Uncertainty in critical source area predictions from watershed-scale hydrologic models.

Watershed-scale hydrologic models are frequently used to inform conservation and restoration efforts by identifying critical source areas (CSAs; alternatively 'hotspots'), defined as areas that export relatively greater quantities of nutrients and sediment. The CSAs can then be prioritized or 'targeted' for conservation and restoration to ensure efficient use of limited resources. However, CSA simulations from watershed-scale hydrologic models may be uncertain and it is critical that the extent and implications of this uncertainty be conveyed to stakeholders and decision makers. We used an ensemble of four independently developed Soil and Water Assessment Tool (SWAT) models and a SPAtially Referenced Regression On Watershed attributes (SPARROW) model to simulate CSA locations for flow, phosphorus, nitrogen, and sediment within the ~17,000-km2 Maumee River watershed at the HUC-12 scale. We then assessed uncertainty in CSA simulations determined as the variation in CSA locations across the models. Our application of an ensemble of models - differing with respect to inputs, structure, and parameterization - facilitated an improved accounting of CSA prediction uncertainty. We found that the models agreed on the location of a subset of CSAs, and that these locations may be targeted with relative confidence. However, models more often disagreed on CSA locations. On average, only 16%-46% of HUC-12 subwatersheds simulated as a CSA by one model were also simulated as a CSA by a different model. Our work shows that simulated CSA locations are highly uncertain and may vary substantially across models. Hence, while models may be useful in informing conservation and restoration planning, their application to identify CSA locations would benefit from comprehensive uncertainty analyses to avoid inefficient use of limited resources.

Evaluating management options to reduce Lake Erie algal blooms using an ensemble of watershed models.

Reducing harmful algal blooms in Lake Erie, situated between the United States and Canada, requires implementing best management practices to decrease nutrient loading from upstream sources. Bi-national water quality targets have been set for total and dissolved phosphorus loads, with the ultimate goal of reaching these targets in 9-out-of-10 years. Row crop agriculture dominates the land use in the Western Lake Erie Basin thus requiring efforts to mitigate nutrient loads from agricultural systems. To determine the types and extent of agricultural management practices needed to reach the water quality goals, we used five independently developed Soil and Water Assessment Tool models to evaluate the effects of 18 management scenarios over a 10-year period on nutrient export. Guidance from a stakeholder group was provided throughout the project, and resulted in improved data, development of realistic scenarios, and expanded outreach. Subsurface placement of phosphorus fertilizers, cover crops, riparian buffers, and wetlands were among the most effective management options. But, only in one realistic scenario did a majority (3/5) of the models predict that the total phosphorus loading target would be met in 9-out-of-10 years. Further, the dissolved phosphorus loading target was predicted to meet the 9-out-of-10-year goal by only one model and only in three scenarios. In all scenarios evaluated, the 9-out-of-10-year goal was not met based on the average of model predictions. Ensemble modeling revealed general agreement about the effects of several practices although some scenarios resulted in a wide range of uncertainty. Overall, our results demonstrate that there are multiple pathways to approach the established water quality goals, but greater adoption rates of practices than those tested here will likely be needed to attain the management targets.

Assessment of Soybean Lodging Using UAV Imagery and Machine Learning.

Plant lodging is one of the most essential phenotypes for soybean breeding programs. Soybean lodging is conventionally evaluated visually by breeders, which is time-consuming and subject to human errors. This study aimed to investigate the potential of unmanned aerial vehicle (UAV)-based imagery and machine learning in assessing the lodging conditions of soybean breeding lines. A UAV imaging system equipped with an RGB (red-green-blue) camera was used to collect the imagery data of 1266 four-row plots in a soybean breeding field at the reproductive stage. Soybean lodging scores were visually assessed by experienced breeders, and the scores were grouped into four classes, i.e., non-lodging, moderate lodging, high lodging, and severe lodging. UAV images were stitched to build orthomosaics, and soybean plots were segmented using a grid method. Twelve image features were extracted from the collected images to assess the lodging scores of each breeding line. Four models, i.e., extreme gradient boosting (XGBoost), random forest (RF), K-nearest neighbor (KNN) and artificial neural network (ANN), were evaluated to classify soybean lodging classes. Five data preprocessing methods were used to treat the imbalanced dataset to improve classification accuracy. Results indicate that the preprocessing method SMOTE-ENN consistently performs well for all four (XGBoost, RF, KNN, and ANN) classifiers, achieving the highest overall accuracy (OA), lowest misclassification, higher F1-score, and higher Kappa coefficient. This suggests that Synthetic Minority Oversampling-Edited Nearest Neighbor (SMOTE-ENN) may be a good preprocessing method for using unbalanced datasets and the classification task. Furthermore, an overall accuracy of 96% was obtained using the SMOTE-ENN dataset and ANN classifier. The study indicated that an imagery-based classification model could be implemented in a breeding program to differentiate soybean lodging phenotype and classify lodging scores effectively.

Projecting the effects of agricultural conservation practices on stream fish communities in a changing climate.

How anticipated climate change might affect long-term outcomes of present-day agricultural conservation practices remains a key uncertainty that could benefit water quality and biodiversity conservation planning. To explore this issue, we forecasted how the stream fish communities in the Western Lake Erie Basin (WLEB) would respond to increasing amounts of agricultural conservation practice (ACP) implementation under two IPCC future greenhouse gas emission scenarios (RCP4.5: moderate reductions; RCP8.5: business-as-usual conditions) during 2020-2065. We used output from 19 General Circulation Models to drive linked agricultural land use (APEX), watershed hydrology (SWAT), and stream fish distribution (boosted regression tree) models, subsequently analyzing how projected changes in habitat would influence fish community composition and functional trait diversity. Our models predicted both positive and negative effects of climate change and ACP implementation on WLEB stream fishes. For most species, climate and ACPs influenced species in the same direction, with climate effects outweighing those of ACP implementation. Functional trait analysis helped clarify the varied responses among species, indicating that more extreme climate change would reduce available habitat for large-bodied, cool-water species with equilibrium life-histories, many of which also are of importance to recreational fishing (e.g., northern pike, smallmouth bass). By contrast, available habitat for warm-water, benthic species with more periodic or opportunistic life-histories (e.g., northern hogsucker, greater redhorse, greenside darter) was predicted to increase. Further, ACP implementation was projected to hasten these shifts, suggesting that efforts to improve water quality could come with costs to other ecosystem services (e.g., recreational fishing opportunities). Collectively, our findings demonstrate the need to consider biological outcomes when developing strategies to mitigate water quality impairment and highlight the value of physical-biological modeling approaches to agricultural and biological conservation planning in a changing climate.

The hydrologic model as a source of nutrient loading uncertainty in a future climate.

Hydrologic models are applied increasingly with climate projections to provide insights into future hydrologic conditions. However, both hydrologic models and climate models can produce a wide range of predictions based on model inputs, assumptions, and structure. To characterize a range of future predictions, it is common to use multiple climate models to drive hydrologic models, yet it is less common to also use a suite of hydrologic models. It is also common for hydrologic models to report riverine discharge and assume that nutrient loading will follow similar patterns, but this may not be the case. In this study, we characterized uncertainty from both climate models and hydrologic models in predicting riverine discharge and nutrient loading. Six climate models drawn from the Coupled Model Intercomparison Project Phase 5 ensemble were used to drive five independently developed and calibrated Soil and Water Assessment Tool models to assess hydrology and nutrient loadings for mid-century (2046-2065) in the Maumee River Watershed,the largest watershedsdraining to the Laurentian Great Lakes. Under those conditions, there was no clear agreement on the direction of change in future nutrient loadings or discharge. Analysis of variance demonstrated that variation among climate models was the dominant source of uncertainty in predicting future total discharge, tile discharge (i.e. subsurface drainage), evapotranspiration, and total nitrogen loading, while hydrologic models were the main source of uncertainty in predicted surface runoff and phosphorus loadings. This innovative study quantifies the importance of hydrologic model in the prediction of riverine nutrient loadings under a future climate.