Key points are not available for this paper at this time.
Study region: The Pearl River Basin in South China. Study focus: Infrared-based satellite rainfall monitoring faces challenges due to limitations in cloud-top observations, as geostationary meteorological satellites cannot directly measure sub-cloud precipitation, leading to uncertainties during heavy rainfall. To address this, we fused FY-4A satellite IR precipitation with gauge observations and ERA5 reanalysis data to improve the detection of heavy rainfall. Multiple fusion approaches are evaluated, from basic statistical correction to advanced spatial models, to quantify their trade-offs across different rainfall intensity levels. An independent validation framework is designed to assess how well these fusion models generalize to unobserved locations. New hydrological insights for the region: The fused dataset demonstrated robust performance, with a Probability of Detection (POD) exceeding 0.90 for general rainfall (>0.1 mm). For heavy rainfall (>50 mm), the POD stabilized at approximately 0.52 in independent tests, slightly outperforming the IMERG product while offering a finer resolution. This improvement stems from the fusion process compensating for the IR saturation effect, enabling the capture of heavy rain bands missed by the raw satellite product. However, performance degraded for highly localized, sub-grid scale storms. This limitation highlights two future needs: developing higher-resolution (1–2 km) satellites over land, and fusing data from multiple polar-orbiting microwave satellites for ungauged oceanic regions.
Wang et al. (Thu,) studied this question.