Remote sensing analysis of rice disease stresses for farm pest management using wide-band airborne data

Zhihao Qin, Minghua Zhang, T. Christensen, Wenjuan Li, Huajun Tang
IGARSS 2003. 2003 IEEE International Geoscience and Remote Sensing Symposium. Proceedings (IEEE Cat. No.03CH37477)  
On-farm pest management and crop protection strongly depend on diagnosis of crop disease stress in the fields. In this paper, we first examine the applicability of broadband highspatial-resolution ADAR (airborne data acquisition and registration) remote sensing data in visible and near infrared regions for rice disease detection and then develop an approach to explore the applicability. Experiments were conducted in 1999 on a largescale rice field with sheath blight infection in central
more » ... , USA. Based on the measured symptoms, a comprehensive ground disease index (DI) was constructed to indicate the infection severity. Correlations between ground data and image data were analyzed with attempt to develop an applicable method for remote sensing of the rice disease. The results indicate that the broadband remote sensing imagery has valuable capability of application. Some image indices such as the RI 14 , SDI 14 and SDI 24 have a correlation of above 0.62, hence are valuable for rice disease identification. A method based on the indices has been developed. Validation with the ground data indicates that the method has an average accuracy of above 70% for classification. The standard estimate error of the method is ~13%. In spite of this encouraging result, we also realize that it is really difficult to discriminate the healthy plants from light infection ones (DI<20%) because of their heavily overlapping in the estimated image indices. Identification is much more accurate when infection reaches to mediate-tosevere levels (DI>35%). High spectral resolution remote sensing imagery with more bands and narrower bandwidth is required for remote sensing diagnosis of crop disease stress.
doi:10.1109/igarss.2003.1294393 fatcat:juwx5qeuznb3vic3lfpqe734im