Climate models simulate precipitation and its extremes under climate change (nonstationarity). However, they often have systematic biases that require correction before practical use at local scales. Conventional correction approaches for extreme precipitation do not deal with climate change very well, as they lack explicit and continuous nonstationarity treatment (are, in fact, stationary or quasi-stationary) and are challenged by scarce extreme-event data and high uncertainty. We propose a novel bias correction approach for extreme precipitation that explicitly models continuous nonstationarity due to climate change and leverages information from both ordinary and extreme events. Specifically, we introduce nonstationary quantile mapping and propose incorporating the simplified Metastatistical Extreme Value (SMEV) distribution. We demonstrate the superiority of the proposed method through a simulation study and real-world applications using high-resolution-regional and coarse-resolution-global climate models. Nonstationary quantile mapping reflects nonstationarity more realistically, but when relying on extreme-event records only, faces large estimation errors and uncertainty due to data limitations. These issues, common in conventional approaches, are effectively mitigated by using the SMEV distribution. The proposed nonstationary quantile mapping leveraging the SMEV distribution offers lower estimation error, approximate unbiasedness, reduced uncertainty, and improved representation of the entire distribution, especially with typical long datasets. Other methods may perform competitively with short samples, but they exhibit large biases in quantile-quantile matching due to bypassing nonstationarity modelling. The proposed method avoids these biases, aligns better with their theoretical functioning, and enhances the correction of extremes under climate change.
Growing concerns about the adverse environmental effects of agriculture have led to the establishment of various government-led or partnership programs to incentivize farmers to implement agricultural conservation practices or beneficial management practices (BMPs) such as conservation tillage, nutrient management, and riparian and wetland restoration for improving water quality and other benefits. In this presentation I introduce the development of a GIS-based fully distributed hydrologic model, namely Integrated Modelling for Watershed Evaluation of BMPs (IMWEBs), for evaluating the water quality benefits (sediment and nutrient reductions) at site, field, farm, watershed, and river basin scales. The IMWEBs characterizes climate, runoff, sediment, plant growth, and nutrient processes from overland to streams to the watershed out. In addition, the IMWEBs characterizes crop management (e.g., crop rotation and tillage), manure and nutrient management (e.g., manure setback and catch basin), riparian and surface water management (e.g., riparian buffer and wetland restoration), wintering site management (e.g, alternating wintering site annually), pasture management (e.g., rotational grazing) and marginal cropland management (e.g., conversion to tame and native perennials). The IMWEBs utilizes climate, topography, soil, landuse, land management and other data for setup and uses flow and water quality monitoring data for calibration and validation. The calibrated IMWEBs can be then used to estimate location-specific watershed quality benefits for various BMP scenarios. The IMWEBs has been applied to quantify water quality benefits of existing and future agricultural conservation practices for evaluating agri-environmental program performance and planning further investments on these programs in provinces of Alberta, Manitoba and Ontario in Canada.
Peaks over threshold flood events in seventy years of data from 202 reference hydrometric stations from cold regions in Canada and the United States were separated into nival, mixed, and pluvial flood types using clustering on the unit circle, climatology, and median daily flows. These groups were then tested for trends over time, with climate indices, with mean annual temperature and with mean annual precipitation. Mann-Kendall trend tests showed few significant trends in flood magnitude. However, using logistic regression, significant changes in flood type fraction were found over time, with annual mean temperature, and with annual precipitation. Nival events decreased in frequency over the seventy-year period in 16% of sites, while mixed and pluvial events increased at 5% and 12% of sites, respectively. These changes indicate a shift from nival events towards more pluvial dominated systems. Fewer significant changes in flood type fraction were found with analysis against four climate indices. Flood frequency analysis using a combined distribution approach with the three flood types resulted in larger magnitude design flow estimates (median increase of 20 – 30 %) in comparison with the results from considering the data to be from a single population.
Isotope tracers can benefit hydrologic modeling, by adding observational data relating to groundwater-surface water interactions, evaporation and water ages as a supplement to flow data. However, to realize these benefits, an isotope tracer model must be linked to the hydrologic model. A key barrier to more wide-spread application isotope tracers in hydrologic modeling is the considerable effort required to add an isotope tracer simulation to an existing model, which requires an uncommon overlapping expertise in both hydrologic model development and isotope tracer science. To reduce the barriers to entry, a model agnostic isotope tracer simulator (MAITsim) has been developed, which can simulate two stable isotope tracers (deuterium and oxygen-18) in association with a wide range of hydrologic models. _x000D_ _x000D_ MAITsim runs as a post-processing model using outputs from a hydrologic model as inputs, such that only the model specific linkage needs to be set up to simulate both flow and isotope tracers. This numerically stable tracer simulator is compatible with any flux-state model with unidirectional flow paths, as it uses no pre-determined spatial sub-divisions (any combination of soil layers, sub-catchments and hydrologic response units can be linked to MAITsim). The model includes both mixing and evaporative fractionation and its equivalence to a previously published embedded isotope tracer model has been verified. This new open-source model can be used to improve modeling of surface-groundwater interactions or as a template to embed stable isotope tracer simulations in more hydrologic models.
Understanding the impact of climate model resolution on the representation of extreme hydrometeorological events is essential for improving flood risk assessment. This study evaluates the added value of convection-permitting climate simulations by comparing three hourly Weather Research and Forecasting (WRF) model simulations, driven by ERA-Interim reanalysis, at spatial resolutions of 4 km, 25 km, and 50 km. The ability of these simulations to reproduce observed extreme summer-fall precipitation and annual maximum summer and fall flood events is assessed across 12 southern Quebec watersheds (61–1550 km²). The ERA5-Land dataset serves as a reference for bias correction and hydrological model calibration. Streamflows are simulated using the GR5dt hydrological model at hourly time step. Preliminary results indicate that all three bias-corrected WRF simulations perform well in capturing extreme precipitation and flood events. However, the convection-permitting 4 km simulation better reproduces flood volume and peak flows in larger watersheds (>750 km²). These findings suggest that hourly-scale hydrological modeling is key for accurately representing summer-fall extreme events. Further analysis is ongoing to refine the comparison between the three WRF simulations and assess their implications for flood modeling.
Hydrological forecasting plays a critical role in water resources management and public safety. Although physically based models (PBM) have been widely used for hydrological forecasting, their applications in data-sparse high-latitude regions presents significant challenges. Deep learning methods such as long short-term memory (LSTM) integrating lagged observations have shown promises for streamflow prediction, but their effectiveness for forecasting remains uncertain. This study compared the performance of an LSTM model to that of a semi-distributed PBM, HYDROTEL, trained with the same historical data and forced with the same meteorological ensemble forecasts, to forecast short-range (0-14 days lead) streamflow and inflow at four sites in Yukon Territory, Canada. The LSTM model used Data Integration (DI) of recent observed flow and inflow, with or without DI of Snow Water Equivalent (SWE), while HYDROTEL used the same flow/inflow data as part of a Data Assimilation (DA)an Ensemble Kalman Filtering (EnKF) framework. Results demonstrated that for all lead times, the LSTM-DI model outperformed that of HYDROTEL-EnKF at three sites. When incorporating SWE data, the LSTM-DI model further surpassed HYDROTEL-EnKF at all sites. Compared to the version integrating only flow/inflow data, integrating SWE data offered meaningful advantages, particularly for predicting spring peak inflows at two sites, while performing similarly for the other sites. This study highlights the reliability of LSTMs for short-range hydrological forecasting, and the added value of including snow information on the forecasting skill in data-sparse high-latitude regions.
Among numerous drought indices, the Standardized Precipitation Index (SPI) and Standardized Precipitation Evapotranspiration Index (SPEI) are widely used for meteorological droughts, while the Standardized Streamflow Index (SSI) is commonly employed for hydrological droughts. These indexes are typically calculated for various timescales (from a week to several years). As drought characteristics are sensitive to the indexes and the timescales, their knowledge-informed selections are desired. Furthermore, their selections should consider seasonality, as the cause of hydrological drought can vary in different seasons, especially for streams fed by rainfall and snowmelt/ice-melt. Therefore, in this study, we demonstrate the differences in the characteristics of SPI- and SPEI-based droughts between summer and winter. We then propose a selection approach to determine the suitable time scale for SSI using the total water storage anomaly based on the notion that watershed storage modulates hydrological drought, and then the suitable meteorological drought index (SPI or SPEI) and its time scale, according to the correlation between meteorological and hydrological droughts for each season. The case study is conducted in the head watershed (above Banff) of the Bow River in Alberta. The results indicate significant differences between SPI- and SPEI-based droughts in terms of severity and duration, particularly during summer, as well as the drought onset time. The timescales of 3- and 6-month are most appropriate for SSI, while 12- and 18-month SPI is most suitable for meteorological droughts in summer and winter, respectively. This study offers valuable insights for improved drought monitoring.
Intensity-Duration-Frequency (IDF) curves describe the relationship between rainfall intensity, duration, and return period at a given location. They are crucial for infrastructure design, flood risk assessment, and stormwater management but are typically available only at specific measurement sites. A previous approach was developed to interpolate IDF curves by incorporating past climatology into a hierarchical Bayesian model using a Gaussian Markov random field. While this method improved accuracy and computational efficiency, its application was limited to Eastern Canada. This talk presents the integration of the Canadian Surface Reanalysis (CaSR) for interpolating extreme precipitation characteristics across Canada. The use of this reanalysis has been shown to enhance predictive performance, enabling more accurate IDF curve interpolation across Canada. By refining IDF curve estimation, this research supports improved planning, infrastructure resilience, and more effective water management strategies.
The data used to calibrate hydrologic models are often captured at spatial scales that differ from the model being developed. This is especially true for snow-related datasets used in watershed-scale hydrologic models. High-quality datasets like snow pillows and snow courses capture snowpack processes at a single point, which have the potential of introducing localized, site-specific effects that aren't relevant at larger scales. Conversely, regional-scale data products, like remote sensing, may be too coarse and fail to sufficiently capture local nuances. Accurate snowpack models that offer a good match to snow accumulation and melt observations improve the representation of watershed conditions and enhance predictive power when assessing future climate impacts. An approach that appropriately considers a combination of at-a-point data and regional data products is crucial for developing accurate hydrologic models across Canada._x000D_ _x000D_ A recent study in the eastern Rockies provided insights into snowpack modelling challenges and opportunities. A distributed snowpack model covering 4,000 km² was developed using various direct observations and remote sensing data. This presentation will discuss available data sources for practitioners in British Columbia and approaches to selecting and utilizing calibration, verification, and validation datasets. It will also present strategies to adapt snow data observations from various scales for use in watershed models. The presentation will illustrate how using both snow depth and snow-water equivalent (SWE) data can result in a robust calibration and provide examples of how this approach improves the representation of the hydrologic response in the study area._x000D_
Hydrologic and human systems are deeply interconnected, but their modelling has traditionally been siloed. As the need for more holistic approaches grows, convergent and transdisciplinary integration is emerging as a key frontier in hydrologic research. This presentation provides an overview of this evolving endeavor through the lens of Razavi et al. (2025; Convergent and transdisciplinary integration: On the future of integrated modeling of human‐water systems. Water Resources Research, 61, e2024WR038088. https://doi.org/10.1029/2024WR038088), highlighting key frontiers that distinguish surface and groundwater hydrology, engineering, social sciences, economics, Indigenous and place-based knowledge, and other interconnected natural systems such as the atmosphere, cryosphere, and ecosphere. We argue that a fundamental gap persists: hydrologic models often disregard management interventions, while water resources management models rarely account for hydrologic feedbacks. This disconnect can lead to inaccurate predictions and suboptimal decision making, particularly in flood forecasting and reservoir operations. For instance, the state of the art in flood research and inundation mapping is often limited to a “weak coupling” of models of hydrology, reservoir systems, and river hydrodynamics, typically involving the addition of simplistic representations of reservoirs and water withdrawals to hydrologic and hydrodynamic simulations, which may not adequately capture management complexities. To illustrate these challenges and opportunities, we leverage new features of the Raven watershed modelling framework, applied to the Assiniboine River Basin, to assess how integrated modelling can improve predictive capabilities and better support decision making.
Southern Alberta has experienced significant water shortages and prolonged drought during the past few years. A lack of snowfall during winter or early melting of snowpack due to warmer winter temperatures and earlier spring periods can cause snow drought. Cool and dry Chinook winds blown from western Canada increase the snow melt in southern Alberta. This research focuses on an automated classification scheme for the types of snow droughts that can occur in southern Alberta during 1980-2023. The standardized SWE index (SWEI) is applied over 6 months from October to May to identify snow drought years. Snow drought years are then classified into three categories namely, warm, warm and dry, and dry. The ratio between mean Snow Water Equivalent (SWE) and cumulative precipitation (P) is used where P is utilized as a proxy to represent the effect of temperature in snow melt. K – means clustering method is used for the classification. Further, the effect of elevation on snow drought is investigated as topography changes substantially from the high-elevation Rocky Mountains to the southern Alberta plains in the east. Snow Time-series: Building Yearly Targeted Ensembles (FROSTBYTE) workflow will be used to gap-fill the Canadian historical SWE dataset (CANSWE). The frequency and intensity of snow drought are greater in the plains than in mountainous areas. The snow drought classification workflows and standard SWEI calculation are developed and available in the UC-HAL GitHub repository.
Droughts, driven by hydrometeorological processes that reduce precipitation and water availability, are extensive natural hazards causing significant socioeconomic impacts worldwide. Southern Alberta, part of the Canadian Prairies, has a history of recurrent droughts, including notable events like the "Dust Bowl" of the 1930s and the 2001–2002 drought, which led to over $5.6 billion in GDP losses in Canada. Effective drought management relies on their monitoring and forecasting based on indices that quantify drought severity, timing, duration, and location. Among the many indices developed, the Standardized Precipitation Index (SPI), Standardized Precipitation Evapotranspiration Index (SPEI), Palmer Drought Severity Index (PDSI), and Reconnaissance Drought Index (RDI) are widely used, each suited to specific conditions and datasets. A index suitable at particular region for particular purpose may not be suitable for other region due to inherent complexity of drought process, and varying geophysical and climatic conditions. This study seeks to identify the most effective drought index for Southern Alberta by comparing four indices against historical drought records and hydrometeorological data from 37 stations over six decades. The intercomparison was assessed using correlation and Cohen’s kappa analysis, while their strength to reflect other aspects of drought was validated by comparing against streamflow and groundwater level data at multiple timescales. Furthermore, the drought condition reflected by the indices were compared against historical drought maps to assess their effectiveness. The findings will contribute to actionable measures for drought preparedness and mitigation in the region.
The Ontario hydrological modelling assessment project applies the Raven Hydrological modelling framework in multiple configurations to simulate the hydrologic behaviour across all of Ontario. This large-scale modelling effort is applied to the nine regions of Ontario and here we present comparative modelling results for the 35,500 km2 SE region, draining to the Ottawa River._x000D_ _x000D_ Our comparative results focus on streamflow and water level prediction. We benchmark our semi-distributed hydrologic models in Raven against a lumped Raven hydrologic model. We compare our semi-distributed hydrologic model against our vector-based routing-only Raven model and both models have 1) the same lake and river routing networks (1057 subbasins, 311 lakes) and 2) are ultimately forced with the same meteorological inputs (Canadian Surface Reanalysis or CaSR ver 2.1). In the case of our routing model, the surface runoff and baseflow forcings to Raven were taken as Environment and Climate Change Canada’s (ECCCs) calibrated GEM-Hydro model outputs produced by forcing GEM-Hydro with CaSR v2.1. _x000D_ _x000D_ Results show the value of distributed modelling with detailed treatment of lakes relative to lumped modelling and also show the improvements associated with a full hydrologic model versus a routing-only model constrained to utilize ECCC land surface hydrologic model fluxes. Furthermore, we show the differences between a regionally calibrated hydrologic model and watershed specific calibration of the same hydrologic model and then demonstrate how we combine these two models together, along with historical flows and lake levels, to create a high resolution 39-year streamflow and lake level reanalysis for the SE region.
Global warming is altering the water cycle, driving changes in the intensity, seasonality and timing of hydro-climatic extremes such as flooding and extreme precipitation. These shifts have significant implications for soil erosion, landslides, and debris flow, requiring a comprehensive analysis of historical trends and future projections. This research integrates historical observations, advanced statistical techniques, and climate model projections to assess large-scale changes in hydro-climatic extremes. Results reveal notable shifts in the timing and seasonality of extreme precipitation, with future projections indicating an intensification of rarest events. These findings highlight the need for improved downscaling techniques and bias correction methods to enhance predictions at finer spatial and temporal scales. This research underscores the critical role of robust statistical frameworks in understanding and mitigating the impacts of extreme events in a changing climate. Particularly as a key driver of hydrological prediction, and understanding how representing precipitation extremes translates to hydrological response.
British Columbia has suffered multiple record-setting summer droughts in the last decade. Historically low summer streamflows have devastated many fish populations, threatened Indigenous cultural and ceremonial uses of water, and forced the province to curtail water use for irrigation, with serious economic consequences. However, unlike other jurisdictions with similar problems, British Columbia has no seasonal streamflow drought forecast system to provide advanced warning to water managers, conservation groups, Indigenous rights-holders, and farmers of the likelihood of extreme drought conditions. _x000D_ We have developed seasonal (1 to 6 month lead time) streamflow forecast models for 207 unregulated, gauged streams in British Columbia, which predict the probability of reaching drought levels 1 through 5 in the late summer. The models assimilate real-time streamflow and snowpack data with 6-month meteorological forecasts from eight weather forecast centres around the world. We compare regression techniques that have been used for decades with Long Short-Term Memory models. For both, we find that most of the uncertainty in the hydrologic prediction derives from uncertainty in the meteorological forecasts. Forecast skill on April 1st is only marginally better than a reference (climatological) forecast but improves considerably throughout the spring and summer as the lead time decreases._x000D_ We will be disseminating forecasts on a monthly basis throughout spring and summer 2025, via an interactive web application at https://sruzzante.shinyapps.io/BC-streamflow-drought-forecast/.
A storm rating scale based on impacts for a warming climate has been developed using precipitation intensity duration frequency (IDF) and Atmospheric River (AR) concepts. Initially devised as a five-level scale, it incorporates antecedent moisture conditions and sensitivity to water input rate and duration. Our ECAR team of Environment and Climate Change Canada, Vancouver BC, along with the Climate Research Division in Victoria, BC, led this initiative with assistance from numerous Canadian scientists despite challenges presented by COVID pandemic isolation. Dr. Matthias Jakob of BGC Engineering, the BC River Forecast Center, and AR experts from NOAA-USGS and UC Santa Barbara also provided significant contributions._x000D_ The ECAR scale is grounded in the concept of Atmospheric Rivers, which involves vertically integrated vapor transport and storm duration, linking precipitation intensity and duration to impacts based on engineering criteria, local hazards, and expertise. This storm rating scale holds considerable potential in improving natural hazards forecasting, providing "Heads up" advisories days to weeks in advance. Projections indicate that future storm impacts will be more severe based on Atmospheric Rivers, with a warmer climate expected to bring more intense and frequent storms than our current engineering can withstand._x000D_ To address increasingly frequent and intense storms, the scale now includes eight levels, although level 5 remains catastrophic according to current engineering design criteria and historical IDF statistics. Findings suggest that substantial enhancements in emergency designs and infrastructure are necessary to adapt to the anticipated changes resulting from ongoing global warming.
Hydrological modelling is a valuable tool to support sustainable and resilient water management, particularly as we adapt to climate change. The calibration of hydrological models, however, can be a difficult and daunting task. This process varies significantly depending on the modeller, available data, model objectives, calibration parameters, and the optimization algorithms used, among others. This results in a wide range of model parameter sets that often lack reproducibility and consistency._x000D_ _x000D_ This research proposes an agnostic strategy to create workflows to calibrate hydrological models, with the aim to generalize the calibration process to create nearly replicable calibrated models. This strategy can be used to calibrate lumped, semi-distributed, or fully distributed models. The framework is tested using the MESH (Modélisation Environnementale communautaire - Surface Hydrology) model to calibrate a vector-based model for the province of Alberta. The results are compared by analyzing the sensitivity of the calibration to different variables within the strategy, showing the potential of this approach to enhance reproducibility and model practices compared to traditional strategies.
Increasing reliance on the rivers of southern Alberta for irrigation and other uses has heightened vulnerability to hydrological drought. While water supplies have been monitored since the early 20th century, these hydrometric records do not capture the full range in the severity and duration of hydrological drought as revealed by previous research on the paleohydrology of the South Saskatchewan River Basin (SSRB). Using a two-stage approach to modeling runoff from tree-ring data, we reconstructed 1000 years of annual and seasonal streamflow at 75 gauges in the SSRB. The two stages – single site reconstructions and principal component analysis – take full advantage of our network of 599 tree-ring chronologies to develop multiple linear regression models that account for up to 80% of the variance in naturalized streamflow. These long runoff records underscore the risks in relying solely on instrumental data for informing water resource policy, management practices and planning. They also provide critical context for assessing uncertainty in the climate model projections that are used to force hydrological models for the prediction of hydrological drought. Research suggests that natural hydroclimatic variability, as recorded by radial tree growth, is the dominant source of uncertainty for the model projection of water balance variables.
Agricultural production in BC's Delta region relies heavily on irrigation water from the Fraser River, which is managed by the City of Delta through an extensive network of canals, gates, and pumps. While the quantity of water is most often adequate, the salinity at the intake points is influenced by the interface of tidal water with fresh river water. When river flows are low, the “salt wedge” tends to creep further up the channel, resulting in water with higher salt concentration, posing a significant risk of crop damage if used for irrigation. This problem has been exacerbated by the impacts of climate change, which have made seasonal river flows increasingly unpredictable, particularly during summer months when water demand is at its peak. To address these growing concerns, the Delta Farmers Institute partnered with Peak HydroMet Solutions to install a network of real-time salinity monitoring stations throughout the region. An online portal has been developed to provide farmers and water managers with information needed to take proactive measures to effectively manage water quality. The implementation of this highly successful program has demonstrated the valuable role that monitoring technologies can play in water management. By leveraging accurate and timely data, farmers can adjust their irrigation practices to mitigate the effects of high salinity levels, thereby protecting their crops and ensuring sustainable production. Additionally, the collaboration between various stakeholders, including farmers, technology providers, and local governments, underscores the importance of partnerships in addressing environmental challenges.
The goal of achieving higher water use efficiency in irrigated agriculture hinges on the ability of irrigators to apply the right amount of water at the right time and right location to a crop. In large-scale field agriculture the challenge is to have access to data of the right granularity to optimize this decision-making process. _x000D_ Here, we present temporal and spatial data collected at irrigated fields of the Lethbridge Polytechnic Research and Demonstration Farm. These datasets include high-frequency timeseries of volumetric water content, matric potential, and evapotranspiration; high-resolution maps of brightness temperature (as a proxy of volumetric water content); and low-frequency / low-resolution data series of manual soil moisture assessment and estimated crop water use._x000D_ We compare error margins of the different datasets to identify where relevant gains in irrigation scheduling can be made. We also evaluate the precision and error margins of the datasets to those of irrigation equipment to see what is achievable at the field scale._x000D_ These analyses form a roadmap for work needed to advance data driven irrigation management.
Climate risk assessments are deemed essential, especially in an alarming shift in climate patterns, increasing the risk of an uncertain future. Smart planning is necessary to mitigate the impacts of climate change and adapt to a sustainable environment. As water scarcity is one of the notable consequences, effective and efficient management of water resources is crucial. Agriculture, the largest consumer of freshwater resources, should account for proper irrigation management to sustain crop production. In this study, long-term irrigation requirements were projected using downscaled data from CMIP-6 models. These models utilize Shared Socio-economic Pathways (SSPs), which explore the effect of socio-economic trends on greenhouse gas emission concentration. Precipitation and temperature data from 27 models were downloaded and analyzed to estimate irrigation requirements under two emission scenarios (SSP2-4.5 and SSP5-8.5). Several irrigation triggers were established and tested through the AquaCrop model to determine the best adaption strategies corresponding to the emission scenarios. Model uncertainty is represented by presenting the 25th, median and 75th percentile of irrigation requirement for four distinct timelines: historical (1981-2010), near-term (2011-2040), mid-term (2041-2070) and end of century (2071-2100). The results from this study will be highly beneficial for stakeholders involved in water budgeting and allocation.
Climatic changes and intensive agriculture bring into question the sustainability of Canada's food production while jeopardizing the health of the freshwater ecosystems on which we all rely. _x000D_ To address these challenges, watershed managers in the Canadian Prairies have access to more raw data than ever, thanks to collaborations between citizen scientists, governments, academic institutions, private industry, and environmental non-profits. In addition, there is a growing number of modeling tools available to interpret these data. However, applying and optimizing these models can be time-consuming and computationally demanding, such that decision-ready data remains unavailable to many stakeholders. _x000D_ The Prairie Watershed Analytics (PWA) project is addressing this challenge by developing a free, open, and automated protocol for fast and easy deployment of hydrologic models specially tailored to prairie watersheds. Prioritizing access and usability, PWA leverages public hydrological datasets, open-source software, and widely available phosphorus data collected by organizations like the Lake Winnipeg Foundation. Using machine learning techniques, the project helps users to understand the hydrology of their local watershed, along with nutrient sources and sinks. _x000D_ This modeling method is designed to be replicated to watersheds across the prairies at low cost. The PWA project seeks to elevate collaborative decision-making processes, offering insights to support agricultural water stewardship programs, forecasting climate scenarios, and informing long-term watershed planning.