An application of time-lag regression technique for assessment of groundwater fluctuations in a regulated river basin: a case study in Northeastern Thailand

Abstract

Time series techniques have been applied to a wide variety of academic fields, particularly in the fields of economics, environment, and hydrology. This research work deals with application of the time-lag multiple linear regression technique to predicting the groundwater level and salinity fluctuation in a saline but irrigated area in Northeastern Thailand. Regulating gates are constructed to prevent flooding of the river downstream and to provide the surrounding upstream areas with sufficient irrigation water. The technique is essentially based on the lagged correlation between leading variables, i.e. rainfall and river stages, and their corresponding responses, i.e. groundwater depth and salinity. While the rainfall is simulated with a multiple sinusoidal model, the river stage model comprises two components: a single sinusoidal component and a multiple linear regression component implemented with independent variables, i.e. rainfall, regulating gate outflow, and irrigation supply. The shallow groundwater and groundwater salinity fluctuation in the irrigated area are shown by cross-correlation analysis to be dependent upon the surface water regulation and prior rainfall and thus can be simulated by a multiple linear regression with lagged time dependence. Once properly calibrated and verified, the model can perform better matching to the observed data vis-a-vis the conventional multiple linear regression model. The proposed model can also be deployed for forecasting and management of the future groundwater system, especially in situations in which costlier and more complex numerical modeling techniques are inapplicable.