Remote Estimation of Vegetation Parameters using Narrowband Sensor for Precision Agriculture in Arid Environment

Document Type : Original Article

Author

Soil and Water Department, Faculty of Agriculture, Suez Canal University, 41522, Ismailia, Egypt

Abstract

The performance of hyperspectral (Hyperion) sensor to evaluate and propose potential vegetation indices and to discriminate land cover classes in arid and semi-arid areas was evaluated. The requirements for extracting indices from Hyperion images are to be first compensated for atmospheric correction using FLAASH model. The minimum noise fraction transformation was applied to reduce the data noise and for extracting the extreme pixels. Some pure pixel endmember for the target types and backgrounds was used in this study to account for the Spectral Angle Mapping and the arrived results were validated with field study. Different classes of vegetation wereidentified from the different images: plantation, grassland, crops and trees at initial and senescence stage. The spectral signature for the differentvegetation is identified and used for spectral library generationand image classification. Vegetation, which is in the senescence stage,can bedistinguished visually from the Hyperion image. The overall accuracy was 96%, 67%, and 91%, for the 177039 image, 177045 image, and 175042image, respectively. Therefore, Hyperion data havehigh capability for crop types mapping and extraction of useful quantitative information for the purpose of crop monitoring and management. 30 vegetation indices have been assessed in this study. Four indices were proposedto support precision agriculture in an arid environment, which isvegetal cover density (VCDI), vegetal water content (VWCI), vegetal water stress (VWSI), and vegetal discriminating (VDI).Thermal imagerycan help in the detection of different infection in different crops from the first hours ofsuccessful germination of conidiospores. The temperature difference allowed the discrimination between infected and healthy leaves before the appearance of visible necrosis on leaves.

Keywords