Changes In Version 4 MODLAND PGE’s

List of affected PGE’s: 07, 08, 11, 12, 13, 14, 15, 16, 21, 22, 23, 24, 25, 25Q, 29, 30, 31, 33, 34,

35, 36, 37, 38, 40, 41, 42, 43, 44, 45, 46, 60, 65, 66, 67, 72, 80 and 82

Ground Control Point matching : PGE 60

PGE60: Ground Control Point matching code (MOD03_CP)

- Started using MOD09 (atmospherically corrected surface reflectance) product for island
matching instead of MOD02HKM.

- Added island matching statistics to report.

- Changed coefficients for land/ water discrimination.

Snow products : PGE’s 07, 43, 45, 46 and 67

PGE07: Level 2 Snow (MOD10_L2)

- Add a new SDS “Snow Cover Reduced Cloud” that uses a subset of the MOD35_L2 cloud
tests to maximize the snow extent mapping. This SDS is not used in downstream products.

- Adjust the snow temperature threshold to detect fractional snow cover.

- Delete the summary statistics local attributes from the “Snow Cover” SDS.

PGE43: Daily snow cover (MOD10A1):

- Use a new scoring algorithm to select pixels from the Level 2G product. The new algorithm
selects observations based on their acquisition time (nearest to the local noon), distance from
nadir and observation coverage.

- Add support for the SIN grid.

PGE45: 8-day snow (MOD10A2):

- Add support for the SIN grid.

PGE46: Daily snow CMG (MOD10C1):

- Improve mapping the terminator for a smooth day – night transition.

- Change the calculation of snow cover percent to base on CMG cell counts.

- Use an impossible snow occurrence mask to eliminate some false detections.

PGE67: 8-Day snow CMG (MOD10C2):

- Improve mapping the terminator for a smooth day – night transition.

- Change the calculation of snow cover percent to base on CMG cell counts.

- Use an impossible snow occurrence mask to eliminate some false detections.

Sea ice products: PGE’s 08 and 44

PGE08: Level 2 sea ice (MOD29):

- Minor metadata changes.

PGE44: Daily day and night sea ice (MOD29P1D and MOD29P1N):

Figure 1: The ice surface temperature parameter (MOD29P1D) for tile h07v08 from day 2002209
produced using the collection 3 scoring algorithm (left) and collection 4 algorithm (right).

Surface reflectance products: PGE’s 11 and 21

PGE11: Level 2 surface reflectance (MOD09)

- Implement an internal cloud screening algorithm to be instead of the official MODIS cloud mask
in the aerosol retrieval and atmospheric correction processes. The internal cloud mask results are
passed to downstream products for use when appropriate.

- Implement a geometrically based cloud shadow screening. Results are passed in the QA bits to
downstream products.

- Improve the aerosol retrieval over bright target and in the presence of clouds and snow.

- Improve the aerosol interpolation scheme.

- Set cirrus bits in the QA field based on a band 26 threshold approach.

- Remove band 26 destriping since this is done in L1B for collection 4.

- Do not use the noisy detector 0 in band 5 for aerosol retrieval to remove a striping artifact.

- Create an intermediate 0.05 deg CMG surface reflectance product.

PGE21: 8-day Land surface reflectance (MOD09A1)

- Use MOD09 internal cloud mask in the observation selection process.

- Change the compositing scheme to select clear observations with maximum observation coverage
that are closest to the nadir view.

- Add logic to handle inconsistencies in the input data. This will avoid ungraceful exit of the code if
bad inputs are used.

- Add support for the SIN grid.

Fire products: PGE’s 30, 29 and 80

PGE30: Level 2 Fire (MOD14)

- Improved contextual algorithm that offers superior sensitivity to smaller, cooler fires, and yields
fewer blatant false alarms. This change addresses the issues of false detection in some deserts and
sparsely vegetated areas (northern Ethiopia, Middle East and central India), and the frequent
non-detection of relatively small fires.

- Use the HDF SDS compression feature to reduce the size of the L2 fire product.

- This algorithm has been tested extensively and has been running in the MODIS Rapid Response
System for the past six months.

PGE29: Daily and 8-day Fire (MOD14A1 and MOD14A2)

- Add support for the SIN grid.

PGE80: Daily intermediate burned area product (MODHDFSR and MODQDFSR)

- Abandoned filtering data based on the land/water mask due to inaccuracies in the inland water mask.

- Modified the clear and good observations filter parameters.

- Added an 8-bit QA SDS.

- Add support for the SIN grid.

Land surface temperature products: PGE’s 16 and 31

PGE16: Level 2 and daily level 3 land surface temperature (MOD11_L2, MOD11A1 and MOD11B1)

- update the look-up tables (LUT) used in the day/night LST algorithm.

- process lake pixels in clear-sky when cloud mask reports a clear confidence of 66% and higher.

- Use the coarse resolution MODIS 16-day BRDF Albedo Parameters product (MOD43B1C) as input.

- Separate the range of viewing zenith angles into five sub-ranges (0-24, 24-38, 38-49, 49-58, and
58-65 degrees) instead of four.

- Implement parallel processing of data in odd and even days to double the production rate and the
storage of interim results for the day/night algorithm.

- Incorporate a split-window method into the day/night algorithm to ensure that the retrieved
emissivities can be used by split-window algorithms.

- Add support for the SIN grid.

PGE31: 8-day land surface temperature (MOD11A2)

- Add support for the SIN grid.

L2G pointers, geoangles and science products: PGE 12, 13, 14 and 15

PGE12: Daily pointers and geoangles (MODPTQKM, MODPTHKM, MODPT1KD, MODPT1KN,
MODMGGAD, MODMGGAN, MODPTPGD, MODPTPGN, MODMGPGD and MODMGPGN)

- Added capability to create additional temporary files when file size exceeds 1.75 GB to prevent
reaching the 2GB file size limit on Linux machines.

- Fixed glitches which occur when the scan barely intersects a tile.

- Fixed a problem caused by miscalculating the pixel size of the first sample of a scan.

- Added support for the SIN grid.

PGE13: Daily L2G surface reflectance and fire (MOD09GQK, MOD09GHK, MOD09GST,
MOD14GD, MOD14GN)

- Set collection 4 meta data.

- Add support for the SIN grid.

PGE14: Daily L2G snow (MOD10L2G)

- Set collection 4 meta data.

- Add support for the SIN grid.

PGE15: Daily L2G sea ice (MOD29PGD and MOD29PGN)

- Set collection 4 meta data.

Aggregated surface reflectance products: PGE 22

PGE22: Daily aggregated surface reflectance and texture (MODAGAGG and MODAGTEX)

- QA adjustments to accommodate upstream L2G changes

- Bug fix to accommodate problems with input data that forced algorithm to exit

- Change to output QA generation

- Add support for the SIN grid

BRDF/Albedo products: PGE’s 23, 24, 65 and 82

PGE23: 16-day BRDF/Albedo (MOD43B1, MOD43B2, MOD43B3 and MOD43B4)

- Update the static BRDF database.

- Update LUT’s.

- Use of the LiTransit BRDF model instead of the Walthall model.

- Add support for the SIN grid.

PGE24: 16-day Albedo CMG (MOD43C1)

- Change resolution from 0.25 deg to 0.05 deg.

- Add support for the SIN grid.

PGE65: 16-day BRDF/Albedo parameters CMG (MOD43C2)

- Change resolution from 0.25 deg to 0.05 deg.

- Add support for the SIN grid.

PGE82: 16-day NBAR CMG (MOD43C3)

- Change resolution from 0.25 deg to 0.05 deg.

- Add support for the SIN grid.

Vegetation Indices products: PGE’s 25, 25Q and 35

PGE25/25Q: 16-day 500m/250m NDVI and EVI (MOD13A1 and MOD13Q1)

- QA adjustments to accommodate upstream changes.

- Bug fix to prevent the code from exiting ungracefully when presented with abnormal inputs.

- Major change to the algorithm to adopt a weighted average scheme in processing observations
overlapping the grid cells. This change is expected to improve the spatial coherence/integrity of
the data. Figure 2 illustrates the effect of this change on the NDVI.

- Add support for the SIN grid.

Figure 2: samples illustrating improvement in the NDVI as a result of the weighted average scheme
(left = collection 3, right = collection 4). These samples are extracted from tile h20v11 from the
16-day period starting 2001129.

PGE35: 16-day 1km NDVI and EVI (MOD13A2)

- QA adjustments to accommodate upstream changes.

- Add support for the SIN grid.

Landcover products: PGE’s 40, 41 and 42

PGE40: 32-day Landcover (MOD12M1)

- Add support for the SIN grid.

PGE41: 96-day Landcover (MOD12Q1)

- Add support for the SIN grid.

PGE42: 96-day Landcover CMG (MOD12C1 and MOD12C2)

- Change resolution from 0.25 deg to 0.05 deg.

LAI/FPAR products: PGE’s 33 and 34

PGE33: Daily LAI/FOPAR (MOD15A1)

- Changed the entire SCF_QC scheme as follows:

o 000 - Main algorithm used, no saturation

o 001 - Main algorith used with saturation

o 010 - Main failed due to bad geometry, backup used

o 011 - Main failed due to sth else, backup used

o 100 - Couldn't even retrieve pixel (NoLand, bad L1B, .. )

- MODLAND_QC bits now get aggregated from both MODAGAGG ch1 (red) and ch2 (nir)

- Added code to handle the change in MODAGAGG's AggrQC:

o WaterVapor -> InternalCloudMask

o Ozone -> InternalFireMask

o AdjClouds -> InternalSnowMask

- Since MODAGAGG doesn't give us AdjacentCloud bit any more use AgrInternalSnowMask as the
replacement in the FparLai_Extra_QC.

- Added options to use Landcover_Type_3 of MOD12Q1 which will eliminate the need for biome
crosswalk. We can still use Landcover_Type_1 by setting the appropriate field in the PCF file.

- Added FPAR_PCF_PROCESS_ENV variable as new metadata field.

- initialized all the SDS layers to fill value at the beginning (it was set to 0 until now).

- Fixed the DeadDetector problem, where DeadDetector bit got set even when DD < 50%

- Added biome type for each pixel as a FparExtra_QC bit 7 this will be than used in compositing in
MOD15A2 to eliminate the problem of unreasonably high LAIs in biomes 1 – 4.

- Added support for the SIN grid.

PGE34: 8-Day LAI/FOPAR (MOD15A2)

- Changed the SCF_QC scheme in accordance with MOD15A1.

- Changed the compositing strategy as follows:
We composite not over 8 values but only over best quality values. So if there are any pixels produced
by main algorithm at best quality, we pick best out of those, if there are none, we try to find all pixels
produced with saturation and pick one of them and so on.

- Added internal switch for switching between old and new compositing scheme. This switch is an
argument to lcmp_xlate_Composite() and can have values COMPOSITE_QA_ALL (= old) or
COMPOSITE_QA_BEST_ONLY (= new).

- Fixed an issue reported by Boston, when lai values were unreasonably high when saturation occured
over biomes <1,4>. To fix this, the approach was to pass an information from MOD15A1 on the
biome type of a given pixel. If the biome is in <1,4> interval, saturation occured and values are too high,
the result is set to Fill, and isQaBuNotPossible is set in the SCF_QA. The information was passed as
a bit 7 in the FparExtra_QC bitfield. This bit was not previously used. As a side efect we can track the
occurences in the output MOD15A2 file: Whenever lai and fpar are fill values, SCF_QA is set to
isQaBuNotPossible (100), and bit 7 in the FparExtra_QC is 0, and all the other bits in FparExtra_QC
are 1, the above situation occured and was corrected.

- Added LCMP_PCF_PROCESS_ENV variable as new metadata field.

- Fixed the problem of GEOANYABNORMAL not being set correctly.

- Added support for the SIN grid.

PSN/NPP products: PGE’s 36, 37 and 38

PGE36, 37 and 38: Daily, 8-day and yearly PSN/NPP (MOD17A1,MOD17A2 and MOD17A3)

- Created new NPP_BPLUT according to numbers supplied by Rama and Maosheng it is part of new
Ancillary file MOD17_ANC_RI8.hdf

- Added a new SDS to MOD17A2 (PsnNet). This contains the 8-day accumulated GPP-RM.

- Added a new SDS to MOD17A1 (PsnNetSum_8day_1km). This is the daily accumulation of GPP - RM
that is then used for PsnNet in MOD17A2. Unlike the GPP - RM field in the MOD17A1, the
PsnNetSum_8day_1km gets reset every 8-days

- The data type of the AnnSum_Mr_1km has been changed from uint16 to int32 to avoid overflows, and
the scale factor for the AnnSum_Mr_1km gridfield has been changed to 0.01

- The old N_Yd_Committed [1200x1200] 2D int16 gridfield is now obsolete and has been deprecated.
Its function is taken over by the ndays_completed[366] global file attribute

- Fixed the problem in psn_xlate, where the NonVeg pixels didn't get set at all in the PsnNet gridfield

- corrected {enc_gain,enc_offset} assignment where enc_gain was incorrectly used for both parameters in
isNpp stanza in psn_xlate_encode()

- Fix the problem with yearly NPP which were getting computed incorrectly due to an error in
Landcover handling

- Updated MUM API to version 2.6

- Modified psn_grid.c so that we can now handle different resolution DAO data (the new GEOS-4
will be 1.25x1.0 deg )

- Changed handling of DAO conversion to different projections. Now we should be able to handle both
SIN and ISIN depending on the projection of the input MOD15A2

- Updated the QA scheme to reflect the changes in MOD15A1 and MOD15A2

- Fixed the incorect unit conversion of DAO data

- Fixed problem with GEOANYABNORMAL metadata field

- Added ProcessingEnvironment LUN as 800550

- Fixed problem in MOD_PR17A1 where the intermediate file was not being updated properly

- Deprecated obsolute tile mbr table in ancillary MOD17_ANC_RI8.hdf

- Added support for the SIN grid.

Vegetation Cover Conversion (VCC) products: PGE’s 66 and 72

PGE66: 32-day 250m VCC product (MOD44B)

- New PGE in MODAPS.

PGE72: 16-day VCC intermediate product (MOD44C)

- Fixed the “left edge effect” problem where fill value was stored instead of good value in the
left edge of the tile

- Minor bug fixes.

- Added support for the SIN grid.