Accurate forecasting of groundwater levels is of primary importance for the sustainable management of water resources, particularly in semi-arid regions that continuously face high agricultural and industrial water demands. This study attempts to model the groundwater level dynamics at the Gozar-Abbas-Ali piezometric well in the northern Kerman aquifer, located in Iran, using time series techniques. The fit and validation of three stochastic models were tested in this study: the autoregressive (AR), the autoregressive moving average (ARMA), and the autoregressive integrated moving average (ARIMA) models. Among these models, ARIMA (5,2,5) was found as the most suitable for prediction accuracy. The trend analysis indicated a continued decline in the groundwater levels, mainly attributable to excessive extraction for agricultural use during the period of 2002 to 2010. Conversely, a decrease in agricultural land, traced through NDVI analysis on remote sensing data, led to the transient slowing down of groundwater depletion from 2010 to 2018, coinciding with the discharge of treated wastewater for artificial recharge. Since 2018, groundwater depletion had resumed more speedily due to water being diverted for large-scale industrial purposes, exacerbating the stress on the aquifer. The integration of regression between ARIMA-predicted groundwater levels and NDVI-derived land-use forecasts predicts further decline from cultivated land from 57.34% of the study area in 2002 to 23.71% in 2026 and to 9.63% in 2036. This observation highlights the intertwined relationship between agricultural expansion, industrial water demand, and groundwater sustainability, warranting the immediate need for integrated water management in arid and semi-arid regions.