Exploring Temporal Dynamics of Spatially-Distributed Groundwater Levels by Integrating Time Series Modeling with Geographic Information System

Abstract

This study developed a novel framework for integrating time series modeling with geographic information system (GIS). For the first time, procedures of four statistical tests, i.e., t-test of stationarity, cumulative deviation test of homogeneity, autocorrelation technique of persistence, and variance-corrected Mann–Kendall test of trend, are implemented in GIS platform to enable use of raster dataset. Application of developed framework is demonstrated by exploring time series characteristics of pre- and post-monsoon groundwater levels in an Indian arid region. Raster dataset of 22-year (1996–2017) groundwater levels are generated using four best-fit geostatistical models, according to mean absolute error, root mean square error, correlation coefficient and modified index of agreement. Increasing groundwater level trends in central and southern parts are attributed to abrupt change points in annual rainfall that enhanced groundwater recharge. The developed framework can be adopted in other parts of the world to explore groundwater-level dynamics in spatially-distributed manner.