USFS Landscape Change Monitoring System v2020.5

USFS/GTAC/LCMS/v2020-5
Dataset Availability
1985-06-01T00:00:00Z - 2020-09-30T00:00:00
Dataset Provider
Earth Engine Snippet
ee.ImageCollection("USFS/GTAC/LCMS/v2020-5")
Tags
change change-detection forest gtac landcover landsat-derived landuse lcms redcastle-resources rmrs sentinel2-derived time-series usda usfs

Description

This product is part of the Landscape Change Monitoring System (LCMS) data suite. It shows LCMS-modeled change, land cover, and/or land use classes for each year.

LCMS is a remote sensing-based system for mapping and monitoring landscape change across the United States. Its objective is to develop a consistent approach using the latest technology and advancements in change detection to produce a "best available" map of landscape change.

Outputs include three annual products: change, land cover, and land use. Change relates specifically to vegetation cover and includes slow loss, fast loss (which also includes hydrologic changes such as inundation or desiccation), and gain. These values are predicted for each year of the Landsat time series and serve as the foundational products for LCMS. Land cover and land use maps depict life-form level land cover and broad-level land use for each year.

Because no algorithm performs best in all situations, LCMS uses an ensemble of models as predictors, which improves map accuracy across a range of ecosystems and change processes (Healey et al., 2018). The resulting suite of LCMS change, land cover, and land use maps offer a holistic depiction of landscape change across the United States over the past four decades.

Predictor layers for the LCMS model include annual Landsat and Sentinel 2 composites, outputs from the LandTrendr and CCDC change detection algorithms, and terrain information. These components are all accessed and processed using Google Earth Engine (Gorelick et al., 2017).

To produce annual composites, the cFmask (Zhu and Woodcock 2012), cloudScore, and TDOM (Chastain et al., 2019) cloud and cloud shadow masking methods are applied to Landsat Tier 1 and Sentinel 2a and 2b Level-1C top of atmosphere reflectance data. The annual medoid is then computed to summarize each year into a single composite.

The composite time series is temporally segmented using LandTrendr (Kennedy et al., 2010; Kennedy et al., 2018; Cohen et al., 2018).

All cloud and cloud shadow free values are also temporally segmented using the CCDC algorithm (Zhu and Woodcock, 2014).

The raw composite values, LandTrendr fitted values, pair-wise differences, segment duration, change magnitude, and slope, and CCDC September 1 sine and cosine coefficients (first 3 harmonics), fitted values, and pairwise differences, along with elevation, slope, sine of aspect, cosine of aspect, and topographic position indices (Weiss, 2001) from the National Elevation Dataset (NED), are used as independent predictor variables in a Random Forest (Breiman, 2001) model.

Reference data are collected using TimeSync, a web-based tool that helps analysts visualize and interpret the Landsat data record from 1984-present (Cohen et al., 2010).

Additional Resources

Contact sm.fs.lcms@usda.gov with any questions or specific data requests.

  • Breiman, L., 2001. Machine Learning. Springer, 45(3): 261-277 doi:10.1023/a:1017934522171
  • Chastain, R., Housman, I., Goldstein, J., Finco, M., and Tenneson, K., 2019. Empirical cross sensor comparison of Sentinel-2A and 2B MSI, Landsat-8 OLI, and Landsat-7 ETM top of atmosphere spectral characteristics over the conterminous United States. In Remote Sensing of Environment. Science Direct, 221: 274-285 doi:10.1016/j.rse.2018.11.012
  • Cohen, W. B., Yang, Z., and Kennedy, R., 2010. Detecting trends in forest disturbance and recovery using yearly Landsat time series: 2. TimeSync - Tools for calibration and validation. In Remote Sensing of Environment. Science Direct, 114(12): 2911-2924 doi:10.1016/j.rse.2010.07.010
  • Cohen, W. B., Yang, Z., Healey, S. P., Kennedy, R. E., and Gorelick, N., 2018. A LandTrendr multispectral ensemble for forest disturbance detection. In Remote Sensing of Environment. Science Direct, 205: 131-140 doi:10.1016/j.rse.2017.11.015
  • Gorelick, N., Hancher, M., Dixon, M., Ilyushchenko, S., Thau, D., and Moore, R., 2017. Google Earth Engine: Planetary-scale geospatial analysis for everyone. In Remote Sensing of Environment. Science Direct, 202: 18-27 doi:10.1016/j.rse.2017.06.031
  • Healey, S. P., Cohen, W. B., Yang, Z., Kenneth Brewer, C., Brooks, E. B., Gorelick, N., Hernandez, A. J., Huang, C., Joseph Hughes, M., Kennedy, R. E., Loveland, T. R., Moisen, G. G., Schroeder, T. A., Stehman, S. V., Vogelmann, J. E., Woodcock, C. E., Yang, L., and Zhu, Z., 2018. Mapping forest change using stacked generalization: An ensemble approach. In Remote Sensing of Environment. Science Direct, 204: 717-728 doi:10.1016/j.rse.2017.09.029
  • Kennedy, R. E., Yang, Z., and Cohen, W. B., 2010. Detecting trends in forest disturbance and recovery using yearly Landsat time series: 1. LandTrendr - Temporal segmentation algorithms. In Remote Sensing of Environment. Science Direct, 114(12): 2897-2910 doi:10.1016/j.rse.2010.07.008
  • Kennedy, R., Yang, Z., Gorelick, N., Braaten, J., Cavalcante, L., Cohen, W., and Healey, S., 2018. Implementation of the LandTrendr Algorithm on Google Earth Engine. In Remote Sensing. MDPI, 10(5): 691 doi:10.3390/rs10050691
  • Weiss, A.D., 2001. Topographic position and landforms analysis Poster Presentation, ESRI Users Conference, San Diego, CAZhu, Z., and Woodcock, C. E. (2012). Object-based cloud and cloud shadow detection in Landsat imagery. In Remote Sensing of Environment. Science Direct, 118: 83-94 doi:10.1016/j.rse.2011.10.028
  • Zhu, Z., and Woodcock, C. E., 2014. Continuous change detection and classification of land cover using all available Landsat data. In Remote Sensing of Environment. Science Direct, 144: 152-171 doi:10.1016/j.rse.2014.01.011

Bands

Bands

Name Pixel Size Description
Change 30 meters

Final thematic LCMS change product. A total of three change classes (slow loss, fast loss, and gain) are mapped for each year. Each class is predicted using a separate Random Forest model, which outputs a probability (proportion of the trees within the Random Forest model) that the pixel belongs to that class. Because of this, individual pixels have three different model outputs for each year. Final classes are assigned to the change class with the highest probability that is also above a specified threshold. Any pixel that does not have any value above each class's respective threshold is assigned to the Stable class.

Land_Cover 30 meters

Final thematic LCMS land cover product. A total of 14 land cover classes are mapped on an annual basis using TimeSync reference data and spectral information derived from Landsat imagery. Each class is predicted using a separate Random Forest model, which outputs a probability (proportion of the trees within the Random Forest model) that the pixel belongs to that class. Because of this, individual pixels have 14 different model outputs for each year, and final classes are assigned to the land cover with the highest probability. Seven of the 14 land cover classes indicate a single land cover, where that land cover type covers most of the pixel's area and no other class covers more than 10% of the pixel. There are also seven mixed classes. These represent pixels in which an additional land cover class covers at least 10% of the pixel.

Land_Use 30 meters

Final thematic LCMS land use product. A total of 6 land use classes are mapped on an annual basis using TimeSync reference data and spectral information derived from Landsat imagery. Each class is predicted using a separate Random Forest model, which outputs a probability (proportion of the trees within the Random Forest model) that the pixel belongs to that class. Because of this, individual pixels have 6 different model outputs for each year, and final classes are assigned to the land use with the highest probability.

Change_Raw_Probability_Slow-Loss 30 meters

Raw LCMS modeled probability of Slow Loss. Defined as: Slow Loss includes the following classes from the TimeSync change process interpretation-

  • Structural Decline - Land where trees or other woody vegetation is physically altered by unfavorable growing conditions brought on by non-anthropogenic or non-mechanical factors. This type of loss should generally create a trend in the spectral signal(s) (e.g. NDVI decreasing, Wetness decreasing; SWIR increasing; etc.) however the trend can be subtle. Structural decline occurs in woody vegetation environments, most likely from insects, disease, drought, acid rain, etc. Structural decline can include defoliation events that do not result in mortality such as in Gypsy moth and spruce budworm infestations which may recover within 1 or 2 years.

  • Spectral Decline - A plot where the spectral signal shows a trend in one or more of the spectral bands or indices (e.g. NDVI decreasing, Wetness decreasing; SWIR increasing; etc.). Examples include cases where: a) non-forest/non-woody vegetation shows a trend suggestive of decline (e.g. NDVI decreasing, Wetness decreasing; SWIR increasing; etc.), or b) where woody vegetation shows a decline trend which is not related to the loss of woody vegetation, such as when mature tree canopies close resulting in increased shadowing, when species composition changes from conifer to hardwood, or when a dry period (as opposed to stronger, more acute drought) causes an apparent decline in vigor, but no loss of woody material or leaf area.

Change_Raw_Probability_Fast-Loss 30 meters

Raw LCMS modeled probability of Fast Loss. Defined as: Fast Loss includes the following classes from the TimeSync change process interpretation-

  • Fire - Land altered by fire, regardless of the cause of the ignition (natural or anthropogenic), severity, or land use.

  • Harvest - Forest land where trees, shrubs or other vegetation have been severed or removed by anthropogenic means. Examples include clearcutting, salvage logging after fire or insect outbreaks, thinning and other forest management prescriptions (e.g. shelterwood/seedtree harvest).

  • Mechanical - Non-forest land where trees, shrubs or other vegetation has been mechanically severed or removed by chaining, scraping, brush sawing, bulldozing, or any other methods of non-forest vegetation removal.

  • Wind/ice - Land (regardless of use) where vegetation is altered by wind from hurricanes, tornados, storms and other severe weather events including freezing rain from ice storms.

  • Hydrology - Land where flooding has significantly altered woody cover or other Land cover elements regardless of land use (e.g. new mixtures of gravel and vegetation in and around streambeds after a flood).

  • Debris - Land (regardless of use) altered by natural material movement associated with landslides, avalanches, volcanos, debris flows, etc.

  • Other - Land (regardless of use) where the spectral trend or other supporting evidence suggests a disturbance or change event has occurred but the definitive cause cannot be determined or the type of change fails to meet any of the change process categories defined above.

Change_Raw_Probability_Gain 30 meters

Raw LCMS modeled probability of Gain. Defined as: Land exhibiting an increase in vegetation cover due to growth and succession over one or more years. Applicable to any areas that may express spectral change associated with vegetation regrowth. In developed areas, growth can result from maturing vegetation and/or newly installed lawns and landscaping. In forests, growth includes vegetation growth from bare ground, as well as the over topping of intermediate and co-dominate trees and/or lower-lying grasses and shrubs. Growth/Recovery segments recorded following forest harvest will likely transition through different land cover classes as the forest regenerates. For these changes to be considered growth/recovery, spectral values should closely adhere to an increasing trend line (e.g. a positive slope that would, if extended to ~20 years, be on the order of .10 units of NDVI) which persists for several years.

Land_Cover_Raw_Probability_Trees 30 meters

Raw LCMS modeled probability of Trees. Defined as: The majority of the pixel is comprised of live or standing dead trees.

Land_Cover_Raw_Probability_Tall-Shrubs-and-Trees-Mix 30 meters

Raw LCMS modeled probability of Tall Shrubs and Trees Mix (SEAK Only). Defined as: The majority of the pixel is comprised of shrubs greater than 1m in height and is also comprised of at least 10% live or standing dead trees.

Land_Cover_Raw_Probability_Shrubs-and-Trees-Mix 30 meters

Raw LCMS modeled probability of Shrubs and Trees Mix. Defined as: The majority of the pixel is comprised of shrubs and is also comprised of at least 10% live or standing dead trees.

Land_Cover_Raw_Probability_Grass-Forb-Herb-and-Trees-Mix 30 meters

Raw LCMS modeled probability of Grass/Forb/Herb and Trees Mix. Defined as: The majority of the pixel is comprised of perennial grasses, forbs, or other forms of herbaceous vegetation and is also comprised of at least 10% live or standing dead trees.

Land_Cover_Raw_Probability_Barren-and-Trees-Mix 30 meters

Raw LCMS modeled probability of Barren and Trees Mix. Defined as: The majority of the pixel is comprised of bare soil exposed by disturbance (e.g., soil uncovered by mechanical clearing or forest harvest), as well as perennially barren areas such as deserts, playas, rock outcroppings (including minerals and other geologic materials exposed by surface mining activities), sand dunes, salt flats, and beaches. Roads made of dirt and gravel are also considered barren and is also comprised of at least 10% live or standing dead trees.

Land_Cover_Raw_Probability_Tall-Shrubs 30 meters

Raw LCMS modeled probability of Tall Shrubs (SEAK Only). Defined as: The majority of the pixel is comprised of shrubs greater than 1m in height.

Land_Cover_Raw_Probability_Shrubs 30 meters

Raw LCMS modeled probability of Shrubs. Defined as: The majority of the pixel is comprised of shrubs.

Land_Cover_Raw_Probability_Grass-Forb-Herb-and-Shrubs-Mix 30 meters

Raw LCMS modeled probability of Grass/Forb/Herb and Shrubs Mix. Defined as: The majority of the pixel is comprised of perennial grasses, forbs, or other forms of herbaceous vegetation and is also comprised of at least 10% shrubs.

Land_Cover_Raw_Probability_Barren-and-Shrubs-Mix 30 meters

Raw LCMS modeled probability of Barren and Shrubs Mix. Defined as: The majority of the pixel is comprised of bare soil exposed by disturbance (e.g., soil uncovered by mechanical clearing or forest harvest), as well as perennially barren areas such as deserts, playas, rock outcroppings (including minerals and other geologic materials exposed by surface mining activities), sand dunes, salt flats, and beaches. Roads made of dirt and gravel are also considered barren and is also comprised of at least 10% shrubs.

Land_Cover_Raw_Probability_Grass-Forb-Herb 30 meters

Raw LCMS modeled probability of Grass/Forb/Herb. Defined as: The majority of the pixel is comprised of perennial grasses, forbs, or other forms of herbaceous vegetation.

Land_Cover_Raw_Probability_Barren-and-Grass-Forb-Herb-Mix 30 meters

Raw LCMS modeled probability of Barren and Grass/Forb/Herb Mix. Defined as: The majority of the pixel is comprised of bare soil exposed by disturbance (e.g., soil uncovered by mechanical clearing or forest harvest), as well as perennially barren areas such as deserts, playas, rock outcroppings (including minerals and other geologic materials exposed by surface mining activities), sand dunes, salt flats, and beaches. Roads made of dirt and gravel are also considered barren and is also comprised of at least 10% perennial grasses, forbs, or other forms of herbaceous vegetation.

Land_Cover_Raw_Probability_Barren-or-Impervious 30 meters

Raw LCMS modeled probability of Barren or Impervious. Defined as: The majority of the pixel is comprised of 1) bare soil exposed by disturbance (e.g., soil uncovered by mechanical clearing or forest harvest), as well as perennially barren areas such as deserts, playas, rock outcroppings (including minerals and other geologic materials exposed by surface mining activities), sand dunes, salt flats, and beaches. Roads made of dirt and gravel are also considered barren or 2) man-made materials that water cannot penetrate, such as paved roads, rooftops, and parking lots.

Land_Cover_Raw_Probability_Snow-or-Ice 30 meters

Raw LCMS modeled probability of Snow or Ice. Defined as: The majority of the pixel is comprised of snow or ice.

Land_Cover_Raw_Probability_Water 30 meters

Raw LCMS modeled probability of Water. Defined as: The majority of the pixel is comprised of water.

Land_Use_Raw_Probability_Agriculture 30 meters

Raw LCMS modeled probability of Agriculture. Defined as: Land used for the production of food, fiber and fuels which is in either a vegetated or non-vegetated state. This includes but is not limited to cultivated and uncultivated croplands, hay lands, orchards, vineyards, confined livestock operations, and areas planted for production of fruits, nuts or berries. Roads used primarily for agricultural use (i.e. not used for public transport from town to town) are considered agriculture land use.

Land_Use_Raw_Probability_Developed 30 meters

Raw LCMS modeled probability of Developed. Defined as: Land covered by man-made structures (e.g. high density residential, commercial, industrial, mining or transportation), or a mixture of both vegetation (including trees) and structures (e.g., low density residential, lawns, recreational facilities, cemeteries, transportation and utility corridors, etc.), including any land functionally altered by human activity.

Land_Use_Raw_Probability_Forest 30 meters

Raw LCMS modeled probability of Forest. Defined as: Land that is planted or naturally vegetated and which contains (or is likely to contain) 10% or greater tree cover at some time during a near-term successional sequence. This may include deciduous, evergreen and/or mixed categories of natural forest, forest plantations, and woody wetlands.

Land_Use_Raw_Probability_Non-Forest-Wetland 30 meters

Raw LCMS modeled probability of Non-Forest Wetland. Defined as: Lands adjacent to or within a visible water table (either permanently or seasonally saturated) dominated by shrubs or persistent emergents. These wetlands may be situated shoreward of lakes, river channels, or estuaries; on river floodplains; in isolated catchments; or on slopes. They may also occur as prairie potholes, drainage ditches and stock ponds in agricultural landscapes and may also appear as islands in the middle of lakes or rivers. Other examples also include marshes, bogs, swamps, quagmires, muskegs, sloughs, fens, and bayous.

Land_Use_Raw_Probability_Other 30 meters

Raw LCMS modeled probability of Other. Defined as: Land (regardless of use) where the spectral trend or other supporting evidence suggests a disturbance or change event has occurred but the definitive cause cannot be determined or the type of change fails to meet any of the change process categories defined above.

Land_Use_Raw_Probability_Rangeland-or-Pasture 30 meters

Raw LCMS modeled probability of Rangeland or Pasture. Defined as: This class includes any area that is either a.) Rangeland, where vegetation is a mix of native grasses, shrubs, forbs and grass-like plants largely arising from natural factors and processes such as rainfall, temperature, elevation and fire, although limited management may include prescribed burning as well as grazing by domestic and wild herbivores; or b.) Pasture, where vegetation may range from mixed, largely natural grasses, forbs and herbs to more managed vegetation dominated by grass species that have been seeded and managed to maintain near monoculture.

Change Class Table

Value Color Description
1 3d4551 Stable
2 f39268 Slow Loss
3 d54309 Fast Loss
4 00a398 Gain
5 1B1716 Non-Processing Area Mask

Land_Cover Class Table

Value Color Description
1 005e00 Trees
2 008000 Tall Shrubs & Trees Mix (SEAK Only)
3 00cc00 Shrubs & Trees Mix
4 b3ff1a Grass/Forb/Herb & Trees Mix
5 99ff99 Barren & Trees Mix
6 b30088 Tall Shrubs (SEAK Only)
7 e68a00 Shrubs
8 ffad33 Grass/Forb/Herb & Shrubs Mix
9 ffe0b3 Barren & Shrubs Mix
10 ffff00 Grass/Forb/Herb
11 AA7700 Barren & Grass/Forb/Herb Mix
12 d3bf9b Barren or Impervious
13 ffffff Snow or Ice
14 4780f3 Water
15 1B1716 Non-Processing Area Mask

Land_Use Class Table

Value Color Description
1 efff6b Agriculture
2 ff2ff8 Developed
3 1b9d0c Forest
4 97ffff Non-Forest Wetland
5 a1a1a1 Other
6 c2b34a Rangeland or Pasture
7 1B1716 Non-Processing Area Mask

Collection Properties

Collection Properties

Name Type Description
study_area STRING

LCMS currently covers CONUS and Southeastern Alaska. It will expand to include all US states and territories in the near future.

Possible values: 'SEAK' or 'CONUS'

Terms of Use

Terms of Use

The USDA Forest Service makes no warranty, expressed or implied, including the warranties of merchantability and fitness for a particular purpose, nor assumes any legal liability or responsibility for the accuracy, reliability, completeness or utility of these geospatial data, or for the improper or incorrect use of these geospatial data. These geospatial data and related maps or graphics are not legal documents and are not intended to be used as such. The data and maps may not be used to determine title, ownership, legal descriptions or boundaries, legal jurisdiction, or restrictions that may be in place on either public or private land. Natural hazards may or may not be depicted on the data and maps, and land users should exercise due caution. The data are dynamic and may change over time. The user is responsible to verify the limitations of the geospatial data and to use the data accordingly.

These data were collected using funding from the U.S. Government and can be used without additional permissions or fees. If you use these data in a publication, presentation, or other research product please use the following citation:

USDA Forest Service. 2021. USFS Landscape Change Monitoring System version 2020.5. Salt Lake City, Utah.

Citations

Citations:
  • USDA Forest Service. 2021. USFS Landscape Change Monitoring System version 2020.5. Salt Lake City, Utah.

Explore in Earth Engine

var dataset = ee.ImageCollection('USFS/GTAC/LCMS/v2020-5');

var lcms = dataset
    .filter(ee.Filter.and(
      ee.Filter.eq('year', 2020),  // range: [1985, 2020]
      ee.Filter.eq('study_area', 'CONUS')  // or 'SEAK'
    ))
    .first();


Map.addLayer(lcms.select('Land_Cover'), {}, 'Land Cover');
Map.addLayer(lcms.select('Land_Use'), {}, 'Land Use');
Map.addLayer(lcms.select('Change'), {}, 'Change');

Map.setCenter(-98.58, 38.14, 4);