Probabilistic estimation of recharge and its sensitivity to changing precipitation regimes for the Urucuia aquifer, Brazil

Abstract

Given the rapid expansion of irrigated agriculture and climate change, it is essential to understand the dynamics of groundwater recharge formation for sustainable management of water resources. This study focuses on a portion (≈ 20,000 km²) of the Urucuia aquifer in northeastern Brazil, for which we use daily hydrometeorological records at five stations over several decades to infer monthly statistics (probabilities and intensities) of precipitation P. We further apply different existing deterministic and probabilistic / stochastic soil water balance approaches to estimate recharge R for the past and a limited number of possible future scenarios. Long-term average recharge estimates range from 257 to 440 mm/year (east to west) with recharge events concentrated exclusively in the rainy summer months. For annual averages in mm/day, we identify the empirical relationship R = P – 2 as a simple and accurate approximation for the region. We further conclude that quantification of recharge as a percentage of precipitation is inappropriate and that seasonal variability with soil water carry-over between seasons cannot be neglected. Future scenarios with reduced average precipitation are estimated to affect groundwater recharge over-proportionally and stronger than soil water content.

Keywords: climate change, evapotranspiration, groundwater, soil water balance, stochastic simulation.