This Level-3 Equal-Area Scalable Earth-Grid (EASE-Grid) Brightness Temperature data set, collected since 09 July 1987, is a part of the NOAA/NASA Pathfinder Program. The data set consists of gridded data from the Special Sensor Microwave/Imager (SSM/I) and the Special Sensor Microwave Imager/Sounder (SSMIS) in three equal-area projections: Northern Hemisphere, Southern Hemisphere, and full global. The data gridding technique maximizes the radiometric integrity of the original brightness temperature values, maintains high spatial and temporal precision, and involves no averaging of original swath data. Spatial coverage is global and a spatial resolution of 25 km is available for all channels. Channels at 85 and 91 GHz are also provided at 12.5 km. For a given projection, there are 18 brightness temperature files per day and two corresponding time files. Data are contained in flat binary files and are available via FTP as processing is completed.
DMSP SSM/I-SSMIS Pathfinder Daily EASE-Grid Brightness Temperatures, Version 2
|Temporal Resolution:||1 day|
|Platform(s)||DMSP, DMSP 5D-2/F11, DMSP 5D-2/F13, DMSP 5D-2/F8, DMSP 5D-3/F17|
|Data Contributor(s):||Richard Armstrong, Mary J. Brodzik|
|Metadata XML:||View Metadata Record|
As a condition of using these data, you must cite the use of this data set using the following citation. For more information, see our Use and Copyright Web page.Armstrong, R., K. Knowles, M. Brodzik, and M. A. Hardman. 1994, updated 2016. DMSP SSM/I-SSMIS Pathfinder Daily EASE-Grid Brightness Temperatures, Version 2. [Indicate subset used]. Boulder, Colorado USA. NASA National Snow and Ice Data Center Distributed Active Archive Center. doi: http://dx.doi.org/10.5067/3EX2U1DV3434. [Date Accessed].
Detailed Data Description
The Level-3 Equal-Area Scalable Earth-Grid (EASE-Grid) Brightness Temperature data set is collected as part of the NOAA/NASA Pathfinder Program. The data set consists of gridded data from the Special Sensor Microwave/Imager (SSM/I) and the Special Sensor Microwave Imager/Sounder (SSMIS) in three equal-area projections: Northern Hemisphere, Southern Hemisphere, and full global. The data gridding technique maximizes the radiometric integrity of the original brightness temperature values, maintains high spatial and temporal precision, and involves no averaging of original swath data. The spatial coverage is global and begins 09 July 1987; processing is ongoing. The spatial resolution is 25 km for all channels; the 85 and 91 GHz channels are also provided at a 12.5 km resolution. There are 18 brightness temperature files per day for a given projection and two corresponding time files. Data are contained in flat binary files and are available via FTP as processing is completed.
Note: The data format information in this document represents the data in its native format as it is archived at NSIDC. If you have downloaded the data using Polaris, please consult the 00README file located in the tar file for information on the data format operations that were performed on this data set.
Brightness Temperature Files
Brightness temperature data are contained in flat binary files (little-endian) with one grid per file consisting of 2-byte integer arrays (721x721) of brightness temperatures in tenths of kelvins.
Each brightness temperature file represents gridded data for a single channel and polarization; they are derived from either ascending or descending orbits (for example, 37 GHz, horizontal, ascending) for one day. There are 18 brightness temperature files per day for each projection.
Time Series Files
Time series data are contained in 1-byte, unsigned integer arrays consisting of Coordinated Universal Time (UTC) in tenths of hours. Each time series file represents the corresponding time of the swath sample used for the interpolation of the given grid cell, for either ascending or descending orbits for that day. There are two time files per day (ascending and descending passes) for a given projection, both at a 25 km resolution.
Beginning with F17 files, time data are contained in 2-byte, signed integer arrays (721x721). Time data are minutes since 00:00 Coordinated Universal Time (UTC), or midnight, of the date of the enclosing file. Each time file represents the corresponding time of the swath sample used for the interpolation of the given grid cell, for either ascending or descending orbits for that day. There are two time files per day (ascending and descending passes) for a given projection, both at 25 km resolution.
Data on the FTP site are divided into three subdirectories,
global, north, and
south. Each of these directories is further subdivided into directories labeled for each year, from 1987 to the most current year's processing. Figure 1 displays the directory structure.
Brightness Temperature Files
This section explains the file naming convention used for the brightness temperature files with an example.
Example File Name:
Refer to Table 1 for the valid values for the file name variables listed above.
||DMSP platform ID (
||EASE-Grid ID (
||3-digit day of year|
||Direction of pass (
||Data version number (example:
||Channel (GHz) and polarization (
||Identifies this as a gzipped file|
Time Series Files
This section explains the file naming convention used for the time series files with an example.
Example File Name:
Refer to Table 2 for the valid values for the file name variables listed above.
||DMSP platform ID (
||EASE-Grid ID (
||3-digit day of year|
||Direction of passes (
||Data version number (example:
||indicates the contents are time data|
||Identifies this as a gzipped file|
Gzipped brightness temperature data files range in size from 25 KB to 3.4 MB, and gzipped time files range in size from 1.1 KB to 36 KB.
These data files are provided in three different equal-area, spatial coverages: Northern Hemisphere azimuthal, Southern Hemisphere azimuthal, and Global cylindrical. Refer to Figures 2, 3, and 4 for maps showing he three different coverages. Refer to the Grid Extent Table on the EASE-Grid: A Versatile Set of Equal-Area Projections and Grids Web page for specific latitude and longitude values.
The spatial resolution for the 19 GHz, 22 GHz, and 37 GHz channels is 25 km. There are two different spatial resolutions for the 85 GHz and 91 GHz channels: 25 km and 12.5 km.
The SSM/I EASE-Grids are a set of three equal-area projections: two azimuthal equal-area projections, one for the Northern and one for the Southern Hemisphere; and a Global cylindrical equal-area projection. Refer to the EASE-Grid: A Versatile Set of Equal-Area Projections and Grids Web page for more information on the EASE-Grid.
The EASE-Grid dimensions are 721 columns by 721 rows. For more details on the EASE-Grid, please refer to the Versions page.
Coverage begins 09 July 1987 and is ongoing. Refer to the SSM/I-SSMIS Data Availability Web page for specific dates.
The goal of the SSM/I Pathfinder product team is to produce a continuous time series of SSM/I data using a single consistent processing and interpolation scheme. Data are made available on the FTP site as processing is completed. For a complete list of missing dates, refer to the SSM/I-SSMIS Data Availability Web page. A brief list of some of the larger gaps in the data are explained in Table 3.
85 GHz channel April 2008 to present
Beginning with data in the late spring of 2008, the 85 GHz horizontal data and some Southern Hemisphere 85 GHz vertical data files contain zeros for the brightness temperatures. This means that either no brightness temperatures were measured or the brightness temperatures failed quality control procedures from Remote Sensing Systems that NSIDC uses in our preprocessing. The affected periods of 85 GHz horizontal files are shown in Table 3.
|Region||Pass||Polarization||Day of Year (ddd)|
|Northern||Descending||Horizontal||118 - 128|
|Southern||Ascending||Horizontal||123 - 127|
|Southern||Descending||Horizontal||114 - 116, 118, 120, 123 - 182|
|Global||Ascending||Horizontal||128 - 182|
|Global||Descending||Horizontal||117 - 127, 129 - 182|
|Global||Descending||Vertical||173, 175, 176, 179-182|
Alaska and Canadian Prairies, 1994 to May 1995
Substantial amounts of swath data over Alaska and the Canadian Prairies are missing beginning early in 1994 until May 1995. During this period, the data tape recorder on the DMSP-F11 failed. As a result, it was necessary to download data to ground stations more frequently than usual. Data download and acquisition could not occur simultaneously, consequently data gaps exist in the EASE-Grid data for Alaska and the Canadian Prairies from early 1994 until data acquisition by the DMSP-F13 SSM/I began in May 1995.
There are usually 18 brightness temperature files per day for a given projection, except from 01 February 1989 through 31 December 1991. For this period, no data exist for the 85 GHz channel due to degradation of this channel after heating cycles during Northern Hemisphere winters that resulted in increased solar illumination on the SSM/I instrument (Wentz 1992). Only 10 brightness temperature files are available per day during these dates, and all files are in the 25 km resolution grid.
The data are daily, separated by ascending and descending passes.
The parameter of this data set is brightness temperature.
Theoretically, brightness temperature is the effective temperature of a blackbody radiating the same amount of energy per unit area at the same wavelengths as the observed body. Empirically, brightness temperature is the apparent radiant temperature of a non-blackbody determined by measurement with an optical pyrometer or radiometer. The brightness temperature (Tb) at a given wavelength (λ) is the product of the physical temperature (Tp) and the emissivity (ε) at the given wavelength of the surface viewed by the radiometer. Refer to Equation 1.
Tb(λ) = ε(λ)Tp (Equation 1)
Equation 1 is the Rayleigh-Jean approximation of Plank's law for the passive microwave region of the electromagnetic spectrum. It is an approximation and does not take into account effects of the atmosphere on the microwave radiation.
Brightness temperature data values are scaled by 10; divide the stored values by 10 to get kelvins. Values range from 550 (representing 55.0 K) to 3200 (representing 320.0 K); missing data are indicated by the value 0.
Time values range from 0 (representing 0000 UTC or midnight) to 239 hours (representing 23:54 UTC or 23.9 hours); missing data are indicated by the value 255. Beginning with F17 time files, the missing data value is the minimum 2-byte integer, or -32768. The legitimate, non-missing value of 0 is midnight of the enclosing day, and a non-missing value of 1439 is one minute prior to midnight the following day.
Latitude and longitude values vary depending upon the grid used. Values are in decimal degrees scaled by 100000; divide the stored values by 100000 to get actual values. Latitude values range from -9000000 to 9000000, and longitude values range from -18000000 to 18000000. Missing data values are indicated by the value 1431655765.
Table 4 summarizes these data ranges.
|Parameter||Unit of Measurement||Data Range||Missing Data Value|
|Brightness Temperatures (Tb)||Tenths of kelvins||550 (representing 55.0 K) to 3200 (representing 320.0 K)||0|
|Time||Tenths of hours since midnight of the date of the enclosing file||0 (representing 00:00 UTC or midnight) to 239 (representing 23:54 UTC or 23.9 hours)||255|
|Time (F17 files only)||Minutes since midnight of the date of the enclosing file||0 (representing 00:00 UTC or midnight) to 1439 (representing 23:59 UTC||-32768|
|Latitude/Longitude||Hundred thousandths of degrees (1 meter precision)||Latitude values range from -9000000 to 9000000 (representing -90 to 90))
Longitude values range from -18000000 to 18000000 (representing -180 to 180)
Figure 5 shows a sample browse image for 01 January 2008 for the Northern Hemisphere at 25 km, 37 GHz.
Measurement Error for Parameters
Geolocation errors in input Remote Sensing Systems (RSS) swath data are no more than 10 km, "although there may be exceptions" (Sharon Tremble, RSS, e-mail to M. J. Brodzik, 18 January 1996). Additional error introduced by nearest-neighbor interpolation from the over-sampled array is approximately 6 km for the 25 km grids and 3 km for the 12.5 km grids.
Each brightness temperature and time file is visually inspected by data center operators before being archived and distributed.
Software and Tools
Geolocation tools for this data set are available via the EASE-Grid Data Geolocation Tools Web page.
The approximate yearly volume of this data is 6.2 GB
Data Acquisition and Processing
The SSM/I and SSMIS instruments measure passive microwave radiances. For a detailed description of how SSM/I obtains its measurements, please see NSIDC's SSM/I Instrument Description Web page for information. For more information about the SSMIS instrument, please see the SSMIS Instrument Description Web page.
For a detailed discussion of the theory behind the acquisition methods used here, please refer to the following articles:
Poe, G. A. 1990. Optimum Interpolation of Imaging Microwave Radiometer Data. IEEE Transactions on Geoscience and Remote Sensing 28(5):800-810.
Galantowicz, J. F. and A. W. England. 1991. The Michigan Earth Grid: Description, Registration Method for SSM/I Data and Derivative Map Projections. Technical Report 027396-2-T. Radiation Laboratory Dept. of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor.
The input orbital brightness temperature data for this product was ingested via the RSS software programs DECODE and QUAL1 (Wentz 1993). Switches were used to include the following options:
- Along-scan bias corrections were turned on.
- Sensor intercalibrations (adjusting non-F8 SSM/I antenna temperatures to correspond to F8 SSM/I antenna temperatures) were turned off.
- Data values during times that RSS has determined to be "periods of erroneous data," and identified in RSS files BADLOC.D08, BADLOC.D11, BADLOC.D13, were discarded prior to gridding and interpolation.
- Groups of scans that RSS identified as corrupted by calibration error problems between 09 October 1990 and 29 August 1992 and identified in RSS files BADCAL.D08 and BADCAL.D11, were discarded prior to gridding and interpolation.
- After 29 August 1992 calibration errors detected in the RSS DECODE processing were discarded prior to gridding and interpolation.
Over the course of a day, points at mid-to-high latitudes will be observed from multiple orbits. For a given grid cell location, only observations from a single orbit were used in the Backus-Gilbert interpolation. In order to ensure the most consistent local observation time at each location, we chose the sample from the orbit whose local time was closest to the equatorial crossing times shown in Table 5. Actual equatorial crossing times do change slowly over the life of sun-synchronous satellites such as the DMSPs. The choice of these particular times was based on nominal crossing times at launch.
The antenna pattern coefficients used in the Backus-Gilbert interpolation were provided to NSIDC by John Galantowicz (Galantowicz 1995, Appendix C). The interpolated brightness temperatures represent optimally filtered data, that is, they represent what the sensor would have measured had it been directed at the center of the fixed grid cell. Depending on one's purposes, the Backus-Gilbert method can be tuned to enhance resolution, or reduce noise, but both cannot be achieved simultaneously. Galantowicz chose to generate coefficients that would minimize noise (ones that produced a pattern with lowest relative side-lobes). Refer to Figures 6 and 7.
The 19 GHz pattern has the largest footprint of the four SSM/I frequencies, and the interpolated patterns of the other channels are able to fit it without high side-lobes. If the 19 GHz pattern were interpolated to a significantly smaller desired pattern--that is, either the 37 GHz or 85 GHz pattern--then the best achievable interpolated pattern would be distorted and have high side-lobe levels (Galantowicz 1995).
The resulting EASE-Grid brightness temperatures for all channels in the 25 km grids represent the Effective Field Of View (EFOV) at -3 db of the 19 GHz vertically polarized channel, and the EASE-Grid brightness temperatures for 85 GHz channels in the 12.5 km grids represent the EFOV of the 85 GHz vertically polarized channel. Users of the SSM/I EASE-Grid data can make inter-channel comparisons and develop geophysical algorithms based on the assumption that the gridded data represent the brightness of the same geographical area. For further details, refer to Galantowicz (1995), Galantowicz and England (1991) , and Poe (1990).
The NOAA/NASA Pathfinder Program is designed to provide scientists with time series of global-scale remote sensing data ahead of the EOS satellite launches. The Pathfinder concept involves careful reprocessing of existing data sets and then making them readily available as high quality products for global change research. Since the polar regions hold special significance for global change research, the Polar Pathfinders have established a cooperation to maximize the scientific potential of polar data.
Binary data arrays contain spatially interpolated data. The data gridding technique maximizes the radiometric integrity of the original brightness temperature values, maintains high spatial and temporal precision, and involves no averaging of original swath data. Backus-Gilbert optimal interpolation is used to artificially increase, by 16 times, the density of brightness temperature measurements in the satellite swath reference frame with a sample interval of 25 km for 19, 22, and 37 GHz, and 12.5 km for 85 GHz. This process uses actual antenna patterns to create the over-sampled array, and the net effect is as if the additional samples had been made by the satellite radiometer itself because the beam patterns and spatial resolutions of the interpolated data approximate those of the original samples. This method is based on the earlier work of Galantowicz and England (1991) and Poe (1990). The brightness temperature for a given EASE-Grid cell is obtained from the over-sampled array by the nearest neighbor method.
The Backus-Gilbert technique also allows resolution enhancement but with a noise penalty. To avoid the noise penalty, the brightness temperatures for all channels in the 25 km grids were interpolated to the EFOV at -3 db of the 19 GHz vertically polarized channel. Brightness temperatures for 85 GHz channels in the 12.5 km grids were interpolated to the EFOV of the 85 GHz vertically polarized channel. However, beginning with the processing of the DMSP-F17 SSMIS sensor, EASE-Grid brightness temperature fields are gridded using an Inverse Distance Squared method instead of the Backus-Gilbert interpolation that had been used for the earlier sensors. Refer to the Processing Step Change for the DMSP-F17 SSMIS Sensor section of the document for additional information.
As the first step in the processing of Pathfinder data, a common Benchmark Period (April 1987 to November 1988) was chosen to facilitate the analysis and comparison of individual Pathfinder data products. Note that the SSM/I Benchmark Period is somewhat shorter (August 1987 to November 1988) than the Pathfinder Benchmark Period because the data record started with the launch of the DMSP-F8 satellite in July 1987.
Processing of the data has progressed over the course of several years. The goal of the product team has been to produce a consistent time series of continuous, gridded SSM/I data. However, over the course of the operational processing we have chosen at certain times to change some parts of the processing code. We have only made such changes after careful consideration of the consequences to the time series and then, only when the changes will result in an overall better quality time series. The following changes have been made.
- We determined through visual inspections of processed data that some data for 28 June 1989 were mislocated. As a result, we added the following line to our BADLOC.D08 file in order to eliminate the erroneous orbital data:
1989 179 8.0 1989 179 8.2
- Due to an error in the processing code, time data files for the F8 data (August 1987 through December 1991) contain times that were inadvertently truncated to tenths of hours, instead of rounded to the nearest tenth of an hour. This problem was corrected beginning with data for January 1992.
- The treatment of out of bounds input data changed between the F8 and F11 data. For F8 data, input brightness temperatures that were out of bounds (defined to be less than 55.0 K or greater than 320. 0 K) were replaced with the average of the two closest in bound brightness temperatures along the same scan. Beginning with F11 data (January 1992), we eliminated this treatment and set the out of bounds value to an invalid brightness temperature. The Backus-Gilbert interpolation for any EASE-Grid pixel was only performed if both of the following conditions were true:
- The interpolation coefficient corresponding to any missing brightness temperature must have been smaller than a missing coefficient threshold of 0.0005.
- The sum of the interpolation coefficients corresponding to valid brightness temperatures must have been within 1.0000 ± 0.0005 (which is the sum of all 16 coefficients).
- There are no 85 GHz data for 01 February 1989 through 31 December 1991; see related note.
Beginning with processing for the DMSP-F17 SSMIS sensor, EASE-Grid brightness temperature fields are gridded using an inverse distance squared method instead of the Backus-Gilbert interpolation that had been used for the earlier sensors. The Backus-Gilbert method requires analysis of the antenna pattern of the sensor to derive weighting coefficients, yet the required analysis has not yet been performed for SSMIS. Instead, the inverse distance squared method performs a weighted average of swath measurements whose center locations are less than a specified distance from the center of the grid cell. The distance is 1.5 times the nominal cell scale, or ~37.6 km (~18.8) for the 25 km (12.5 km) grids.
Since the two interpolation methods differ, there is a difference in the brightness temperature fields. NSIDC has investigated the difference and has found that most differences are within +/- 1 K, and the vast majority of grid cells have differences within 2 K. However, in regions with steep brightness temperature gradients, the differences can be upwards of 20 K. These regions include:
- coast lines,
- edges of swaths where overlap between swaths occurs, which is most noticeable near the poles (poleward of 60 degrees latitude), and,
- to a lesser degree, in mountainous regions.
These differences may be positive or negative and are one or two grid cells wide.
Users who make use of this extended EASE-Grid brightness temperature time series should be cautious using SSMIS data, particularly in the high difference regions noted above. NSIDC will investigate options for providing fully consistent EASE-Grid brightness temperatures as resources allow.
These data were acquired using the SSM/I instrument on the DMSP-F08, -F11, and -F13 platforms, as well as the SSMIS instrument on the DMSP-F17 platform. For more information about the SSM/I instrument, please refer to the SSM/I Instrument Description Web page. For more information about the SSMIS instrument, refer to the SSMIS Instrument Description Web page.
The source for the raw antenna temperature and brightness temperature data for this data set is Remote Sensing Systems (RSS), Santa Rosa, California Wentz 1993).
Table 6 outlines the processing and version history for this product. Note: Currently there is an overlap in the data for V1 and V2 from 14 December 2006 through 29 April 2009.
|Data Version||Platform||Temporal Range||Source Data Version||Description of Changes|
|V2||F17||14 Dec 2006 — most current processing date||RSS V7||
|V1||F13||03 May 1995 — 31 Dec 2007||RSS V4||N/A|
|V1||F11||03 Dec 1991 — 30 Sep 1995||RSS V3||N/A|
|V1||F8||09 Jul 1987 — 31 Dec 1991||RSS V3||N/A|
|V1||F8||Not available||N/A||Original version of data.
Note: V1 was not indicated in Version 1 file names.
Intercomparison of F13 and F17 Data
NSIDC conducted an intercomparison of F13 with F17 data during the 01 January 2007 through 31 December 2007 period. The vast majority of differences are between 0.5 K and 2 K. Some larger differences, up to 10 K, are found primarily in regions of sharp gradients of brightness temperatures, along coasts and the ice edge, likely due to the changes in geolocation. Smaller biases of 0.5 K to 2 K in some channels, such as 19V, 19H, 22V, and 37H, are likely due to the cross-sensor calibration.
References and Related Publications
Contacts and Acknowledgments
Richard L. Armstrong
CIRES, 449 UCB
University of Colorado
Boulder, CO 80309-0449 USA
Special thanks to the following:
- Dr. Tony England, Dr. John Galantowicz, and Ed Kim for their ongoing research efforts; for consistent support and input during all phases of this project, including prototyping and development; and for producing the Antenna Pattern Correction coefficients used in the interpolation procedure.
- Molly Hardman for software development of the first revisions of the processing software.
- Dr. Ted Habermann for the development of the FREEFORM and makeHDF tool set.
- Ray Kokaly for technical support and comments.
- Ken Knowles for technical support, gridding and overlay software support, and advice and comments.
- Mark Ohrenschall for support with the FREEFORM and makeHDF tool set.
- To members of the product team at the NSIDC DAAC including Renea Ericson, Rachel Hauser, David Hoogstrate, Tracy Thrasher-Hybl, Xiaoming Li, Mike Machowski, Alex Machado, John Maurer, Heidi Schumacher, Annette Varani, I-Pin Wang, Jason Wolfe, Fran Coloma, Dave Korn, Michelle Holm, Jonathan Kovarik, Peter Gibbons, Ann Windnagel, Donna Scott, and Karla LeFevre, whose constant attention to detail and professional energies have made possible the continuing availability of this high-quality data set.
- Annette Varani for documentation support, developing the EASE-Grid Sampler area of the Equinox Sampler, and much patience involved in the packaging artwork.
DOCUMENT CREATION DATE
DOCUMENT REVISION DATE
July 2016 - put into Drupal, made minor updates
NSIDC User Services
Phone: 1 303 492-6199
Fax: 1 303 492-2468
National Snow and Ice Data Center
CIRES, 449 UCB
University of Colorado
Province or State: CO
Postal Code: 80309-0449