Causes of groundwater patterns over multi-decadal scales: Climate or Agriculture?

In general, groundwater depletion has been attributed to excessive irrigation to meet global food demands. However, data-driven, causal mechanisms of long-term groundwater patterns have not been assessed formally. Here, we build more than 50 vector autoregressive models to conduct a Granger causality analysis to reveal the "causes" of groundwater patterns in the rice-producing counties of Louisiana. Trend analysis exhibited a decline up to 6m in groundwater level over 50yrs. We found that no single cause explained groundwater patterns of all counties. Causal linkages were noted between groundwater level and area harvested, number of irrigation wells, and palmer drought severity index. To counter droughts, farmers relied heavily on groundwater and dramatically increased the number of irrigation wells. This adaptation had no net effect on the production of rice over the last 50yrs. In other words, the additional groundwater pumping kept production from diminishing due to changes in climate, but it did not result in greater rice production. Causal linkages were absent between groundwater and many drivers where significant correlations were noted, pointing to the importance of using robust causal relationships over illusive correlations to detect a cause. Overall, in addition to irrigation, farmers and decision-makers need to carefully evaluate all possible causes that may accelerate the groundwater decline. These results advance our understanding of groundwater dynamics and reveal a key connections between food and groundwater.