A Hydroinformatic Approach to Assess Interpolation Techniques in High Spatial and Temporal Resolution

One of the problems which often arises in hydrologic and hydraulic design is the estimation of data at a given site, where data are missing or the site is ungauged. Estimates can be made by spatial interpolation of data available at other sites. Geographic information systems offer a number of embedded ready-to-use spatial interpolation techniques. It is the intent of this paper to document the development of a GIS-based tool to compare the applicability of various interpolation techniques for estimating rainfall for a selected area in Ontario, Canada. The interpolation techniques in the ArcView GIS® are: IDW, Spline (Tension and Regularized), 2nd Order Polynomial, and Kriging (Ordinary and Universal). One-minute rainfall data for sixteen events were extracted from ten rainy days during the summer of 1989. The test case is an 8 km by 8 km area located in the Hamilton-Wentworth region. It is covered by a nine-gauge network. The comparison shows that the IDW and Kriging_Gaussian are the best and worst techniques, respectively. Results also point out that some gauges are more important than others. The results, however, can also be considered a useful warning of the instability of interpolation techniques with limited/sparse data, particularly in the context of commercial software.