We have produced an uncertainty product for the Special Sensor Microwave Imager Sounder (SSMIS) radiometer data. The goal of this effort was to perform a detailed uncertainty analysis on SSM/I data and to add uncertainty estimates to each and every measurement of surface wind speed, columnar atmospheric water vapor, columnar cloud liquid water, and rain rate. This uncertainty information is available as an additional product for download and use with the SSMIS bytemap files. At this time, only F17 uncertainty files are produced.
The RSS F17 SSMIS Daily data files now have a companion file that contains a quantitative short-term uncertainty for each radiometer ocean measurement. The uncertainty file is the same format as the RSS F17 bytemap files, except that instead of containing values for wind/vapor/cloud and rain, the file contains uncertainties values for each of these ocean parameters. Users seeking uncertainty information should download the uncertainty file in addition to the F17 data file.
Uncertainty Data File Description
The binary uncertainty data files contain estimated input-induced uncertainty for each F17 SSM/I geophysical product. At this time, we only provide files for the F17 SSMIS Daily data. No uncertainty estimate data files are available for the other SSM/I or SSMIS, or for AMSR-E, WindSat, or the time-averaged (3-day, weekly and monthly) data files.
F17 is an operational instrument for which we produce daily geophysical retrieval files in near-real time. However, due to processing constraints, we will run the uncertainty program monthly, providing one month batches of associated uncertainty files by the 10th day of the following month (April data files available by May 10th).
A paper describing the uncertainty analysis is under review at Journal of Geophysical Research, Atmospheres.
Using the uncertainty estimates
The values in the uncertainty files are the 1-sigma input-induced uncertainty determined by propagating estimated uncertainties in the input parameters (brightness temperatures, SST, wind direction, and earth incidence angle) through our retrieval algorithm. This type of uncertainty does not contain uncertainty from other sources, such as algorithm errors or differences due to spatial-temporal mismatch between the satellite measurements. Because of this, these uncertainty estimates likely underestimate the true uncertainty by a substantial amount. That said, these maps of input-induced uncertainty do contain useful information, including information about regions where the uncertainty is likely to be larger than typical values due to the presence of rain, thick clouds, or unfavorable wind direction.
In general, larger input-induced uncertainties occur when the geophysical parameters are relatively poorly constrained by the measurements. These conditions are also very likely to lead to larger uncertainty caused by algorithm errors. Thus the input-induced uncertainty maps contain information about possible algorithm errors. Therefore, we make the following recommendation: to obtain an estimate of the total uncertainty, scale the input-induced uncertainties in these files by a factor of 2.0. This is obviously an ad-hoc action that is unsupported by rigorous analysis, but it does result in an uncertainty estimate reflecting the likely geographical distribution of uncertainty, and results in a value consistent with the total uncertainty found by intercomparing our satellite data with other measurements of the same parameter, and then sharing the between-measurement difference variance approximately equally between the two sources.
Gridded Binary Uncertainty Files
The daily uncertainty files for F17 SSMIS contain uncertainty estimates mapped to a regular grid complete with data gaps between orbits, and exactly correspond to the daily retrieval maps for F17. The uncertainty files are meant to be used in conjunction with the F17 geophysical retrieval files. Each binary uncertainty data file has the same file structure consisting of 10 0.25 x 0.25 degree grid (1440,720) byte maps. Five local morning (descending) uncertainty maps in the following order, Time (T), 10 meter Surface Wind Speed (W), Atmospheric Water Vapor (V), Cloud Liquid Water (L), and Rain Rate (R), are followed by five local evening (ascending) maps in the same order. The gridded data are assembled from swath data using the exact same methods used to create the gridded geophysical retrieval files. The use of two maps per day, one morning and one evening passes, prevents overlap or averaging of data from greatly varying times of day. However, data mapped onto each of the maps are overwritten at high latitudes where successive orbits cross and at the "seam" or region where the last orbit of the day overlaps the first orbit of the day. The files contain time values of the data for that cell representing the time in minutes UTC or fractional hour of day.
Table 1: RSS SSMIS Uncertainty Product details. Measurements are provided in the order contained in the binary data files.
|Description||Scale||Valid Data Range|
|TIME||Time||Minutes since midnight GMT
Fractional hour of day GMT
|0 to 1440
0.0 to 24.0
|Uncertainty in 10-m surface wind speed||Input-induced uncertainty in the 10-meter surface wind speed||0.01||0. to 2.5 m/s|
|VAPOR Uncertainty||Uncertainty in columnar atmospheric water vapor||Input-induced uncertainty in total gaseous water contained in a vertical column of the atmosphere||0.01||0. to 2.5 mm
|CLOUD Uncertainty||Uncertaintly in columnar cloud liquid water||Input-induced uncertainty in total cloud liquid water contained in a vertical column of the atmosphere||0.001||0. to 0.25 mm|
|RAIN Uncertainty||Uncertainty in rain rate||Input-induced uncertainty in rate of liquid water precipitation||0.002||0. to 0.5 mm/hr|
The data values between 0 and 250 need to be scaled to obtain meaningful geophysical data. To scale the data, multiply by the scale factors listed in the table above. No offsets are used. Note that the scale factors differ between the uncertainty data and the geophysical retrieval data. The read routines provided in the support directory scale the data and return real values. Gridded binary data values above 250 have been reserved for the following:
|0 to 250||=||valid geophysical data|
|251||=||missing wind speed due to rain,
missing water vapor due to heavy rain
|253||=||observations exist, but are bad (not used in composite maps)|
The daily uncertainty maps are stored in appropriate year and month sub-directories. The file names have the following naming conventions:
Where yyyy, mm, dd, and vv stand for:
|yyyy||year||1998, 1999, 2000, etc.|
|mm||month||01 (Jan), 02 (Feb), etc.|
|vv||version||v7 = version 7|
The center of the first cell of the 1440 column and 720 row map is at 0.125 E longitude and -89.875 latitude. The center of the second cell is 0.375 E longitude, -89.875 latitude.
There are gaps within these data. If you select a date for which no data is available, either a list of acceptable dates will appear, or a blank map with text stating "Data not available" will be posted. Please select another date if this occurs.
Uncertainty binary file read routines and verification files are available on our ftp server in the ssmi/ssmi_support directory. The routines exist for IDL, MatLab, Fortran, Python and C++.
Related Data Sets
The uncertainty data files are to be used in conjunction with the RSS F17 SSMIS geophysical retrieval files.
These SSMIS data are produced by Remote Sensing Systems with support from the NASA Earth Science ERRORs Project. The funding supports the analysis of radiometer data quality and distribution of these data for use. We are grateful to NASA for continued support.
How to Cite These Data
Continued production of this data set requires support from NASA. We need you to be sure to cite these data when used in your publications so that we can demonstrate the value of this data set to the scientific community. Please include the following statement in the acknowledgement section of your paper:
"SSMIS data are produced by Remote Sensing Systems and sponsored by the NASA Earth Science ERRORs Project and are available at www.remss.com. "