Pre-harvest cotton acreage and production forecast using remote sensing techniques.

摘要

Cotton production forecast (kharif 2003) was attempted using remote sensing techniques under the Crop Acreage and Production Estimation Project funded by the Ministry of Agriculture. The 4 major cotton-growing districts, namely Hisar, Fatehabad, Sirsa and Jind, which accounted for 83% of cotton acreage and 87% of total cotton production in Haryana, India, were selected. District level cotton acreages were estimated using the Indian Remote Sensing Satellite digital data of LISS-III of the first week of October 2003, coinciding with flowering to boll formation stage of cotton. Ground truth data were collected near synchronous to satellite pass. The maximum likelihood supervised classification and complete enumeration approach was used for district level cotton acreage estimation. The cotton yields were estimated using trend yield and agrometeorological data along with the crop condition term to account for damage due to pests/diseases. The yield estimates were revised using the meteorological data up to September end, which showed a decline in cotton yields mainly due to higher relative humidity during the first fortnight of July and higher minimum temperature during the first fortnight of June. The state level cotton acreage and production were estimated using extrapolation factors computed from data of the previous seven years. The performance of these models was evaluated by computing relative deviation with BES estimates. State level cotton acreage of 591 thousand hectare and production of 1391 thousand bales showed an increase of 11% in acreage and decrease in production by 1.0%, respectively, compared to the Department of Agriculture estimates.