Globally, many rivers are experiencing declining water quality, for example, with altered levels of sediments, salts, and nutrients. Effective water quality management requires a sound understanding of how and why water quality differs across space, both within and between river catchments. Land cover, land use, land management, atmospheric deposition, geology and soil type, climate, topography, and catchment hydrology are the key features of a catchment that affect: (1) the amount of suspended sediment, nutrient, and salt concentrations in catchments (i.e., the source), (2) the mobilization ,and (3) the delivery of these constituents to receiving waters. There are, however, complexities in the relationship between landscape characteristics and stream water quality. The strength of this relationship can be influenced by the distance and spatial arrangement of constituent sources within the catchment, cross correlations between landscape characteristics, and seasonality. A knowledge gap that should be addressed in future studies is that of interactions and cross correlations between landscape characteristics. There is currently limited understanding of how the relationships between landscape characteristics and water quality responses can shift based on the other characteristics of the catchment. Understanding the many forces driving stream water quality and the complexities and interactions in these forces is necessary for the development of successful water quality management strategies. This knowledge could be used to develop predictive models, which would aid in forecasting of riverine water quality.WIREs Water2018, 5:e1260. doi: 10.1002/wat2.1260This article is categorized under: Science of Water > Hydrological Processes Science of Water > Water Quality

Understanding the factors that influence temporal variability in water quality is critical for designing water quality management strategies. In this study, we explore the key factors that affect temporal variability in stream water quality across multiple catchments using a Bayesian hierarchical model. We apply this model to a case study data set consisting of monthly water quality measurements obtained over a 20‐year period from 102 water quality monitoring sites in the state of Victoria (Southeast Australia). We investigate six water quality constituents: total suspended solids, total phosphorus, filterable reactive phosphorus, total Kjeldahl nitrogen, nitrate‐nitrite (NO _x ), and electrical conductivity. We find that same‐day streamflow has the greatest effect on water quality variability for all constituents. Additional important predictors include soil moisture, antecedent streamflow, vegetation cover, and water temperature. Overall, the models do not explain a large proportion of temporal variation in water quality, with Nash‐Sutcliffe coefficients lower than 0.49. However, when considering performance on a site‐by‐site basis, we see high model performance in some locations, with Nash‐Sutcliffe coefficients of up to 0.8 for NO _x and electrical conductivity. The effect of the temporal predictors on water quality varies between sites, which should be explored further for potential spatial patterns in future studies. There is also potential for further extension of these temporal variability models into a predictive spatiotemporal model of riverine constituent concentrations, which will be a useful tool to inform decision making for catchment water quality management.

This study uses water‐quality data collected over 20 years, from 102 predominantly rural sites across Victoria, Australia, to further our understanding of spatial variability in riverine water quality. We focus on concentrations of total suspended solids, total phosphorus, filterable reactive phosphorus, total Kjeldahl nitrogen, nitrate/nitrite (NO_x), and electrical conductivity. We used an exhaustive search approach to identify the linear models that best link catchment characteristics to time‐averaged constituent concentrations. We ran additional analyses to (1) assess the performance of these models under drought conditions, and (2) understand the key drivers of site‐level variability (standard deviations) of constituent concentrations. Natural catchment characteristics appear to have a greater effect on spatial differences in average constituent concentrations. Performance of the statistical models of time‐averaged constituent concentrations varied, and spatial variability in mean electrical conductivity levels could be more readily explained by catchment characteristics compared to more reactive nutrients. Notwithstanding, the models performed relatively well under varying hydrologic conditions for most constituents. As such, these models provide an insight into the key factors affecting spatial variability in average stream water‐quality conditions. We also identified that hydrologic, climatic, and topographic characteristics of the catchment helped explain the spatial variability in temporal changes in constituents. After calibration and validation, these models of both average water quality and variability in water quality could be used to forecast stream water‐quality responses to future land use, climate, or soil and land management changes.

Our current capacity to model stream water quality is limited – particularly at large spatial scales across multiple catchments. To address this, we developed a Bayesian hierarchical statistical model to simulate the spatiotemporal variability in stream water quality across the state of Victoria, Australia. The model was developed using monthly water quality monitoring data over 21 years and across 102 catchments (which span over 130 000 km²). The modeling focused on six key water quality constituents: total suspended solids (TSS), total phosphorus (TP), filterable reactive phosphorus (FRP), total Kjeldahl nitrogen (TKN), nitrate–nitrite (NO_x) and electrical conductivity (EC). The model structure was informed by knowledge of the key factors driving water quality variation, which were identified in two preceding studies using the same dataset. Apart from FRP, which is hardly explained (19.9 %), the model explains 38.2 % (NO_x) to 88.6 % (EC) of the total spatiotemporal variability in water quality. Across constituents, the model generally captures over half of the observed spatial variability; the temporal variability remains largely unexplained across all catchments, although long-term trends are well captured. The model is best used to predict proportional changes in water quality on a Box–Cox-transformed scale, but it can have substantial bias if used to predict absolute values for high concentrations. This model can assist catchment management by (1) identifying hot spots and hot moments for waterway pollution; (2) predicting the effects of catchment changes on water quality, e.g., urbanization or forestation; and (3) identifying and explaining major water quality trends and changes. Further model improvements should focus on the following: (1) alternative statistical model structures to improve fitting for truncated data (for constituents where a large amount of data fall below the detection limit); and (2) better representation of nonconservative constituents (e.g., FRP) by accounting for important biogeochemical processes.

A new method, which integrates multivariable consisting of soil moisture (SM) active passive (SMAP)-derived SM and vegetation optical depth, the water seasonality, geolocation, digital elevation model (DEM), slope, and biomass as inputs and adopts the technique of bootstrap aggregation of regression trees (BARTs), is proposed for retrieving monthly surface water fraction (SWF) at a spatial resolution of 0.025° from cyclone global navigation satellite system (CYGNSS) data. The model is trained using surface water microwave product series (SWAMPS) data with a coarser resolution of 25 km and then applied to CYGNSS data with an enhanced resolution of 0.025° to generate high-resolution water maps. The resulting CYGNSS SWF (CSWF) maps are evaluated by comparing them with other water data sources, namely, SWAMPS, global surface water (GSW), and global surface water dynamics (GLADs), as well as ground measurements. A quadruple collocation (QC) analysis indicates that the CSWF results exhibit the lowest error variance among the four SWF datasets. Furthermore, additional testing with water level (WL) measurements demonstrates a strong correlation with station data and clear seasonal patterns. Notably, the CSWF estimates significantly improve spatial coverage compared to both optical data (GSW and GLAD) with enhanced spatial resolution and the coarser SWAMPS data. This study underscores the effectiveness and efficiency of CSWF estimates, highlighting their potential as a valuable complement to existing microwave- and optical-based surface water products.