Search for: All records

Post-fire flooding and debris flows are often triggered by increased overland flow resulting from wildfire impacts on soil infiltration capacity and surface roughness. Increasing wildfire activity and intensification of precipitation with climate change make improving understanding of post-fire overland flow a particularly pertinent task. Hydrologic signatures, which are metrics that summarize the hydrologic regime of watersheds using rainfall and runoff time series, can be calculated for large samples of watersheds relatively easily to understand post-fire hydrologic processes. We demonstrate that signatures designed specifically for overland flow reflect changes to overland flow processes with wildfire that align with previous case studies on burned watersheds. For example, signatures suggest increases in infiltration-excess overland flow and decrease in saturation-excess overland flow in the first and second years after wildfire in the majority of watersheds examined. We show that climate, watershed and wildfire attributes can predict either post-fire signatures of overland flow or changes in signature values with wildfire using machine learning. Normalized difference vegetation index (NDVI), air temperature, amount of developed/undeveloped land, soil thickness and clay content were the most used predictors by well-performing machine learning models. Signatures of overland flow provide a streamlined approach for characterizing and understanding post-fire overland flow, which is beneficial for watershed managers who must rapidly assess and mitigate the risk of post-fire hydrologic hazards after wildfire occurs. 

The protection of headwater streams faces increasing challenges, exemplified by limited global recognition of headwater contributions to watershed resiliency and a recent US Supreme Court decision limiting federal safeguards. Despite accounting for ~77% of global river networks, the lack of adequate headwaters protections is caused, in part, by limited information on their extent and functions—in particular, their flow regimes, which form the foundation for decision-making regarding their protection. Yet, headwater streamflow is challenging to comprehensively measure and model; it is highly variable and sensitive to changes in land use, management and climate. Modelling headwater streamflow to quantify its cumulative contributions to downstream river networks requires an integrative understanding across local hillslope and channel (that is, watershed) processes. Here we begin to address this challenge by proposing a consistent definition for headwater systems and streams, evaluating how headwater streamflow is characterized and advocating for closing gaps in headwater streamflow data collection, modelling and synthesis. 

Abstract Hydrologic signatures are quantitative metrics that describe a streamflow time series. Examples include annual maximum flow, baseflow index and recession shape descriptors. In this paper, we use machine learning (ML) to learn encodings that are optimal ML equivalents of hydrologic signatures, and that are derived directly from the data. We compare the learned signatures to classical signatures, interpret their meaning, and use them to build rainfall‐runoff models in otherwise ungauged watersheds. Our model has an encoder–decoder structure. The encoder is a convolutional neural net mapping historical flow and climate data to a low‐dimensional vector encoding, analogous to hydrological signatures. The decoder structure includes stores and fluxes similar to a classical hydrologic model. For each timestep, the decoder uses current climate data, watershed attributes and the encoding to predict coefficients that distribute precipitation between stores and store outflow coefficients. The model is trained end‐to‐end on the U.S. CAMELS watershed data set to minimize streamflow error. We show that learned signatures can extract new information from streamflow series, because using learned signatures as input to the process‐informed model improves prediction accuracy over benchmark configurations that use classical signatures or no signatures. We interpret learned signatures by correlation with classical signatures, and by using sensitivity analysis to assess their impact on modeled store dynamics. Learned signatures are spatially correlated and relate to streamflow dynamics including seasonality, high and low extremes, baseflow and recessions. We conclude that process‐informed ML models and other applications using hydrologic signatures may benefit from replacing expert‐selected signatures with learned signatures. 

Abstract Dominant processes in a watershed are those that most strongly control hydrologic function and response. Estimating dominant processes enables hydrologists to design physically realistic streamflow generation models, design management interventions, and understand how climate and landscape features control hydrologic function. A recent approach to estimating dominant processes is through their link to hydrologic signatures, which are metrics that characterize the streamflow timeseries. Previous authors have used results from experimental watersheds to link signature values to underlying processes, but these links have not been tested on large scales. This paper fills that gap by testing signatures in large sample data sets from the U.S., Great Britain, Australia, and Brazil, and in Critical Zone Observatory (CZO) watersheds. We found that most inter‐signature correlations are consistent with process interpretations, that is, signatures that are supposed to represent the same process are correlated, and most signature values are consistent with process knowledge in CZO watersheds. Some exceptions occurred, such as infiltration and saturation excess processes that were often misidentified by signatures. Signature distributions vary by country, emphasizing the importance of regional context in understanding signature‐process links and in classifying signature values as “high” or “low.” Not all signatures were easily transferable from single, small watersheds to large sample studies, showing that visual or process‐based assessment of signatures is important before large‐scale use. We provide a summary table with information on the reliability of each signature for process identification. Overall, our results provide a reference for future studies that seek to use signatures to identify hydrological processes.