Journal of The American Water Resources Association最新文献

This study evaluates National Water Model (NWM) performance in low-order montane catchments across the northeastern United States by comparing retrospective simulations to measured observations. To address deficiencies, we develop a machine learning (ML) correction model for selected sites using LightGBM, a different approach from conventional bias correction methods. Montane, low-order streams play a crucial role in water quality and flood generation but pose challenges for streamflow prediction and are under-represented in the national streamgaging network. NWM provides streamflow forecasts across the United States; yet a focused assessment of its performance in these settings has not been comprehensively undertaken. Results indicate NWM performance varied seasonally, with the best performance during the fall and particularly poor performance during snowmelt, spring runoff, and high flow events, with a tendency towards flow underestimation. The ML correction model markedly improved hourly streamflow prediction accuracy based on continuous time series and runoff event-based metrics. Including antecedent water level measurements as input, even from distant sites, greatly improved model performance, demonstrating the potential to improve predictions by deploying supplemental low-cost water level sensors. We demonstrate that NWM performance can be improved in these complex watersheds using ML tools. This approach could be implemented elsewhere to improve NWM streamflow predictions.

Arkansas is a leading state in groundwater use and application in the United States, as well as a top agricultural producer with a history of irrigated farming dating back over a century. Extensive monitoring of the primary irrigation water source, the Mississippi River Valley Alluvial Aquifer (alluvial aquifer), has shown a history of groundwater decline and only recent recharge. The objective of this study was to report the findings of a survey of producers in the region overlying the alluvial aquifer to determine the likelihood of adopting specific irrigation practices shown to either promote conservation of water or increase water use efficiency. This was completed using the Theory of Planned Behavior (TPB). Three models were developed to determine the adoption likelihood of tailwater recovery and surface storage, implementation of soil moisture sensors, and implementation of surge irrigation. Results show that portions of the TPB were present within each model, but that the strongest predictors were often prior adoption of other farm water management practices. It is suggested that, while social profiling may be a valuable tool to identify producers inclined to adopt farm water management practices, focus should be placed on individuals who have already adopted other practices.

Accurate quantification of sediment occurrence in river basins is essential for establishing sediment management strategies. This study enhanced the simplified Bagnold method within the SWAT model to enhance simulation of sediment dynamics due to the rise and fall of streamflow. The sediment simulation efficacy of the enhanced Bagnold method in the SWAT was evaluated by comparison of estimated daily sediment with the original Bagnold method in the SWAT model. During the rise of streamflow, the original Bagnold method in the SWAT model yielded an R² of 0.86 and an NSE of 0.85, but these indicators declined to an R² of 0.49 and an NSE of −14.88 in simulations of sediment during the fall of streamflow. In contrast, the enhanced Bagnold method in the SWAT showed superior calibration performance, achieving an R² of 0.94 and an NSE of 0.92 during the rise of streamflow, and an improved R² of 0.68 and an NSE of 0.6 during the fall of streamflow. Notably, the enhanced Bagnold method in the SWAT provided a more accurate sediment prediction during increases in streamflow, with an overestimation of only 25.1% relative to observed data. This marks a significant improvement over the original model, which overestimated sediment by 443.5%. As shown in this study, streamflow variability (changes in rise/fall) needs to be considered in sediment simulations, enhancing model accuracy and informing effective sediment management strategy development.

We present methods to reconstruct historical chlorophyll a and surface water temperatures from satellite-based remote sensing products for Blue Mesa Reservoir, Colorado, to support algal bloom monitoring. A machine learning model was trained to construct chlorophyll a concentrations from Sentinel-2 satellite imagery and in situ measurements of chlorophyll a concentrations (out of bag RMSE = 1.9 μg/L, R² = 0.63) and reconstruct summertime chlorophyll a concentrations over the entire reservoir from 2016 through 2023. Concurrently, we developed an approach to retrieve remotely sensed water temperatures from the Landsat collection 2 provisional surface temperature product (MAE = 0.6°C) and reconstructed summertime surface water temperature records from 2000 through 2023. Finally, we demonstrate how the reconstructed chlorophyll a and temperature records can yield insight on reservoir dynamics. The chlorophyll a records indicate that algal blooms have a consistent spatial pattern across multiple years, initiating in the eastern end of the reservoir and spreading to the west over time. Water temperatures increased at a linearized rate of 0.3°C per decade from 2000 through 2023 and were inversely proportional to reservoir water surface elevation. Finally, mean summer remotely sensed chlorophyll a concentration had a moderately positive correlation with mean summer remotely sensed water temperature.

Interannual variability of streamflow will increase under a future climate, but at the regional scale, there is uncertainty regarding changes in drought severity, and in particular, changes in extreme hydrological drought that could necessitate new water supply infrastructure. This is due to the wide range of regional projections for precipitation and the challenge of estimating statistics in a nonstationary climate. We assess changes in annual streamflow in the Potomac River Basin using a nonparametric approach based on a climate response function and the K-nearest neighbor method, which is relied on to construct time series of sufficient length to compute extreme quantile values. Our results indicate that future Potomac River flows will be impacted by “hot drought”, that is, increasing drought severity caused by rising temperatures coupled with natural variability in precipitation. Average precipitation is projected to increase in the Potomac basin by 9%–12% in the period 2039–2069 and by 11%–16% by 2070–2099. Average streamflow increases more modestly, by 4%–7% in 2039–2069 and by 2 to 9% in 2070–2099, whereas annual flows in an extreme drought year decrease by 3 to 26% in 2039–2069 and by 2%–49% in 2070–2099, assuming a medium sensitivity of flow to temperature. Our approach can provide multi-model consensus inputs for water supply planning models to support decision-making regarding new infrastructure.

Flood-mitigation reservoirs have long been known to impact pollutant transport by retaining or removing incoming sediment and nutrients. However, historical reductions in these systems have rarely been well quantified. In this study, we used water quality data to estimate inputs and outputs of total suspended solids (TSS), two phosphorus (P) forms, and three nitrogen (N) forms in three Iowa reservoirs (Coralville, Red Rock, and Saylorville). We also explored the influence of reservoir residence times on removal rates. Annual residence times were largely consistent across the basins, ranging from roughly 6 to 100 days (mean of 19 days). Between 2001 to 2023, most TSS (~ 80%) entering the reservoirs was retained. This sedimentation corresponded to average volume losses in the reservoirs' normal storage pools of 0.37%–0.85%/year. About 40% of P and 12% of N were likewise reduced—driven mainly by decreases in particulate P and nitrate. Residence time appeared unrelated to removal rates of TSS and particulate nutrient forms, but longer residence times coincided with increased nitrate loss. Reservoir impact on statewide nutrient export was significant, with loads in Iowa's major rivers being reduced by 9.8% (for P) and 4.7% (for N) due to reservoir capture. These findings suggest that reservoir operators may be able to facilitate further nitrate removal by lengthening storage durations without incurring additional sedimentation or generating other nutrient forms.

This document examines the Chesapeake Bay watershed response to nutrient and sediment reduction efforts under the Clean Water Act's total maximum daily load (TMDL) regulation. As the 2025 Chesapeake Bay TMDL deadline approaches, water quality goals remain unmet, primarily because of nonpoint source pollution, the largest remaining source of nutrients and sediment, and the primary obstacle to meeting the TMDL. We focus on the factors influencing the gap between the expected effect of management to reduce nonpoint source loads reaching the Bay and empirical evidence suggesting that decades of effort have not produced the expected improvement. This gap may be caused by both insufficient scale and type of implemented water quality management practices and by an overestimation of practice effectiveness. Reasons water quality goals remain unmet include legacy nutrients and lag times masking or delaying the effects of management efforts, areas with large nutrient mass imbalances contributing disproportionate loads, and the difficulty of incentivizing behavior change in voluntary nonpoint source programs. Closing the response gap may require fundamental changes to nonpoint source programs. Apart from seeking additional funding, nonpoint source programs could develop policies to more effectively incentivize behavior change, identify and target treatment of high loading areas with appropriate management actions, and address nutrient mass imbalances.