Effects of Soils and Seasonal Time on Accuracy of Crop Identification in Southwestern Ontario From 1974 Landsat-1 Imagery
Year: 1977
Type: article
Abstract: Imagery from the earth resources satellite, Landsat-1 for the dates of 2 and 20 June, and 8 July 1974, supplemented by airborne photography (1:80,000, and enlarged to 1:3,000), was used to classify crops in Kent County, Ontario. The County was stratified into five areas according to the Canadian Land Inventory Capability classification system. A minimum of three to five fields per crop was selected from each area, using the previous year’s photography, for training prior to 2 June. The percentage of fields accurately identified from the satellite data varied widely among sites, dates and crops. In general, identification accuracy of crops > 80% was achieved by stratifying the area according to soils and using appropriate dates for classifying each crop. Classification accuracy was higher with a layered classifier (developed manually) than with a minimum distance or maximum likelihood classifier. The best date for classifying barley was 2 June (100%); hay, 20 June (87%); winter wheat, 20 June to 8 July (87%); and corn, 8 July (91%). The highest level of accuracy of classification for all crops and dates occurred on the clay loam–clay soils (89%) and the lowest on the sandy loam series (78%). A multiple regression analysis of the values for the average-feature vectors computed from the radiation values of the four bands of Landsat-1 imagery showed that 69% of the variation in the vector values from the 213 fields of the five test sites for 2 and 20 June was attributable to variation in the soil, ground cover, and solar radiation rather than to crop, land class and growth stage.
DOI: 10.4141/cjps77-083
Language: English
Subfield: Ecology
Field: Environmental Science
Domain: Physical Sciences
SDGs: Zero hunger, Life on land
Open Access status: bronze
OpenAlex ID: W2115158337
