Point-to-area validation of passive microwave satellite precipitation with shipboard disdrometers

The high spatial and temporal variability of precipitation poses a substantial challenge to its global monitoring, particularly over the global oceans. The insufficient coverage of surface-based measurements adds to this problem. These data-sparse areas profit from passive microwave (PMW) satellite sensors, which essentially contribute to the estimation of global long-term precipitation estimates. However, the uncertainty of PMW satellite products demands a validation that requires high-quality surfacebased precipitation reference data over the ocean. The Ocean Rainfall And Ice-phase precipitation measurement Network (OceanRAIN) collects high-resolution point-like precipitation data from long-term equipped RVs for the validation of low-resolution PMW satellite precipitation products. Deployed on research vessels (RVs), these optical disdrometers designed for ocean usage are suitable instruments to accurately record oceanic precipitation. This thesis addresses the emerging point-to-area problem and data-imposed validation constraints on the example of collocated PMW-based Hamburg Ocean Atmosphere Parameters and Fluxes from Satellite data (HOAPS) and OceanRAIN data. The thesis presents three essential aspects for representative satellite validation: the precipitation phase (PP) distinction (rain, snow and mixed-phase), the point-to-area precipitation representation and the instrument/algorithm sensitivity. For the PP distinction, a newly developed algorithm is presented that predicts the PP using the air temperature, relative humidity and the 99th percentile of the particle diameter provided by the particle size distribution (PSD) of the disdrometer. Incorporating the PSD mainly adds to the high accuracy of 91% for a rain–snow distinction. However, including the important mixed-phase precipitation yields 81% in accuracy for the PP separation. The point-to-area problem is addressed by applying two statistical adjustments derived from subtropical radar data to ensure that the along-track data represents the areal precipitation distribution in a PMW satellite pixel. The first statistical adjustment uses the average precipitation event duration to correct underrepresented precipitation rates by sampled short events (1–3 min) and overrepresented precipitation rates of long events (5–20 min). The second adjustment takes the median-normalized precipitation rate to mainly correct cases of strong precipitation overrepresentation on the ship track with respect to the area. Both adjustments reduce the sum of squared errors for the track–area precipitation difference by 86%.