Resource Estimate: Generating Polar Stereographic Brightness Temperature Grids for the Polar Regions from Additional SSM/Is
During the PoDAG meeting held in Boulder, Colorado in June 2004, members of PoDAG requested NSIDC conduct an assessment of the resources required to generate polar stereographic brightness temperature grids from additional Defense Meteorological Space Program (DMSP) Special Sensor Microwave/Imager (SSM/I) sources (e.g., DMSP SSM/I F11 orbital data).
NSIDC currently obtains the DMSP SSM/I F13 orbital antenna temperature (TA) files to produce the standard research-quality polar stereographic brightness temperature (TB) and sea ice concentration grids from Remote Sensing Systems (RSS), Incorporated. NSIDC pays RSS $5,000 per year of DMSP SSM/I F13 TA data. We receive four 8mm tapes per year from RSS, each tape containing the last quarter (i.e., three months) of data observed by the SSM/I F13 instrument.
NSIDC recently contacted RSS to obtain a cost estimate for the entire collection of orbital SSM/I TA data. The period of coverage for each SSM/I sensor is listed in Table 1.
Table 1
F08 SSM/I / Jul 1987 / Dec 1991F10 SSM/I / Dec 1990 / Nov 1997
F11 SSM/I / Dec 1991 / May 2000
F13 SSM/I / May 1995 / present
F14 SSM/I / May 1997 / present
F15 SSM/I / Dec 1999 / present
If NSIDC were to obtain the data from RSS in the normal manner in which we obtain the F13 antenna temperature data, the cost would be $5000/year/sensor. RSS quoted NSIDC $140,000 as the cost to acquire the F10, F11, F14, and F15 antenna temperature records, in order to complete SSM/I TA collection at NSIDC. The price estimate from RSS is based on sending the data to NSIDC on 8mm media.
Assume NSIDC intended to extend the polar stereographic brightness temperature grids collection for SSM/I F11. NSIDC currently archives the SSM/I F11 data for the period of record beginning with December 1991 extending through September 1995. The cost to procure the SSM/I F11 TA for the period October 1995 through May 2000 (the end of the F11 data record) is approximately $22,500. The estimated level of effort to store the data into the archive, to generate the brightness temperature, to store the gridded products into the archive, and to generate a CD master is eighteen person-days (~8 hours staff time between ingest of three months of RSS data through the generation of a CD master for the quarter year), which equates to approximately seventy (70) days of elapsed time to complete the end-to-end process for the 18 quarters (October 1995 through May 2000) of the SSM/I F11 data stream.
NSIDC generates 300 CDs for each quarter of the SSM/I polar stereographic brightness temperature grids. Historically, it takes about two days to replicate the data from a CD master onto 300 CD-Rs. The estimate to replicate 300 CDs for each of the remaining quarters of SSM/I F11 time-series is six weeks.
Materials to produce and to package the 300 CDs for each of the quarterly SSM/I CD-R volumes is roughly $300, or approximately $5,400 to package the 18 volumes expected to be generated for the remainder of the SSM/I F11 TA data record. Packaging includes the cost for the CD-R media, label printing supplies, jewel cases, and the jewel case inserts. These costs do not reflect the labor to package the jewel cases, nor the cost to ship the product to users of the SSM/I Brightness Temperature Grids for the Polar Regions time-series. The estimated resources to complete the SSM/I F11 brightness temperature grids collection for the polar regions is $27,900 for data and media, and approximately 306 person-hours including both professional and student labor. Some of the packaging and labor costs could be reduced if the data were distributed on DVD media or made available solely via the NSIDC ftp site.
One issue that should be considered by PoDAG is whether NSIDC should conduct a reprocessing campaign of the standard SSM/I products (polar stereographic and/or EASE grids) now that RSS has performed a consistent inter-sensor calibration for the current suite of SSM/Is.
RSS developed a new calibration algorithm in 2002. The algorithm was based on the microwave sounding unit (MSU) calibration technique. The new algorithm includes “a correction for the radiometer non-linearity, which depends on the hot load temperature… It also depends on orbit position because it appears that some of the SSM/Is experience a slight deviation from a geodetic orbit. The satellite offsets were also adjusted to obtain better inter-satellite agreement. In addition, the orientation of the SSM/I (F08, F13, F14) relative to the spacecraft was slightly adjusted to fix geolocation problems.” RSS released the Version 5 SSM/I TA data in July 2002.
RSS can provide NSIDC with the entire collection (Table 1) of their Version 5 inter-sensor calibrated SSM/I TA data on a 250 GB disk drive for $50,000. Future updates to the F13, F14, and F15 can be received annually on a disk drive for $6,000 per year. The annual update would include one year of data for each of the three SSM/I data streams, which would normally cost NSIDC $15,000 if we were to receive a year of data for each of the three sources on 8mm media. In addition to the data provided by RSS on disk being inter-calibrated, the data on the disk drive is arranged differently than the data received on 8mm. The data on 8mm is stored in files containing four days of data for each source sensor; the data on disk is stored in files containing data for a single orbit for each source sensor.
Should NSIDC investigate the differences in generating brightness temperatures (and sea ice concentrations) using the Version 5 data as input into our processing stream? Are the differences between Version 4 and Version 5 of the SSM/I TA data significant to warrant reprocessing?
Can NSIDC wait to process a year of standard SSM/I products after receiving annual updates from RSS?
Will the near real-time products generated by NSIDC daily using the NOAA-generated orbital brightness temperature products (TDRs) satisfy the needs of the community until an annual update of the standard products are generated using inter-calibrated data from RSS?