Search for: All records

Water temperature can vary substantially even across short distances within the same sub-watershed. Accurate prediction of stream water temperature at fine spatial resolutions (i.e., fine scales, ≤ 1 km) enables precise interventions to maintain water quality and protect aquatic habitats. Although spatiotemporal models have made substantial progress in spatially coarse time series modeling, challenges persist in predicting at fine spatial scales due to the lack of data at that scale. To address the problem of insufficient fine-scale data, we propose a Multi-Scale Graph Learning (MSGL) method. This method employs a multi-task learning framework where coarse-scale graph learning, bolstered by larger datasets, simultaneously enhances fine-scale graph learning. Although existing multi-scale or multi-resolution methods integrate data from different spatial scales, they often overlook the spatial correspondences across graph structures at various scales. To address this, our MSGL introduces an additional learning task, cross-scale interpolation learning, which leverages the hydrological connectedness of stream locations across coarse- and fine-scale graphs to establish cross-scale connections, thereby enhancing overall model performance. Furthermore, we have broken free from the mindset that multi-scale learning is limited to synchronous training by proposing an Asynchronous Multi-Scale Graph Learning method (ASYNC-MSGL). Extensive experiments demonstrate the state-of-the-art performance of our method for anti-sparse downscaling of daily stream temperatures in the Delaware River Basin, USA, highlighting its potential utility for water resources monitoring and management. 

Abstract Groundwater discharge to streams is a nonpoint source of nitrogen (N) that confounds N mitigation efforts and represents a significant portion of the annual N loading to watersheds. However, we lack an understanding of where and how much groundwater N enters streams and watersheds. Nitrogen concentrations at the end of groundwater flowpaths are the culmination of biogeochemical and physical processes from the contributing land area where groundwater recharges, within the aquifer system, and in the near-stream riparian area where groundwater discharges to streams. Our research objectives were to quantify the spatial distribution of N concentrations at groundwater discharges throughout a mixed land-use watershed and to evaluate how relationships among contributing and riparian land cover, modeled aquifer characteristics, and groundwater discharge biogeochemistry explain the spatial variation in groundwater discharge N concentrations. We accomplished this by integrating high-resolution thermal infrared surveys to locate groundwater discharge, biogeochemical sampling of groundwater, and a particle tracking model that links groundwater discharge locations to their contributing area land cover. Groundwater N loading from groundwater discharges within the watershed varied substantially between and within streambank groundwater discharge features. Groundwater nitrate concentrations were spatially heterogeneous ranging from below 0.03–11.45 mg-N/L, varying up to 20-fold within meters. When combined with the particle tracking model results and land cover metrics, we found that groundwater discharge nitrate concentrations were best predicted by a linear mixed-effect model that explained over 60% of the variation in nitrate concentrations, including aquifer chemistry (dissolved oxygen, Cl⁻, SO₄²⁻), riparian area forested land cover, and modeled physical aquifer characteristics (discharge, Euclidean distance). Our work highlights the significant spatial variability in groundwater discharge nitrate concentrations within mixed land-use watersheds and the need to understand groundwater N processing across the many spatiotemporal scales within groundwater cycling.