Doctor of Philosophy (PhD)


Geography and Anthropology

Document Type



Land subsidence, defined as a land sinking or a gradual inward caving of land, presents a common disturbance observed in many areas of the world. In Louisiana, this specific problem posed a serious threat to the populace living there. Considered by denizens to be an adverse impact of land use, the extant Louisiana subsidence causes serious problems that tend to worsen, such as excessive wetland formation or land loss. Unless researchers find appropriate treatments to address this increasingly serious problem, the present issues will be exacerbated. To visualize the spatio-temporal subsidence patterns, this study used data collected by high-precision GPS stations and processed high-accuracy land elevation data in coastal Louisiana by means of a GIS-based spatio-temporal data model. I used the Kriged Kalman Filter (KKF) to map the spatial temporal field of land elevation change in southern Louisiana from 2011 to 2013, which showed a clear subsidence area after 2012. The coincidence of the Bayou Corne Sinkhole enabled a validation of the GPS data and the spatio-temporal data model. In addition, the spatial pattern for subsidence was predicted by Regression-Kriging and based on observed GPS data in tandem with the data on contributing subsidence factors. The prediction results using Regression-Kriging had high and acceptable accuracy. I applied the geographically weighted regression (GWR) model to show the spatial heterogeneity of contributing factors to subsidence in the study site. The statistical results showed that spatial heterogeneity for the data of contributing factors vi would be useful to recognize the agglomeration of communities in the study area. The regionalization work of these contributing factors could also be helpful to form location-based subsidence mitigation policies. This research contributes to the knowledge of GIS data modeling by incorporating a spatio-temporal interpolation—the Kriged Kalman filter (KKF)—into mapping and monitoring the land elevation change. This technique overcomes the problems of traditional spatial interpolation methods that disregard the time dependency of the geospatial data. The second contribution of this research is to predict the spatial pattern of subsidence using the information in regard to the subsidence factors at GPS stations. A cross-scale subsidence prediction, drawn solely on point based data from GPS stations, was made possible by Regression-Kriging. The third contribution of this research is that the spatial statistical models used for data analysis enable location-based policy-making. In other words, the local government can embrace smart policies that are specifically effective for certain regions to prevent further land loss or subsidence.



Document Availability at the Time of Submission

Release the entire work immediately for access worldwide.

Committee Chair

Wang, Lei