Design of optimal precipitation sensor networks is a common topic in hydrological literature, however this is still an open problem due to lack of understanding of some spatially variable processes, and assumptions that often cannot be verified. Among these assumptions lies the homoscedasticity of precipitation fields, common in hydrological practice. To overcome this, it is proposed a local intensity-variant covariance structure, which in the broad extent, provides a fully updated correlation structure as long as new data are coming into the system. These considerations of intensity-variant correlation structure will be tested in the design of a precipitation sensor network for a case study, improving the estimation of precipitation fields, and thus, reducing the input uncertainty in hydrological models, especially in the scope of rainfall-runoff models.
Chacon-Hurtado, Juan Carlos; Alfonso, Leonardo; and Solomatine, Dimitri P., "Precipitation Sensor Network Optimal Design Using Time-Space Varying Correlation Structure" (2014). CUNY Academic Works.