Sarima-based streamflow forecasting in the San Juan River, Arid Andes

Abstract

River water is a vital resource for human consumption, industrial use, power generation, and recreational activities, among other uses. A number of techniques are currently available to forecast streamflow at different temporal scales, although such approaches usually require different types of input data. The aim of this study is to analyze the long-term hydrological behavior of the San Juan River, located in the Arid Andes of Argentina, a region facing severe water scarcity. A Seasonal Autoregressive Integrated Moving Average (SARIMA) statistical model was applied to project future annual streamflow volume (Hm3) dynamics up to the 2070-2071 hydrological cycle. The results indicate a general downward trend. Over the next decade, critical low-flow periods are expected, similar to those observed in recent years. For the 2026-2029 period, projected flows are slightly above the average total annual consumption level (1,200 Hm³), yet insufficient to fully restore ecosystems affected by previous prolonged droughts. In contrast, for the 2030-2035 period, forecasts suggest a sharp decline in streamflow to around 680 Hm³, nearly half the current annual consumption. These findings provide a solid basis for developing adaptive strategies to manage potential future water availability scenarios effectively.