A knowledge based spatial model for utilizing point and nested areal observations: A case study of annual runoff predictions in the Voss area
share
In this study, annual runoff is estimated by using a Bayesian geostatistical model for interpolating hydrological data of different spatial support. That is, streamflow observations from catchments (areal data), and precipitation and evaporation data (point data). The model contains one climatic spatial effect that is common for all years under study, and one year specific spatial effect. The climatic effect provides a framework for exploiting sparse datasets that include short records of runoff, and we obtain a model that can be used for spatial interpolation and for gaining knowledge about future annual runoff. The model's ability to predict annual runoff is investigated using 10 years of data from the Voss area in Western Norway and through a simulation study. On average we benefit from combining point and areal data compared to using only one of the data types, and the interaction between nested areal data and point data gives a geostatistical model that takes us beyond smoothing. Another finding is that climatic effects dominates over annual effects in Voss. This implies that systematic under- and overestimation of runoff often occur, but also that short-records of runoff from an otherwise ungauged catchment can lead to large improvements in the predictability of runoff.
READ FULL TEXT