Aqua Collection 4 PGE changes

 

 

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

Sea ice products: PGE’s 08 and 44

Surface reflectance products: PGE’s 11, 21 and 75

Fire products: PGE’s 30, 29 and 80

Land surface temperature products: PGE’s 16, 31, 32, 58 and 59

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

Aggregated surface reflectance products: PGE 22

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

Vegetation Indices products: PGE’s 25, 25Q, 26, 27 and 35

LAI/FPAR products: PGE’s 33 and 34

PSN/NPP products: PGE’s 36, 37 and 38

Vegetation Cover Conversion (VCC) products: PGE 72

 

 

 

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

 

PGE07: Level 2 Snow (MYD10_L2)

-         MODIS snow algorithm was modified to use band 7 data in place of band 6 in MOD_PR10 for Aqua processing only due to the large number of disfunctional detectors in band 6.  Critera used in the snow algorithm were adjusted for band 7 data . 

-         Due to the problem of increased errorneous snow detection specific to the NDSI/NDVI decision region of the algorithm using band 7 data, that test was disabled.  Code for that test remains in the algorithm however the result is always negative for snow. 

(see figures 1 and 2).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


Figure 1:  Example of improved quality between the previous and new version of MYD10_L2 Aqua algorithms.  Northern central Northwest Territories shown in this 400x400 pixel subset.  Condition of low illumination near the change to night.  Key: white - snow, green - land, sea-green - inland water, pink - cloud, blue - ocean.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


Figure 2.  Example of continuity and improved quality between the previous and new version of MYD10_L2 Aqua algorithms.  The snow extent mapped (colors in each image) is extremely similar but the V004 is slightly improved as there are few instances of erroneous snow (medium green) mapped in the southwestern quarter of Alaska. 

 

 

PGE43: Daily snow cover (MYD10A1):

-         No algorithm change.

 

PGE45: 8-day snow (MOD10A2):

PGE46: Daily snow CMG (MOD10C1):

PGE67: 8-Day snow CMG (MOD10C2):

-         No algorithm change.

-         Set the RangeBeginningTime (RangeEndingTime) metadata to 00:00:00 (23:59:59).

 

 

 

Sea ice products: PGE’s 08 and 44

 

PGE08: Level 2 sea ice (MYD29):

-          An Aqua specific version of MOD_PR29 was developed that uses Aqua band 7 data instead of band 6 data for the sea ice by reflectance part of the algorithm. This change removed the striping due to band 6 bad detectors from the sea ice product. (see figures 3 and 4).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


Figure 3:  Example of greatly improved quality between the previous and new version of MYD29 sea ice Aqua algorithms.  Image is a reduced resolution swath of MYD29 covering the Arctic Ocean and coastline of Alaska, on right side of the image to Victoria Island on the left side of the image.  Key; white - sea ice, blue - ocean, pink - cloud, green - land, orange - no-decision.

 

 

 

MYD29P1D.A2003200.h08v07.004
 

MYD29P1D.A2003200.h08v07.003
 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


Figure 4.  Example of greatly improved quality between MYD29P1D generated from the previous MYD29 and from the new version of MYD29 sea ice Aqua algorithms.  Image is a reduced resolution tile of MYD29P1D  covering the Arctic Ocean and northern coastline of Alaska.  Key; white - sea ice, blue - ocean, pink - cloud, green - land, orange - no-decision.

 

PGE44: Daily day and night sea ice (MYD29P1D and MYD29P1N):

-         No algorithm change.

 

 

 

Surface reflectance products: PGE’s 11, 21 and 75

 

PGE11: Level 2 surface reflectance (MYD09)

-         Improve cloud shadow detection algorithm by using water vapor information to refine the cloud height  estimate.

-         Improve the code efficiency by precomuting certain parameters.

-         Simplify ceratin parts of the code, especially I/O.

 

PGE21: 8-day Land surface reflectance (MYD09A1)

-         Improve data handling at high solar zenith.

-         Refine the cloud and cloud shadow detection over snow.

-         Exclude observations with a view zenith angle > 60 deg.

 

PGE75: Daily CMG Land surface reflectance (MYD09CMG)

-         No algorithm change.

 

 

 

Fire products: PGE’s 30, 29 and 80

 

PGE30: Level 2 Fire (MOD14)

-         No algorithm change.

 

PGE29: Daily and 8-day Fire (MYD14A1 and MYD14A2)

-         No algorithm change.

 

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

-         No algorithm change.

 

 

Land surface temperature products: PGE’s 16, 31, 32, 58 and 59

 

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

 

-         No algorithm change.

 

PGE31: 8-day land surface temperature (MYD11A2)

PGE32: Daily CMG land surface temperature (MYD11C1)

PGE58: 8-Day CMG land surface temperature (MYD11C2)

PGE59: Monthly CMG land surface temperature (MYD11C3)

-         Set the RangeBeginningTime (RangeEndingTime) metadata to 00:00:00 (23:59:59).

 

 

 

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)

-         No algorithm change.

 

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

-         No algorithm change.

 

PGE14: Daily L2G snow (MOD10L2G)

-         No algorithm change.

 

PGE15: Daily L2G sea ice (MOD29PGD and MOD29PGN)

-         No algorithm change.

 

 

Aggregated surface reflectance products: PGE 22

 

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

-                Fixed a problem that caused data drops (FILL VALUE) in some high latitude pixels/tiles. The problem is related to inconsistent band quality across observations/orbits. In some instances, an observation/orbit contains only one/few good band(s) that is subsequently chosen for output by the algorithm after being ranked high (we only used QA filter). The fix added a Surface Reflectance validity check/filter for each band and will favor observations/orbits that contain all/mostly valid bands.
-                Fixed a bug that caused PGE22 algorithm (collection 4) to output bad L/W masks for some pixels in high latitude tiles. In some instances the Land/Water mask was mislabeled over few pixels leading to problems in downstream products.  Only vertical tiles 00 and 01 were affected.  This bug was due to an initialization error that only affected high latitude tiles (More than 4 orbits).
-                Fixed a bug that caused the program to exit when input tiles have different file formats.  The algorithm was adjusted to use a minimum format of LH2G_ONELAYERONLY (one later) when file formats are not the same.
 

    

                                                                  Before                                                 After                             Before After difference

 
   
                                                                                                                                               Before (Zoom)                                    After (Zoom)
 

Figure 5:  MODAGG RGB (Bands 1, 4, and 3) Image (Tile h17v01). 2001-193

 

 

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

 

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

-         No algorithm change.

 

PGE24: 16-day Albedo CMG (MOD43C1)

-         No algorithm change.

 

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

-         No algorithm change.

 

PGE82: 16-day NBAR CMG (MOD43C3)

-         No algorithm change.

 

 

 

Vegetation Indices products: PGE’s 25, 25Q, 26, 27 and 35

 

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

-         No algorithm change

 

PGE26: Monthly 1km NDVI and EVI (MOD13A3)

-         New algorithm

-         This product is a calendar month temporal average of the 16 days 1km VI product (MOD13A2) and it contains the 11 SDSs as in the input.

      

                                 Monthly 1km VI                                                       Input 16 day 1km VI

Figure 6: 1km monthly VI product (left) and input16 days 1km product (Right).

PGE27/: 16-day 0.05o NDVI and EVI (MOD13C1)

-                     New algorithm

-                     It ingest the global 16 day 1km VI tiles (MOD13A2)

-                     Geographic projection

-                     The product SDSs list is below:

  • CMG 0.05deg 16 days NDVI  - Recalculated from averaged input reflectance
  • CMG 0.05deg 16 days EVI - Recalculated from averaged input reflectance
  • CMG 0.05deg 16 days NDVI Quality  - Reflecting the dominant QA from input
  • CMG 0.05deg 16 days EVI Quality - Reflecting dominant QA from input
  • CMG 0.05deg 16 days red reflectance - Spatially averaged from input
  • CMG 0.05deg 16 days NIR reflectance  - Spatially averaged from input
  • CMG 0.05deg 16 days blue reflectance - Spatially averaged from input
  • CMG 0.05deg 16 days MIR reflectance - Spatially averaged from input
  • CMG 0.05deg 16 days average view zenith angle - Spatially averaged from input
  • CMG 0.05deg 16 days average sun zenith angle - Spatially averaged from input
  • CMG 0.05deg 16 days average relative azimuth angle - Spatially averaged from input
    The above three SDSs contain averaged geometric viewing angles, and are not considered scientifically correct, but are output for completeness and will give an idea of how these angles change.
  • CMG 0.05deg 16 days NDVI standard deviation
  • CMG 0.05deg 16 days EVI standard deviation

The standard deviations in SDS 12 and 13 is the statistical standard deviation of the finer resolution input VI (CMG pixels are about 5x5 input pixels).

  • CMG 0.05deg 16 days Percent cloud cover – based on finer resolution input pixels
  • CMG 0.05deg 16 days Percent with vegetation

SDS 14 and 15 give percentage of contributing input pixels that were deemed cloudy (based on input QA) or vegetated (based on a VI baseline derived from the MODIS). The vegetation percentage can for example be used to identify changes in vegetation once time series data becomes available.

 

 

                                 CMG NDVI                                           CMG NDVI standard deviation

Figure 7:  Sample MODIS VI CMG product

 

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

-         No algorithm change

-         Speckle in northern latitude changed as a result of fixing the surface aggregation algorithm (Figure 8)

 

 

                           Before                                                             After

Figure 8:  MOD13A2 (1km 16 day VI) showing the speckle resulting from the problematic input data and the same image after using the corrected input...

 

 

 

LAI/FPAR products: PGE’s 33 and 34

 

PGE33: Daily LAI/FOPAR (MOD15A1)

-                Introduced biome dependent uncertainties (read from Bu_anc_constants).
-                Handle varying length interval on NDVI.
-                Use new LookUp Tables to compute the LAI for biome 5 (broadleaf forest). Example 9 illustrates the improvement resulting from this change. The example shows the use of main and backup algorithms for tile h11v09 (Amazonia) where over 95% of the tile is covered by biome 5. The histograms show the significant drop in using the ‘back up’ algorithm (42% -> 11%). The class labeled ‘saturated’ still uses the main algorithm and lead to high LAI (> 4). The quality of the observations labeled ‘saturated’ is better than the quality of those labeled ‘back up’.
 

SCF_QC, AQUA, collection 3

SCF_QC, AQUA, collection 4

Figure 9: LAI QA layer for tile h11v09 produced by version 3 and 4 of PGE33. The changes made in the AQUA reprocessing version (4) lead to a significant increase in the use of the ‘main’ algorithm at the detriment of the ‘backup’ algorithm.

 

 

PGE34: 8-Day LAI/FOPAR (MOD15A2)

-         Compositing change for biomes 5 and 6 (Join main and saturation into one level).

-         Set the RangeBeginningTime (RangeEndingTime) metadata to 00:00:00 (23:59:59).

 

 

 

PSN/NPP products: PGE’s 36, 37 and 38

 

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

-         Set the RangeBeginningTime (RangeEndingTime) metadata to 00:00:00 (23:59:59).

 

 

 

Vegetation Cover Conversion (VCC) products: PGE 72

 

PGE72: 16-day VCC intermediate product (MOD44C)

-         No algorithm change.