Tess A. Russo

Machine learning algorithms for modeling groundwater level changes in agricultural regions of the U.S.

Climate, groundwater extraction, and surface water flows have complex nonlinear relationships with groundwater level in agricultural regions. To better understand the relative importance of each driver and predict groundwater level change, we develop a new ensemble modeling framework based on spectral analysis, machine learning, and uncertainty analysis, as an alternative to complex and computationally expensive physical models. We apply and evaluate this new approach in the context of two aquifer systems supporting agricultural production in the United States: the High Plains aquifer (HPA) and the Mississippi River Valley alluvial aquifer (MRVA). We select input data sets by using a combination of mutual information, genetic algorithms, and lag analysis, and then use the selected data sets in a Multilayer Perceptron network architecture to simulate seasonal groundwater level change. As expected, model results suggest that irrigation demand has the highest influence on groundwater level change for a majority of the wells. The subset of groundwater observations not used in model training or cross-validation correlates strongly (R > 0.8) with model results for 88 and 83% of the wells in the HPA and MRVA, respectively. In both aquifer systems, the error in the modeled cumulative groundwater level change during testing (2003–2012) was less than 2 m over a majority of the area. We conclude that our modeling framework can serve as an alternative approach to simulating groundwater level change and water availability, especially in regions where subsurface properties are unknown.

Hyporheic zone influences on concentration‐discharge relationships in a headwater sandstone stream

Complex subsurface flow dynamics impact the storage, routing, and transport of water and solutes to streams in headwater catchments. Many of these hydrogeologic processes are indirectly reflected in observations of stream chemistry responses to rain events, also known as concentration-discharge (CQ) relations. Identifying the relative importance of subsurface flows to stream CQ relationships is often challenging in headwater environments due to spatial and temporal variability. Therefore, this study combines a diverse set of methods, including tracer injection tests, cation exchange experiments, geochemical analyses, and numerical modeling, to map groundwater-surface water interactions along a first-order, sandstone stream (Garner Run) in the Appalachian Mountains of central Pennsylvania. The primary flow paths to the stream include preferential flow through the unsaturated zone (“interflow”), flow discharging from a spring, and groundwater discharge. Garner Run stream inherits geochemical signatures from geochemical reactions occurring along each of these flow paths. In addition to end-member mixing effects on CQ, we find that the exchange of solutes, nutrients, and water between the hyporheic zone and the main stream channel is a relevant control on the chemistry of Garner Run. CQ relationships for Garner Run were compared to prior results from a nearby headwater catchment overlying shale bedrock (Shale Hills). At the sandstone site, solutes associated with organo-mineral associations in the hyporheic zone influence CQ, while CQ trends in the shale catchment are affected by preferential flow through hillslope swales. The difference in CQ trends document how the lithology and catchment hydrology control CQ relationships.

America's water: Agricultural water demands and the response of groundwater

Agricultural, industrial, and urban water use in the conterminous United States (CONUS) is highly dependent on groundwater that is largely drawn from nonsurficial wells (>30 m). We use a Demand-Sensitive Drought Index to examine the impacts of agricultural water needs, driven by low precipitation, high agricultural water demand, or a combination of both, on the temporal variability of depth to groundwater across the CONUS. We characterize the relationship between changes in groundwater levels, agricultural water deficits relative to precipitation during the growing season, and winter precipitation. We find that declines in groundwater levels in the High Plains aquifer and around the Mississippi River Valley are driven by groundwater withdrawals used to supplement agricultural water demands. Reductions in agricultural water demands for crops do not, however, lead to immediate recovery of groundwater levels due to the demand for groundwater in other sectors in regions such as Utah, Maryland, and Texas.

Assessment of Agricultural Water Management in Punjab, India, Using Bayesian Methods

The success of the Green Revolution in Punjab, India, is threatened by a significant decline in water resources. Punjab, a major agricultural supplier for the rest of India, supports irrigation with a canal system and groundwater, which is vastly overexploited. The detailed data required to estimate future impacts on water supplies or develop sustainable water management practices is not readily available for this region. Therefore, we use Bayesian methods to estimate hydrologic properties and irrigation requirements for an under-constrained mass balance model. Using the known values of precipitation, total canal water delivery, crop yield, and water table elevation, we present a method using a Markov chain Monte Carlo (MCMC) algorithm to solve for a distribution of values for each unknown parameter in a conceptual mass balance model. Model results are used to test three water management strategies, which show that replacement of rice with pulses may be sufficient to stop water table decline. This computational method can be applied in data-scarce regions across the world, where integrated water resource management is required to resolve competition between food security and available resources.

Controls on nitrogen transformation rates on restored floodplains along the Cosumnes River, California

Levee construction results in the systematic replumbing of river systems and reduces the frequency of floodplain inundation, which impacts nutrient delivery and transformations in floodplains. Floodplain restoration via levee removal affects downstream water quality by restoring soil microbial metabolic pathways such as denitrification, anaerobic ammonium oxidation (anammox), and dissimilatory nitrate reduction to ammonium (DNRA). Although these metabolisms are important for the nitrogen cycle, few studies have quantified the contribution of all three pathways to nitrate retention or loss in restored floodplains. The objectives of this study were to quantify the relevance of denitrification, anammox and DNRA to nitrogen retention, characterize the hydrologic conditions most favorable to each pathway, and estimate the potential for floodplain restoration to improve nitrogen cycling in the Cosumnes River watershed. To address these goals, we simulated flood conditions in soil mesocosms collected from two floodplains where levees were breached in 1997 and 2014 along the Lower Cosumnes River in the San Joaquin Basin of California. River water enriched with K15NO3 tracer was pumped into each mesocosm at a constant rate for a period of 3 months. Samples were collected from the surface water and soil pore water for measurements of NO3-, NO2-, and NH4+ concentrations, and δ15N of dissolved gases (N2 and N2O). To the best of our knowledge, this study reports the highest relative contribution to N2 production due to anammox for freshwater systems (41 to 84%) to date. High anammox rates were associated with heterogeneous grain size distribution across depth and high nitrification rates. We quantify the capacity of restored floodplain soils with distinct textural and chemical characteristics to retain or release nitrogen during large and small floods in a particular water year.