Improving Warm Rainfall Detection And Rainfall Estimation Of A Multiple Satellite Based Rainfall Retrieval Algorithm PDF Download

In respond to the community demands, combining microwave (MW) and infrared (IR) estimates of precipitation has been an active area of research since past two decades. The anticipated launching of NASA's Global Precipitation Measurement (GPM) mission and the increasing number of spectral bands in recently launched geostationary platforms will provide greater opportunities for investigating new approaches to combine multi-source information towards improved global high resolution precipitation retrievals. After years of the communities' efforts the limitations of the existing techniques are: (1) Drawbacks of IR-only techniques to capture warm rainfall and screen out no-rain thin cirrus clouds; (2) Grid-box- only dependency of many algorithms with not much effort to capture the cloud textures whether in local or cloud patch scale; (3) Assumption of indirect relationship between rain rate and cloud-top temperature that force high intensity precipitation to any cold cloud; (4) Neglecting the dynamics and evolution of cloud in time; (5) Inconsistent combination of MW and IR-based precipitation estimations due to the combination strategies and as a result of above described shortcomings. This PhD dissertation attempts to improve the combination of data from Geostationary Earth Orbit (GEO) and Low-Earth Orbit (LEO) satellites in manners that will allow consistent high resolution integration of the more accurate precipitation estimates, xxii directly observed through LEO's PMW sensors, into the short-term cloud evolution process, which can be inferred from GEO images. A set of novel approaches are introduced to cope with the listed limitations and is consist of the following four consecutive components: (1) starting with the GEO part and by using an artificial-neural network based method it is demonstrated that inclusion of multi-spectral data can ameliorate existing problems associated with IR-only precipitating retrievals; (2) through development of Precipitation Estimation from Remotely Sensed Information using Artificial Neural Network - Multi-Spectral Analysis (PERSIANN-MSA) the effectiveness of using multi-spectral data for precipitation estimation are examined. In comparison to the use of a single thermal infrared channel, using multi-spectral data has a potential to significantly improve rain detection and estimation skills; (3) a method proposed to integrate the previously developed cloud classification system (PERSIANN CCS) with PERSIANN-MSA. Through the integration, PERSIANN-MSA benefits from both cloud-patch classification capability as well as multi-spectral information to culminate the GEO-based precipitation estimation techniques; (4) finally, a new combination technique that incorporates multi-sensor information is developed. The technique is called REFAME, short for Rain Estimation using Forward Adjusted advection of Microwave Estimates. REFAME allows more consistent integration of MW VIS/IR information through hybrid advection and adjustment of MW precipitation rate along cloud motion streamlines obtained from a 2D cloud tracking algorithm using successive GEO/IR images. Evaluated over a range of spatial and temporal scales it is demonstrated that REFAME is a robust technique for real-time high resolution precipitation estimation using multi-satellite information.