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
Collection 006 Update
Collection 051 Update
Collection 005 Update
Known Problems
Filename Convention
Flow Diagram

Known Problems

Collection 051

    Bad data in a subset of Collection 051

    An error involving incorrect Clear Sky Radiances was found in a subset of Terra Collection 051. This adversely affected all cloud-related parameters in both Level-2 and Level-3 for the impacted subset. This bad data has been reprocessed and new corrected Collection 051 data is now available for download from LAADS; however users of MODIS Terra Collection 051 data need to check their previously downloaded archive of data to make sure none of their data is bad.

    How To Identify Bad C051 Data

    All Cloud-related (MOD06_L2, MODATML2, MOD08_D3, MOD08_E3, and MOD08_M3) Collection 051 Terra HDF files with a data date prior to April 1, 2010 (2010 091) and a processing date prior to September 28, 2010 (2010 271) is bad. Note that this error did not impact Terra C051 HDF files with data dates on or after April 1, 2010 (2010 091); nor did this error impact any Aerosol or Water Vapor SDS's. Both the data date and production date are a part of the HDF filename. A new corrected Collection 051 Terra reprocessing effort began again on calendar date September 28, 2010 (2010 271) and is now completed.

    Users of cloud-related HDF product files should replace any bad data files with the latest reprocessed C051 data from the LAADS website, the distributor of MODIS data.

    It should be noted that any Collection 051 HDF data file with a data date on or after April 1, 2010 (2010 091) was not impacted by this problem and is considered correct.

    Below is a sample HDF file name, along with the 4 keys on how it can be identified as bad data:

    View the MODIS Data Processing & Availability Calendar to see the status of data availabile in various Collections.

    To see how good C051 data compares with bad C051 data, see the bottom of the C051 page in the products section, linked here: Comparison of Good vs. Bad C051 Data

Collection 005

Listed below are known problems associated with each MODIS Atmosphere data product in Collection 005.

Aerosol Product (04_L2)

    Uninitialized, Non-Fill Data Occasionally Contaminates Deep Blue Aerosol parameters. In certain circumstances, data with very high solar zenith angles (SZA) can cause uninitialized, non-fill data to be placed into the MYD04_L2 file. Fill data is appropriate for Deep Blue retrieval at very high SZA (above 72 degrees). The latest version of Deep Blue corrects this defect. To date, a limited number of days which are effected by this problem were found. Problematic data is typically found in the November to March time period, and in polar regions (Canada, Siberia, Alaska, Greenland, Antarctica) In any given month, there may be a single day exhibiting the problem, or there may be a large number of days.
    Problematic: [PGE = PGE04 v5.3.7] [Period = 8/2002-10/2007] [Platform = Aqua]
    Corrected: [PGE = PGE04 v5.3.8] [Period = 10/2007-current] [Platform = Aqua]
    Information provided by Jeremy Warner (NASA/GSFC).

Water Vapor Product (05_L2)

    Near-IR Water Vapor Problematic over Bright Clouds. The near-IR water vapor values over some very bright cloud pixels are in error because the channels used in the retrieval algorithm are saturated over bright clouds. In order to fix the problem, QA parameters need to be utilized so that saturated pixels are properly identified and flagged.
    Problematic: [PGE = PGE05 all] [Period = 2/2000-current] [Platform = Terra & Aqua]
    Corrected: [T.B.D]
    Information provided by Dr. Bo-Cai Gao (Navy NRL).

Cloud Product (06_L2)

    Cloud Optical Properties

    Click for zoom Striping in Cloud Optical Property parameters. Between March 2007 and August 2007 in Terra data only, a problem with detector 6 of band 29 caused single scanline "stripes" (every 5th line, but not always continuous across the scanline) in the following Level 2 parameters: Cloud Optical Thickness, Cloud Effective Radius, Cloud Water Path, and QA Flags related to cloud phase. This was caused by a change to the handling of uncertainty indices for band 29 in the Level-1B data to compensate for the fact that in 2007 on Terra MODIS band 29 detector 6 went dead. Since band 29 was being used for phase determination for thin cirrus clouds, striping in Level-2 Cloud Optical products resulted. Every 6th line of every granule that contains any thin cirrus is affected by cloud phase being erroneously turned to water and the subsequent retrieval failing in most cases.
    Problematic: [PGE = PGE06 v5.11.0] [Period = 3/2007-8/2007] [Platform = Terra]
    Corrected: [PGE = PGE06 v5.12.4] [Period = 9/2007-current]
    Information provided by Dr. Steven Platnick and Gala Wind (NASA/GSFC).

    Click for PDF Cloud Optical Thickness Retrieval over Land Dropout. For all cloudy MODIS data over land (both Terra and Aqua) from launch through August 2007, errors in the cloud optical propery retrieval algorithm caused small displacements in the distribution of Cloud Optical Thicknesses. A number of software bugs in the Rayleigh correction portion of the Cloud Optical Properties code caused dropouts in Cloud Optical Thickness ~11.5 for liquid water clouds and ~8.5 for ice clouds. Since a number of other parameters are dependent on the Cloud Optical Thickness retrieval, including Cloud Effective Radius, Cloud Water Path, and QA Flags related to Cloud Optical Properties Cloud Phase; these were also impacted to a small degree. Small impacts were also felt in downstream L3 products related to these parameters (including Cloud Optical Properties Cloud Fraction). Every retrieval over land was affected. For more detail on this problem and its impact, see the Cloud Optical Thickness over Land Retrieval Problem document [View PDF]. A fix was implemented in all MODIS data (both Terra and Aqua) starting in September 2007.
    Problematic: [PGE = PGE06 v5.11.0] [Period = 2/2000-8/2007] [Platform = Terra & Aqua]
    Corrected: [PGE = PGE06 v5.12.4] [Period = 9/2007-current]
    Information provided by Dr. Steven Platnick, Paul Hubanks, Gala Wind (NASA/GSFC).

    Click for Zoom Missing Blocks (Retrieval Failure) in Cloud Optical Properties. The holes (missing blocks) come from fill values in the GDAS dataset causing the Cloud Top Properties retrieval to abort. And if there is no cloud top retrieval (CTP, CTT) , then there will be no attempted retrieval of Cloud Optical Properties. This missing block issue was most pronounced in high latitudes with more than one box missing in many cases. However, this issue does occasionally occur in other global regions. Click on the thumbnail image to view a before and after (fix implementation) from a sample Greenland granule. A fix was implemented in all MODIS data (both Terra and Aqua) starting in September 2007.
    Problematic: [PGE = PGE06 v5.11.0] [Period = 2/2000-8/2007] [Platform = Terra & Aqua]
    Corrected: [PGE = PGE06 v5.12.4] [Period = 9/2007-current]
    Information provided by Dr. Steven Platnick and Gala Wind (NASA/GSFC).

    Cloud Top Properties

    Click for Zoom Missing Blocks (Retrieval Failure) in Cloud Top Properties. A problem in the Cloud Top Retrieval algorithm caused retrieval failures when corresponding 1-degree resolution Global Data Assimilation System (GDAS) temperature and moisture profiles were very cold and dry. Sometimes non-physical values of relative humidity are found leading to gaps (missing 1 degree blocks) in the retrievals. The problem occurs mostly over snow and ice covered Arctic and Antarctic regions (e.g. Greenland) in conditions of polar night and twilight. Changes were made in the Cloud Top Properties retrieval algorithm and implemented in September 2007 to better handle the problematic moisture profiles. Shown in the accompanying figures are cloud top pressure retrievals for an Aqua granule beginning at 02:00 UTC on December 1, 2004 that includes northeastern Siberia near the Arctic Ocean. Clear areas and missing data are indicated in blue. Notice the several 1-degree blocks in the "before" image that are replaced by retrievals in the "after" image.
    Problematic: [PGE = PGE06 v5.11.0] [Period = 2/2000-8/2007] [Platform = Terra & Aqua]
    Corrected: [PGE = PGE06 v5.12.4] [Period = 9/2007-current]
    Information provided by Dr. Paul Menzel and Richard Frey (University of Wisconsin).

    Click for zoom Missing First Pixel in all Scanlines for Cloud Phase Infrared. The first IR cloud phase value in each scanline across a granule (i.e.: the first column) was erroneously set to a fill value in all Cloud Phase Infrared related SDSs. The viewing zenith angle limit was incorrectly set to 65 degrees beyond which a phase retrieval was not performed. The error should have mimimal effect on users since the problem occurs only on the very edge of the scan. There is minimal effect on statistics since the problem is found in only 406/(406*270) values per granule, or < 0.4% of the total. As a result of this error, the first column of the MOD06CT "Cloud_Phase_Infrared" parameter is set to the FillValue. This also impacts all the Cloud Phase Infrared QA Flags. This viewing zenith angle check will be eliminated in the Collection 006 delivery.
    Problematic: [PGE = PGE06 v5.11.0] [Period = 2/2000-current] [Platform = Terra & Aqua]
    Corrected: [To be corrected in Collection 006]
    Information provided by Richard Frey (University of Wisconsin).

    CTP and CTT Too Low in Inversion Conditions. There is a known problem of cloud heights being too high (and therefore cloud top pressure and temperature too low numerically) in atmospheric inversion conditions when the IR window channel retrieval method is used instead of CO2-slicing. It is most frequent in the trade wind stratus regimes off the coasts of California, Peru, Namibia, etc., but can also occur over land at night and in polar night conditions. The current cloud top pressure algorithm is highly dependent on low vertical and spatial resolution input GDAS temperature and moisture profiles that sometimes do not accurately capture the details of the inversions. Several options are being explored to ameliorate the problem. Validation of potential algorithm changes will be accomplished via space-borne lidar cloud height retrievals (CALIPSO).
    Problematic: [PGE = PGE06 v5.11.0] [Period = 2/2000-current] [Platform = Terra & Aqua]
    Corrected: [T.B.D.]
    Information provided by Richard Frey (University of Wisconsin).

    Cirrus Reflectance

    Cloud Reflectance Too High over High Elevation Surfaces. Cirrus Reflectance values over high terrain (such as Andes Moutains and Tibet Plateau) in certain seasons are too large because of insufficient water vapor absorption at 1.38 micron when the atmosphere is very dry. Techniques have been developed to mask out the surface-contaminated pixels and future corrections are anticipated.
    Problematic: [PGE = PGE06 all] [Period = 2/2000-current] [Platform = Terra & Aqua]
    Corrected: [T.B.D]
    Information provided by Dr. Bo-Cai Gao (Navy NRL).

L3 Daily, Eight Day, and Monthly Products (08_D3, 08_E3, 08_M3)

    Upstream L2 Errors Propagate into L3. There are no known L3 unique errors. However, errors in the upstream L2 data products (see entries above) will propagate into the L3 products.
    Information provided by Paul Hubanks (NASA/GSFC).