MODIS Atmosphere
MODIS Atmosphere Home Products  Images  Validation  News  Staff  Forum  Reference  Tools  Help
Aerosol Product  Water Vapor Product  Cloud Product  Atmosphere Profile Product  Cloud Mask Product   Joint Atmosphere ProductLevel-2 Products
Daily Global Product  Eight-Day Global Product  Monthly Global ProductLevel-3 Products       Filled Land Surface Albedo ProductFilled Normalized Difference Vegetative Index ProductOne-Minute Land Ecosystem Classification ProductLevel-3 Ancillary Land Surface Products
MOD07_L2 (Atmosphere Profile Product)
Introduction
Format & Content
Grids & Mapping
DOI
Sample Images
QA Information
Product Modification History
Acquiring Data
Filename Convention
Analysis Tools
Theoretical Basis
Validation
Detailed File Spec
Algorithm Software
Production Plan
Support Team

HISTORY

General PGE Information

The MOD07_L2 and MYD07_L2 products are produced from the Program Executable (PGE) identified as PGE03. To determine the current PGE03 version that has completed GDAAC operations implementation, visit the GDAAC Promotion History page. For information on reconciling Algorithm process version and PGE version, visit the PGE03 Version History File.


Tool/Command for PGE Determination

One may identify the maturity and quality of an HDF data product by finding the PGE (program executable) version that was used to produce the file. This information can be obtained directly from the HDF file itself by using the HDF utility ncdump. The (UNIX) command for finding the PGE version that was used to create a particular HDF file is

ncdump -h *.hdf | awk '/PGEVERSIONCLASS/,/END_GROUP/'

One may also find PGE version information in the ascii met (metadata) file associated with a particular HDF file.


PGE03 Algorithm History - Product Impact Changes

PGE03 Version Detail
Product impact changes only
High Impact = High Impact     Moderate Impact = Moderate Impact     Low Impact = Low Impact
(Dates show approximate starting production date)

PGE03 v4.4.0 includes:       ( Aqua Forward & Reprocess: ~1.Jan.2004 )

  • Atmosphere Profile (v4.1.1)
      Low Impact ECS Metadata Change. This version of Atmosphere Profile is promoted to Collection 4 (Aqua). The science algorithm is identical to v4.1.0, but the ECS metadata interface is changed slightly to make it compatible with Collection 004 (Aqua).
  • Cloud Mask (v4.3.0)
      High Impact Daytime Clouds Over Ocean Algorithm Enhancement.
      • Lowered band 2 cloud test thresholds in order to detect very small element cumulus and thin cirrus clouds, which are sometimes characterized by very low reflectance values.
      • Lowered 3.9 - 11 Ám test threshold in order to greatly reduce oval-shaped false cloud signals within sunglint regions.

      High Impact Non-Cloud Obstruction Algorithm Improvement. A completely new algorithm that now uses band 2 and 3 reflectances along with band 1 variability test. The previous version reported too much false smoke.
      Moderate Impact Updated Spectral Response Files. Updated spectral response files and temperature ranges in planck and brightness functions.
  • Run Time Scripts
      Low Impact Build and Runtime Changes.
      • Modified code such that if an ancillary file (such as NISE snow extent) exists but cannot be opened or read, execution is immediately terminated. If a file is not found, however, the program will continue. This is done so that the production rules may be over- ridden by operations, if desired. The production rules were modified so that any files actually necessary for execution will be required to be present.
      • ECS Core attribute VersionId (the Collection identifier) was changed from 3 to 4 in all Aqua metadata configuration files (MCFs).
      • Attribute ProcessingEnvironment was added to the Archive metadata record. It identifies the computer host and operating system that produced the product. MODAPS (MODIS Adaptive Processing System) currently operates on IRIX and Linux hosts.
      • The SDPTK (Toolkit) configuration was upgraded from 'SDPTK5.2.8v3_64_f77' to 'SDPTK5.2.9v1_64_f77'.
      • Ancillary data products SEA_ICE and NISE are classified as required inputs in this PGE03 version. In each case, the production rule performs an initial search over a 6-day centered interval. If unsuccessful, the search is expanded to 14 days. If still unsuccessful, a 'DeletePGE' request is issued.


Start of Aqua Collection 004


PGE03 v4.3.0 includes:       ( Terra Forward: ~3.October.2003       Aqua Forward: ~3.October.2003 )

  • Atmosphere Profile (v4.1.0)
      High Impact Changes to the Regresion Coefficients & Training Data.
      • Updated the forward model codes and coefficients used in computing radiances from the training profiles with the latest version using the HITRAN2000 spectral database.
      • Modified the TPW integration routine to handle the levels closest to the surface better.
      • Removed band 24 from the regression equation. Due to the improved emissivity data used in training the regression, we no longer need to use the difference 25-24, and removing this differencing reduces noise due to striping. Band 24 also introduced uncertainty because of a solar reflective component to the signal that was particularly obvious in calculating radiance observed minus calculated bias for this band.
      • Improved emissivity assigned to profiles in training data. Using emissivity data from UCSB's MODIS emissivity library, representative emissivities for ocean, land, desert, and snow/ice were developed. These emissivities were applied to the training data, +/- a random number of standard deviation from 0.01 to 0.03 depending on variability within the ecosystem data base. Emissivities were limited to 0.99. Previously, only two different emissivities were assigned to the profiles: 0.95 for 13-9 Ám, 0.84 for 4.75 to 3 Ám , and linear in between.
      • New frequencies were used to better represent the emissivity spectra. Old wavelengths were 13, 11, 9, 4.75, 4, 3.5, and 3 Ám . The range of wavelengths for the MODIS bands used in this algorithm is 14.3-4.4 Ám , so some of these frequencies were not relevant to our data. New wavelengths are 14.3, 10.8, 9.3, 8.3, 7.6, 5.0, and 4.3 Ám (700, 926, 1075, 1205, 1316, 2000, 2326 wavenumbers). These were chosen at inflection points of significant changes in emissivity spectra.
      High Impact Changes to the Atmosphere Profile Science Code.
      • mod07_compute_products.f -
        Skin temperature retrieved by regression is now being saved into the 'Surface_Temperature' variable. Previously the ancillary input GDAS surface temperature had been saved to this variable.
        Bias corrections are now read in from an ascii file, instead of hardwired into the code. An 'init_bias' parameter is set so we only read the bias files once. Depending on the flag set in the bias input file, the radiance bias corrections can be a linear function of GDAS TPW (flag = 2) or MODIS brightness temperature (flag = 1). If any other flag is given, no biases are applied to the radiances. This replaces the earlier latitude-dependent bias corrections. Bias slope and y-intercept are read in for each band and applied to the GDAS PWAT or brightness temperature. An optional min and max bias can also be input as 3rd and 4th columns in the bias input file.
        To reduce noise due to detector striping, arrays of detector flags (10 for each band) are read in from an ascii file. If a detector for a specific band is flagged with a 0, radiances from this detector are skipped in the retrievals.
        Parameter averaging: Now, only those zenith angle, latitude, surface pressure, elevation, and GDAS TPW values corresponding to pixels with box_flag == 1 (clear sky) are averaged. Previously, the center value in a 5x5 retrieval area was used, regardless of which pixels were actually used in the retrieval. This is important when some pixels in a 5x5 are not retrieved.
        Added a provision to only skip the first 10 pixels of each radiance scan line if the data was from Terra (included platform_name.inc to access platform_name variable). These pixels are skipped on Terra due to errors introduced by a cross-talk correction. Problem does not exist on Aqua.
      • modis_ges101.f -
        A change was made in the way surface level is handled when integrating moisture profiles to obtain TPW. Included variable for retrieved surface skin temperature. New coefficient version number in check of file header. Band 24 was removed from the regression, as well as the predictor BT25-24.
      • Other files -
        New subroutine, get_coeff.f to read bias values and detector flags. Subroutine get_sfcpres_pwat.f replaces old get_sfcpres.f. New adds GDAS TPW. Subroutine tprecw_new2.f replaces tprecw.f for new surface handling in TPW integration. Note that layer PW is still integrated using precwv.f. Small changes made in the following files to accommodate the new pwat1 variable (GDAS TPW):
        mod07_check_box.f - QC check of pwat (0-110 kg/m^2)
        mod07_get_data.f - added pwat1, and new call to get_sfcpres_pwat instead of get_sfcpres
        mod07_data.inc - added pwat1 (GDAS precipitable water for TPW dependent biases) to mod07 common block
        PCF file numbers added to mod07_pcfnum.inc for bias and detector files, each for Aqua and Terra. New numbers assigned are 420015-420018. Lines were added to MOD_PR07.pcf and MYD_PR07.pcf defining bias and detector flag files and their PCF numbers. Copied version of MOD_PR07.f from current ops. Hucek made changes 12/02. Copied version of mod07_meta.inc from current ops. Hucek changed 12/02.
      Moderate Impact New ancillary files for bias and detector flags.
      • Ascii files containing new radiance bias values and detector flags were added. PCF file numbers 420015-420018 were assigned to these files and added to MOD_PR07.pcf.template. Filenames are: terra_bias.dat, aqua_bias.dat terra_det.dat, aqua_det.dat
        a) format of bias file
        LINE 1: String containing platform name, version number, and date Current version number for both Aqua and Terra is 2.0. LINE 2: String containing platform name: Terra or Aqua (caps important). LINE 3: Number of bands, number of columns, flag to indicate bias is a function of MODIS BT of that band (1) or NCEP-GDAS TPW (2). if any other flag is given, no biases are applied to the radiances.
        Biases are OBSERVED minus CALCULATED in degrees K.
        Number of columns does not including band number itself If 2 columns after band number, these columns are: slope, y-intercept If 4 columns after band number, these columns are: slope, y-intercept, min, max Bias determined by: bias(band X) = slope * TPW + y-intercept [or BT of band X instead of TPW] then bias values > max are set equal to max and similar for min
        b) format of detector file
        LINES 1-2 same as for bias file above LINE 3: number of bands, number of detectors, flag not currently used LINES 4-end: band number, detector flag for each detector detector flag = 1 for good, or = 0 to skip
      Low Impact Changes to code build and runtime environment.
      • The MOD_PR07 build script (MOD_PR07.mk) was modified to be able to run under SDPTK 5.2.7v3 and above, and it is backward compatible to earlier versions. A reference to the HDF5 dynamic shared runtime libraries is required for successful compilation and linking with the later versions.
      • In the runtime environment, the path to dynamically linked shared libraries (specified in environment variables LD_LIBRARY_PATH, LD_LIBRARYN32_PATH, and LD_LIBRARY64_PATH) must include the HDF5 runtime library for the later SDPTK versions. Atmosphere codes use LD_LIBRARYN32_PATH or LD_LIBRARY64_PATH on IRIX, and LD_LIBRARY_PATH on Linux.

    PGE03 v4.2.1 includes:       ( Aqua Forward Only!: ~10.May.2003 )

    • Cloud Mask (v4.2.1)
        Moderate Impact Snow Detection. Changes made to the cloud mask snow detection algorithm that increase the number of clear, snow/ice pixels found.
        • Modified Band 7 Reflectance Threshold. Changed band 7 (Aqua algorithm only) reflectance threshold.
        • Added 3.7-11 Ám Test. Added 3.7-11 Ám test to the 8.5-11 Ám cloud test in the snow detection algorithm (snow_mask.f).
    • Cloud Mask (v4.2.0)
        Moderate Impact Snow Detection. Additional tests for screening out cloudy pixels incorrectly identified as snow.
        • Cloud/Snow Differentiation. Added switch for Aqua vs. Terra processing. Use band 6/7 for calculation of NDSI for Terra/Aqua and adjusted thresholds accordingly. Added reflectance threshold test for band 6/7 for Terra/Aqua.
        Low Impact Additional Changes.
        • Greenland Elevation Adjustment. Defined small portion of Greenland to be "high elevation" even though it is below 2000 meters. This is to eliminate false clouds generated by the 1.38 Ám test.
        • Day/Land/Clear-Sky Test. Adjusted daytime, land, clear sky restoral test thresholds.
        • Logic Fix for Polar Daytime Processing. Fixed logic error in polar, daytime processing - very small impact on results.
        • Updated File Spec Description Attribute . Updated the Cloud_Mask description attribute in the MOD35_L2 CDL file (MOD35_L2.CDL.fs). This is reflected in the science S/W by an update to the src_UW shared code library module get_spec_info.f.
    • RunTime Scripts
        Low Impact Threshold Version Update. Cloud Mask thresholds were updated to version 5.

    PGE03 v4.1.2 includes:       ( Terra Forward: ~16.Jan.2003       Aqua Forward: ~16.Jan.2003 )

    • Atmosphere Profile (v4.0.4)
        Low Impact Ancillary Input Removal. NCEP Reynolds sea surface temperature (REYNSST) and TOVS ozone (OZ_DAILY) products were removed as ancillary inputs.
    • Cloud Mask (v4.1.2)
        Low Impact Ancillary Input Removal. NCEP Reynolds sea surface temperature (REYNSST) and TOVS ozone (OZ_DAILY) products were removed as ancillary inputs.

    PGE03 v4.1.1 includes:       ( Terra Forward: Changes promoted in v4.1.2       Aqua Forward: Changes promoted in v4.1.2 )

    • Atmosphere Profile (v4.0.3)
        Low Impact Ancillary (NCEP) Input Changes. The Reynolds sea surface temperature file (REYNSST) can now be accessed in either the original ascii format or the new binary format. (NCEP plans to cease production of the ascii format in January 2003.) Both big-endian (IRIX) or little-endian (Linux) host processing machines are supported.
    • Cloud Mask (v4.1.1)
        Low Impact Ancillary (NCEP) Input Changes. The Reynolds sea surface temperature file (REYNSST) can now be accessed in either the original ascii format or the new binary format. (NCEP plans to cease production of the ascii format in January 2003.) Both big-endian (IRIX) or little-endian (Linux) host processing machines are supported.

    PGE03 v4.1.0 includes:       ( Terra Forward: ~04.Nov.2002       Aqua Forward: ~10.Oct.2002 )

    • Cloud Mask (v4.1.0)
        High Impact Low Cloud Determination Improvement. The nighttime land 3.9-11 Ám brightness temperature difference (BTD) low cloud test threshold is now a function of the 11-12 Ám BTD. This allows better discrimination between low clouds and clear sky especially through humid atmosphere.
        High Impact Mid-Level Cloud Detection Improvement. A new 7.3-11 Ám BTD test for mid-level clouds has been included for nighttime land scenes. This was done because clouds with temperatures in the 240-260K range are sometimes very difficult to detect without solar illumination.
        High Impact Low-Cloud/Snow Discrimination Improvement. The polar night 3.9-11 Ám cloud test for snow/ice surfaces will also be used for other polar night land surfaces not explicitly listed as snow-covered by ancillary data. This results in the identification of more low clouds.
        High Impact Clear Sky Determination Improvement. A nighttime land 11 Ám clear-sky restoral test was introduced. It is similar to the daytime 11 Ám clear-sky restoral test but uses lower thresholds. This aids in identifying clear-sky pixels.
        Low Impact Additional Bands in Radiance/Reflectance Files. Clear-sky radiance/reflectance files (MODCSR_G files) now contain more bands. All bands are recorded with the exception of ocean color. These granule-based files are small and the change will have little impact on storage space or processing time.
        Low Impact NISE Ancillary Product Search Window Extension. The NISE ancillary product search window was extended from a 6 to a 14 day centered interval.

    PGE03 v4.0.1 includes:       ( Terra Forward: Changes promoted in v4.1.0       Aqua Forward: ~24.Jun.2002 )

    • Atmosphere Profile (v4.0.1)
        High Impact Total Totals Bug Fix. Fixed an algorithm bug, where (previously) the total totals atmospheric instability index was being overwritten with the direct water vapor retrieval result. Note that this bug fix was also simultaneously implemented in PGE03 v3.2.1, which, at the time, was being used for Terra Forward processing. This version of PGE03 (v4.0.1) was used in Aqua Forward processing only until 11/2002. PGE03 v4.0.0 was actually developed prior to v3.2.1.

    PGE03 v4.0.0 includes:       ( Terra Forward: Changes promoted in v4.1.0       Aqua Forward: Changes promoted in v4.0.1 )

    • Cloud Mask (v4.0.0)
        High Impact Reduction of False Cloud in Sunglint Regions. Band 2 ocean threshold in sun-glint areas is now dynamically calculated based on the value of the sun-glint angle. Reduces false cloudy designations and improves clear vs. low cloud discrimination.
        High Impact Reduction of False Clear over Heavy Vegetation. Coastal NDVI threshold for clear-sky restoral test changed from 0.3 to 0.4. Reduces false clear determinations in heavily vegetated river basins.
        High Impact Reduction of False Cloud in Arid Regions. Added new clear-sky restoral tests for non-snow covered land surfaces - 3.7-3.9 and 3.9-11 Ám brightness temperature difference and 1.2/0.55 ratio tests. Reduces false cloud and uncertain determinations in arid and semi-arid regions.
        High Impact Thin Cirrus Threshold. Set new 1.38 Ám thresholds for thin cirrus tests. There are separate thresholds for land, water, and snow/ice surfaces. These thresholds assume corrected band 26 data.
        High Impact Deep Polar Inversion Clear/Cloud Discrimination Improvement. A 13-11 Ám clear-sky restoral test was added for polar night conditions along with a new 11-4 Ám cloud test with dynamically calculated thresholds. This test replaces the original. Improves detection of low clouds (11-4 Ám) and clear skies in areas characterized by deep polar inversions (13-11 Ám).



    Start of Terra Collection 004 Start of Aqua Collection 003


    PGE03 v3.2.1 includes:       ( Terra Forward: ~01.Jun.2002 )

    • Atmosphere Profile (v3.1.1)
        High Impact Total Totals Bug Fix. Fixed an algorithm bug, where (previously) the total totals atmospheric instability index was being overwritten with the direct water vapor retrieval result.
        Low Impact Volcano Algorithm Dropped. A call to the volcano alert algorithm was dropped because near real-time data was not expected to be available.

    PGE03 v3.2.0 includes:       ( Terra Forward: ~01.Apr.2002 )

    • Atmosphere Profile (v3.1.0)
        High Impact High Total Precipitable Water Over Deserts - Eliminated. There are significant updates in this version of the atmosphere profile algorithm that greatly improve the quality of retrieved products, particularly total precipitable water vapor (TPW). In particular, the extremely large TPW values that were previously observed over deserts (e.g. North Africa) have been eliminated. Specific changes made include:
        • Include Global Radiance Bias Corrections. Biases were separated into three latitude zones: tropical (lat < 30), mid-latitude (lat 30-50), polar (lat > 50), separated by land and ocean. The biases were expressed as observed minus calculated brightness temperature and were based on global data from four days in June 2001 (153-156). The new 101 level model with updated constants was used to compute the predicted radiance. NCEP-GDAS profiles were used for temperature and moisture profiles and surface pressure. Above GDAS, climatological T and W was used. Below GDAS, T was extended by using GDAS SfcT at GDAS SfcP and mixing ratio was extrapolated. Climatological ozone (by month and latitude) was used. Skin temperature and emissivity was computed by regression from MODIS radiances. Desert and antarctica cases were excluded from the dataset because of known problems with the emissivity calculation (desert) and cloud mask (Antarctica). Because of the use of climatological ozone, biases for band 30 were set to zero. Also, biases from band 27 were set to zero. Calculated biases for this band had a very high standard deviation. Comparison with MODIS and the CART site microwave water radiometer and radiosonde showed improvement with band 27 bias = 0. Also polar biases were set to zero because significant variation was found between the night and day biases and the cloud mask may have trouble in these regions, particularly at night.
        • Skip Retrieval For First Ten Pixels In Each Scan Line. The first 10 radiance pixels in each scan line are not included in the retrieval. There is an error introduced in the radiances by the cross-talk correction that caused a fairly significant variation in the TPW values for the first two columns of 5x5 retrievals.
        • Cap Brightness Temperatures. Limited the maximum brightness temperature for any band used in the retrieval algorithm to 330K.

        Moderate Impact Regression Coefficient Improvement. Many changes in the regression coefficient files were made including:
        • Modify The Retrieval Predictors. Now uses 11 predictors. Instead of using the brightness temperature from individual bands 24,25,27-36, the new predictors include the brightness temperature *difference* between bands 25-24, plus the individual bands 27 through 36.
        • Use 7 Brightness Temperature Classes. The training data set has been partitioned into 7 groups by the 11 Ám (band 31) brightness temperature. When the retrieval is performed, it uses only that portion of the training data set corresponding to the same 11 Ám BT as the MODIS observation. Classes are <245, 245-269, 269-285, 285-294, 294-300, 300-310, and >310.
        • Improve Training Data Over Desert Regions. After partitioning the training data into brightness temperature classes, very few training data existed for the very warm surfaces (the last two BT classes) although plenty of desert areas observed by MODIS had 11 Ám brightness temperature > 300K. To remedy this, 900 new radiosondes from the northern African deserts from all 12 months of 2001 were added to the training data set. Skin temperature for the training data was determined by adding a random positive number between 0 and about +15C to the lowest radiosonde level temperature.
        • New NeDT and corrected reference BT. Old code used the instrument spec NeDT multiplied by 3 for the instrument noise. The new NeDT was chosen as the highest between (instrument spec)x3 or the new values provided by Chris Moeller. These high values were necessary to help reduce the noise in the retrievals due to striping effects.
        • Improve Reference Brightness Temperature. The old algorithm used a constant reference brightness temperature (250K) over all bands when computing dB/dTb_ref. This has been replaced by an array of the instrument spec reference temperatures.
        • Add Direct Total Water Vapor. Added a training variable for a new parameter: total water vapor retrieved directly. The original total water vapor is integrated from retrieved profiles. For this additional parameter, the training profiles are integrated and the logarithm of the TPW is stored. A check for saturation is made of all the profiles before integrating. This new parameter replaces the microwave emissivity in the retrieval array.
        • Improve Forward Model. Updated the forward model used in computing radiances from the training profiles with the most recent version. This new version has the capability of running for individual detectors or for the average (average was used in creating the coefficient file). It also can be used for either Aqua or Terra and includes updated values to certain constants. Effective central wavenumbers for Terra were updated to be consistent with the operational retrieval code and new central wavenumbers were included for Aqua.

        Low Impact Aqua Modifications. Logic was added to stage unique sets of coefficient files for Aqua and Terra processing. Aqua regression coefficients for the atmosphere profile algorithm were designated v1; Terra coefficients were updated to v3.
    • Cloud Mask (v3.1.1)
        Moderate Impact Bug Fixes. A number of minor bug fixes were implemented, including:
        • Nighttime Ocean - Granule Edge Fix. A bug fix was implemented that adversely affected nighttime ocean edge-of-granule results.
        • Band 26 Reflectance Check. An additional check on value of band 26 reflectances was added.
        • 250m Sun-Glint Dark Spots. Minor change in 250m processing was implemented. Now the retrieval algorithm uses only the darkest pixels within the sun-glint regions.
        • Sun-Glint NDSI Test For High-Sun Polar Regions. Modified sun-glint check in NDSI test for polar regions during times of highest sun.
        • NISE Time Period Check. Modified time period check for NISE data sets. NISE data must be within one week of granule time or data will not be used.

        Low Impact Aqua Modifications. The final modifications were made in order to process both Terra and Aqua data. The changes include: 1). Planck function and radiance-to-brightness temperature codes now include response functions for Aqua data and 2). Unique threshold files for Aqua and Terra. Aqua thresholds for the cloud mask algorithm were designated v1, while the Terra thresholds were updated to v7.
    • RunTime Scripts
        Low Impact Production Rules. Production rules for running PGE03 were changed for the following ancillary products: GDAS_0ZF, REYNSST, SEA_ICE and NISE.
        • If the current GDAS_0ZF is unavailable, wait until killed or forced to run without GDAS_0ZF by operator.
        • If current REYNSST is unavailable, use prior week. If prior week is also unavailable, use the next week (only possible in reprocessing). If next week is unavailable, extend search to a centered 42 day interval for nearest produduct. If still unavailable, run without REYNSST.
        • Search a centered 6-day interval for nearest SEA_ICE product. If unavailable, wait until killed or forced to run by operator. In the latter case, first extend the search to a centered 28-day interval.
        • Search a centered 6-day interval for nearest NISE product. If unavailable, wait until killed or forced to run without NISE by operator.

    PGE03 v3.1.0 includes:       ( Terra Forward: ~15.Dec.2001 )

    • Atmosphere Profile (v3.0.3)
        Low Impact Port To Linux Fix. The makefile logic was changed to test first for a linux environment (i.e. query the BRAND environment variable) and, if not linux, default to IRIX compilation. This delivery allows sucessful compilation at the Goddard DAAC (GDAAC) where BRAND is not defined.
    • Cloud Mask (v3.1.0)
        High Impact Shallow Lake/River Processing Path Update. The cloud mask algorithm over shallow lakes and rivers was modified to use the coast processing path. This provides more accurate cloud results over these regions.
        High Impact NDVI Cloud Mask Confirmation. An NDVI final confidence confirmation test is performed for all daytime pixels where the coast processing path is used and IR tests did not find evidence of high clouds. This test is also performed in shallow ocean cases.
        High Impact 1km Cloud Mask Result Used As Input to 250m Mask. Cloud mask results from 1km pixels are used as ancillary input to the 250m cloud mask. Separate decision trees are used for land and water. Over water, the decision tree dictates that the 250m mask copy the 1km result for sun-glint, ice, and pixels with ambiguous band 2 reflectances. Over land, the decision tree dictates that the 250m mask copy the 1km result if the pixel is clear, otherwise the Global Environment Monitoring Index (GEMI) and band 2 reflectance threshold are used (where appropriate).
        Moderate Impact More Pixels Processed When Band 2 Is Saturated. Band 2 reflectances are considered to be high valid values when error codes are 65528 (1km) or 65533 (250m) and band 1 reflectances are valid (no error code).
        Moderate Impact NDVI Final Confidence Test Result Added to Cloud Mask SDS. Bit 22 (previously a spare) of the cloud mask now reports the result of the NDVI final confidence confirmation test.
        Low Impact Port To Linux Fix. The makefile logic was changed to test first for a linux environment (i.e. query the BRAND environment variable) and, if not linux, default to IRIX compilation. This delivery allows successful compilation at the Goddard DAAC (GDAAC) where BRAND is not defined.

    PGE03 v3.0.1 includes:       ( Terra Forward: ~15.Oct.2001 )

    • Runtime Scripts
        Low Impact Ancillary Input Requirement Change. The ancillary inputs OZ_DAILY and NISE were changed from required to optional.

    PGE03 v3.0.0 includes:       ( Terra Forward: ~15.Apr.2001 )

    • Atmosphere Profile (v3.0.0)
        High Impact Water Vapor Retrieval Improvement. This version uses an updated set of regression coefficients from Li (SSEC) to improve the water vapor retrieval. Specifically, retrieved mixing ratios are now lower than before. This change was made to fix a wet bias seen in the MODIS moisture retrievals. A check for saturated water vapor mixing ratios was also added. With these changes, retrieved values can no longer exceed the saturation value.
        Moderate Impact Input Surface Pressure Improvement. Previous versions used pressure at mean sea level from GDAS1 as ancillary input to the retrieval. This version was updated to use the surface pressure from GDAS1 (which includes topography effects) instead. In addition, surface pressures are now bilinearly interpolated to MODIS resolution, rather than using the nearest 1 degree resolution value which tended to leave 1 degree artifacts in the retrieved values.
        Low Impact First MODIS Consistent Year Delivery. This delivery was used to run in the first MODIS Consistent Year processing effort. Reprocessed and forward processed data using frozen PGEs are planned for the data year November 1, 2000 to October 31, 2001. MODIS products including the Atmospheric Profiles will be stored in MODIS Data Collection 003.
    • Cloud Mask (v3.0.0)
        High Impact High Elevation Flag and Adjustment. Added elevation information to algorithm. A high elevation flag is set when elevation is >2000 meters. Turned off 1.38 Ám test at high elevations (>2000 meters).
        High Impact Bug Fix - Snow Detection. Fixed code bug in snow detection subroutine.
        High Impact Cloud Detection Over Snow - Threshold Adjustment. Adjusted thresholds for cloud detection over snow-covered surfaces. Created separate thresholds for cloud detection over Antarctica.
        High Impact Arid Clear Sky Confidence Refinement. Refined final clear-sky confidence confirmation test for arid regions including an elevation adjustment for the 11 Ám brightness temperature threshold.
        Moderate Impact Arid Ecosystem Expansion. Expanded list of arid ecosystems.
        Low Impact QA Bits - File Spec Documentation Update. Added documentation for QA bits 26 through 28 in the cloud mask CDL file spec (MOD35.V2.CDL).
        Low Impact First MODIS Consistent Year Delivery. This delivery was used to run in the first MODIS Consistent Year processing effort. Reprocessed and forward processed data using frozen PGEs are planned for the data year November 1, 2000 to October 31, 2001. MODIS products including the Cloud Mask will be stored in MODIS Data Collection 003.
    • Runtime Scripts
        Low Impact Ancillary Search Interval Expansion. The production rules for ancillary products OZ_DAILY and NISE were expanded to select the nearest granule within a 4-day search interval centered on the MODIS 5-minute Collection time.



    Start of Terra Collection 003


Next Page
DATA