AVI 080 Project "ACROSS"
| |
AVI 080 Project "ACROSS" |
|
"Analyzed Climatology of Rainfall Obtained from Satellite and Surface data
for the Mediterranean Basin"
by M.C. Todd, E.C. Barrett and M.J. Beaumont
This report summarises the work carried out by the Centre of Remote Sensing (CRS) - University of Bristol - under task 2.1 - Processing of SMMR images over the established window within the Mediterranean area for the period 1978-1987. The work focused on processing satellite data from the Scanning Multichannel Microwave Radiometer (SMMR) for retrieval of rainfall estimates over the Mediterranean Sea.
The procedure for acquisition and preparation of SMMR satellite data are described in detail by Todd et al. (1996a) in the report for Task 1.3; the development of the techniques for estimation of rainfall isdescribed in detail by Todd et al. (1996b) in the report for Task 1.2. The procedures were conducted as described in these Reports with the exception of further developments and refinements of the techniques as described in the present Report.
The data from the entire SMMR operation period of November 1978 - August 1987 inclusive are made available to the ACROSS project archive, with the exception of a few monthly periods where either no data are available from the US National Space Service Data Center, or Temperature Calibrated Tapes in CRS archive are corrupted.
SMMR data were extracted from the archive for a selected region with boundaries of 12°W and 46°E and 29°N to 48°S. The boundaries of the 37 Ghz Tb scan mean, maximum, minimum and standard deviation required careful “fine-tuning” to ensure that as much corrupt data as possible was identified and removed.
SMMR and SSM/I sensors data are recorded at different spatial resolutions; in addition both sensors have elliptical footprint areas. A reprojection procedure was devised which maintained as far as possible the spatial structure of satellite data. Estimates of rainfall were produced from the raw unprojected, cleaned data files using the algorithm described in this Report, and were reprojected to an equal area UTM (Universal Transverse Mercator) projection with a grid resolution of 25 km. Using the UTM projection artificial artefacts resulting from reprojection and resampling will be minimised and rainfall estimates will be as accurate as possible. For display purposes only the rainfall products were reprojected to a linear latitude / longitude projection with a resolution of 1/16°N: this is used as a temporary format until the format of the final Atlas products is decided.
Todd et al. (1996b) describe the process of algorithm calibration, validation and selection: the report describes the selected H37C algorithm and its performance assessed. However further research resulted in a number of changes to the finding of the referred task 1.2 report. First an error was found in the validation procedure for SSM/I rainfall algorithms: the error relates to the performance of the 37 GHz polarisation difference algorithm calibrated using the histogram optimisation method (PD37B). A validation of the PD37B algorithm rainfall estimates in comparison with coincident estimates for the Meteo-France Nimes radar was conducted and the revised results show that the PD37B algorithm produces rainfall statistics which match those of Meteo-France radar estimates marginally better than the H37C algorithm.
Todd et al. (1996b) describe the correction applied to rainfall estimates to account for bias in the FRONTIERS radar estimation of rainfall. An additional correction was applied to SMMR estimates to account for the effect of “partial beam-filling” by non-uniform rain rates.
Using the PD37 algorithm with the corrections described, rainfall estimate products were produced on a UTM projection then reprojected to a linear latitude / longitude projection for the following quantities and integration periods:
Mean monthly rainfall for some selected months (mm) is illustrated here as a sample product.
Analysis of the full products database, and rainfall anomalies in particular, will be conducted once the processing of SSM/I data is complete.
