Precipitation Data Sets: Overview & Comparison table

Precipitation (rain, snow, hail, ...) is one of the key components of the hydrological cycle. Its societal importance cannot be over stated. For climate research, precipitation is one of the key terms for balancing the energy budget, and one of the most challenging aspects of climate modeling. Hence, high quality estimates of precipitation's distribution, amounts and intensity are essential. However, the very nature of precipitation and the limitations of the observing system make the quantification of precipitation challenging.  Unlike (say) temperature which has a high degree of spatial and temporal correlation, precipitation can be fractal in space and discontinuous in time. Further, regional variations in topography can affect precipitation amounts significantly. Most precipitation data sets may be categorized into one of three broad categories: gauge data sets (e.g. CRU TS, GPCC, APHRODITE, PREC/L), satellite-only data sets (e.g., CHOMPS) and merged satellite-gauge products (e.g. GPCP, CMAP, TRMM 3B42). 

Even with the near-global coverage of satellites, most satellites fly over a region only twice per day, potentially missing precipitation events. For this reason, many data sets combine observations from multiple satellite platforms that carry passive microwave and/or infrared instruments.  Infrared sensors are used to estimate cloud top temperatures, which must be calibrated to some other precipitation estimate. The microwave-based algorithms derive the precipitation signal from both scattering and emission but only the scattering signal is useful over land because of strong variations in surface emissivity that distort the emission. Microwave-only products include CHOMPS and the SSM/I datasets; PERSIANN and CMORPH combine information from passive microwave and infrared.

Climate-quality, gauge-based data sets can be difficult to construct due to the widely distributed and heterogeneous nature of the source data. Moreover, wind and evaporation effects on the gauge measurments, typically resulting in under-catch, need to be considered.

Despite many advances in observing systems and algorithms over the years, validating global precipitation observations, for example through the energy budget approach, has proven challenging. Adler et al (2012) estimate the global (ocean+land)-mean precipitation as 2.6 ± 7% mm/day. Over oceans only,Behrangi et al (2014),  'estimate for 3-yr (2007–09) near-global (80S-80N) oceanic mean precipitation rate is ~2.94 mm/day which is about 9% than the CPC/CMAP rate (2.68 mm/day) and 4% higher than the GPCP rate (2.82 mm/day)'.