Search

The feasibility and utility of using satellite data and computer-aided remote sensing analysis techniques to conduct range inventories were tested. This pilot study was focused over a 250,000 acre site in Galveston and Brazoria Counties along the Texas Gulf Coast. Rectified enlarged aircraft color infrared photographs of this site were used as the ground truth base. The different land categories were identified, delineated, and measured. Multispectral scanner (MSS) bulk data from LANDSAT-1 was received and analyzed with the Image 100 pattern recognition system. Features of interest were delineated on the image console giving the number of picture elements classified; the picture elements were converted to acreages and the accuracy of the technique was evaluated by comparison with data base results for three test sites. The accuracies for computer aided classification of coastal marshes ranged from 89% to 96%.

The Edit 9 forest scene, a computer processing technique, and its capability to map timber types in the Sam Houston National Forest, are evaluated. Special efforts were made to evaluate existing computer processing techniques in mapping timber types using ERTS-1 and aircraft data, and to provide an opportunity to open up new research and development areas in forestry data.

There are no author-identified significant results in this report.

The author has identified the following significant results. An assumption that short prairie grass and salt grass could be differentiated on aircraft photographs was inaccurate for the Weld County site. However, rangeland could be differentiated using procedure 1 from LACIE. Estimates derived from either random or systematic sampling were satisfactory. Level 1 features were separated and mapped, and proportions were estimated with accompanying confidence statements.

The author has identified the following significant results. It was determined that hardwood in Warren County, Pennsylvania could best be inventoried in May. The acreage estimate was less than 3% different from Forest Service estimates.

The author has identified the following significant results. An area inventory was performed for three southeast Texas counties (Montgomery, Walker, and San Jacinto) totaling 0.65 million hectares. The inventory was performed using a two level hierarchy. Level 1 was divided into forestland, rangeland, and other land. Forestland was separated into Level 2 categories: pine, hardwood, and mixed; rangeland was not separated further. Results consisted of area statistics for each county and for the entire study site for pine, hardwood, mixed, rangeland, and other land. Color coded county classification maps were produced for the May data set, and procedures were developed and tested.