Intercomparison of satellite-based and model-based rainfall analyses

Abstract

Four satellite rain estimations based on microwave (MW), infrared (IR) or combined MW-IR techniques are compared with the BOlogna Limited Area Model (BOLAM) rain forecast for a severe weather event (8–13 November 2001) over the western Mediterranean Sea.

Two of the investigated multi-channel MW rainfall algorithms use data from the Tropical Rainfall Measuring Mission (TRMM). The Frequency Difference Algorithm relies on data from the TRMM Microwave Imager (TMI) and the other one combines data from the Precipitation Radar (PR) with those from the nine-channel radiometer TMI, called PR Adjusted TMI Estimations of Rainfall (PATER) algorithm. The pure IR Rain Estimator uses geostationary IR METEOSAT data and the combined Naval Research Laboratory algorithm uses both MW data from low orbiting satellites and IR data from the geostationary orbit.

Validation results, computed over a common grid, which is independent of the different field of view sizes of the applied data sets, indicate that there is generally a better performance for heavy rain (> 6 mm h⁻¹) than for light rain (<1 mm h⁻¹). Both MW algorithms perform rather similarly, although PATER shows some rain detection problems due to thick aerosol loads originating from the desert. The BOLAM model presents a good agreement with the MW and only a minor location error of a heavy rain area was detected. Both IR-based algorithms have problems in identifying the correct rainy areas compared to MW. Overall, the results suggest that there are advantages in combining both techniques – the well-known rain physics of the MW channels with the high temporal resolution of IR algorithms – to retrieve precipitation from satellite data.