Research in Satellite-Aided Crop Forecasting,
Abstract
Developments of efficient and accurate automated procedures for analysis of multitemporal Landsat multispectral scanner data to extract information on crop area and production can provide a greatly improved capability for practical and affordable use, on a global basis, without requiring ground observations. Evaluations of procedures developed specifically to estimate non-U.S. spring small grains area show accuracies of less than 10 percent relative difference to U. S. Department of Agriculture reference statistics for North Dakota in 1978 and good comparison with 9000 square miles of observations over four states (Montana, North Dakota, South Dakota, and Minnesota) and Saskatchewan, Canada during years (1976-79). Processing a 5x6-nautical-mile sample site requires a few minutes manual time and a few minutes central processing unit time on an AS-3000 computer. Evaluations of summer crop, corn, and soybeans area estimates show unbiased summer crops estimates in the U.S. central corn belt but significant bias in 1 of 2 years for area estimates of corn and soybeans. We expect, based on results to date, to achieve a highly automated corn/sorghum/soybean area estimation procedure that is applicable to Argentina. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1982
- Accession Number
- ADP001992
Entities
People
- J. D. Erickson
- J. L. Dragg
- M. C. Trichel
- R. M. Bizzell
Organizations
- National Aeronautics and Space Administration