SPARROW modeling: Estimating nutrient, sediment, and dissolved solids transport

Featured: Projecting future flow in Southwest streams

Featured: Projecting future flow in Southwest streams

Streamflow in the Southwestern U.S. is projected to decrease by as much as 36–80% by the end of this century, reports a new study by the U.S. Geological Survey. These decreases could challenge our ability to meet future water demand in this region.

Learn More

Featured: Nutrient yields in the Mississippi/Atchafalaya River Basin

Featured: Nutrient yields in the Mississippi/Atchafalaya River Basin

new USGS study uses SPARROW modeling to estimate total nitrogen (N) and total phosphorus (P) yields from catchments throughout the Mississippi/Atchafalaya River Basin U.S. Results could assist nutrient reduction strategies.

Learn More

Science Center Objects

SPARROW (SPAtially Referenced Regression On Watershed attributes) models estimate the amount of a contaminant transported from inland watersheds to larger water bodies by linking monitoring data with information on watershed characteristics and contaminant sources.  Interactive, online SPARROW mapping tools allow for easy access to explore relations between human activities, natural processes, and contaminant transport.

Integration of monitoring and modeling is critical to our future understanding and management of the Nation’s water quality. Monitoring is the direct observation, often over time, of water-quality properties and characteristics, and models are tools for interpreting these observations. 


SPARROW models are used to estimate long-term average values of water characteristics, such as the amount of a contaminant that is delivered downstream, based on existing monitoring data, location and strength of contaminant sources, and characteristics of the landscape.

SPARROW models can help managers:

  • Determine options for reducing loads of contaminants
  • Design strategies for protection or to meet regulatory requirements
  • Predict changes in water quality that might result from management actions
  • Identify gaps and priorities in monitoring network design


USA Map showing Northeast, Southeast, Midwest, Southwest, and Pacific regions

The regions delineated for use in Spatially Referenced Regressions on Watershed Attributes (SPARROW) models in the conterminous United States


SPARROW mappers are interactive tools that allow the user to explore river nutrient loads and yields and the importance of different sources of contaminants in a particular river basin.

Data can be visualized using maps and interactive graphs and tables, and rankings can be shown by catchment, watershed, and state. Modeling results can be exported as an Excel spreadsheet, CSV file, or a geospatial dataset.

New mappers, representing circa 2012 source inputs, are available for 5 regions of the conterminous United States. The Mappers replace the SPARROW Decision Support System (Booth and others, 2011).

Nutrients and the Nation’s Estuaries

Access maps of watershed nutrients flowing to the Nation’s estuaries and download data tables of nutrient sources and loads. Compare nutrient sources and watersheds that contribute elevated nutrient loads to downstream receiving waters, such as the Southeast Atlantic and Gulf of Mexico, inland and coastal waters of the Northeast, the Upper Mississippi and Great Lakes, Puget Sound and the Northwest coast, and the California coast.

Estuary and bird

The San Francisco Bay Estuary.


National Models

SPARROW models are unique in that they retain the spatial detail of underlying data sets while extending over areas as large as the conterminous United States. This allows the simultaneous assessment of water-quality conditions in many water bodies. National SPARROW models have been developed for a number of water-quality constituents including nutrients and total dissolved solids.


Regional Models

SPARROW models are flexible—they can be applied to any region where there are specific needs for water-quality information and where data to support modeling are abundant.  Five new regional models of streamflow, total nitrogen, total phosphorus and suspended sediment have been developed for the conterminous United States.  Other regional models have been developed previously for the Chesapeake Bay, Mississippi River and the Great Lakes watersheds.


International Models

SPARROW models can be applied in any part of the world where sufficient data are available to support model development. Examples include models developed for New Zealand to identify the primary sources of nutrients to streams (Alexander and others, 2002), and a joint U.S.-Canadian effort to build nutrient models for the entire Great Lakes watershed, to better understand nutrient loading to the lakes (Robertson and others, 2019).


Applications of SPARROW models

Once built and calibrated, SPARROW models can be applied in a variety of ways to better understand the environmental factors affecting water-quality conditions in streams.

  • SPARROW models were applied at the national scale to estimate natural background levels of nutrients to help guide the potential development of nutrient criteria in streams.
  • A model developed for the upper Midwest was used to identify the benefits of management practices designed to limit the amount of agricultural nutrients reaching streams (Garcia and others, 2016).
  • A Chesapeake Bay SPARROW model was used to identify those areas that export nitrogen to streams with the greatest efficiency (Ator and Garcia, 2016).
  • A SPARROW model was used to simulate impacts of climate change on phosphorus load to Lake Michigan (Robertson and others, 2016).


Research using SPARROW models

SPARROW models can be used as tools in research to better understand the environmental processes that affect water-quality conditions. A SPARROW model for the Mississippi drainage was used to evaluate the role of stream size on denitrification and attenuation of nitrogen levels with transport downstream. Previous studies based on a limited number of sites had hypothesized that the rate of denitrification was significantly greater in smaller streams due to greater sediment contact. This was confirmed by the SPARROW model for which the results were consistent with the other studies, but based on a much larger area and many more measurement locations (Alexander and others, 2000).


The national data bases used in the SPARROW models have value in themselves and can be used for other scientific evaluations. Attributes, such as point sources discharges, agricultural fertilizer / manure nutrients, atmospheric deposition, climate, geology / soils, land cover, hydrologic characteristics and physical characteristics, are available for all catchments in the RF1 and NHDPlus digital stream networks.

Stream network datasets and watershed attribute data

  • The EPA RF1 data set (with attributes) is defined at the 1:500K scale and has stream catchments that are on average 130 km2.
  • The NHDPlus data set (with attributes Version 1.1 and Version 2.1) is defined at the 1:100K scale and has stream catchments that are on average 3 km2.

Descriptions of these data sets and how they are used in SPARROW models can be found in Preston and others (2011).


What’s New …

  • RSPARROW, now available on the USGS GitLab repository, provides the first open-source version of the USGS SPARROW water-quality model, with new features that improve the utility of the model for conducting studies of contaminants in surface waters and informing water resource management decisions.  RSPARROW extends the capabilities of the current proprietary SAS SPARROW version and streamlines user and R developer access to SPARROW modelling technology.
  • Although SPARROW models are typically based on a single time period, dynamic versions of SPARROW models are being developed to take advantage of new data sets. Dynamic SPARROW models will account for temporary storage of contaminants and to simulate seasonal variations. They also will allow simulation of contaminant loads through time to estimate the time needed for management actions on the land to affect loads in streams.