Land Cover Trends Project

Methods

Overview/Summary  •   Ecoregion Framework  •   Sampling Strategy  •   LandSat Source Data   •   Classification System  •   Interpretation Process  •   References


Interpretation Process

LULC delineations for each sample block began with the creation of a baseline reference LULC dataset. The 1992 date was usually the starting point due to the availability of the 30 meter resolution 1992 National Land Cover Dataset (NLCD) (Vogelmann, 2001). The detailed NLCD classes were first collapsed to match the more general land-cover classification system used by the Land Cover Trends project. The NLCD data were then manually edited on the computer screen using ERDAS IMAGINE software, using on screen interpretation methods while looking at 1992 Landsat TM data along with aerial photographs (National Aerial Photography Program and National High Altitude Program) and other ancillary aids (Google Earth, topographic maps, National Wetland Inventory datasets, etc.) serving to assist and improve interpretations (table 3). This editing (i.e., cleanup) procedure was done because NLCD data were created using automated spectral image processing procedures and were not meant for use in local-scale assessments. Additionally, Land Cover Trends aimed to improve upon NLCD by explicitly mapping land disturbance events.

LULC for the 1986 and 2000 periods was then backward- or forward-classified using the 1992 land-use/land-cover data as a template. For example, creation of the 2000 LULC product began by making an exact copy of the 1992 LULC product. This copy served as a baseline for the 2000 LULC product, with identified changes between 1992 and 2000 manually edited into the copied image. This baseline 2000 LULC product, the 1992 Landsat imagery, and the 2000 Landsat imagery were displayed on-the computer screen, allowing the analyst to pan through the entire area covered by the sample block, examining the 1992 and 2000 Landsat imagery and any relevant aerial photography for valid LULC changes between the two dates. Any identified LULC changes were manually digitized on-screen, and the land use or land cover attributed on the 2000 LULC map.

Upon completion of the 2000 LULC product, the same procedures were used to create the 1986 LULC map using 1992 to back-classify change. The process then repeated, with the 1986 LULC product used to map change in 1980, and the resulting 1980 product used to map change in 1973. This manual process eliminates errors that may arise between independently created LULC products for two dates and then differencing them in a subsequent change analysis. Classification errors are greatly reduced because only manually identified, delineated, and coded LULC changes are delineated during this phase.

A traditional accuracy assessment was not conducted for the interpreted blocks. However, quality-assurance/quality-control measures consisted of a formal block review conducted by the national research team where each LULC date for each sample block was visually inspected for errors and informally compared to ancillary data. Where blocks bordered one another, an examination of LULC adjacency was also conducted. The image interpreter was required to correct all errors following each review and document the corrections that were made.

Flow Chart

Flow diagram depicting the block interpretation procedure used to develop the Land Cover Trends dataset. The manual interpretation process starts with the development of the 1992 baseline LULC. The resulting LULC map was then used to forward-and-back classify areas that changed.



Data sources and dates of ancillary data acquired to aid interpreters in manually delineating land use and land cover from Landsat imagery

[For example, aerial photography was acquired for each sample block to provide a high resolution data source to aid difficult interpretations. Availability of ancillary data varied, yet many datasets covered large spatial extents. *Indicates aerial photography or satellite imagery.]

Data Source Data Provider Dates
7.5 Minute Topographic Maps U.S. Geological Survey Variable
National Wetland Inventory (NWI) U.S. Fish and Wildlife Service Variable
National Wetland Inventory (NWI) U.S. Fish and Wildlife Service Variable
Monitoring Trends in Burn Severity U.S. Geological Survey 1984-present
National High Altitude Program (NHAP)* U.S. Geological Survey 1978-1980s
National Aerial Photography Program (NAPP)* U.S. Geological Survey 1987-1990s
National Land Cover Dataset (NLCD) U.S. Geological Survey 1992, 2000
Google Earth Imagery* Google 1990s - present


LandSat and LULC images

An example of the 5 dates of Landsat imagery and corresponding land-use/land-cover data available for each sample block. The example sample block is located in the Ouachita Mountains ecoregion (samp36_0180), and shows forest cutting between 1973 and 2000 (green-forest/woodland; yellow-grassland/shrubland; magenta-mechanical disturbance; blue-water).

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo USA.gov logo U.S. Department of the Interior | U.S. Geological Survey
URL: http://landcovertrends.usgs.gov/main/interp.html
Page Contact Information: WRG Web Team
Page Last Modified: Thursday, 07-Jul-2016 11:19:52 EDT (mg)