Previous Updates

Data Products Update - 10 January 2003

ECS Collection 4 Processing

Collection 4 processing of the MODIS snow and sea ice data products has begun. Collection 4 processing will generate a consistent data collection from March 2000 to the present using version 4 algorithms. Two processing streams are being run, one for forward processing and one for reprocessing. Forward processing commenced on 1 January 2003. Reprocessing began on 21 December 2002. Reprocessing began with MODIS data acquired on 24 February 2000 and will move forward to reprocess the data to 31 December 2002.

The MODIS snow and sea ice algorithms have been revised to correct some known problems and include some improvements in the algorithms and data products for Collection 4. A synopsis of the revisions for each product is given in this news posting. The snow user guide has been revised to describe the Collection 4 algorithms and data products.

MOD10_L2
The estimated surface temperature screen cutoff in the algorithm was raised to 283K. That was done to allow for detection of snow in mixed pixels, especially in the transition seasons. Evaluation of MOD10_L2 data revealed that the previous temperature screen cutoff of 277K was too restrictive in situations of mixed pixels. The temperature screen remains effective at eliminating false snow detection on bright, warm targets.

In addition to the standard snow map there is a new snow map that is made with a subset of cloud mask spectral tests from the cloud mask product (MOD35_L2). That new snow map is stored in the SDS “Snow Cover Reduced Cloud”. In some situations the subset cloud mask alleviates snow/cloud confusion and allows for better snow mapping. Depending on the situation the snow map may show great or minor improvement. However, in some cloud situations cloud may be falsely identified as snow. This new snow map was include for those who work on a local to regional scale, and as a research product for continued investigation of snow/cloud confusion.

The land/water mask used in processing is known to be inaccurate in places around the globe. One inaccuracy that affects the snow and sea ice products is the coastline of Antarctica. Ice shelves are mapped as land in the land/water mask, thus calving of icebergs may appear in the snow product rather than in the sea ice product. A more accurate land/water mask is being developed within the MODIS land discipline group and an improved Antarctic coastline database as been identified. The more accurate land/water mask will be phased into use during Collection 4 processing.

MOD10A1
Is now in Sinusoidal projection. Previous collections used the Integerized Sinusoidal projection. The change in projection caused very little change in the snow maps.

Two changes are likely to be made in the MOD10A1 product during Collection 4 processing. A new daily observation selection algorithm that includes solar elevation as a selection factor is being developed. That improvement should alleviate a bias in the current selection observation that results in patches of nighttime data to be mapped near the north polar regions from late summer through autumn.

The snow albedo algorithm and product should be completed and integrated into the MOD10A1 product in the near future. This snow albedo product will be a daily snow albedo and corresponds with the MOD10A1 snow map. Albedo will not be calculated for other surface features. The MODIS albedo product (MOD43) has a temporal resolution of sixteen days and albedo for all surface features, including snow, are given in that product.

MOD10C1
Improved mapping transition from day to night in the polar regions. A smoother transition boundary is generated.

To alleviate false snow detection in regions of the world where snow is an extremely rare climatological event, a snow impossible mask was applied. Any cells in the mask will not be mapped as snow.

Antarctica is mapped as always snow-covered. That was done to enhance the quality of the global snow map. Areas of Antarctica where snow cover may diminish and increase with the seasons should be observed with the MOD10_L2 or MOD10A1 snow products. Also, see the note about the land water mask in the MOD10_L2 section.

MOD10A2
Is now in Sinusoidal projection. Previous collections used the Integerized Sinusoidal projection. The change in projection caused very little change in the snow maps.

MOD10C2
Improved mapping transition from day to night in the polar regions. A smoother transition boundary is generated.

To alleviate false snow detection in regions of the world where snow is an extremely rare climatological event, a snow impossible mask was applied. Any cells in the mask will not be mapped as snow.

Antarctica is mapped as always snow-covered. That was done to enhance the quality of the global snow map. Areas of Antarctica where snow cover may diminish and increase with the seasons should be observed with the MOD10_L2 or MOD10A2 snow products. Also, see the note about the land water mask in the MOD10_L2 section.

MOD29
No significant changes were made in the algorithm or product for Collection 4. Discrimination of clouds over ice surfaces during the day and night continues to be problematic.

MOD29P1D
A new scoring algorithm for selecting the observation for a day was integrated into the sea ice algorithm. In this scoring algorithm, solar elevation is a factor in selecting the observation, the observation with the highest score based on solar elevation (highest solar elevation is favored), distance from nadir (close to nadir is favored) and observation coverage in the grid cell (greatest coverage favored) is selected as the observation for the day. That scoring is applied to reflectance data then the corresponding ice surface temperature observation is also selected. In the case of only night, thermal only, data, solar elevation is not a factor in the scoring algorithm.

A great improvement in the quality of sea ice maps is realized with the new scoring algorithm.

MOD29P1N
No significant changes were made in the algorithm or product for Collection 4.

Data Products Update - 21 August 2002

MOD10C1 DIMENSION FAULT

The dimension fault in the dimension data of the structural metadata (StructMetadata.0) was fixed. MOD10C1 product files from 24 July 2002 (2002 205) have the correct dimension setting in the structural metadata (StructMetadata.0).

MOD10C1 PERCENTAGE OF SNOW REPORTED

A change in the percentage of snow reported in the MOD10C1 product was made in Version 4 of the product. Daily snow cover is now reported in the range of 0-100% in the "Day_CMG_Snow_Cover" SDS. This change became effective on 24 July 2002 (2002 205). Prior to that date daily snow cover was restricted to the range of 40-100%.

A caveat regarding the 0-100% snow cover. The user needs to interpret the accuracy and worth of using the lower snow cover percentages. Often the lower snow cover percentages are the result of false snow detections made at the start of the product processing sequence, MOD10_L2 product that have been propagated to the CMG product. Some snow is likely to occur in areas where snow is an impossibility. Where snow is likely to occur small amounts of snow maybe shown but, snow may or may not actually have been present. The user is cautioned to use climatology or other data sets to ascertain the probability of snow shown in the CMG when its occurrence is suspicious.

In the near future a mask of where snow is an impossibility will be added to the product to aid in filtering out occurrences of snow errors of commission.

MOD10C2 DIMENSION FAULT

There is an error in the structural metadata (StructMetadata.0) of the MODIS/Terra Snow Cover 8-Day L3 Global 0.05Deg CMG products. This error affects all V003 granules distributed by the NSIDC DAAC and relates to geolocating and sub setting the data.

There is an error in the dimension settings in the StructMetadata.0 that causes software tools that read the StructMetadata.0 for dimension data for geolocation and sub setting to fail. The geolocation-based functions of the HDF-EOS API and tools such as the UAH Subsetter that are based on the HDF-EOS API are known to fail.

Software tools that do not use the StructMetadata.0 are not affected and the data arrays can be properly visualized. The fix to the dimension error has been made and will be implemented in Collection 4 processing (V004) scheduled to begin in October 2002.

Data Products Update - 15 July 2002

MOD10_L2 IMPACTED BY SURFACE TEMPERATURE SCREEN

A revised version of the MODIS snow algorithm with a surface temperature screen has been generating MOD10_L2 snow products since 3 October 2001. It has been discovered that the surface temperature screen in that version is too restrictive. The problem is most notable in spring and autumn when the temperature screen prevents snow from being detected in mixed pixel conditions. The effect is variable depending on conditions, e.g. the mix of snow with rocks or vegetation, it may prevent a small amount or a large amount of snow extent from being mapped.

Comparisons of that version of MOD10_L2 with previous time periods or with snow maps from other sources will be affected by this problem. It appears that the effect is greatest in late spring and autumn in the Northern Hemisphere. The effect also carries into the daily snow products, MOD10A1 and MOD10C1 and the eight-day products, MOD10A2 and MOD10C2.

Adjustments to the surface temperature screen in MOD10_L2 are being investigated. A change to alleviate this problem will be implemented in the next version of the algorithm which will be used in the ECS Collection 4 processing scheduled to begin 1 October 2002. Reprocessing with the new version will also begin at that time and the MOD10_L2 version with the surface temperature problem will be replaced with the new version.

MOD10C1 DIMENSION FAULT

It has recently come to our attention that there is an error in the structural metadata (StructMetadata.0) of the MODIS/Terra Snow Cover Daily L3 Global 0.05Deg CMG and the MODIS/Terra Snow Cover 8-Day L3 Global 0.05Deg CMG products. This error affects all V003 granules distributed by the NSIDC DAAC and relates to geolocating and subsetting the data.

There is an error in the dimension settings in the StructMetadata.0 that causes software tools that read the StructMetadata.0 for dimension data for geolocation and subsetting to fail. The geolocation-based functions of the HDF-EOS API and tools such as the UAH Subsetter that are based on the HDF-EOS API are known to fail.

Software tools that do not use the StructMetadata.0 are not affected and the data arrays can be properly visualized.

The fix to the dimension error has been made and will be installed in V003 after further review and approval. That fix will also be implemented in Collection 4 processing (V004) scheduled to begin in October 2002.

Website Update - 1 February 2002

Website name changed to http://modis-snow-ice.gsfc.nasa.gov. This new URL is meant to be more user friendly, and also follows a common naming convention being used by MODIS related websites. See the MODIS Organigram [ http://modis.gsfc.nasa.gov/cgi-bin/texis/organigram/weblinks ] for a full index of MODIS related websites.

MODIS Snow Products Update - 6 December 2001

Image of snow extent in mid-US from November 26-28, 2001. Click on image for larger image and more information. Also, visit our new Direct Broadcast section.

MODIS Snow Products Update - 11 October 2001

Visit the MOD10C1 and MOD10C2 snow data products web pages to view images and download data for 2001 and 2002 MOD10C2 snow products.