Skip to main content
U.S. flag

An official website of the United States government

New high-resolution, land-use and change data improves decision-making in the Chesapeake Bay watershed Completed

By Chesapeake Bay Activities September 27, 2022

Issue
Informing restoration across the nearly 64,000 square miles of the Chesapeake Bay watershed is an enormous challenge that requires detailed and accessible landscape data. Among the most pressing challenges being addressed by the Chesapeake Bay Program (CBP) partnership are:
→ Water pollution from agriculture and animal waste,
→ Conversion of critical habitat like forest and marshes for new development,
→ Stormwater runoff and water pollution from increasing impervious surfaces, and
→ Sea level rise caused by climate change.

The CBP needed improved land data for multiple purposes, including documenting changes in vital habitat (such as forests and wetlands), modeling of water-quality conditions, and assessing the effects of development.

New Chesapeake Bay Program Land Use/Land Cover Product

With support from the CBP, the USGS, Chesapeake Conservancy and the University of Vermont developed an improved 1-meter resolution land use/land cover (LULC) data set for the Chesapeake Bay watershed. 

The improved data have been almost a decade in the making. In December of 2010, the U.S. Environmental Protection Agency (EPA) developed a plan to reduce levels of nitrogen and phosphorus in the Chesapeake Bay known as the Total Maximum Daily (TMDL). However, the TMDL did not specify how to reduce the nutrient levels, therefore, CBP updated its Watershed Model—which uses land use data—to set targets and allocations for the TMDL. 

For the early iterations of the model, the team used 30-m resolution LULC data from the National Land Cover Database (NLCD) as finer-scale LULC data were too expensive. While the NLCD proved effective on a broad scale, these data could be inaccurate at local levels. A common complaint amongst local stakeholders was that the land use was ‘simply wrong’, thus generating skepticism about other datasets informing the TMDL, as well as the land use data, explained Peter Claggett, Coordinator, Chesapeake Bay Program Land Use Workgroup and Research Geographer, U.S. Geological Survey. Responding to these concerns in 2016, the USGS led the first effort to map 16 land uses at 1-m resolution throughout the watershed. When those data became available, applications to inform other goals outlined in the 2014 Chesapeake Bay Watershed Agreement were identified.

The new and improved LULC data are derived from 2017 and 2018 aerial imagery and include a revision to the previous 2013–14 high-resolution imagery to ensure their classification schemes are consistent and the data comparable. As a result, the new data product accurately captures changes in land use and land cover throughout the Chesapeake Bay watershed from 2013–14 to 2017–18 at a 1-m resolution. 

And it’s not only the spatial resolution of the imagery derived from the USDA’s National Agriculture Imagery Program (NAIP) that make these new data so powerful; it’s the combination of high-resolution imagery with ancillary data. After acquiring NAIP imagery, the team added object height information derived from LiDAR imagery, planimetric data (photo-derived images of roads, structures, and impervious surfaces taken from a plane), and land parcels to build a detailed land use classification. Physical context of features identifiable in NAIP imagery was especially useful for interpreting land use classes such as forest and turf grass. After two-years and an investment of nearly $3 million, the team had 1-meter resolution LULC with 54 classes (e.g., timber harvest, cropland, etc.) which is quite unique as there are no known regional-scale LULC datasets that have this level of detail for so many land-use classes allowing for more focused restoration and conservation efforts.

 

Major Findings

The new land data provide two primary improvements: (1) better characterization of the types of land use, and (2) land-change analysis.

Characterization of the types of land use

The new 2017–18 dataset includes a more detailed LULC classification with enhanced technical and scientific information that was not previously available for the original 2013–14 data products.

Impervious surface and tree canopy are the two groups of classes with the most improved accuracy because of the improved quality of ancillary data and increased availability of high-quality LiDAR imagery used. The new 2017–18 and revised 2013–14 datasets contain 54 LU/LC classes grouped into a general 18-class schema. The major types of land uses identified in this dataset are development, production, natural (forest-related), water, and wetlands. This new release includes 38 more LU/LC classes than the original 2013–14 data product, which have revealed patterns and processes that were not evident before. 

Some of the findings from the new classifications include: 

  • Transitional lands mapped for the first time. Low, herbaceous vegetation and barren lands can represent one of the highest polluting land uses (such as corn production) or one of the lowest (such as fallow lands). Transitional lands include areas under construction, timber harvests, natural succession, and barren water margins. Capturing these land uses is necessary for monitoring LU/LC change and understanding its significance for water quality, watershed health, and vital habitats.
  • Forests more accurately defined and mapped. Experts paid particular attention to accurately distinguish between forests and other trees by accounting for edge effects (such as alterations in microclimate and species composition along the edge of wooded areas). If a wooded area is at least one acre in size with a width of at least 72 meters, it was classed as forest. The results align with estimates from the U.S. Forest Service’s’ Forest Inventory Assessment, which provides evidence of the accuracy of this new dataset. Identifying forested areas more precisely allows managers to consider the habitat values and effects of fragmentation when considering the role of trees for protecting water and air quality.
  • Turf grass is the largest “crop” in the watershed. Improved ancillary data and decision rules resulted in a more accurate turf grass classification, which includes all low vegetation close to structures and within and around airports, golf courses, cemeteries, recreational fields, surrounding residential and commercial structures, and under trees in residential yards. There are 2.92 million acres of turf grass and tree canopy over turf grass in the Bay watershed, which is greater than the approximately 1.8 to 1.9 million acres of corn, the most extensive agricultural crop in the watershed. 
Land-change analysis

The spatial resolution of these new datasets and the improved land classification allows managers to understand and track changes in land use in the watershed. Some examples of change between 2013–2014 and 2017–2018 include the four topics listed below:

  • New development increased across the watershed. Development increased by an additional 131,000 acres. Impervious surfaces increased by 50,700 acres, 26% of which were associated with the intensification of already developed areas (e.g., turf grass transitioning to structures, driveways, roads, etc.). More impervious surfaces results in additional pollutant runoff during storms and increases the temperature of water running into streams. Stormwater runoff is the fastest growing source of pollution in the Chesapeake Bay by delivering nutrients, sediments, and other chemical contaminants into rivers and streams. New development also includes changes in pervious cover such as the conversion of forests and farmland to turf grass, construction, solar fields, and utility and road rights-of-way. There was an increase of 67,600 acres of turf grass and tree canopy over turf grass in the watershed. The largest increase in turf grass was from pervious developed (PDEV) to turf grass (TURF, 15,300 acres), likely associated with new construction while the largest increase in tree canopy over turf grass originated from forests indicative of new development in forested areas. 
    Three side-by-side panels to show land-use-land-change in Charles County, MD
    Sources/Usage: Public Domain. View Media Details
    New development (red, black, yellow) near the towns of Waldorf and La Plata in Charles County, Maryland from 2014-2018. Dynamic LULC Change, Chesapeake Conservancy
  • Trees affected by development. In areas already developed in 2013-14, tree canopy decreased by 10,400 acres. New development resulted in the conversion of 61,900 acres of forests (FORE) and other tree canopy (TCOT) and may have adversely affected an additional 31,000 acres of forest and other tree canopy through soil compaction and disturbance.

  • Timber harvest produced the largest land change in the watershed. A total of 305,000 acres of forest (FORE) have been cleared primarily due to timber harvest (HARF), accounting for 176,000 acres. An additional 81,500 acres of forest transitioned to natural succession (NATS) which may also reflect timber harvest activity. Other contributors to forest clearing include development, agriculture, and solar fields. The largest gain in forest resulted from natural succession (NATS, 76,400 acres).

    Side-by-side panels to show forest harvested for timber in southwest Cumberland County, Virginia from 2014-2018
    Sources/Usage: Public Domain. View Media Details
    Forest harvested for timber (brown) in southwest Cumberland County, Virginia from 2014-2018. Dynamic LULC Change, Chesapeake Conservancy

  • Solar fields are an emerging land use. Solar fields are included in the productive class, along with cropland, pasture, and extractive lands. This is the first time that artificial intelligence has used to identify solar fields in the CBP’s LULC dataset. Land use for solar fields increased by 1,012 acres and is expected to continue as demand for clean energy increases. Some of the largest solar fields have occurred in previously forested areas. Solar fields are one component of the pervious developed (PDEV) land use which increased by 12,600 acres.

Side-by-side panels to show forest harvested for timber for the development of solar fields in southwest Cumberland County, V
Sources/Usage: Public Domain. View Media Details
Forests in this area of Louisa County, Virginia were recently cleared for timber (brown) and for the development of solar fields (yellow). Dynamic LULC Change, Chesapeake Conservancy

Management Implications

These new LULC datasets are transformational because they have the potential to change the way restoration and conservation actions are planned and implemented, from an opportunistic approach to one that is more geographically focused and therefore more effective and efficient.  LULC data inform many of the outcomes in the 2014 Chesapeake Bay Watershed Agreement and will serve as the basis for developing the next generation of CBP watershed and land-change models. These LULC data are intended to be updated every 5 years through 2030. The next iteration of land use/land cover and change data is expected to be released 2024, representing the 2021-22 conditions and change from 2017-18 to 2021-22.

Some examples of how CBP stakeholders can use the new high-resolution data for management applications of the Watershed Agreement Goals and Outcomes include: 

  • Water Quality: identifying opportunities to implement water-quality practices (e.g., riparian forest buffers, urban tree planting, stream restoration) where they may be most effective; and informing stormwater management plans.  
  • Land Conservation and Healthy Watersheds: Identifying land conservation opportunities; monitoring, maintaining, and determining new conservation easements; identifying potential healthy and vulnerable watersheds; characterizing the rate of farmland and forest conversion; and informing land use planning decisions.
  • Vital Habitat: assessing changes in forests, community tree canopy, and measuring the extent and rate of change in impervious surfaces that can impact streams, wetlands, and other important aquatic habitats.

 

For Further Information

Data and maps made readily available for use to better understand and inform policy decisions and educate the public about land use and land change across the Chesapeake Bay watershed.

This science summary was prepared in collaboration between the USGS and the University of MD Center for Environmental Science, Integration and Application Network. More information on UMCES-IAN can be found at: Integration and Application Network | University of Maryland Center for Environmental Science.

Contact Peter Claggett for more information, his CBP email is pclagget@chespeakebay.net, and his USGS contact information is listed above.

  • Overview

    Issue
    Informing restoration across the nearly 64,000 square miles of the Chesapeake Bay watershed is an enormous challenge that requires detailed and accessible landscape data. Among the most pressing challenges being addressed by the Chesapeake Bay Program (CBP) partnership are:
    → Water pollution from agriculture and animal waste,
    → Conversion of critical habitat like forest and marshes for new development,
    → Stormwater runoff and water pollution from increasing impervious surfaces, and
    → Sea level rise caused by climate change.

    The CBP needed improved land data for multiple purposes, including documenting changes in vital habitat (such as forests and wetlands), modeling of water-quality conditions, and assessing the effects of development.

    New Chesapeake Bay Program Land Use/Land Cover Product

    With support from the CBP, the USGS, Chesapeake Conservancy and the University of Vermont developed an improved 1-meter resolution land use/land cover (LULC) data set for the Chesapeake Bay watershed. 

    The improved data have been almost a decade in the making. In December of 2010, the U.S. Environmental Protection Agency (EPA) developed a plan to reduce levels of nitrogen and phosphorus in the Chesapeake Bay known as the Total Maximum Daily (TMDL). However, the TMDL did not specify how to reduce the nutrient levels, therefore, CBP updated its Watershed Model—which uses land use data—to set targets and allocations for the TMDL. 

    For the early iterations of the model, the team used 30-m resolution LULC data from the National Land Cover Database (NLCD) as finer-scale LULC data were too expensive. While the NLCD proved effective on a broad scale, these data could be inaccurate at local levels. A common complaint amongst local stakeholders was that the land use was ‘simply wrong’, thus generating skepticism about other datasets informing the TMDL, as well as the land use data, explained Peter Claggett, Coordinator, Chesapeake Bay Program Land Use Workgroup and Research Geographer, U.S. Geological Survey. Responding to these concerns in 2016, the USGS led the first effort to map 16 land uses at 1-m resolution throughout the watershed. When those data became available, applications to inform other goals outlined in the 2014 Chesapeake Bay Watershed Agreement were identified.

    The new and improved LULC data are derived from 2017 and 2018 aerial imagery and include a revision to the previous 2013–14 high-resolution imagery to ensure their classification schemes are consistent and the data comparable. As a result, the new data product accurately captures changes in land use and land cover throughout the Chesapeake Bay watershed from 2013–14 to 2017–18 at a 1-m resolution. 

    And it’s not only the spatial resolution of the imagery derived from the USDA’s National Agriculture Imagery Program (NAIP) that make these new data so powerful; it’s the combination of high-resolution imagery with ancillary data. After acquiring NAIP imagery, the team added object height information derived from LiDAR imagery, planimetric data (photo-derived images of roads, structures, and impervious surfaces taken from a plane), and land parcels to build a detailed land use classification. Physical context of features identifiable in NAIP imagery was especially useful for interpreting land use classes such as forest and turf grass. After two-years and an investment of nearly $3 million, the team had 1-meter resolution LULC with 54 classes (e.g., timber harvest, cropland, etc.) which is quite unique as there are no known regional-scale LULC datasets that have this level of detail for so many land-use classes allowing for more focused restoration and conservation efforts.

     

    Major Findings

    The new land data provide two primary improvements: (1) better characterization of the types of land use, and (2) land-change analysis.

    Characterization of the types of land use

    The new 2017–18 dataset includes a more detailed LULC classification with enhanced technical and scientific information that was not previously available for the original 2013–14 data products.

    Impervious surface and tree canopy are the two groups of classes with the most improved accuracy because of the improved quality of ancillary data and increased availability of high-quality LiDAR imagery used. The new 2017–18 and revised 2013–14 datasets contain 54 LU/LC classes grouped into a general 18-class schema. The major types of land uses identified in this dataset are development, production, natural (forest-related), water, and wetlands. This new release includes 38 more LU/LC classes than the original 2013–14 data product, which have revealed patterns and processes that were not evident before. 

    Some of the findings from the new classifications include: 

    • Transitional lands mapped for the first time. Low, herbaceous vegetation and barren lands can represent one of the highest polluting land uses (such as corn production) or one of the lowest (such as fallow lands). Transitional lands include areas under construction, timber harvests, natural succession, and barren water margins. Capturing these land uses is necessary for monitoring LU/LC change and understanding its significance for water quality, watershed health, and vital habitats.
    • Forests more accurately defined and mapped. Experts paid particular attention to accurately distinguish between forests and other trees by accounting for edge effects (such as alterations in microclimate and species composition along the edge of wooded areas). If a wooded area is at least one acre in size with a width of at least 72 meters, it was classed as forest. The results align with estimates from the U.S. Forest Service’s’ Forest Inventory Assessment, which provides evidence of the accuracy of this new dataset. Identifying forested areas more precisely allows managers to consider the habitat values and effects of fragmentation when considering the role of trees for protecting water and air quality.
    • Turf grass is the largest “crop” in the watershed. Improved ancillary data and decision rules resulted in a more accurate turf grass classification, which includes all low vegetation close to structures and within and around airports, golf courses, cemeteries, recreational fields, surrounding residential and commercial structures, and under trees in residential yards. There are 2.92 million acres of turf grass and tree canopy over turf grass in the Bay watershed, which is greater than the approximately 1.8 to 1.9 million acres of corn, the most extensive agricultural crop in the watershed. 
    Land-change analysis

    The spatial resolution of these new datasets and the improved land classification allows managers to understand and track changes in land use in the watershed. Some examples of change between 2013–2014 and 2017–2018 include the four topics listed below:

    • New development increased across the watershed. Development increased by an additional 131,000 acres. Impervious surfaces increased by 50,700 acres, 26% of which were associated with the intensification of already developed areas (e.g., turf grass transitioning to structures, driveways, roads, etc.). More impervious surfaces results in additional pollutant runoff during storms and increases the temperature of water running into streams. Stormwater runoff is the fastest growing source of pollution in the Chesapeake Bay by delivering nutrients, sediments, and other chemical contaminants into rivers and streams. New development also includes changes in pervious cover such as the conversion of forests and farmland to turf grass, construction, solar fields, and utility and road rights-of-way. There was an increase of 67,600 acres of turf grass and tree canopy over turf grass in the watershed. The largest increase in turf grass was from pervious developed (PDEV) to turf grass (TURF, 15,300 acres), likely associated with new construction while the largest increase in tree canopy over turf grass originated from forests indicative of new development in forested areas. 
      Three side-by-side panels to show land-use-land-change in Charles County, MD
      Sources/Usage: Public Domain. View Media Details
      New development (red, black, yellow) near the towns of Waldorf and La Plata in Charles County, Maryland from 2014-2018. Dynamic LULC Change, Chesapeake Conservancy
    • Trees affected by development. In areas already developed in 2013-14, tree canopy decreased by 10,400 acres. New development resulted in the conversion of 61,900 acres of forests (FORE) and other tree canopy (TCOT) and may have adversely affected an additional 31,000 acres of forest and other tree canopy through soil compaction and disturbance.

    • Timber harvest produced the largest land change in the watershed. A total of 305,000 acres of forest (FORE) have been cleared primarily due to timber harvest (HARF), accounting for 176,000 acres. An additional 81,500 acres of forest transitioned to natural succession (NATS) which may also reflect timber harvest activity. Other contributors to forest clearing include development, agriculture, and solar fields. The largest gain in forest resulted from natural succession (NATS, 76,400 acres).

      Side-by-side panels to show forest harvested for timber in southwest Cumberland County, Virginia from 2014-2018
      Sources/Usage: Public Domain. View Media Details
      Forest harvested for timber (brown) in southwest Cumberland County, Virginia from 2014-2018. Dynamic LULC Change, Chesapeake Conservancy

    • Solar fields are an emerging land use. Solar fields are included in the productive class, along with cropland, pasture, and extractive lands. This is the first time that artificial intelligence has used to identify solar fields in the CBP’s LULC dataset. Land use for solar fields increased by 1,012 acres and is expected to continue as demand for clean energy increases. Some of the largest solar fields have occurred in previously forested areas. Solar fields are one component of the pervious developed (PDEV) land use which increased by 12,600 acres.

    Side-by-side panels to show forest harvested for timber for the development of solar fields in southwest Cumberland County, V
    Sources/Usage: Public Domain. View Media Details
    Forests in this area of Louisa County, Virginia were recently cleared for timber (brown) and for the development of solar fields (yellow). Dynamic LULC Change, Chesapeake Conservancy

    Management Implications

    These new LULC datasets are transformational because they have the potential to change the way restoration and conservation actions are planned and implemented, from an opportunistic approach to one that is more geographically focused and therefore more effective and efficient.  LULC data inform many of the outcomes in the 2014 Chesapeake Bay Watershed Agreement and will serve as the basis for developing the next generation of CBP watershed and land-change models. These LULC data are intended to be updated every 5 years through 2030. The next iteration of land use/land cover and change data is expected to be released 2024, representing the 2021-22 conditions and change from 2017-18 to 2021-22.

    Some examples of how CBP stakeholders can use the new high-resolution data for management applications of the Watershed Agreement Goals and Outcomes include: 

    • Water Quality: identifying opportunities to implement water-quality practices (e.g., riparian forest buffers, urban tree planting, stream restoration) where they may be most effective; and informing stormwater management plans.  
    • Land Conservation and Healthy Watersheds: Identifying land conservation opportunities; monitoring, maintaining, and determining new conservation easements; identifying potential healthy and vulnerable watersheds; characterizing the rate of farmland and forest conversion; and informing land use planning decisions.
    • Vital Habitat: assessing changes in forests, community tree canopy, and measuring the extent and rate of change in impervious surfaces that can impact streams, wetlands, and other important aquatic habitats.

     

    For Further Information

    Data and maps made readily available for use to better understand and inform policy decisions and educate the public about land use and land change across the Chesapeake Bay watershed.

    This science summary was prepared in collaboration between the USGS and the University of MD Center for Environmental Science, Integration and Application Network. More information on UMCES-IAN can be found at: Integration and Application Network | University of Maryland Center for Environmental Science.

    Contact Peter Claggett for more information, his CBP email is pclagget@chespeakebay.net, and his USGS contact information is listed above.