1 PSR/A Multiband Polarimetric Imaging During Wakasa Bay (WBAY03) Field Campaign - Data Delivery Report to NASA PIs - B. Stankov, A.J. Gasiewski (PI), M. Klein, V. Leuski, V. Irisov, and B. Weber NOAA Environmental Laboratory Division of Microwave Systems Development 325 Broadway R/ET1 Boulder, CO 80305-3328 (303) 497-7275 [email protected]
39
Embed
PSR/A Multiband Polarimetric Imaging During Wakasa Bay ...nsidc.org/sites/nsidc.org/files/files/asmr_validation/0267_WBAY03... · MATLAB Software and Directory Structure for Displaying
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
PSR/A Multiband Polarimetric Imaging During Wakasa Bay (WBAY03)
Field Campaign
- Data Delivery Report to NASA PIs -
B. Stankov, A.J. Gasiewski (PI), M. Klein, V. Leuski, V. Irisov, and B. Weber
NOAA Environmental Laboratory
Division of Microwave Systems Development 325 Broadway R/ET1
6. Data Structure and Format……………………………………………………………………. 20
7. Data Examples………………………………………………………………………………… 21
8. MATLAB Software and Directory Structure for Displaying PSR/A Data………………………. 38
3
1. Background. In response to the need to estimate rainfall for both climate prediction and severe weather forecasting, ETL has developed a sensor for accurate calibration of existing and planned satellite microwave rainfall instruments. The sensor, based on the ETL Polarimetric Scanning Radiometer (PSR) system, is the first airborne conical-scanning radiometer system, and is able to image rainfall in regions inaccessible by the land-based rain radars, e.g. WSR-88D radar. The system was first flown for rainfall measurement in 1998 and provided images of the rainbands of Hurricane Bonnie at landfall on the North Carolina coastline. As the result of a joint U.S.-Japanese collaboration involving NOAA, NASA, and the Japanese Space Agency NASDA, the PSR is providing data for calibration of the NASA-NASDA AMSR-E sensor on the NASA Aqua satellite over the Sea of Japan and Pacific coastal region east of Japan. Currently, NOAA/NESDIS plans to use the AMSR-E data in operational algorithms and as a means of improving rainfall algorithms in preparation for the launch of NPOESS at the end of the decade. Data from high-resolution airborne instruments such as the PSR is critical for both on-orbit assessment of the performance of sensors such as AMSR-E, AMSU, and CMIS, as well as improving the accuracy of operational satellite rainfall algorithms. The resolution of the PSR is at least a factor of ten greater than that of satellites, and thus provides a means of resolving the structure of rain-producing frontal systems that are otherwise impossible to observe from space. Improved rainfall estimation is essential for quantitative precipitation forecasting (e.g., flash flood, hurricanes at landfall, other high-impact weather such as coastal storms), drought monitoring and prediction, and global climate change assessment. This document provides information for the WBAY03 team scientists about the first version of processed PSR/A data collected during the field campaign and delivered to the National Snow and Ice Data Center (NSIDC) for archiving as part of the WBAY03 combined dataset and to the Principal Investigator, Tom Wilheit in May 2004.
2. Measurements. The PSR configuration on the NASA P-3 N426NA aircraft during the 2003 Wakasa Bay experiment uses the NOAA PSR/A scanhead (Table. 1) operated in both conical and cross-track scanned mode. The conical mode uses an incidence angle of 55 degrees from nadir. The purposes of collecting this data are:
• To study the impact of melting level on passive microwave rain signatures collected coincidentally with multiband (Ku, Ka, and W) scanning Doppler rain/cloud radars. The joint data provides a unique basis for algorithm improvement for NASA TRMM and algorithm development for the future NASA Global Precipitation Mission (GPM).
• To provide high resolution multiband microwave radiometric imagery for: • AMSR-E calibration and rain rate/snow retrieval validation studies. • NOAA-NASA Joint Center for Satellite Data Assimilation (JCSDA) radiative transfer
PSR imaging occurred at medium altitude (~22,000' MSL) during flight lines crossing maritime and orographic precipitation. Cases of both snow and moderate to light rain were observed with melting levels from the surface up to ~8,000'. Several low-altitude (~700' MSL) lines were also flown for surface emission studies. Extensive thermal stabilization of PSR/A radiometers, refurbishment of the PSR/A 18/21 and 89 GHz receivers, and installation of new 37 GHz LNAs account for much improved stability.
3. Flight Summaries. The following flight summaries (Table 2) provide basic information on the details of each flight.
1. January 14th, 2003: ~6.0-hour snow showers mapping mission over Sea of Japan (Fig. 1). PSR/A scanhead was not operating due to a problem with the 89 GHz receiver.
2. January 15th, 2003: ~4-hour snow showers mapping mission over the mountains of Honshu, just north of Tokyo and along the Sea of Japan coast (Fig. 2). PSR/A scanhead operated well but the tempscan data were not recorded. Data processing for this flight will require more attention and thus we are not providing them in this data delivery.
3. January 19, 2003: ~7.0-hour rainstorm mapping mission. Rainstorm started south of Honshu and moved northeastward to Western Pacific while deepening explosively. Seven observation lines at various altitudes and with PSR/A scanning in cross-track or conical mode were completed.
4. January 21, 2003: ~6.0-hour mission to observe frontal rain bands associated with east-moving low pressure area that formed over the Kanto plain during night with rain and snow reported in local area. Even reports of thunder-snow were recorded. The low moved rapidly over the Western Pacific Ocean generating copious rain.
5. January 23, 2003: ~4.0-hour mission to observe eastward moving cold front and frontal rain/snow bands. A low had passed out of the Yellow sea, across Kyushu and along the southern edge of Honshu with a cold front trailing out of the low. The low tracked to the NE and deepened explosively.
6. January 26, 2003: ~1.5-hour mission to observe land background in clear air and with conical scanning mode during the AMSR-E overpass.
7. January 27, 2003: ~5.0-hour Wakasa Bay rain data flight to observe light rain and snow over Sea of Japan during the AMSR-E overpass.
8. January 28, 2003: ~5.5-hour over roughly the same area as on 27 January 2003 to observe snow over water that was a result of strong onshore flow with scattered snow showers all along the central Sea of Japan coast of Honshu.
9. January 29, 2003: ~4.5-hour over roughly the same area as on 27 and 28 January 2003 to observe snow over water. Strong northwesterly flow over all of the Sea of Japan produced extensive areas of heavier snow off the coast of Honshu.
5
10. January 30, 2003: ~5.0-hour over roughly the same area as on 27, 28, and 29 January 2003 to observe snow over water. However, snow was observed in the Wakasa Bay but did not extend too far over the Sea of Japan.
11. February 1, 2003: ~1.0-hour Wakasa Bay data flight to observe snow covered land background in clear air and using conical scanning mode.
12. February 3, 2003: ~4.0-hour flight over the SE Pacific Ocean to observe rain over the water that was associated with a low southwest of Kanto region.
Table 2. Summary of WBAY03 Flights
Exp. Name
Region Observed Date Obs. Times
(UTC) # of Flight
Lines
%Time PSR
Operated
Flight Altitude
(km MSL) Sea of Japan 1/14/03 02:51-09:18 1, 1 0 7, 0.15 Honshu 1/15/03 03:03-06:49 4 80 7
2, 0.15 Western Pacific 1/21/03 02:05-08:26 2, 1, 1 95 7, 3, 0.15 Western Pacific 1/23/03 06:23-10:54 2, 2 100 7, 3 Honshu 1/26/03 03:27-04:48 4 100 7 Sea of Japan 1/27/03 02:08-07:34 5, 1 100 7, 5 Sea of Japan 1/28/03 02:25-08:07 5, 1 100 7, 5 Sea of Japan 1/29/03 02:14-06:37 5, 1 100 7, 5 Sea of Japan 1/30/03 02:06-07:19 5, 1 100 7, 5 Honshu 2/01/03 03:14-04:31 2 100 7 Western Pacific 2/03/03 02:00-06:24 3, 1 100 7, 0.15
WBAY03
TOTAL 62:17 60
Figure 1. NASA P-3 Flight tracks on 14 January 2003 over Sea of Japan. Grey –the entire flight path; Red –data collected with PSR/A flown in a straight and level (SL) line; Green –data collected with PSR/A flown in a constant angle turns (CAT); Black lines – projection of flight lines onto the surface.
6
Figure 2. NASA P-3 flight tracks on 15 January 2003 over Honshu mountains and Sea of Japan coast. Line colors as in Fig. 1.
Figure 3. NASA P-3 Flight tracks on 19 January, 2003 over Western Pacific. Line colors are the same as in Fig. 1; Bold black numbers are serial numbers associated with each line based on the navigational system data; Purple letters and numbers refer to the line numbers designated in flight, where the letters C and X refer to conical and X-track mode.
7
Figure 4. NASA P-3 Flight tracks on 21 January, 2003 over Western Pacific. Line colors are the same as in Fig. 1; Numbers are the same as in Fig. 3.
Figure 5. NASA P-3 Flight tracks on 23 January, 2003 over Western Pacific. Line colors are the same as in Fig. 1; Numbers are the same as in Fig. 3.
Figure 6. NASA P-3 Flight tracks on 26 January, 2003 over Honshu. Line colors are the same as in Fig. 1; Numbers are the same as in Fig. 3.
8
Figure 7. NASA P-3 Flight tracks on 27 January, 2003 over Sea of Japan. Line colors are the same as in Fig. 1; Numbers are the same as in Fig. 3.
Figure 8. NASA P-3 Flight tracks on 28 January, 2003 over Sea of Japan. Line colors are the same as in Fig. 1; Numbers are the same as in Fig. 3.
9
Figure 9. NASA P-3 Flight tracks on 29 January, 2003 over Sea of Japan. Line colors are the same as in Fig. 1; Numbers are the same as in Fig. 3.
Figure 10. NASA P-3 Flight tracks on 30 January, 2003 over Sea of Japan. Line colors are the same as in Fig. 1; Numbers are the same as in Fig. 3.
10
Figure 11. NASA P-3 Flight tracks on 1 February, 2003 over Honshu. Line colors are the same as in Fig. 1; Numbers are the same as in Fig. 3.
Figure12. NASA P-3 Flight tracks on 3 February, 2003 over Western Pacific. Line colors are the same as in Fig. 1; Numbers are the same as in Fig. 3. 4. Data Status. We have recorded an extensive and valuable data set during the WBAY03 experiment. In addition, we analyzed 9 out of 10 data flights during which the PSR/A scanhead operated normally. We found data to be excellent at the instrument level and very good at the first processing level, i.e., data we are providing with this document. However, there are some remaining
11
problems that need to be resolved through algorithm, software improvements, and further data study. Resolution of these problems requires data reprocessing with new and improved processing software and that entails efforts beyond the scope of this project. Thus, the data delivered at this time still suffer from some known problem as addressed in Fig. 13. Here we show two flight lines from two
Figure 13. Differences in vertical and horizontal polarization TB at nadir for all channels. a) 28 January 2003 line 2, clear above and scattered clouds below. b) 3 February 2003 line 3, largely clear below. During both lines the PSR/A scanhead operated in cross-track mode. separate flights, the January 28th (Fig. 13a) and the February 3rd (Fig. 13b) February, 2004 flights. The first flight was conducted over the Sea of Japan in the light snow conditions and the second one was conducted over SE Pacific Ocean under almost clear conditions below the aircraft. When observing at nadir, the horizontal and vertical polarization brightness temperatures observed at the same frequency should be the same, thus their difference should be zero. In Fig. 13 we plotted those differences for each channel and for the elevation angles less than 0.5º, the minimal elevation angles for the maneuvers. The TB difference is approaching zero for most of the channels, with the exception on 21.5 and 89 GHz radiometers. In order to resolve the problem a further study of data, instrument performance, and calibration algorithm is required.
a. b.
12
5. Flight Catalogs. Table 4 shows the maneuver ID example. It consists of several fields of information separated by dots. For example, for the maneuver ID WB03.DF013.M1313.SLM.T we have the following interpretation:
Table 4. Maneuver ID interpretation WB03 WBAY03 DF013 13th data flight of the PSR during WBAY03 M1313 maneuver serial number SLM maneuver type: straight and level, at middle-altitude
(SLL - straight and level at low altitude SLH - straight and level at high altitude)
T maneuver type (transect)
Tables 5a-14b provide a description of maneuvers listed by PSR serial numbers. For each flight there are four tables. The first of the two tables shows maneuvers ID and maneuvers description and the second one shows times and locations of start and end of each maneuver. Maneuvers are determined by the PSR processing system and each one is assigned an ID, i.e., a serial number.
6. Data Structure and Format. Level2.3a data are provided to the WBAY03 scientists for use in inter-comparisons with surface-based and satellite-based data collected during the January/February 2003 field campaign. Level2.3a stands for PSR data processed to level 2.3 and a letter “a” stands for subset of PSR data provided to outside users.
Tables 15 and 16 describe the meaning of the various data planes within any PSR/A WBAY03 level 2.3a data file. Indices in a 3-D PSR data matrix indicate scan, sample (within a scan), and data type (explained below). All brightness temperatures are calibrated and in units of Kelvin. The data are geolocated with respect to the sea level surface.
The data is provided in two formats: 1) MATLAB and 2) Binary with an ASCII header file. MATLAB files (e.g., L23assss.mat, where ssss is the maneuver serial number) contain data organized in the following manner:
Table 15. PSR/A Data type (i.e. order of variables in the third dimension). Data Plane Description
sceneL23a(:,:,1) 10.7v GHz vert. polarization brightness temperature (TB) sceneL23a(:,:,2) 10.7h GHz horiz. polarization brightness temperature (TB) sceneL23a(:,:,3) 18.7v GHz vert. polarization brightness temperature (TB) sceneL23a(:,:,4) 18.7h GHz horiz. polarization brightness temperature (TB) sceneL23a(:,:,5) 21.5v GHz vert. polarization brightness temperature (TB) sceneL23a(:,:,6) 21.5h GHz horiz. polarization brightness temperature (TB) sceneL23a(:,:,7) 37.0v GHz vert. polarization brightness temperature (TB) sceneL23a(:,:,8) 37.0h GHz horiz. polarization brightness temperature (TB) sceneL23a(:,:,9) 89.0v GHz vert. polarization brightness temperature (TB)
21
sceneL23a(:,:,10) 89.0h GHz horiz. polarization brightness temperature (TB) sceneL23a(:,:,11) 10.0 um Infra red brightness temperature(TB) sceneL23a(:,:,12) Scanhead encoder position azimuth sceneL23a(:,:,13) Scanhead encoder position elevation sceneL23a(:,:,14) High data rate atitude data pitch sceneL23a(:,:,15) High data rate atitude data roll sceneL23a(:,:,16) Latitude sceneL23a(:,:,17) Longitude sceneL23a(:,:,18) Heading sceneL23a(:,:,19) Altitude (ft) sceneL23a(:,:,20) Ambient Temperature sceneL23a(:,:,21) Ground Speed sceneL23a(:,:,22) Hardware trigger value sceneL23a(:,:,23) Time stamp (sec from the beginning of the day) sceneL23a(:,:,24) True azimuth angle sceneL23a(:,:,25) True elevation angle sceneL23a(:,:,26) Polatization angle sceneL23a(:,:,27) Pixel terrain geolocated latitude (dd.ddd) sceneL23a(:,:,28) Pixel terrain geolocated latitude (dd.ddd)
Table 16. Additional variables present in each of the PSR/A WBAY03 level 2.3a file ssn Serial number of maneuver (see CLPX02 data flight catalog) firstday Julian day on which the flight began scanhead Type of PSR scanhead (e.g., PSRA for CLPX02) description Text string indicating PSR maneuver ID
Binary files L23assss.bin contain data written out as a long line of sequential points of binary data in the order described in Tables 15 and 16. Each binary file has an ASCII header file L23assss.txt associated with it.
An example of the ASCII header files follows:
PSR/A Julian day at the beginning of the flight: 28 PSR scanhead type: PSRA Maneuver serial number: 1308 Size of sceneL23a data matrix is (numscans,numsamples,numchannels): sceneL23a(486,125,28) Binary data file: C:\PSR\WBAY03\2003_0128\level2.3a\SL\L23a1308.bin is written as real*8 (double) i.e. floating point; 64 bits
7. Data Examples. Figures 14-23 represent image examples of the PSR/A data collected and delivered to the NASA PIs. Here, we chose to show only the 37 GHz horizontal polarization channel images for all the lines covered during the field campaigns described in Section 3. However, the MATLAB software to create images (Section 8) from the provided data and some additional jpeg images are included in the data delivery.
22
WBAY03 2003_0119
23
Figure 14. PSR/A lines for 19 January 2003, over Western Pacific. Only 37.0 GHz TBs shown
WBAY03 2003_0121
24
Figure 15. PSR/A 37.0 GHz channel data obtained during the 21 January 2003 flight over Western Pacific.
WBAY03 2003_0123
25
Figure 16. PSR/A 37.0 GHz channel data obtained during the 23 January 2003 flight over Western Pacific
WBAY03 2003_0126
Figure 17. PSR/A 37.0 GHz channel data obtained during the 26 January 2003 flight over Honshu.
26
WBAY03 2003_0127
Figure 18. PSR/A 37.0 GHz channel data obtained during the 27 January 2003 flight over Sea of Japan.
27
WBAY03 2003_0128
28
Figure 19. PSR/A 37.0 GHz channel data obtained during the 28 January 2003 flight over Sea of Japan.
29
WBAY03 2003_0129
30
31
32
Figure 20. PSR/A 37.0 GHz channel data obtained during the 29 January 2003 flight over Sea of Japan.
WBAY03 2003_0130
33
34
Figure 21. PSR/A 37.0 GHz channel data obtained during the 30 January 2003 flight over Sea of Japan.
35
WBAY03 2003_0201
36
Figure 22. PSR/A 37.0 GHz channel data obtained during the 1 February 2003 flight over Honshu mountain.
WBAY03 2003_0203
37
Figure 23. PSR/A 37.0 GHz channel data obtained during the 3 February 2003 flight over Western Pacific.
38
8. MATLAB Software and Directory Structure for Displaying PSR/A Data. To enable the PSR/A data users to quickly view provided data, we included with the data a folder “display_l23a” that contains MATLAB version 6.1 m-files necessary to render brightness temperature maps from the PSR/A, level 2.3a data. We also included a MATLAB file “ReadBinL23aFile.m” which is an example of how to read binary and text files assuming that this will aid the user in designing their own data-reading routine using their software of choice. The calibrated 2.3a data are assumed to be organized in a series of subdirectories with the top one being the experiment directory. Thus, we have:
experiment_directory\yyyy_mmdd\level2.3a\sl\ where: “experiment_directory” is the root name of the main directory (e.g. WBAY03), “yyyy_mmdd” is a subdirectory indicating the year, month, and day of data, “level2.3a” is a subdirectory referring to the data level, “sl” is a subdirectly referring to straight and level maneuvers. “L23axxxx.mat” is a data file of type “.mat” i.e. MATLAB, corresponding to a maneuver with
serial number xxxx. “L23axxxx.bin” is a binary data file corresponding to a maneuver with serial number xxxx. It
contains data identical to the data contained in “.mat” file and is associated with header file “L23axxxx.txt”
Information pertaining to each individual maneuver can be found in Section 5 of this document.
MATLAB data and m-files are organized as described above. The user should create a PSR directory on their computer and copy the content of ETL Web files into this directory. The following two steps should be done next:
1) the m-file named “setrootdir.m” needs to be edited, with the variable “rootdir” changed to indicate the path to the directory “experiment_directory” , and
2) the MATLAB path needs to be modified (appended) using the “set path” command to include the subdirectory of mfiles contained in “display_23a”.
To run the display m-files a log file “WBAY03L23a.log” should be located in the “experiment_directory”. Issuing the command “mapl23a” in the Matlab command window will start the display code. The program will first ask which scanhead do you want and then it will show the available dates for display. The program then lists all available maneuvers from the selected flight code by their serial numbers and queries the user to select which file(s) is(are) to be displayed. The user can select the maneuver(s) by either the listed ordinal number or the associated WBAY03 maneuver serial number, for example, “sxxxx”. A group of maneuvers can be selected by indicating the range of ordinal numbers, for example, “2:13”. If more than one maneuver is selected, the program will ask if user wants to spatially interpolate between adjacent maneuvers (e.g., using kridging) or overlay them on top of each other. If interpolation is selected, the program will automatically interpolate between the end of one maneuver and beginning of the next. When the default option (overlay) is selected, no interpolation is performed and the data are overlaid, possibly overwriting data from previous maneuvers.
After loading data for selected maneuvers, the program queries the user for the channel (or set of channels) to be displayed. Several channel grouping options are provided in the command line. The next variables that can be selected are the minimum and maximum brightness temperatures for the range of the color map. If the minimum color temperature is defined by user, he/she will also be
39
asked for the maximum, otherwise the program will automatically assign those values. If auto-range calculation is selected the program will attempt to fit a Gaussian probability distribution function to the brightness temperature histogram, and compute the color range individually for each channel using the Gaussian parameters along with a range factor. The range factor sets the color range relative to the mean by the indicated number of standard deviations. The range factor defaults to 0.6, but can be modified according to the needs of the user. Autocorrelation is useful for scenes wherein the brightness temperatures mostly fall within a narrow range of values.
Proceeding, the user is able to display only a portion of the maneuver by selecting the scans to be displayed from all the scans available in the selected set of maneuvers. Here, for conical scanning, one full scan means one full rotation around azimuth axis, and includes front and back looks. Next, the user can choose to produce either individual maps (i.e., one image for each channel) or composite image of all channels and looks in a single map. Finally, the user has the option of selecting new latitude and longitude corners. If the user has installed the MATLAB Mapping Toolbox, he/she will also be asked to choose whether to include lines of individual U.S. states on the final map. Finally, we also provide a function “save_figures.m” that will save displayed figures in jpeg mode.