Research on rice acreage estimation in fragmented area based on decomposition of mixed pixels

• Published in: International archives of the photogrammetry, remote sensing and spatial information sciences. Vol. XL-7/W3; no. 7; pp. 133 - 139
• Main Authors: Zhang, H., Li, Q. Z., Lei, F., Du, X., Wei, J. D.
• Format: Journal Article Conference Proceeding
• Language: English
• Published: Gottingen Copernicus GmbH 28.04.2015; Copernicus Publications
• ISSN: 2194-9034; 1682-1750; 2194-9034
• DOI: 10.5194/isprsarchives-XL-7-W3-133-2015

Rice acreage estimation is a key aspect to guarantee food security and also important to support government agricultural subsidy system. In this paper, we explored a sophisticated method to improve rice estimation accuracy at county scale and we developed our approach with China Environment Satellite HJ-1A/B data in Hunan Province, a fragmented area with complex rice cropping patterns. Our approach improved the estimation accuracy by combing supervised and unsupervised classification upon decomposition of mixed pixels model, and the rice estimation results, validated by ground survey data, showed a close relationship (RMSE≈3.40) with survey figures, the estimated accuracy (EA) reached 83.74% at county level according to the sub-pixel method, and the accuracy can be increased about 12% compared to the pure-pixel method. The results suggest that decomposition of mixed pixels method has great significance to the improvement of rice acreage estimation accuracy, and can be used in mountainous and broken planting area.