Prediction spatial data by inverse distance weighting, surface trend analysis, and GWR: A comparative study

It provides analytical models for spatial data prediction using Inverse Distance Weighting, Surface Trend Analysis and Geographically Weighted Regression and a comparative analysis of these three methods. The study will investigate the performance of each approach on spatial data prediction on different contexts, and also analyze the quality and efficiency of those methods to process geographic data. In this work, the IDW (inverse distance weight) method is used as an initial study which assumes that the points closer to each other possess more similarity. Finally, surface trend analysis is also employed to interpret the trends and spatial modes of the underlying data. The GWR is utilized for spatially weighted regression analysis, which endows the method with the capacity to account for spatial heterogeneity and to estimate those relationships in different geographies. This work gives a complete comparison between the three methods in accuracy, applications for handling spatial changes, future work and so on. Results indicated that the GWR model was the most accurate for complex spatial data with strong spatial heterogeneity, and that the IDW and Surface Trend Analysis methods were used in some applications. This statement helps to illustrate what can be achieved by each technique in spatial data prediction, and thus is a source of useful and practical knowledge for the practitioners of geography, remote sensing, and resource management.