Correcting position error in rainfall estimates using temporal and spatial warping

An increasing number of satellite-based rainfall estimates, with ever finer resolution, are becoming available. They are particularly valuable in regions with sparse radar and gauge networks. For example, in most of sub-Saharan Africa, the gauge network is not dense enough to represent the high variability of the rainfall during the monsoon season. However, satellite-based estimates can be subject to errors in position and/or timing of the rainfall events, in addition to errors in the intensity.
Many satellite-based estimates use gauge measurements for bias correction. Bias correction methods focus on the intensity errors, and do not correct the position error explicitly. We propose to gauge-adjust the satellite-based estimates with respect to the position and time. We investigate two approaches: spatial and temporal warping. The first one is based on a spatial mapping and correct the spatial position while keeping the time constant. The second uses a temporal mapping and keeps the spatial domain unchanged. The mappings are derived through a fully automatic registration method. That is, only the gauge and satellite-based estimates are needed as inputs. There is no need to manually predefine the rain features.
The spatial and temporal approaches are both applied to a rainfall event during the monsoon season in southern Ghana. The Trans-African Hydro-Meteorological Observatory (TAHMO) gauge network is used to gauge-adjust the IMERG-Late (Integrated Multi-Satellite Retrievals for GPM) satellite-based estimates. The two approaches are evaluated with respect to the timing, the location and the intensity of the rainfall event.