A spatio-temporal correlation technique has been developed to combine satellite rainfall measurements using the spatial and temporal correlation of the rainfall fields to overcome problems of sparse and infrequent measurements, while at the same time accounting for the measurements' accuracies. This technique estimates instantaneous rainfall with desired temporal sampling using only currently available satellite measurements with the goal of estimating 3-hour total rainfall accumulations at various spatial scales.

The technique uses weighted mean to combine the measurements, adjusting the weights to the temporal correlation length of the measured rainfall field, and to the instrument accuracies. The relationship between the temporal and spatial correlation of the rainfall field is exploited to provide information about rainfall beyond instantaneous measurements. This information, depending on the nature of the rainfall field, can be accurate for prolonged time periods. It is shown that slow changing rainfall fields (i.e. stratiform-like rain) have high values of spatial correlation coefficients, and temporal correlation lengths as long as 60min. While, on the other hand, fast changing rainfall fields (i.e. convective-like rain) tend to have low spatial correlations, and temporal correlation lengths as short as 20min.

This technique is developed using synthetic radar data. Nine months of the Operational Program for the Exchange of weather RAdar (OPERA) data is used on grid sizes of 100km, 250km and 500km with pixel resolutions of 8km, 12km and 24km to simulate satellite FOVs, and then applied to the real satellite data over the Southwest region of USA to calculate 3-hour rainfall accumulations. The results are then compared to the simple averaging technique, which takes a simple mean of the measurements as a constant rainfall rate over the entire accumulation period. The comparison is presented as improvements of the total absolute and RMS errors. Using synthetic data, depending on the time separation of the measurements and their accuracy, the technique has shown the potential to bring improvements of up to 40% in absolute, and up to 25% in RMS error. When applied to the real satellite data over the SE-USA, the technique has shown less skill, only 2% to 6% error improvement, which can be explained by the poor temporal sampling of the reference measurements.

This technique is computationally inexpensive and easily applicable to currently used rainfall accumulation methods with linear interpolation between measurements such as CMORPH (Climate Prediction Center's Morphing Technique) and TMPA (The Tropical Rainfall Measuring Mission Multi-satellite Precipitation Analysis).