Stochastic characterization of space-time precipitation: Implications for remote sensing

In this paper the characterization of the space-time spectral characteristics of tropical rainfall fields are studied directly from measurements from ground sensors and by using three stochastic models. The characterization of the spatial-temporal spectra is important for the design of satellite missions and to determine the bias in ground truth experiments. To carry out the latter a scheme is used to compare contemporaneous measurements of rain rate from a single-field-of-view estimate, based on a satellite remote sensor such as a microwave radiometer, with those coming from a point raingauge. Using this scheme the errors are computed for several observed rainfall fields, either from the data estimated spectra or from analytically derived and characterized ground measurements. This quantification provides a lower bound to total errors, since perfect instruments are assumed in this work, and it helps in terms of isolating and evaluating typical biases that might be contaminating retrieval algorithms.