Performance evaluation of deep learning based stream nitrate concentration prediction model to fill stream nitrate data gaps at low-frequency nitrate monitoring basins.

Accurate and frequent nitrate estimates can provide valuable information on the nitrate transport dynamics. The study aimed to develop a data-driven modeling framework to estimate daily nitrate concentrations at low-frequency nitrate monitoring sites using the daily nitrate concentration and stream discharge information of a neighboring high-frequency nitrate monitoring site. A Long Short-Term Memory (LSTM) based deep learning (DL) modeling framework was developed to predict daily nitrate concentrations. The DL modeling framework performance was compared with two well-established statistical models, including LOADEST and WRTDS-Kalman, in three selected basins in Iowa, USA: Des Moines, Iowa, and Cedar River. The developed DL model performed well with NSE >0.70 and KGE >0.70 for 67% and 79% nitrate monitoring sites, respectively. DL and WRTDS-Kalman models performed better than the LOADEST in nitrate concentration and load estimation for all low-frequency sites. The average NSE performance of the DL model in daily nitrate estimation is 20% higher than that of the WRTDS-Kalman model at 18 out of 24 sites (75%). The WRTDS-Kalman model showed unrealistic fluctuations in the estimated daily nitrate time series when the model received limited observed nitrate data (less than 50) for simulation. The DL model indicated superior performance in winter months' nitrate prediction (60% of cases) compared to WRTDS-Kalman models (33% of cases). The DL model also better represented the exceedance days from the USEPA maximum contamination level (MCL). Both the DL and WRTDS-Kalman models demonstrated similar performance in annual stream nitrate load estimation, and estimated values are close to actual nitrate loads.

How does the choice of DEMs affect catchment hydrological modeling?

The digital elevation models (DEMs) are the primary and most important spatial inputs for a wide range of hydrological applications. However, their availability from multiple sources and at various spatial resolutions poses a challenge in watershed modeling as they influence hydrological feature delineation and model simulations. In this study, we evaluated the effect of DEM choice on stream and catchment delineation and streamflow simulation using the SWAT model in four distinct geographic regions with diverse terrain surfaces. Performance evaluation metrics, including Willmott's index of agreement, and nRMSE combined with visual comparisons were employed to assess each DEM's performance. Our results revealed that the choice of DEM has a significant impact on the accuracy of stream and catchment delineation, while its influence on streamflow simulation within the same catchment was relatively minor. Among the evaluated DEMs, AW3D30 and COP30 performed the best, closely followed by MERIT, whereas TanDEM-X and HydroSHEDS exhibited poorer performance. All DEMs displayed better accuracy in mountainous and larger catchments compared to smaller and flatter catchments. Forest cover also played a role in accuracy, mainly due to its association with steep slopes. Our findings provide valuable insights for making informed data selection decisions in watershed modeling, considering the specific characteristics of the catchment and the desired level of accuracy.

A deep learning-based novel approach to generate continuous daily stream nitrate concentration for nitrate data-sparse watersheds.

High-frequency stream nitrate concentration provides critical insights into nutrient dynamics and can help to improve the effectiveness of management decisions to maintain a sustainable ecosystem. However, nitrate monitoring is conventionally conducted through lab analysis using in situ water samples and is typically at coarse temporal resolution. In the last decade, many agencies started collecting high-frequency (5-60 min intervals) nitrate data using optical sensors. The hypothesis of the study is that the data-driven models can learn the trend and temporal variability in nitrate concentration from high-frequency sensor-based nitrate data in the region and generate continuous nitrate data for unavailable data periods and data-limited locations. A Long Short-Term Memory (LSTM) model-based framework was developed to estimate continuous daily stream nitrate for dozens of gauge locations in Iowa, USA. The promising results supported the hypothesis; the LSTM model demonstrated median test-period Nash-Sutcliffe efficiency (NSE) = 0.75 and RMSE = 1.53 mg/L for estimating continuous daily nitrate concentration in 42 sites, which are unprecedented performance levels. Twenty-one sites (50 % of all sites) and thirty-four sites (76 % of all sites) demonstrated NSE > 0.75 and 0.50, respectively. The average nitrate concentration of neighboring sites was identified as a crucial determinant of continuous daily nitrate concentration. Seasonal model performance evaluation showed that the model performed effectively in the summer and fall seasons. About 26 sites showed correlations >0.60 between estimated nitrate concentration and discharge. The concentration-discharge (c-Q) relationship analysis showed that the study watersheds had four dominant nitrate transport patterns from landscapes to streams with increasing discharge, including the flushing pattern being the most dominant one. Stream nitrate estimation impedes due to data inadequacy. The modeling framework can be used to generate temporally continuous nitrate at nitrate data-limited regions with a nearby sensor-based nitrate gauge. Watershed planners and policymakers could utilize the continuous nitrate data to gain more information on the regional nitrate status and design conservation practices accordingly.

Water quality benefits of weather-based manure application timing and manure placement strategies.

The timing of manure application and placement of manure significantly affects manure nutrient use efficiency and the amount of nutrient lost from a field. Application of manure prior to a minimal precipitation period, and manure application through incorporation, reduces risks associated with nutrient loss through surface runoff. The current study aims to explore potential water quality impacts related to manure application strategies on the timing of application and approach (surface broadcasting or incorporation). The Soil and Water Assessment Tool (SWAT) was used to represent manure application scenarios and quantify potential water quality impacts in Susquehanna River Basin located in the Mid-Atlantic region of the United States. A baseline (business-as-usual) scenario was developed with manure application based on crop planting date and manure storage availability, and surface broadcasting as the application approach. The baseline was compared with a strategically timed manure application considering weather forecasting and manure incorporation. The strategic, weather-based manure application approach reduced TN and TP loading at the outlet by 4% and 6%, respectively. Manure incorporation simulations considering low-disturbance injection showed significant reduction of about 19% for TN and 44% for TP at the watershed outlet. Winter closure of manure application could reduce organic nutrient loss. Winter application of manure in 21% of row cropped areas (2% of whole watershed area) increased organic N and P loading by 10% and 4%, respectively, at watershed outlet.

Evaluating water quality benefits of manureshed management in the Susquehanna River Basin.

Manureshed management guides the sustainable use of manure resources by matching areas of crop demand (nutrient sinks) with areas generating livestock manure (nutrient sources). A better understanding of the impacts of manureshed management on water quality within sensitive watersheds is needed. We quantified the potential water quality benefits of manureshed-oriented management through scenario-based analyses in the Susquehanna River Basin (SRB) using the Soil and Water Assessment Tool. Five manureshed management scenarios were developed and compared with a baseline "business-as-usual" scenario. The baseline assumes manure is less transportable, which means some locations have manure application in excess of crop demand. The "watershed nutrient balance" scenarios assume excess manure from surplus locations is transportable and that manure is applied around the SRB based on crop nutrient demand. The "watershed nutrient balance avoiding runoff prone areas" scenarios assume manure is transportable but not applied in vulnerable landscapes of the SRB. Each scenario was evaluated under two application rates considering crop nitrogen demand (N-based) and phosphorus demand (P-based). Phosphorus-based manureshed management was more effective in water quality improvements than N-based management. Phosphorus-based nutrient balance scenarios simulated 3 and 25% reduction in total N (TN) and total P (TP), respectively, from the baseline scenario at the watershed outlet. The N- and P-based scenarios avoiding runoff prone areas simulated 3 and 6% reduction in TN loss and 4 and 25.2% reduction in TP loss, respectively, from the baseline. Overall, the manureshed management scenarios were more effective in improving the quality of local streams in livestock-intensive regions than at the watershed outlet.

Envisioning the manureshed: Toward comprehensive integration of modern crop and animal production.

The specialization and intensification of agriculture have produced incredible gains in productivity, quality, and availability of agricultural commodities but have resulted in the separation of crop and animal production. A by-product of this separation has been the accumulation of manure regions where animal production is concentrated. Enter the "manureshed," an organizing framework for integrating animal and crop production where budgeting of manure nutrients is used to strategically guide their recycling and reuse in agricultural production systems where manure resources are of highest value. To move beyond regional nutrient balance analyses into the transformational realm required to mitigate "wicked" manure problems, manureshed management requires recognition of the challenges to systematically reorganizing resource flows. In better integrating crop and livestock systems, manureshed management must account for the unique nature of managing manure nutrients within individual livestock industries, anticipate trade-offs in substituting manure for commercial fertilizer, promote technologies to refine manure, and engage extensive social networks across scales that range from the farmgate to nation and beyond.

Manuresheds offer a system-level strategy for recovering manure's fertilizer value. Manuresheds address nutrient imbalances and environmental and socioeconomic outcomes. Manuresheds scale from single operations to a "mega-manureshed" transecting the southeastern United States. Manureshed management supports the strategic alignment of technologies, markets, and networks.

Riparian buffer effectiveness as a function of buffer design and input loads.

Although many agricultural watersheds rely heavily on riparian buffer adoption to meet water quality goals, design and management constraints in current policies create adoption barriers. Based on focus group feedback, we developed a flexible buffer design paradigm that varies buffer width, vegetation, and harvesting. Sixteen years of daily-scale nutrient and sediment loads simulated with the Soil and Water Assessment Tool (SWAT) were coupled to the three-zone Riparian Ecosystem Management Model (REMM) to compare the effectiveness of traditional, policy-based buffer designs with designs that are more flexible and integrate features important to local farmers. Buffer designs included (i) 10 m grass, (ii) 15 m grass, (iii) 15 m deciduous trees, (iv) 30 m grass and trees, (v) 30 m grass and trees with trees harvested every 3 yr, and (vi) 30 m grass and trees with grass harvested every year. Allowing harvesting in one zone of the buffer vegetation (either trees or grasses) minimally affected water quality, with annual average percent reductions differing by <5% (p > .05; 76-78% for total nitrogen [TN], 51-55% for total phosphorus [TP], and 68% for sediment). Under the highest input loading conditions, buffers with lower removal efficiencies removed more total mass than did buffers with high removal efficiencies. Thus, by focusing on mass reduction in addition to percent reduction, watershed-wide buffer implementation may be better targeted to TN, TP, and sediment reduced. These findings have important implications for informing flexible buffer design policies and enhanced placement of buffers in watersheds impaired by nutrient and sediment.

Evaluation of bioenergy crop growth and the impacts of bioenergy crops on streamflow, tile drain flow and nutrient losses in an extensively tile-drained watershed using SWAT.

Large quantities of biofuel production are expected from bioenergy crops at a national scale to meet US biofuel goals. It is important to study biomass production of bioenergy crops and the impacts of these crops on water quantity and quality to identify environment-friendly and productive biofeedstock systems. SWAT2012 with a new tile drainage routine and improved perennial grass and tree growth simulation was used to model long-term annual biomass yields, streamflow, tile flow, sediment load, and nutrient losses under various bioenergy scenarios in an extensively agricultural watershed in the Midwestern US. Simulated results from bioenergy crop scenarios were compared with those from the baseline. The results showed that simulated annual crop yields were similar to observed county level values for corn and soybeans, and were reasonable for Miscanthus, switchgrass and hybrid poplar. Removal of 38% of corn stover (3.74Mg/ha/yr) with Miscanthus production on highly erodible areas and marginal land (17.49Mg/ha/yr) provided the highest biofeedstock production (279,000Mg/yr). Streamflow, tile flow, erosion and nutrient losses were reduced under bioenergy crop scenarios of bioenergy crops on highly erodible areas and marginal land. Corn stover removal did not result in significant water quality changes. The increase in sediment and nutrient losses under corn stover removal could be offset with the combination of other bioenergy crops. Potential areas for bioenergy crop production when meeting the criteria above were small (10.88km2), thus the ability to produce biomass and improve water quality was not substantial. The study showed that corn stover removal with bioenergy crops both on highly erodible areas and marginal land could provide more biofuel production relative to the baseline, and was beneficial to water quality at the watershed scale, providing guidance for further research on evaluation of bioenergy crop scenarios in a typical extensively tile-drained watershed in the Midwestern U.S.

Environmental and economic trade-offs in a watershed when using corn stover for bioenergy.

There is an abundant supply of corn stover in the United States that remains after grain is harvested which could be used to produce cellulosic biofuels mandated by the current Renewable Fuel Standard (RFS). This research integrates the Soil Water Assessment Tool (SWAT) watershed model and the DayCent biogeochemical model to investigate water quality and soil greenhouse gas flux that results when corn stover is collected at two different rates from corn-soybean and continuous corn crop rotations with and without tillage. Multiobjective watershed-scale optimizations are performed for individual pollutant-cost minimization criteria based on the economic cost of each cropping practice and (individually) the effect on nitrate, total phosphorus, sediment, or global warming potential. We compare these results with a purely economic optimization that maximizes stover production at the lowest cost without taking environmental impacts into account. We illustrate trade-offs between cost and different environmental performance criteria, assuming that nutrients contained in any stover collected must be replaced. The key finding is that stover collection using the practices modeled results in increased contributions to atmospheric greenhouse gases while reducing nitrate and total phosphorus loading to the watershed relative to the status quo without stover collection. Stover collection increases sediment loading to waterways relative to when no stover is removed for each crop rotation-tillage practice combination considered; no-till in combination with stover collection reduced sediment loading below baseline conditions without stover collection. Our results suggest that additional information is needed about (i) the level of nutrient replacement required to maintain grain yields and (ii) cost-effective management practices capable of reducing soil erosion when crop residues are removed in order to avoid contributions to climate change and water quality impairments as a result of using corn stover to satisfy the RFS.