Evaluation of Alternative Geospatial Models using Image Ranking and Machine Learning: An Application in Shallow Groundwater Recharge and Discharge

摘要

This paper addresses the problem of accurate estimation of geospatial models from a set ofgroundwater recharge and discharge maps and from auxiliary remote sensing and terrestrialraster measurements. The motivation for our work is driven by the cost of field measurements,and by the limitations of currently available physics-based modeling techniques that do notinclude all relevant variables and allow accurate predictions only at coarse spatial scales. Thegoal is to improve our understanding of the underlying physical phenomena and increase theaccuracy of geospatial models-with a combination of remote sensing, field measurements andphysics-based modeling. Our approach is to process a set of recharge and discharge mapsgenerated from interpolated sparse field measurements using existing physics-based models, andidentify the recharge and discharge map that would be the most suitable for extracting a set ofrules between the auxiliary variables of interest and the recharge and discharge map labels. Weimplemented this approach by ranking recharge and discharge maps using information entropyand mutual information criteria, and then by deriving a set of rules using a machine learningtechnique, such as the decision tree method. The novelty of our work is in developing a generalframework for building geospatial models with the ultimate goal of minimizing cost andmaximizing model accuracy. The framework is demonstrated for groundwater recharge anddischarge rate models but could be applied to other similar studies, for instance, to understandinghypoxia based on physics-based models and remotely sensed variables. Furthermore, our key