Recent Updates

Resolution Comparison of Daily Global MODIS Snow Maps - 10 January 2006

The standard MODIS daily global fractional snow (FSC) map product is generated at 0.05 degree resolution (about 5 km at the equator). However current climate models typically run at 0.25 degree to 0.33 degree spatial resolution, a lower resolution than the MODIS daily global FSC map product. In response to those modelers' spatial preferences a special daily global FSC map at 0.25 degree resolution was created from the standard 0.05 degree MODIS daily global snow map product. Comparative analysis of the MODIS daily global FSC maps at 0.05 degree and 0.25 degree resolution reveals that the general pattern of snow extent is very similar between the maps. However, extent of snow especially at high and low fractions and at edges of snow covered areas, can vary considerably. An example of how snow cover mapping changes with grid resolution is shown in the graphic below.

Sierra Nevada Mountain subset from the MODIS 0.05 degree (left) and 0.25 degree (right) CMG snow map.

Workshop on EOS Snow and Ice Products 16-17 November 2004

The First Workshop on Earth Observing System (EOS) Snow and Ice Products was held near Goddard Space Flight Center in Landover, Maryland, on 16-17 November 2004. The intent of the workshop was to bring together current and potential users of EOS snow and ice standard products from the Moderate-Resolution Imaging Spectroradiometer (MODIS), Advanced Microwave Scanning Radiometer - EOS (AMSR-E), Ice, Cloud and land Elevation Satellite (ICESat), Enhanced Thematic Mapper Plus (ETM+), and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) sensors. Visit the following website to view the meeting agenda as well as meeting presentations: http://nsidc.org/events/eos_workshop04/

Snow products improvement examples - Aqua v004 Swath images display the southern half of Greenland. The cloud stripping problems are eliminated by use of band 7 data in version 004 swaths. Arctic Ocean and Northwestern Alaska. New scoring algorithm preferentially selects observations based on local solar noon, near nadir and observation coverage in a grid cell. Eliminated problems with night and low illumination data being selected. Sea ice products improvement examples - Aqua v004 Sea ice by reflectance - stripping problems eliminated by use of band 7 data. Sea ice detection and cloud detection both are affected by the change to band 7 data.

Aqua Collection 4 Snow and Sea Ice Products Released on Friday, 13 February 2004

Aqua MODIS Collection 4 (V004) data is set for public release on Friday, 13 February 2004. Analysis of the Aqua V004 products from a science test run of the products and initial evaluation of the Collection 4 products has been completed. The Aqua V004 data products show significant improvement in quality compared to V003. Most of the improvement in the snow and sea ice products is a result of changing to the use of Aqua MODIS band 7. The reason for that band change is because the numerous non-functional detectors in band 6 degraded the quality of the snow, sea ice and cloud mask products. The cloud mask product, which is used as input, by the snow and sea ice algorithm was also changed to use band 7 data in place of band 6 data. Quality is improved through the sequence of data products from level-2 to level-3 as the quality improvements are passed along in the product sequence. All Aqua V004 snow and sea ice products will be released.

Aqua V004 is being produced in a forward and reprocessing stream. Forward processing began with 2 February 2004 (2004 033). Reprocessing began with 4 July 2002 (2002 185). Aqua snow and sea ice Earth Science Data Types (ESDT) released are; snow; MYD10_L2, MYD10A1, MYD10C1, MYD10A2, MYD10C2, sea ice; MYD29, MYD29P1D, MYD29P1N and MYD29E1D. The ESDTs can be ordered from the NSIDC DAAC.

Terra and Aqua specific versions of the snow (MOD10_L2 and MYD10_L2) and sea ice (MOD29 and MYD29) algorithms were implemented in Aqua V004. Satellite specific versions were required because of the numerous non-functional detectors in MODIS band 6 on Aqua degraded the quality of the products. Reflectance characteristics of snow are similar between MODIS bands 6 (1.6 um) and 7 (2.1 um) so band 7 is a reasonable replacement for band 6 in the snow algorithm. The Aqua snow algorithm was adjusted to band 7 data based on analysis of Terra snow maps. Several selected MOD10_L2 swaths were analyzed to determine snow reflectance values and NDSI values to use with band 7 data. Using the MOD10_L2 snow map statistics for reflectance and NDSI values using band 6 and band 7 were collected and analyzed. From that analysis a band 7 specific version of the algorithm was developed. The band 7 specific version was then applied. In those swaths studied a nearly identical snow map was generated with the band 7 algorithm version as with the standard version. However, it was observed in those and other swaths that there was a tendency for the band 7 algorithm to occasionally produce more erroneous snow results at edges of clouds within the NDSI/NDVI test in the algorithm. To decrease that error the NDSI/NDVI test was disabled, until further analysis can be conducted, in the Aqua version of MYD10_L2.

Products in the Aqua V004 collection are released with a provisional validation status. Provisional status is assigned because analysis has been limited to short temporal periods in the science test and V004 processing to date. As the V004 collection grows in time and more analysis is possible validation status will be revised.

Check the NSIDC website for further information and to order data products.
User Guides will be revised with descriptions of the V004 algorithms and data products.

Daily Snow Mapping Improvement and Snow Albedo - 18 September 2003

Top image is the snow map from MOD10A1, tile h10v04, on 30 January 2002 with snow shown in white. Corresponding snow albedo is shown as shades of gray for the snow albedo range of 0 – 100% in the bottom image.

An improved version of the MODIS daily snow product (MOD10A1) is scheduled to begin production with data from 14 September 2003 onwards in the MODPAPS forward processing stream. (This improved version will not be put in the reprocessing stream, as the Collection 4 reprocessing is nearly complete.) The enhanced product should be available for ordering about 24 September 2003.

A new scoring algorithm for selecting the most favorable observation of the day from several MODIS swath acquisitions was integrated. The scoring algorithm selects the observation of the day based on solar elevation, distance from nadir and amount of coverage in a cell of the sinusoidal projection. Algorithm is weighted to favor the observation obtained about local noontime. Improved continuity in features mapped has been observed. Patches of nighttime data that were mapped from swaths of day and night data at high northern latitudes, about 60 – 70 degrees north, in the boreal summer by the old algorithm have been eliminated. Daytime data are mapped instead so a complete day picture of snow cover is seen. Similar improvement in continuity of features mapped and the elimination of nighttime data are observed in the MOD10C1.

Snow albedo is included as a data array in the MOD10A1 product. Snow albedo is determined for cells in the grid that are mapped as snow. Snow albedo is calculated using the surface reflectance data from MOD09GHK. Also used is the land cover product MOD12Q1 and slope and aspect data. Surface reflectance data from MODIS bands 1-5 and 7 are used. A narrow-to-broadband conversion function is used to calculate a snow albedo. Snow albedo is reported as a range from 0 – 100%. Along coastlines there is disagreement between the slope and aspect data used and the land/water mask carried in the sequence of MODIS products. Special data values are included to identify those situations. Details of the snow albedo algorithm can be found in the snow product user guide.

Release of Aqua MODIS Snow and Sea Ice Products - 20 May 2003

Aqua MODIS snow and sea ice products are scheduled for release on or about 23 May 2003.

Performance of the MODIS on Aqua has been very good, with the exception of the band 6 (1.6 um). Fourteen of the 20 detectors in band 6 are non-functional. The non-functional detectors have impacted the cryospheric snow and sea ice level-2 algorithms and the cloud mask level-2 algorithm because MODIS band 6 is the primary band used in snow/cloud discrimination. The products have been affected in different ways however, the quality of both the cryospheric and cloud mask products has been degraded. Greatest degradation of quality occurred in the cloud mask algorithm over snow where the non-functional data resulted in stripes of cloud over snow, which then appeared in the cryospheric products. The cryospheric algorithms have been less impacted by the non-functional band 6 data because averaged band 6 data is used for the non-functional detectors. Use of averaged data results in a reasonable mapping of snow extent. The cloud mask algorithm has been revised to use MODIS band 7 (2.1 um) in place of band 6 for snow/cloud discrimination. That has resulted in improved quality of the cloud mask and improved quality in the Aqua cryospheric products beginning about 21 May 2003 in Collection 4. Aqua cryospheric products can be released with improved quality at a provisional product quality status with the revised cloud mask as input. Next improvement in Aqua cryospheric product quality status is not expected until analysis of using band 7 in place of band 6 for Aqua MODIS is completed. At some point band 7 will replace band 6 in the Aqua cryospheric algorithms.

Differences in the impact of the MODIS Aqua non-functional band 6 detectors on the cryospheric and cloud mask algorithms are caused by the different resolutions between the products. In L1B processing of the 500m resolution band 6 data the non-functional detectors are flagged and a linear averaged value based on functional detectors adjacent to the non-functional detectors is written for the non-functional detectors. That is done to produce a reasonably good image primarily for visualization purposes. In L1B processing of the 1km data product the MODIS 500m bands are aggregated to 1km however, non-functional detector data is not used in the aggregation. The cryospheric products use the averaged 500m L1B data for the non-functional detectors and flag that data in the QA data array. The cloud mask algorithm uses the 1km L1B product so a significant amount of data was missing and default of cloud was written in the cloud mask causing stripes of cloud over snow in swaths were snow was observed. Cryospheric level-2 algorithms use the cloud mask product thus stripes of cloud appeared over snow cover.

The number of MODIS Aqua non-functional detectors will not decrease, in fact the number may increase over the life of the sensor. An alternative to use of band 6 (1.6 um) for snow mapping and snow/cloud discrimination is the use of band 7 (2.1um). Snow has similar reflectance characteristics in bands 6 and 7 relative to other bands but reflectance in band 7 is lower than in band 6. That difference affects the NDSI by increasing the NDSI values of snow when band 7 is used. The revised cloud mask algorithm uses band 7 instead of band 6 for snow/cloud discrimination with reasonable results for purposes of cloud detection. Analysis of snow mapping using band 7 as a replacement for band 6 in the cryospheric algorithms is in progress. Of concern is that the Aqua cryospheric algorithms using band 7 will have the same quality and consistency that has been achieved with the Terra cryospheric algorithms and products. Revised Aqua cyrospheric products are expected at some point in Collection 4.