The classification is now reading the cloud and its shadows as built up and bare soil Estimating land surface temperature from satellite data will never be perfect hence the need for ground truthing measurements 10 You can further check the accuracy of your land cover classification by performing an accuracy assessment between the classification and the training ROIs Select Post Processing gt Accuracy from the SCP toolbar F 11 Select your classification tif layer as the classification to assess and the ROI shp as the reference shapefile If these do not appear in the dropdown menus Refresh list and check again Press the Calculate error matrix button and select where the error matrix a csv file and the error raster are saved The error matrix will be displayed in the SCP screen and saved in t
7. PKG 2A Ra abc T abe abs abe abc amp Coordinate x Y Si 7a af ae 426102 5111135 Processing Toolbox Y Recently used algorithms amp Shortest Path Polygon centroids Reclassify grid values Raster calculator gt pg GDAL OGR 32 geoalgorithms vV vy y GRASS GIS 7 commands 159 geo amp Models 3 geoalgorithms wy GRASS commands 167 geoalgorit Advanced interface Scale 1 1 161 374 E Render EPSG 32610 4 IV CONVERSION FROM AT SATELLITE TEMPERATURE TO LAND SURFACE TEMPERATURE Now we will convert the At Satellite Brightness Temperature calculated Step III to Land Surface Temperature From the Processing toolbox panel navigate to SAGA gt Grid Calculus gt Raster calculator Press the button to choose the two input layers the emissivity raster Reclassified grid and the thermal Landsat8 band B10 1 Input layers 0 elements selected Formula Result Save to temr Log Help Open output file after running algorithm Under formula write oo Multiple selection RT_LC80460282014315LGN00_B5 TIF EP classification3 tif EPSG 32610 landsat EPSG 32610 RT_LC80460282014315LGN00_B9 TIF EPS RT_LC80460282014315LGN00_B2 TIF EPS RT_LC80460282014315LGN00_B3 TIF EPS errormatrix3 tif EPSG 32610 Reclassified Grid EPSG 32610 RT_LC80460282014315LGN00_B10 TIF EF RT_LC80460282014315LGN00_B11 TIF EF RT_LC8046028201
8. be found at http docs qgis org 2 6 en docs gentle gis introduction introducing gis html ii The user manual created by QGIS authors is also a valuable tool if you decide to continue using QGIS applications http docs qgis org 2 6 en docs training manual 2 In this tutorial we will estimate the land surface temperature over Portland OR using Landsat8 amp imagery and the Semi Automatic Classification Plugin SCP for QGIS There are four phases we will go through a Conversion of raster bands from digital numbers DN to reflectance and At Satellite Temperature b Land cover classification of study area Reclassification of the land cover classification to emissivity values d Conversion from At Surface Temperature to Land Surface Temperature 4 ME41 1 Winter 2015 3 Inthe Landsat package you downloaded from USGS we will use bands 2 7 and 10 The other bands we will not use are designated for coastal aerosol panchromatic and high cloud cover Visit http landsat gsfc nasa gov page_ id 5377 fora review of the bands collected by Landsat8 The following is a list of the bands we will use and their respective spectrum d m omano Band 2 Blue Band 3 Green Band 4 Red Band 5 Near Infrared Band 6 Short Wavelength Infrared 1 Band 7 Short Wavelength Infrared 2 Band 10 Thermal band TIRS 1 Note band 11 is also a thermal band but there is larger uncertainty in its values M
9. calculated as T Tg 1 A Ts p In e J where e wavelength of emitted radiance e p h c s 1 438 10 2 mK e h Planck s constant 6 626 104 34 Js e s Boltzmann constant 1 38 104 23 J K e c velocity of light 2 998 1048 m s The values of A for the thermal bands of Landsat 8 are listed in the following table Sobrino J Jim nez Mufioz J C amp Paolini L 2004 Land surface temperature retrieval from LANDSAT TM 5 Remote Sensing of Environment Elsevier 90 434 440 i Weng Q Lu D amp Schubring J 2004 Estimation of land surface temperature vegetation abundance relationship for urban heat island studies Remote Sensing of Environment Elsevier Science Inc Box 882 New York NY 10159 USA 89 467 483 2 Mallick J Singh C K Shashtri S Rahman A amp Mukherjee S 2012 Land surface emissivity retrieval based on moisture index from LANDSAT TM satellite data over heterogeneous surfaces of Delhi city International Journal of Applied Earth Observation and Geoinformation 19 348 358 ME411 Winter 2015 Lab 4 Download Software and Data Products 1 You will need to install the mapping software QGIS with the GDAL package as well as the python modules Numpy Scipy and Matplotlib a For Windows all needed software can be found here http www qgis org en site forusers download html b For Mac install QGIS and GDAL here http www kyngchaos com software qgis and the modules here
10. new incremental class ID to each ROI and a different macroclass ID MC ID to each unique land cover class Creation of around ten ROIs would give you a good start but you should define as many as needed for a classification that both correctly and completely assigns land cover classes Here are some examples Vegetation MC ID 2 Builtup MC ID 3 ME41 1 Winter 2015 Lab 4 Bare soil MC ID 4 Snow MD ID 5 5 As you are working compare the spectral signatures of each ROI you collect in order to evaluate the spectral similarity In the Signature list table highlight one or more of the signatures and click the button that pulls up the spectral signature plot By checking the box Plot o the standard deviation of each signature is displayed iqnature list ta sf S MCI MC Info a Cho Piot o s 1 4 Bare soa soa t 2 JE Builtup builtup_2 3 s5 Snow snow_1 4 D2 Vegetation veg s 1 Water water_1 Fit to data 6 1 Water water_2 Bitte Dt ets Plot 1 Water 1 water_1 1 Water 1 water_2 2 Vegetation 2 veg_1 3 Builtup 3 builtup_2 4 Bare soil 4 soil_1 5 Snow 5 snow_1 00 7 BBZ 6 Another way to check the accuracy of your creation of signatures is to perform a temporary classification on part of the image Select Spectral Angle Mapping as the classification algorithm in the SCP Classification panel make sure
11. solar zenith angle sz 90 Use For more accurate reflectance calculations per pixel solar angles could be used instead of the scene center solar angle but per pixel solar zenith angles are not currently provided with the Landsat 8 products Conversion to At Satellite Brightness Temperature TIRS band data can be converted from spectral radiance to brightness temperature using the thermal constants provided in the metadata file K2 k gt Me In 1 Ly where T At satellite brightness temperature K Ly TOA spectral radiance Watts m2 srad um K Band specific thermal conversion constant from the metadata K1_CONSTANT_BAND_x where x is the band number 10 or 11 K3 Band specific thermal conversion constant from the metadata K2 _CONSTANT_BAND_ x where x is the band number 10 or 11 Based on these formulas we will be able to concert digital numbers to At Satellite Brightness Temperature For Landsat 8 the K1 and K2 values are provided in the image metafile There are several studies about the calculation of land surface temperature For instance using NDVI for ME41 1 Winter 2015 Lab 4 the estimation of land surface emissivity or using a land cover classification for the definition of the land surface emissivity of each class For instance the emissivity e values of various land cover types are provided in the following table Asphalt 0 942 Therefore the land surface temperature can be
12. 4315LGN00_B6 TIF EPS RT_LC80460282014315LGN00_B7 TIF EPS RT_LC80460282014315LGN00_B4 TIF EPS RT_LC80460282014315LGN00_B1 TIF EPS Cancel Select all Clear selection Toggle selection oo Processing Toolbox gt di GDAL OGR 32 geoalgorithms gt yw GRASS commands 167 geoalgo gt yw GRASS GIS 7 commands 159 g gt amp Models 3 geoalgorithms b Orfeo Toolbox Image analysis gt QGIS geoalgorithms 79 geoalgo v SAGA 237 geoalgorithms Geostatistics gt Grid Analysis v Grid Calculus Function amp Fuzzify Fuzzy intersection and amp Fuzzy union or amp Geometric figures Gradient vector from cart Gradient vector from pola Grid difference Grid division Grid normalisation Grid standardisation Grid volume Grids product Grids sum Metric conversions Polynomial trend from grids Random field ee b 1 10 8 b 14380 In a where a is the emissivity raster and b is the brightness temperature raster wy v voronoi Creates a Voronoi di y v voronoi Creates a Voronoi di gt Orfeo Toolbox Image analysis 83 gt QGIS geoalgorithms 79 geoaigorit gt amp SAGA 237 geoalgorithms gt Scripts 15 geoalgorithms NOTE If the B10 band appears above the reclassified grid during selection you must adjust the formula to match this ordering i e a 1 10 8 a
13. ME411 Winter 2015 Lab 4 ME 411 LAB 4 REMOTE SENSING LAND SURFACE TEMPERATURE Prepared by Katie Fankhauser Luca Congeto Evan Thomas This is the first of a three part lab series for ME 411 Measurement and Instrumentation In this sequence we will cover end to end data collection and analysis for a practical application The overall objective is to develop an algorithm to determine ground surface temperature using satellite imagery and to correlate and calibrate that algorithm against field measurements taken with calibrated thermocouples through a data acquisition system At the end of the sequence you will have an introductory familiarity with Remote Sensing NASA s LandSat 8 satellite imagery products Geographic Information Systems GIS using a free software platform QGIS Thermal infrared spectrum analysis Ground surface temperature derivation Thermocouple design and calibration Data acquisition through a analog to digital A D converter MySQL data storage and retrieval Data calibration statistical analysis and correlations using the R Project for statistical computing Collecting field data for ground truthing Correlation and calibration of ground data against satellite derived data Introduction The term remote sensing usually describes the collection of data by satellites In most cases the remote refers to spectral imagery collected by cameras and other spectral instruments across a broad ra
14. TOA spectral radiance Watts m2 srad um M Band specific multiplicative rescaling factor from the metadata RADIANCE_MULT_BAND_x where x is the band number A Band specific additive rescaling factor from the metadata RADIANCE _ADD_BAND_x where x is the band number Qeal Quantized and calibrated standard product pixel values DN Conversion to TOA Reflectance OLI band data can also be converted to TOA planetary reflectance using reflectance rescaling coefficients provided in the product metadata file MTL file The following equation is used to convert DN values to TOA reflectance for OLI data as follows p MpQcai Ap where pA TOA planetary reflectance without correction for solar angle Note that pA does not contain a correction for the sun angle M Band specific multiplicative rescaling factor from the metadata REFLECTANCE _MULT_BAND_x i https landsat usgs gov Landsat8 Using Product php ME41 1 Winter 2015 Lab 4 where x is the band number Ap Band specific additive rescaling factor from the metadata REFLECTANCE_ADD_BAND_x where x is the band number Qeal Quantized and calibrated standard product pixel values DN TOA reflectance with a correction for the sun angle is then Ov p p cos Jsz sin se where p TOA planetary reflectance Ose Local sun elevation angle The scene center sun elevation angle in degrees is provided in the metadata SUN ELEVATION Os7 Local
15. ake sure the metadata file MTL txt remains with the dataset 4 Installing the SCP Plugin a b Open QGIS In the menu bar choose Plugins then Manage and Install Plugins Lab 4 Semi Automatic Classification Workflow Conversion of raster bands from DN to reflectance Definition of the classification inputs Creation of the ROIs Semi Automatic classification of the study area Calculation of classification accuracy Legend tal Data Calculation of classification statistics Cc Process http fromgistors blogspot com Search for Semi Automatic Classification Plugin and click Install plugin Close and open QGIS again to restart the plugin application View gt Panels gt SCP ROI Creation Do the same for SCP Classification Layers and Toolbox should also be checked if they aren t already l CONVERSION OF RASTER BANDS FROM DN DIGITAL NUMBERS TO REFLECTANCE The SCP automatically converts Landsat Digital Numbers DN dimensionless pixel values to Top of Atmosphere reflectance TOA which is defined as the ratio between the radiation striking a surface and the radiation reflected off of the surface In the same step the SCP performs atmospheric correction using the DOS1 Dark Object Subtraction 1 DOS1 method Conversion and correction are necessary because atmospheric effects such as absorption and scattering affect the electromagnetic energy measured by satellites
16. ant to include bands that are outside of our spectrum of interest Reorder the bands so that they are in increasing order i e B2 will be in the first soot and B10 will be in the last spot From the dropdown Quick wavelength settings menu select Landsat 8 OLI The center wavelength of your bands should have now been adjusted Your screen should match the one below eee EF Semi Automatic Classification Plugin EF Tools i Preprocessing jj Post processing Settings About Select raster bands RT_LC80460282014315LGNO0_B11 TIF Refresh list RT_LC80460282014315LGNO00_B10 TIF Select all RT_LC80460282014315LGNO0_B9 TIF RT_LC80460282014315LGNO0_B7 TIF RT_LC80460282014315LGNO00_B6 TIF Add rasters to set Band set definition F Band name Center wavelength 1 RT_LC80460282014315LGN00_B2 TIF 0 48 Wavelength unit 2 RT_LC80460282014315LGN00_B3 TIF 0 56 um 1 E 6m 3 RT_LC80460282014315LGN00_B4 TIF 0 655 et pee 4 RT_LC80460282014315LGNO00_B5 TIF 0 865 l Ea RT_LC80460282014315LGN00_B6 TIF 1 61 i Clear all 6 RT_LC80460282014315LGN00_B7 TIF 2 2 Import Export Quick wavelength settings Landsat 8 OLI bands 2 3 4 5 6 7 Create virtual raster of band set Create raster of band set stack bands Show docks GJ Quick user guide Online help Close N ME411 Winter 2015 Lab 4 9 Inthe SCP ROI creation panel save a new shapefile where the Regions of Interest in the following step will be stored Click the button Ne
17. ask Reset 1 Macroclass_1 Create vector Classification report CID C Info 1 ji Class_1 Perform classification x Add sig list Layers 12 It is also a good idea to save the entire project at this time Go to Project in the top menu bar and Save as Choose a file where to save your work Remember to frequently save your project throughout this tutorial some of the processing can cause QGIS to crash or freeze at times Il COLLECTION OF ROIS AND SPECTRAL SIGNATURES Regions of Interest ROIs are polygons drawn over homogeneous areas of the image that represent land cover classes ROIs can be drawn manually or with a region growing process i e image segmentation that groups similar pixels and they should account for the spectral variability of land cover classes The Semi Automatic Classification Plugin SCP calculates the spectral signatures which are used by classification algorithms considering the pixel values under each ROI SCP allows for the definition of a Macroclass ID i e MC ID and a Class ID i e C ID for each ROI or spectral signature which are the identification codes of land cover classes Macroclasses allows for the classification of materials that have different spectral signatures therefore are processed individually but belong to the same land cover class thus the same MC ID is assigned to these pixels For instance we could classify grass e g MC ID 1 and C ID 1 and trees e g MC ID 1 an
18. ate crops TIRS will become an invaluable tool for managing water consumption http www nasa gov mission_pages landsat spacecraft index html VMaqKHDF 2E ME411 Winter 2015 Lab 4 In this lab we will take LandSat 8 imagery from the TIRS spectroradiometer for Portland and derive ground surface temperature The USGS provides an introduction to processing LandSat 8 spectral data which we have quoted directly in the box below Using the USGS Landsat 8 Product The standard Landsat 8 products provided by the USGS EROS Center consist of quantized and calibrated scaled Digital Numbers DN representing multispectral image data acquired by both the Operational Land Imager OLI and Thermal Infrared Sensor TIRS The products are delivered in 16 bit unsigned integer format and can be rescaled to the Top Of Atmosphere TOA reflectance and or radiance using radiometric rescaling coefficients provided in the product metadata file MTL file as briefly described below The MTL file also contains the thermal constants needed to convert TIRS data to the at satellite brightness temperature Further details can be found in the LDCM Cal Val Algorithm Description Document and the Landsat 8 Science Users Handbook available from the Landsat website Conversion to TOA Radiance OLI and TIRS band data can be converted to TOA spectral radiance using the radiance rescaling factors provided in the metadata file La M Qeai AL where Ly
19. d C ID 2 as a vegetation macroclass e g MC ID 1 Every ROI or spectral signature should have a unique C ID while the MC ID can be shared with other ROIs In the dock Classification it is possible to choose between MC ID and C ID classification 1 In order to create an ROI choose the next to Create a ROI on the SCP ROI Creation panel You may choose whether to leave the NDVI or EVI measures of vegetation that are respectively either chlorophyll sensitive or based on variations in canopy cursor on as you navigate the map Zoom in the map and click on a blue pixel of the Willamette River After a few seconds a semitransparent polygon will appear over your selection Under the ROI parameters heading ME41 1 Winter 2015 Lab 4 check the Automatic Refresh ROI button and experiment changing the Min ROI size and Range radius to see its effect on the ROI selection 2 Assign a Macroclass ID and Class ID to the selection and write a brief description of the ROI under ROI Signature definition Make sure each land cover class i e water is assigned the same Macroclass ID and that each ROI within this class is given a sequential Class ID 3 In order to save the ROI to the training shapefile click the button Save ROI to shapefile making sure the box next to this button Add sig list is checked The ROI is now saved in the ROI list and the spectral signature is added to the Signature
20. e vegetation comprises about 56 of the total image while being around 30 000 square kilometers in size eee EF Semi Automatic Classification Plugin DF Tools f Pre processing a Band set i Settings About H Accu Fo Land cover ch 3 Classification to v 3 gt Reclassific Select the classification classification3 tif w Refresh lis Use No data value 0 Calculate classification report lass PixelSum Percentage Area metre 2 Cc 1 975938 1 58345891615 878344200 0 2 34320982 55 6857761034 30888883800 0 3 1264566 2 05175770157 1138109400 0 4 24725773 40 1175543072 22253195700 0 5 346042 0 561452971665 311437800 0 Save report to file Show docks J Quick user guide C Online help Close Note These figures were created for the purpose of this tutorial Several more ROIs of each class and consideration of their spectral variability are needed for a better classification Also field data is useful for improving the creation of ROIs and spectral signatures III RECLASSIFICATION OF THE LAND COVER CLASSIFICATION TO EMISSIVITY VALUES 1 The emissivity e values for the land cover classes are provided in the following table these values are only indicative because they should be obtained from field survey Land Surface _ uitup T http www infrared thermography com material 1 htm 2 Inthe Processing Toolbox panel navigate to SAGA gt Grid Tools gt Reclassify grid values In this tool wind
21. he error matrix csv file and the error raster will be loaded in QGIS Each value or color represents the comparison between the user created ROIs and the classification produced by the SCP EF Input ima lt lt band set gt gt gt ier b amp Ei L 9o00 Layors IF orrormatrix tif EB Unchanged 1 E 12 0 24 0 IF classification2 tit VA SAS m o 3 ay wi Go me gt fez 8 PF landsat IF RT_LC80460282 IF RT_Lc80460282 IF RT_LC80460282 IF RT_LC80460282 IF RT_LC80460282 IF RT_LC80460282 IF RT_LC80460282 IF RT_LC80460282 IF RT_LC80460282 IF RT_LC80460282 S of Ss amp EY O A oD os lt 0 a tO Le Us i koe eee IF Semi Automatic Classification Plugir Tools fj Preprocessing ae Band s F Land cover ch Classification r Classifi LT Select the classification to assess classification tif Select the reference shapefile ROLshp Calculate error matrix S F amp eference lassification PixelSum 9 A OONONAON M 3 SPPeOWDOWONNNNN nm am T ON HEON F S20N lt 7 20N lt 0 o I Show docks CA Quick user guide C Online help 12 Scrolling to the bottom of the Error Matrix window shows the accuracy of the classification ERROR MATAIX gt Reference V Classification 1 2 1 9190 1387 2 46 8251 3 0 0 4 0 0 5 0 0 Total 9236 9638 Overall accuracy 92 3166708957 Class 1 producer accuracy 99 5019488956 Class 2 pr
22. list table Define the color of classes that will be used in the classification by double clicking on the Color column in the Signature list of the SCP Classification panel i e use blue for water Dee QGIS 2 4 0 Chugiak SCP ROI creation i a oe EX oe i gt fb aD TO gt Lo de bes Say ix Q x 5 LOT EO EA ewe ee E Foish Bo New sh E i 2 ae Ky Nta san i a I CE ee oll Mi OAT a ono F Open Jesktop Tutorial SIG xmi Save Reset S WCIC MCino CID Cino Color 1 1 Water 1 water1 D A Addtosignatue b k ROl parameters n QB le Tb FE Lhe EE Epon import Range radius Min ROIsize Max ROI width assihcaton algorithm 0 010000 gt 60 gt 100 gt A Minimum Distance 0 0000 C Rapid ROI on band 115 Use Macroclass ID o Ro gt a 2 e i Ol creation j Vey Sze 100 3 Redo U i J ROlcreaton O s s a j Ki FP assihcation style Rada 0 x dee Display cursor for NDVI kd Show RO Select qmi Reset Aassitication output LORE definton inition MC ID IMC Info Apply mask 1 gt Water C Create vector Classification report TID C info Perform classification I ao water_1 Save ROI a Add sig list Unda Stee ee Layers 4 Create multiple ROIs for the remaining land cover classes vegetation builtup bare soil and snow Remember to assign a
23. nge of wavelengths In case the of Earth observation satellites take spectral data reflecting from the atmosphere and the Earth s surface Interpretation of this data often represented as imagery requires an understanding of spectral data and physical properties of the Earth and atmosphere It also often requires calibration against data collected on the Earth s surface or in the atmosphere directly data from sensors that are in situ rather than remote NASA s Earth Observatory website provides an excellent overview of the field of remote sensing and the basic processes involved Please carefully read this page http earthobservatory nasa gov Features RemoteSensing The NASA USGS LandSat program was launched in 1972 and was the first Earth observation satellite not designed for military use The current satellite LandSat 8 was launched in 2013 LandSat 8 has two main instruments the Operational Land Imager visible near IR and short wave IR and the Thermal Infrared Sensor thermal IR LandSat 8 covers every point on Earth every 16 days and has a resolution of 15 100 meters TIRS was added to the LandSat 8 mission when it became clear that state water resource managers rely on the highly accurate measurements of Earth s thermal energy obtained by LDCM s predecessors Landsat 5 and Landsat 7 to track how land and water are being used With nearly 80 percent of the fresh water in the Western U S being used to irrig
24. oducer accuracy 85 6090475202 Class 3 producer accuracy 62 7218934911 Class 4 producer accuracy 97 02479393884 J 4 19 0 21 1 106 17 23 587 0 0 169 605 user accuracy 86 7308418271 user accuracy 99 1825940618 user accuracy 86 1788617886 user accuracy 96 2295081967 Class 5 producer accuracy 100 0 user accuracy 100 0 12 ME411 Winter 2015 Lab 4 In this classification the overall accuracy is around 92 In general classification accuracy gt 80 is considered good It is also useful to consider the error for single classes In the upper half of the image above the number of pixels classified correctly is displayed along the major diagonal As you can see the largest errors are in class 3 bare soil This is also confirmed by a comparison of user and producer accuracy In order to improve the results one would want to collect more ROIs and spectral signatures in the bare soil class paying attention to the spectral similarity with other classes 13 We can also use the SCP tool to calculate the percentage and area of land cover classes While you are still in the Post Processing window select the Classification report tab Select the classification tif and press Calculate classification report After a few seconds the report will generate the percentage and area the area unit is calculated from the image itself of each land cover class in the image In this exampl
25. ow select your classification tif as the Grid under method choose 2 simple table and press the button to the right of the Fixed table 3 X 3 heading 13 ME41 1 Winter 2015 Lab 4 Reclassify grid values Log Heip Grid classification3 tif EPSG 32610 kj ene Method 2 simple table old value for single value change 0 000000 a new value for single value change 1 000000 operator for single value change 0 minimum value for range 0 000000 se wo maximum value for range 1 000000 A new value for range 2 000000 A operator for range 0 lt Lookup Table Fixed table 3 X 3 w operator for table 3 This button will bring up a table with Minimum Maximum the range of the current value and New value columns The minimum and maximum values are based on the Macroclass ID you assigned to each land cover class in previous steps For example water MC ID 1 will have a minimum value of 1 maximum value of 1 9 and a new value of 0 98 e Fill out the table as seen below making sure to leave the first line as unclassified Fixed Table 0 0 0 1 1 9 0 98 Cancel 2 29 0 98 3 3 9 0 94 4 4 9 0 93 5 5 9 0 85 Add row Remove row e o Layer Properties classification3 tif Style Note The value of each land cover class can be aS es confirmed by double clicking the classification em p Oee aw oeio aao A u
26. rface emissivities when possible Homework 5 Assignment Email fankhauk onid oregonstate edu your error matrix csv file and a screen shot of your final image in QGIS as above by February 10 at 10 am 16
27. t yw GRASS commands 167 geoalgo gt yw GRASS GIS 7 commands 159 g gt Models 3 geoalgorithms gt Orfeo Toolbox Image analysis gt QGIS geoalgorithms 79 geoalgo v SAGA 237 geoalgorithms Geostatistics gt Grid Analysis v Grid Calculus Function Fuzzify Fuzzy intersection and Fuzzy union or Geometric figures Gradient vector from cart 23 PF landsat Gradient vector from pola gt E RT_LC80460282014315LGN00_B11 TIF Grid difference gt E RT_LC80460282014315LGN00_B10 TIF Grid division gt E RT_LC80460282014315LGNO00_B9 TIF Grid normalisation Sy BH RT_LC80460282014315LGNO0_B7 TIF amp Grid standardisation gt JR RT_LC80460282014315LGN00_B6 TIF amp Grid volume gt JR RT_LC80460282014315LGN00_B5 TIF drer gt BH RT_LC80460282014315LGNO0_B4 TIF S ri Bends letric conversions gt E RT_LC80460282014315LGNO0_B3 TIF Polynomial trend from grids gt BF RT_LC80460282014315LGNO0_B2 TIF amp Random field gt RT_LC80460282014315LGNO00_B1 TIF S Random terrain generation b Grid Filter SCP Classification Advanced interface Po Coordinate 492014 5026991 Scale 1 1 161 374 Render EPSG 32610 A 4 NOTE Keep in mind that analysis of satellite imagery is best used for relative not exact measurements Thus it is important to perform field surveys to determine actual land cover classification of the area and measure su
28. w shp and select where to save the shapefile for instance ROI shp 10 Create a signature list file where the spectral signatures of the land cover classes will be stored and used to create the classification by clicking the button Save in the SCP Classification panel for example SIG xml 11 In the SCP toolbar the name lt lt band set gt gt is displayed in the Input image combo box The shapefile name is displayed in the Training shapefile combo box and the path to the xml file is displayed in the Signature list file See below QGIS 2 4 0 Chugiak SCP ROI creation fm 3 e Fa P y Training shap wB BARQAS SPPHAPRBPLAws A vrb E New sh jf y o gt i vi abe ba A ane ay he ag t ROI shp 5 woth i GAC CE lt lt band set gt gt U RGI A P ey 5 F us iF E i G a s MIC IC MC Info CID C Inf 00 SCP Classification Va Poi Open Desktop TutoriaV SIGxmi Save Reset Q SOO E S VCIC MCino CID Cinfo Color P Add to signature ba TA B RO parameters la E BS le ke Lk Export import ug Range radius Min ROI size Max ROI width Class caton algorithm 40000 m 1 Q Select classification algorithm Threshold a g 60 Ji 00 jl P Minimum Distance 0 0000 5 Rapid ROI on band 11 2 fs Use Macrociass ID Automatic refresh RO Automatic plot Vv Li Vov Size 100 2 a KN Classincation styie Display cursor for NOVI Show RO Select qmi Reset Classihcation output n LEO MC ID MC Into Apply m
Download Pdf Manuals
Related Search