Forward morphing of passive microwave derived precipitation field with adjusted intensity from GOES information

The anticipated launching of the global precipitation measurement (GPM) mission and the increasing number of available spectral bands from recently launched (e.g, MSG-SEVIRI) and future (e.g, GOES-R-ABI) sensors, will provide greater opportunities for investigating new approaches to combine multi-source information towards improved consistency, accuracy, coverage, and timeliness of high resolution precipitation. The above-mentioned characteristics are important to many hydrologic applications including flood forecasting. The main challenge is how to benefit from the strengths of different types of satellite sensors while minimizing the impacts of their limitations.

A new technique for half-hourly global near real time precipitation estimation is proposed in which relatively high quality precipitation estimates derived from passive microwave (PMW) sensors are both propagated and modified using cloud motion vectors and cloud type classes obtained from high-frequency geostationary infrared images. The morphing procedure involves three major steps: (a) Infrared images are first used to classify clouds into a number of predefined classes using extracted cloud features such as brightness temperature and its gradients both in time and space to consider the dynamic evolution of cloud systems in time and space, (b) For each class, the mean value of observed precipitation rates (MPR) is calculated, (c) In a near-real time process, MPR in conjunction with high resolution- high-frequency infrared-derived cloud motion vectors are used to dynamically modify (morph) the propagated PMW-derived precipitation intensities every 30 minutes.

The proposed method was employed to derive fine scale (0.08°x0.08° lat/long every 30 min) precipitation intensity over the conterminous United States. The results demonstrate significant improvement in forward morphing of PMW precipitation intensities. The observed improvements, which depend on time-distance from previous microwave overpass, can reach to more than 100 % over both simple forward morphing along the motion vectors as well as over simple averaging of the PMW estimates of intensities. More detailed comparative statistics in addition to practical ways of supplementing the proposed method with multi-spectral information and geostationary-derived rain estimate will be discussed in the presentation.