Bias correction is then performed for the raw CMORPH over the entire data period from 1998 to the present. Over land, the bias in the raw CMORPH is removed by matching the probability density function (PDF) of the CMORPH with that of the CPC unified daily gauge analysis in two sequential steps. Bias in the raw CMORPH is first removed using PDF tables established for each 0.25olat/lon grid box and for each calendar day using co-located CMORPH gauge data pairs collected over a sliding window of +/-7 days centered at the target calendar day over a 13-year period from 1998 to 2011 and over a spatial domain centering at the target grid box. The output of this first step is then calibrated against the daily gauge analysis using PDF tables established using data over a 30-day period ending at the target day to remove year-to-year variations of the CMORPH bias.
The bias-corrected CMORPH precipitation estimates are further combined with the gauge analysis through an optimal interpolation (OI) technique, in which the bias-corrected CMORPH is used as the first guess while the gauge data is used as observations to modify the first guess over regions with station coverage. Cross-validation tests and comparisons against independent gauge observations demonstrate improved quality of the bias corrected CMORPH and the gauge CMORPH merged precipitation analyses. Details about the high resolution precipitation analyses and their applications will be reported at the workshop.