A Two-Field Geostatistical Model Combining Point and Areal Observations—A Case Study of Annual Runoff Predictions in the Voss Area

In this study, annual runoff is estimated by using a Bayesian geostatistical model for interpolation of hydrological data different spatial support. That is, streamflow observations from catchments (areal data), and precipitation evaporation (point data). The contains one climatic effect that common all years under year specific effect. Hence, the framework enables quantification variability due to long-term weather patterns processes. This can contribute better understanding biases uncertainties in environmental modeling. By integrated nested Laplace approximations (INLA) stochastic partial differential equation approach (SPDE) modeling, two field computationally feasible fast. suggested tested predicting 10 around Voss Norway through simulation study. We find on average we benefit combining point areal compared only types, interaction between gives takes us beyond smoothing. Another finding when effects dominate over effects, systematic under- overestimation time be expected. On other hand, dominating implies short records an otherwise ungauged catchment lead large improvements predictability runoff.