CLABAUTAIR User Manual
Contents
1. be liable to you for damages including any general special incidental or consequential damages arising out of the use or inability to use the programs 1 2 First steps Under http www smhi se cloud_generator you will find updates for both manual and programs To unpack the file CLABAUTAIR tar gz just copy this file to the directory of your choice in a UNIX LINUX environment and type first gunzip CLABAUTAIR tar gz and then tar xvf CLABAUTAIR tar This will create a directory CLABAUTAIR containing all the necessary files 1 INTRODUCTION 2 Architecture This section give an overview about the structure of program cooperation and program flow Central settings the user wishes are to be placed in one central file communicat dat that all programs will read 2 1 Input files Most of the data files in use are both input and output files because they are produced or modified by one program and read afterwards by the next The files discussed here are files that must exist before of the first program run 2 1 1 communicat dat This is the main file where all user settings are stored in The coding is position based i e there are no keywords needed the meaning of variables is defined by their position in this file Since the requirements are concise this shouldn t be a problem Settings are to be written in lines one to nine This are 1 obsfile 2 Xres yres angres 3 nz corrflag 4 vertical borders 5 lwcres reffr2. CLABAUTAIR User Manual Version 1 0 October 2006 correspond to Ronald Scheirer Sveriges meteorologiska och hydrologiska institut SMHI Forskning och Utveckling Folkborgsvagen 1 601 76 Norrk ping Sverige email ronald scheirer smhi se CONTENTS Contents 1 Introduction LL Biscl aimer sr A oe ea a A aa ta See ted 1D WITS SUC DSS gaar a wey Mere Se a Be ad AT oe 2 Architecture 2 19 Input Dies 44 tl IA AO A EEE 5 Sa DL communi Catsdat it a a are Snap ar ra DAT Eine 1 gt obshle tota a Shake F e 2 1 1 2 Line 2 xres yres angres os sa Bu u 2 1 1 3 Line 3 nz corrflag A II ss ss sot 2 1 1 4 Line 4 vertical borders snus 2 2 rr rr rr rr seen 2 1 1 5 Line 5 lweres reffres 2 1 1 0 Eine G TIR He 3a EEE Rue MN o a ware at ta eek 2 1 1 8 Line 8 vel dir slot time ZELLE Ele Dr prefl g ost ter lr ls is e 2 19 name Ged eee eno aa a er ee 2 2 Prosramsand theirtask 5 a 2222 as si a ee e a Book Avia os SN a a ete O Glas DEEE AS 23262 LINSE ar ee aS ee ae is es ie ee ws a a a Be 22 0 COTES hw ue 29 din Re ew A we ha Ah a DIA RA Sn cane eaten o at tin eaten Un as cus Siete a Gi Sen 22 9 COSCAVC io SIRE RRS RAO ee AS 2 37 Ou tp tfles dl late O lends Oh ES aoa Gl ahd aoe ete Zi SWELOUdSD dat es LA ANS LA an adds Sera hd Risa O0 O0 III O Or OO OT A RE R www CONTENTS 1 Introduction One of the remaining challenges in three dimension
3. al radiative transfer calculation is the generation of realistic cloud fields as input for sophisticated radiative transfer models The algorithm CLABAUTAIR cloud liquid water content and effective radius retrieval by an automated use of aircraft measurements has been invented at the Institut fiir Physik der Atmosphare Deutsches Zentrum fiir Luft und Raumfahrt Oberpfaffenhofen Germany but development is still going on This tool is designed to retrieve the structure of clouds from airborne measurements of microphysical parameters Data from individual flight legs are scanned for characteristic patterns and the autocorrelation functions for several directions are used to extrapolate the observations along the flight path to a full three dimensional distribution of the cloud field Thereby the mean measured profiles of microphysical parameters are imposed to the cloud field by mapping the measured probability density functions onto the model layers 1 1 Disclaimer These programs are distributed as freeware Except when otherwise stated in writing the software packet is provided as is without warranty of any kind either expressed or implied including but not limited to the implied warranties of merchantability and fitness for a particular purpose The entire risk as to the quality and performance of the programs is with you In no event unless agreed to in writing will the author or any other party who may modify and or redistribute the programs
4. ase read Section 2 2 3 for some details 2 1 1 4 Line 4 vertical borders Enter here the nz 1 arbitrarily spaced layer borders real values in km If nz is less than zero type just three real values here The lowest border the upper border and the increment Again in km 2 1 1 5 Line 5 lwcres reffres This is not a critical setting Just give here the bin width for the calculation of the probability density functions of LWC lweres and Reff reffres in real numbers Take care that you not just get one bin filled with measurements and not plenty of bins filled with just one measurement Use simply the same units that are used in the measurements file 2 1 1 6 Line 6 gridflag This is easy An integer that should be 0 if the horizontal positions in the measurements file are Cartesian coordinates use distance in meter separated in North South and East West direction from a certain reference point in the South Westerly corner or 1 if they are given in latitude and longitude 2 1 1 7 Line 7 indflag Another integer flag If set to 0 LWC and Rog are cal culated independently on the same way calculating the autocorrelation functions etc This could be time consuming and may lead into trouble if for example you find two slightly shifted fields for scattered clouds of LWC and Reg due to small differences in their autocorrelation functions or if you use different instruments for deriving LWC and Reff One answer and a proper one at lea
5. ation coefficient as a function of the total shifted distance Since the overlap of both series decreases with increasing lag the user can switch between a periodic shift or just filling empty locations with zeros Except for the unlikely case that a flight leg has exactly the length of an integer multiple of a cloud cloud gap distance the autocorrelation function decreases with displacement from its origin This is not satisfying A possible solution for the future could be a sledge like part of the measurement series to be shifted along almost the whole distance Input The general settings are read from communicat dat and the selected flight legs are taken from output2 dat Output All relevant activities together with some useful results are protocoled to to the file cloud log The autocorrelation functions of LW C s are written to file output3 1 dat and for Rog to output3 2 dat 2 2 4 fill This is the most time consuming program First of all the shifted by the wind vector anomalies are averaged over the respective box sizes where available The starting field initialised this way provides the basis for ongoing calculations If you have strong winds and a small domain or unfavourable distributed measurements severe errors may occur if no or deficient data are left inside the selected domain Empty boxes are selected by random next From the centre of the current box filled boxes 8 2 ARCHITECTURE are searched along the directions determ
6. d cloud log from LAT LON notation to an orthog outputl dat onal metric grid output1b dat lines select contiguous measurements lo communicat dat communicat dat cated on a straight line for calculations output1 dat cloud log of autocorrelation functions output2 dat output2b dat corel calculate lag correlation coefficients communicat dat cloud log along flight legs by shifting a measure output2 dat output3 1 dat ment series against a copy of itself output3 2 dat fill shifted measured anomalies are aver communicat dat cloud log aged over the respective boxes and a outputib dat random seed wheighted interpolation along flight di output3 1 dat output4 1 dat rections over the whole 3D space will output3 2 dat output4 2 dat be done rescale fit the resulting probability density communicat dat cloud log functions of LWCs and Reg to the outputl dat random seed measurements output2b dat wcloud3D dat output4 1 dat output4 2 dat notation to an orthogonal metric grid The most south westerly coordinate is selected for marking the reference corner All later given coordinates are relative coordinates A listing of X 500 m Y 700 m means 700 m north and 500 m to the east of the reference point Input The general settings are read from communicat dat Aircraft data are taken from name red name is specified in communicat dat Output Some general informations are written to communicat dat and all activities are protocol
7. ed to cloud log In output1 dat the measurements in Cartesian coordinates re sorted in layers are stored The same but shifted by the wind vector is found in outputib dat 2 2 Programs and their tasks 7 2 2 2 lines The main purpose of the program lines is to select contiguous measurements located on a straight line for calculations of lag correlation coefficients To this end all measurement coordinates at each level are connected among one another For all data points the found directions are combined to angular bins and measurements outside a preselected time slot are sorted out The eight flight legs found this way providing the most measurements and are directionally separated by angles of at least 15 are selected for the calculation of autocorrelation coefficient Input The general settings are read from communicat dat The re sorted measurement field is read from output1 dat Output Some general informations are written to communicat dat All activities are protocoled to cloud log The points of measurement sorted by flight leg and distance from starting point are stored in output2 dat Some cryptic information on the correla tion of liquid water content and effective radius can be found in output2b dat 2 2 3 corel Autocorrelation function or lag correlation coefficients along flight legs are calculated by this program This is done by shifting a measurement series against a copy of itself and storing the calculated correl
8. es 6 gridflag 7 indflag 8 vel dir slot time 9 preflag 2 1 1 1 Line 1 obsfile This is simply a part of the name of a file containing the aircraft measurements Just enter a string name there CLABAUTAIR expects a file name red to exist in the same directory don t ask for the reasons 4 2 ARCHITECTURE 2 1 1 2 Line 2 xres yres angres This is the place to provide informations on some resolutions you selected Enter two real values for xres and yres to specify the required horizontal resolution xres is assigned to the East West direction and yres to the North South direction both in km The angular resolution angres specifies the angle bin width used to scan the measurements for straight flight legs If it is too small you may get short or even too few lines On the other hand to large values result in blurred legs We found good results with angular resolutions around 2 5 degree By the way an integer times angres should result in a full circle 2 1 1 3 Line 3 nz corrflag The vertical resolution results from settings of nz Place an integer there to give the number of layers The number of vertical borders read from line 4 is nz 1 then If you just have equally sized vertical layers place a negative value here find the reasons in the next paragraph The corrflag is an integer which controls the calculations of autocorrelation functions If this is set to 1 the calculation is switched to a periodic mode ple
9. ined by lines section 2 2 2 Then a weighted av erage over all anomalies found by this means is calculated The autocorrelation coefficients are fixed as weighting factors here If the weighting sum fails to reach a certain threshold initially set to 3 the calculation for this box is postponed Additionally a success rate ST sr calculated bores is calculated periodically In case of small srs over 3 cycles the threshold is reduced temporarily On the other hand if the threshold is reduced and srs are large the threshold is increased again The program ends if all boxes are assigned with calculated anomalies or the threshold has to be reduced too dramatically to expect reasonable results Input The general settings are read from communicat dat and informations on shifted measurements are read from outputib dat The autocorrelation functions of LW C s and Ref are read from output3 1 dat and output3 2 dat respectively Output All rele vant activities together with some useful results are protocoled to to the file cloud log Since a random generator is used here the initialising value is stored in random seed The anomaly field of LW Cs is written to output4 1 dat and for Rag to output4 2 dat 2 2 5 rescale The polish or more precise the roughened finish is performed by this program The aim is to fit the resulting probability density functions of LW Cs and Reg to the measurements To this end the calculated anomalies are assumed t
10. n is the longitude in decimal degree if gridflag is set to 0 lat and lon are expected to contain North South and East West distances in meter from a certain reference point alti the altitude in m lwc the liquid water content in arbitrary units reff the effective radius in arbitrary units and mtime the measurement time in seconds If you want to calculate an other parameter than the LWC say the volume extinction and you think that this parameter should follow similar rules feel free to test it You should replace reff by the same parameter because a correlation may be questionable 2 2 Programs and their tasks In general CLABAUTAIR consists of 5 programs written in FORTRAN90 These pro grams should be started in the order given below sections 2 2 1 to 2 2 5 to make sure that following programs are provided with all required informations from the proper source Table 1 gives an overview on the interaction of the programs 2 2 1 navigat The program navigat fixes the general geometric extensions of the cloud field Ifthe aircraft measurements are not given on a Cartesian grid they are transformed from LAT LON 2 ARCHITECTURE Table 1 The information flow or in other words the interaction between the five programs is given in brief below Program Purpose Reads from Writes to navigat fixes the general geometric extensions communicat dat communicat dat of the cloud field transform positions name re
11. o represent the probability for a local large LWC or Reg First the cloud free fraction of each layer 1 p r z is assigned to the fraction of boxes with the smallest values A random number generator throws the dice for each of the remaining boxes p r 2 X Nyoxes This random number generator is prepared to transform from an equal distribution of numbers to the measured PDFs Finally the largest anomalies in boxes are replaced by the largest values produced by the random number generator Input The general settings are read from communicat dat and original measurements needed for the PDFs are read from outputi dat Foe the connection of LWCs with Reg the informations stored in output2b dat are required Of course the anomaly fields are read from output4 1 dat and output4 2 dat Output All relevant activities together with some useful results are protocoled to to the file cloud log The initialising value of the random number generator is stored in random seed The final three dimensional cloud field is written to wcloud3D dat 2 3 Output files There are several files produced during the program run Normally there is no need to keep all of them 2 3 Output files 9 2 3 1 wcloud3D dat This file follows the requirements for three dimensional cloud files from the libRadtran packages running with the MYSTIC solver There is just one known problem depending on the chosen compiler and its setting Line breaks may appear in the line
12. reserved for initialising vertical borders if it is quite long They have to be removed manually 10 2 ARCHITECTURE
13. st for narrow distributions of Rog is to correlate them with the LWC Choose 1 if you want to do so 2 2 Programs and their tasks 5 2 1 1 8 Line 8 vel dir slot time Four real settings on the wind and time The wind speed in is read from vel the direction in degree from dir A value of 270 means wind is blowing from the West to the East The typical time an aircraft needs to pass one flight leg should be given in seconds here in slot This information is necessary to avoid the assignment of earlier or later measurements with possible different cloud microphysics that may cause discontinuities in the autocorrelation function to a certain flightleg The time in seconds for the calculation of the cloud state is specified in the variable time If less than zero the time centred between the first and the last measurement is taken Note If you use UTC MESZ or Lummerland Time in the measurement file use it also here 2 1 1 9 Line 9 preflag Set this integer flag to 1 if you want cloud free layers between cloudy ones to be calculated first If done so cloudy layers separated by cloud free layers do not affect each other This leads to a cloud field comparable to one following a random maximum overlap assumption 2 1 2 name red This file contains an arbitrary number of aircraft measurements in an arbitrary order Six columns are expected that are lat lon alti lwc reff mtime The meaning of lat is the latitude in decimal degree lo
Download Pdf Manuals
Related Search