Evaluation of a new technique to combined microwave and infrared satellite data for estimation of small-scale rainfall over the global tropics and subtropics

There is increasing demand from the hydrological and climate communities for improved spatial and temporal resolution of rainfall estimates. However, accurate estimation of rainfall at small spatial and temporal scales (1 degree/daily or less) presents problems for satellite based methods. Passive microwave (MW) and active microwave (rain radar) algorithms are able to provide accurate estimation of instantaneous rain rates, but the poor temporal sampling of low earth orbiting satellites makes them most suitable for estimation over longer periods of perhaps a month or more. In contrast, infrared (IR) satellite algorithms benefit from the high temporal sampling of geostationary satellites, but are limited by the weak relationship between IR cloud top radiance and surface rainfall.

Here, we introduce a new algorithm that combines satellite MW data (SSM/I and TRMM) with geostationary IR data. Rainfall estimates over the global tropics and subtropics (40N-40S) are produced at the high temporal frequency of IR data using rainfall information derived from microwave data. The technique uses coincident instantaneous MW and IR observations to provide a spatially and temporally variable calibration of optimum IR threshold temperatures and IR/rain rate relationships. Using data sets of varying spatial and temporal resolution we establish the minimum space/time domains within which calibration can be undertaken. Results show significant variations in IR optimum thresholds and rain rate relationships over space and time, providing a valuable insight into the nature of rainfall processes. Initial results using TRMM microwave imager and precipitation radar data are particularly revealing in this regard. An evaluation of the broad scale structure of cloud/rainfall regimes is presented.

An extensive validation of estimates at temporal scales ranging from instantaneous to monthly is provided. A number of MW algorithms are compared. The new method provides the basis for creation of a 3 hourly global rainfall database, with spatial resolution down to 30km, from 1987 to present. This represents a potentially important new resource for climate studies.