Introduction

The U. S. Geological Survey (USGS) has an ongoing Land Cover Characterization Program (LCCP). The four components of the LCCP are global (1:2,000,000-scale), national (1:100,000-scale), urban (1:24,000-scale), and special projects (various scales). Only the urban and special project components include the collection of historical as well as contemporary land use and land cover (LULC) data.

The urban component requires greater LULC detail because the data are collected as a large-scale product. The source materials used during LULC collection are historical and current aerial photographs, 1-meter resolution digital orthophoto quadrangles (DOQ), and 1:24,000-scale digital line graphs (DLGs). At present, the USGS is collecting temporal LULC data for several metropolitan areas at a minimum mapping unit (MMU) of 2.5 acres. With an MMU of 2.5 acres, a more accurate assessment can be made of the land surface activity for a particular region.

Data Requirements

The demand for large-scale LULC data has increased recently, especially in rapidly growing metropolitan areas. Many State, regional, and local planning agencies require up-to-date LULC information for various applications, including modeling urban growth, determining land suitability for future development, monitoring how land use changes affect the environment, understanding land use patterns, and developing policies that could encourage or discourage certain land use development.

In response to these increasing demands, the USGS began temporal LULC collection for the Middle Rio Grande Basin (Central New Mexico) and Front Range Infrastructure Resources (Colorado) projects in 1996. The time periods of interest for the LULC collection include the 1930's, 1950's, 1970's, and 1990's.

The USGS has also collected contemporary LULC data for the Arroyo Colorado Watershed Project in Texas. The collection criteria for this project were modified to use an MMU of 5 acres and an Anderson level 3 LULC classification system (Anderson, 1976). Additionally, new LULC categories, such as sugar cane and citrus, were collected. The resulting LULC data set will be used to monitor the effects of agricultural pollutants on water quality.

Project Status - Middle Rio Grande Basin LULC Data Collection

A land surface analysis pilot area encompassing the city of Albuquerque, N. Mex., was defined for collecting contemporary and historical LULC data. At present, all temporal LULC periods have been compiled at 1:24,000 scale for 14 of the 28 quadrangles. An additional 14 quadrangles of temporal LULC data will be collected in the near future for the northern half of the pilot area. These additional quadrangles will be merged with the 14 quadrangles already completed for the southern half of the pilot area.

Temporal LULC Mapping Methodology

Procedures involved in compiling the temporal LULC data include (1) obtaining historical aerial photographs, DOQ's, and ancillary data for collecting, interpreting, and classifying the land surface features, (2) scanning, georeferencing, and mosaicking the historical aerial photographs, (3) compiling the LULC data by 7.5-minute USGS quadrangles using USGS-developed software, and (4) merging the 7.5-minute quadrangles into one seamless data set.

Aerial Photograph and Ancillary Source Acquisition
Before the historical aerial photographs for the project area are obtained, the availability of DOQ's is investigated. Typically, a DOQ is only available for the 1990 time period. If a DOQ does not exist for a particular 7.5-minute quadrangle, the USGS or a USGS contractor will create the DOQ to be used for the contemporary LULC collection. For the contemporary and historical time periods, the availability of aerial photographs must also be researched.

Historical aerial photographs for temporal LULC collection need to be at a photographic scale larger than 1:40,000. If the USGS does not possess the necessary imagery, the historical aerial photographs must be acquired from other government or local agencies. The USGS acquired a large number of the historical aerial photographs from the Soil Conservation Service archives. The most desired photographs for LULC collection are color or color infrared. Aerial photographs ideally should be taken during the growing season so that vegetation land cover classes are discernible. Color imagery enhances the ability to differentiate between agricultural land cover features such as hay/pasture and small grains. For the 1930 and 1950 time periods, only black-and-white photographs are available. When black-and-white aerial photographs are used, land surface patterns must be used to identify specific land cover categories.

Acquisition of ancillary material, such as National Wetlands Inventory data, USGS 7.5-minute maps, and the Bureau of Reclamation's Land Use Trend Analysis data, helps enhance the interpretation and classification of the land surface features.

Image Processing - Georectification
After acquiring the aerial photographs and ancillary data sources, the project staff scans the historical aerial photographs for each timeframe. During the scanning, every other photograph in a stereopair is used in order to obtain complete coverage for each 7.5-minute quadrangle.

The scanned aerial photographs are mosaicked and georeferenced using image processing software. As part of the georeferencing process, control points are created by transferring coordinates of well-defined points (that is, road intersections) from a DLG to the scanned image. At least nine georeferenced control points per image are required. In areas of high terrain relief, a minimum of 15 points is recommended. The controlled part of each photograph is then clipped out and mosaicked to adjacent georeferenced aerial photographs. The objective is to create a historical georeferenced image covering each USGS 7.5-minute quadrangle within the project area.

The final processing step is to check each mosaicked image against the DLG georeferenced features and to verify complete quadrangle coverage. Subsequently, the mosaicked image is split into north and south sections to create smaller image files that the USGS LULC collection software can display efficiently.

The USGS can not produce a complete mosaicked image for all quadrangles within a project area. Primary reasons include (1) few or no horizontal control points available on the DLG, (2) terrain too mountainous, or (3) no change in land cover classification between temporal periods in remote areas. For those quadrangles where data gaps exist in the mosaicked image, ancillary sources are used to classify the LULC category.

LULC Collection Criteria
The LULC feature collection requirements include an MMU of 2.5 acres and a minimum polygon width of 125 feet. The identification of LULC categories follows a modified Anderson classification system developed by the USGS Rocky Mountain Mapping Center. The USGS Anderson classification system developed in the 1970's has been expanded to include more category levels (see table 1). Currently, the USGS is mapping LULC features down to level 4 of the classification system. Examples of features collected at level 4 include urban parks, natural grasslands, major retail, light industry, and row crops.

Table 1. A part of the modified Anderson classification system illustrating subcategories included in the "developed" land use category

Level 1	Level 2	Level 3	Level 4
2.0 Developed	2.1 Residential	2.11 Single-family residential
		2.12 Multi-family residential
	2.2 Nonresidential developed	2.21 Commercial/Light Industry	2.211 Major retail
			2.212 Mixed/minor retail and services
			2.213 Office
			2.214 Light industry
		2.22 Heavy industry	2.221 Petrochemical refinery
	2.3 Mixed urban

Temporal LULC Collection
The contemporary LULC features are mapped first using the 1990's vintage DOQ's as the source. Upon completion, the 1990's LULC data set is used as the foundation for collecting the 1970's LULC data set. The completed 1990's LULC data set is modified to represent the 1970's land surface by using the historical image as a backdrop to delineate the geographic extent of the LULC features. The 1950's and 1930's LULC data sets are collected by, once again, using the next most recent time period as the land surface foundation, which is modified by using the historical images to represent the temporal land surface. The temporal LULC collection is performed on a quadrangle-by-quadrangle basis. Using the more recent LULC data set as the foundation for collecting earlier time periods avoids the redundant collection of unchanged LULC features between temporal periods.

LULC Quality Control
Once a specific LULC time period is completed for all quadrangles, the delineation and attribution of all LULC features are checked for consistency and quality assurance and, if necessary, corrected within each quadrangle. In addition, individual quadrangle edges are checked for consistent LULC feature delineation and attribution between adjacent quadrangles. After the quadrangle edges are checked, the individual 7.5-minute quadrangles are merged together to form a seamless LULC data set for each time period of the project area.

The contemporary LULC data are checked before the compilation of the historical LULC data sets. Since the contemporary LULC data are used as the foundation for collecting the historical time periods, a quality-controlled 1990's LULC data set reduces the possibility of correcting the same error for all historical periods, especially in those areas where temporal LULC change has not occurred.

LULC Data Applications

The temporal LULC data will be used for land use change analysis and as an input for predicting future land use for the Middle Rio Grande Basin project area. Public access to the USGS temporal LULC data and historical imagery will be made available on the Middle Rio Grande Basin Web site at http://rockyweb.cr.usgs.gov/public/mrgb/lulc_over.html.

Anderson, James, Hardy, Ernest, Roach, John, and Witmer, Richard. A Land Use and Land Cover Classification System for Use with Remote Sensor Data, U.S. Government Printing Office, Washington D.C., 1976