You are here:
StreamCat and LakeCat: An overview of algorithms, data, and models developed at the US EPA Western Ecology Division to facilitate and advance watershed prediction in the conterminous US.
Citation:
Hill, R., M. Weber, AND S. Leibowitz. StreamCat and LakeCat: An overview of algorithms, data, and models developed at the US EPA Western Ecology Division to facilitate and advance watershed prediction in the conterminous US. To be Presented at 2018 AWRA Spring Specialty Conference: GIS & Water Resources X, Orlando, Florida, April 22 - 25, 2018.
Impact/Purpose:
Over the last four years, scientists at the US EPA’s Western Ecology Division have worked to develop data and methods to improve our ability to model, map, and understand the condition of the Nation’s water resources. These data allow scientists to easily obtain watershed information for streams and lakes across the continental US and are called the StreamCat and LakeCat Datasets, respectively. These datasets include information on the percent of the watershed composed of land uses (e.g., urbanization or agriculture) and natural features (e.g., precipitation and soils). These types of information can be important for understanding human-related effects on the biological, physical, and chemical conditions of water resources. StreamCat and LakeCat data are available for ca. 2.6 million stream segments and ca. 378,000 lakes across the US. This presentation provides and overview of these efforts. It will introduce the algorithms that were critical for developing these datasets to an audience of geospatial experts. In addition, we will introduce a project that has not been presented on before that characterizes the connectivity of ca. 4.6 million wetlands to nearby streams and how these data are being used to understand the important role of wetlands as filters of agricultural fertilizers before they reach streams. StreamCat and LakeCat support the development of robust national maps of stream and lake conditions, which is of interest to the Monitoring Branch within the Office of Water. The data produced by this work have been used to model the biological condition of streams and in modeling chlorophyll a concentrations of lakes across the CONUS. StreamCat contributed to an FY16 deliverable and LakeCat contributed to an FY17 deliverable under SSWR 3.01B.
Description:
Geospatial data and techniques have long been critical to advancing the analysis and management of freshwater ecosystems. However, these data and techniques have often been limited to specific sample sites or regional analyses because of the difficulty associated with generating spatial summaries for hundreds to thousands of nested watersheds. Over the last four years, the US EPA Western Ecology Division has developed algorithms and datasets that characterize landscape factors that influence water bodies across the conterminous US. These datasets and algorithms are or will be publically available and facilitate the acquisition of watershed covariates for modeling and prediction of aquatic conditions. The first of these efforts produced the StreamCat dataset, which uses the National Hydrography Dataset Plus (version 2) to characterize several hundred natural (e.g., climate and soils) and anthropogenic (e.g., agriculture and urbanization) watershed features for ~2.6 million stream segments. We linked these data to an existing stream biological assessment to provide covariates in a model that predicted the probable biological condition at ~1.1 million stream segments. Furthermore, we are using these data to explore the landscape factors associated with the abundance of specific antibacterial-resistance genes within US waterways. Next, we expanded these algorithms developed for StreamCat to delineate and hydrologically link nested lake catchments. This effort produced the LakeCat dataset: watershed summaries of several hundred landscape features for ~378,000 lakes across the conterminous US. We used these data to model and predict the risk of eutrophication at ~291,000 lakes within the US and are currently exploring landscape factors that elevate the risk of harmful algal blooms. Our current project reversed our algorithms to connect flow pathways rather than nested catchments of wetlands. We used this reversed algorithm to connect the pathways of 4.6 million wetlands to the nearest downslope streams. By summarizing soil, topographic, levee, and vegetation information along these pathways, we developed a classification of wetland-stream connectivity for the conterminous US. We are using this characterization to model instream nitrogen and understand the role that wetlands, and their connectivities, play in mitigating nutrients from anthropogenic sources. The ability to summarize and aggregate fine-grained geospatial data in hydrological networks for thousands to millions of water bodies is an important innovation that should improve the way that we apply predictive models, understand factors that influence their conditions, and ultimately manage these important ecosystems. This presentation will provide an overview of these methods and datasets and how we have used them to model the ecological, chemical, or physical condition of the Nation’s water bodies.