Hydrological models for climate-based assessments at the watershed scale: A critical review of existing hydrologic and water quality models

Water availability and extreme events under climate change scenarios in an experimental watershed of the Brazilian Atlantic Forest

Climate change has diversified negative implications on environmental sustainability and water availability. Assessing the impacts of climate change is crucial to enhance resilience and future preparedness particularly at a watershed scale. Therefore, the goal of this study is to evaluate the impact of climate change on the water balance components and extreme events in Piabanha watershed in the Brazilian Atlantic Forest. In this study, extreme climate change scenarios were developed using a wide array of global climate models acquired from the Intergovernmental Panel on Climate Change (IPCC) Sixth Assessment Reports (AR6). Two extreme climate change scenarios, DryHot and WetCool, were integrated into the Soil and Water Assessment Tools (SWAT) hydrological model to evaluate their impacts on the hydrological dynamics in the watershed. The baseline SWAT model was first developed and evaluated using different model performance evaluation metrics such as coefficient of determination (R2), Nash-Sutcliffe (NSC), and Kling-Gupta efficiency coefficient (KGE). The model results illustrated an excellent model performance with metric values reaching 0.89 and 0.64 for monthly and daily time steps respectively in the calibration (2008 to 2017) and validation (2018 to 2023) periods. The findings of future climate change impacts assessment underscored an increase in temperature and shifts in precipitation patterns. In terms of streamflow, high-flow events may experience a 47.3 % increase, while low-flows could see an 76.6 % reduction. In the DryHot scenario, annual precipitation declines from 1657 to 1420 mm, with evapotranspiration reaching 54 % of precipitation, marking a 9 % rise compared to the baseline. Such changes could induce water stress in plants and lead to modifications on structural attributes of the ecosystem recognized as the Atlantic rainforest. This study established boundaries concerning the effects of climate change and highlighted the need for proactive adaptation strategies and mitigation measures to minimize the potential adverse impacts in the study watershed.

Agricultural watershed conservation and optimization using a participatory hydrological approach

Maximizing the impact of agricultural wastewater conservation practices (CP) to achieve total maximum daily load (TMDL) scenarios in agricultural watersheds is a challenge for the practitioners. The complex modeling requirements of sophisticated hydrologic models make their use and interpretation difficult, preventing the inclusion of local watershed stakeholders' knowledge in the development of optimal TMDL scenarios. The present study develops a seamless modeling approach to transform the complex modeling outcomes of Hydrologic Simulation Program Fortran (HSPF) into a simplified participatory framework for developing optimized management scenarios. The study evaluates seven conservation practices in the Pomme de Terre watershed in Minnesota, USA, focusing on sediment and phosphorus pollutant load reductions incorporating farmers' opinions to guide practitioners toward implementing cost-effective CPs. Results show reduced tillage and filter strips are the most cost-effective practices for non-point source pollution reduction, followed by conservation cover perennials. The integration of SAM with HSPF is crucial for sustainable field-scale implementation of conservation practices through enhanced involvement of amateur-modeling stakeholders and farmers directly connected to fields.

Enhancing phosphorus source apportionment in watersheds through species-specific analysis

Phosphorus (P) is a pivotal element responsible for triggering watershed eutrophication, and accurate source apportionment is a prerequisite for achieving the targeted prevention and control of P pollution. Current research predominantly emphasizes the allocation of total phosphorus (TP) loads from watershed pollution sources, with limited integration of source apportionment considering P species and their specific implications for eutrophication. This article conducts a retrospective analysis of the current state of research on watershed P source apportionment models, providing a comprehensive evaluation of three source apportionment methods, inventory analysis, diffusion models, and receptor models. Furthermore, a quantitative analysis of the impact of P species on watersheds is carried out, followed by the relationship between P species and the P source apportionment being critically clarified within watersheds. The study reveals that the impact of P on watershed eutrophication is highly dependent on P species, rather than absolute concentration of TP. Current research overlooking P species composition of pollution sources may render the acquired results of source apportionment incapable of assessing the impact of P sources on eutrophication accurately. In order to enhance the accuracy of watershed P pollution source apportionment, the following prospectives are recommended: (1) quantifying the P species composition of typical pollution sources; (2) revealing the mechanisms governing the migration and transformation of P species in watersheds; (3) expanding the application of traditional models and introducing novel methods to achieve quantitative source apportionment specifically for P species. Conducting source apportionment of specific species within a watershed contributes to a deeper understanding of P migration and transformation, enhancing the precise of management of P pollution sources and facilitating the targeted recovery of P resources.

Long short-term memory models of water quality in inland water environments

Water quality is substantially influenced by a multitude of dynamic and interrelated variables, including climate conditions, landuse and seasonal changes. Deep learning models have demonstrated predictive power of water quality due to the superior ability to automatically learn complex patterns and relationships from variables. Long short-term memory (LSTM), one of deep learning models for water quality prediction, is a type of recurrent neural network that can account for longer-term traits of time-dependent data. It is the most widely applied network used to predict the time series of water quality variables. First, we reviewed applications of a standalone LSTM and discussed its calculation time, prediction accuracy, and good robustness with process-driven numerical models and the other machine learning. This review was expanded into the LSTM model with data pre-processing techniques, including the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise method and Synchrosqueezed Wavelet Transform. The review then focused on the coupling of LSTM with a convolutional neural network, attention network, and transfer learning. The coupled networks demonstrated their performance over the standalone LSTM model. We also emphasized the influence of the static variables in the model and used the transformation method on the dataset. Outlook and further challenges were addressed. The outlook for research and application of LSTM in hydrology concludes the review.