Analysis and comparison of global precipitation datasets with gridded products from passive microwave retrieval algorithms in the GPM era

Precipitation estimation and evaluation of its distribution are key elements in hydrological applications, water management and for improving flood or droughts monitoring. While traditional means of precipitation observation, such as ground-based gauges and radars, are limited in their spatial coverage and are often scarce over some remote regions, satellite-based techniques have allowed precipitation estimation on a global scale with good temporal coverage. The ongoing NASA/JAXA Global Precipitation Measuring mission (GPM) aims to provide instantaneous precipitation estimations with a coverage of less than 1 h over 60 % of the globe and less than 3 h over 80 % of the globe. Therefore the exploitation of the complete GPM constellation of passive microwave (PMW) radiometers is a crucial need. In this context, we have developed two different passive microwave precipitation retrieval algorithms: one is the Cloud Dynamics Radiation Database algorithm (CDRD), a physically ¬based Bayesian algorithm for conically scanning radiometers (i.e., DMSP SSMIS, GMI); the other one is the Passive microwave Neural network Precipitation Retrieval (PNPR) algorithm for cross¬-track scanning radiometers (i.e., NOAA and MetOp¬A/B AMSU-¬A/MHS, and NPP Suomi ATMS). The algorithms, originally created for applications over Europe and the Mediterranean basin and recently modified and extended to Africa and Southern Atlantic for application to the MSG full disk area, provide instantaneous precipitation rates. These algorithms are used in this study to show the potentials and limitations of PMW products to effectively monitor precipitation combining all observations available from cross-track and conical scanning radiometers orbiting around the Earth. In this work we show the analysis of surface precipitation at 0.25° x 0.25° resolution at different time scales for the period 2011-2014. Gridded data are produced by means of bilinear interpolation of the orbital instantaneous precipitation retrievals from each algorithm. A final gridded PMW precipitation product is also obtained by merging retrievals from all available radiometers. The evaluation of consistency of the precipitation patterns and amounts obtained from the two different algorithms has shown a good level of coherence of daily/monthly/seasonal means. Quantitative comparisons with other global gridded dataset of monthly/daily precipitation (single and multiple satellites products, and global raingauge datasets) will be showns. The analysis is carried out in relation to the number of daily overpasses available from each sensor, and to the climatology and environmental conditions in each location.