A data-driven approach to quantifying the correlation between groundwater and wildfire in the United States

Abstract:

Over the last 30 years, wildfires in the United States (US) have increased in number, size, and duration. This trend is expected to continue under warmer and dryer conditions. Wildfires can affect surface water quality and quantity by reducing stream water quality and enhancing annual river flow. However, it is unclear how wildfires affect groundwater and vice versa. This study quantifies the correlation between wildfire activity and groundwater using statistics and big data collection over the continental US. The data include groundwater levels from 13,433 wells that have at least 100 measurements from 1940 to 2018, groundwater use (1985-2015), Monitoring Trends in Burn Severity (1984-2016), PRISM climate (1979-2018), and National Land Cover Database (2016). To quantify the correlation, maps of groundwater depths were firstly generated using the Barnes interpolation for both the shallow wells and the deep wells. Then, the Theil-Sen method was used to estimate the groundwater level trends, and the Mann-Kendall method was used to test for the statistical significance of the estimated trends. Finally, the Vector Autoregression (VAR) and the Granger causality test were utilized to quantify the correlation. The results revealed that changes in groundwater and wildfires are correlated. The impulse response (IR) of wildfires on changes in groundwater level is greatest within the first year and may last up to four years depending on well depth and climate region. However, this IR is very small compared with the IR of precipitation anomalies on changes in groundwater level. The IRs of changes in groundwater level on wildfires are largest within the first year for both aquifer depths.