Comparing Accuracy of Spatial Forecasts
Mandy Hering, Colorado School of Mines
Monday, October 19, 2009
4:00 p.m., 223 Weber
Under a general loss function, a hypothesis test is developed to determine whether a significant difference in the spatial forecasts produced by two competing models exists on average across the entire spatial domain of interest. The null hypothesis is that of no difference, and a spatial loss differential is created based on the observed data, the two sets of forecasts, and the loss function chosen by the researcher. The test assumes only isotropy and short-range spatial dependence of the loss differential but does allow it to be non-Gaussian, non-zero mean, and spatially correlated. Constant and non-constant spatial trends in the loss differential are treated in two separate cases. Monte Carlo simulations illustrate the size and power properties of this test, and an example based on daily average wind speeds in Oklahoma is used for illustration. The test is also compared to a wavelet-based method presented by Shen et al. (2002) that is designed to test for a spatial signal at every location in the domain.