Application of Drone Data to Assess Damage Intensity of Bacterial Leaf Blight Disease on Rice Crop in Indonesia
1C. Hongo, 2S. Isono, 3G. Sigit, 3B. Utoyo, 1E. Tamura
1. Center for Environmental Remote Sensing, Chiba University
2. Graduate School of Science and Engineering, Chiba University
3. Regional Office of Food Crops Service West Java Province, Indonesia
The Government of Indonesia has launched agricultural insurance program since 2016. A key in agricultural insurance is damage assessment which is required to be as precise, quick, quantitative and inexpensive as possible. Current method is to inspect the damage by human eyes of specialist having experiences. This method, however, costs much and is difficult to estimate disease infected fields precisely in wide area. So, there is increasing need to develop effective, simplified and low cost method using remote sensing and GIS, covering wide area. With this background, we conducted research on development of new method using drone data for evaluation of damage ratio caused by bacterial leaf blight (BLB) in West Java, Indonesia.
During the time between dry season of 2019 and 2020, damage assessment of BLB was conducted by damage assessors using current assessment method. Drone-imagery was acquired using the sequoia camera mounted on the DJI phantom before assessing BLB by damage assessors. The ortho-mosaic images were created and then reflectance value was calculated for green band, red band red-edge band and near infrared band. Normalization procedure was applied to all observation bands using r0 (=[Green + Red + Red-edge + NIR]/4). Each band value was divided by r0, and normalized green band (Ngreen), normalized red band (Nred), normalized red-edge band (Nred-edge) and normalized near infrared band (N near infrared) were obtained.
The relationships between Ngreen, Nred, N red-edge, NNIR, NDVI, GNDVI, RGI and the BLB damage assessment result by damage assessor were analyzed to derive the estimation equation of BLB damage assessment. Comparing the correlation coefficient between before and after normalization of the reflectance, all the correlation coefficients between normalized reflectance and BLB damage intensity were higher than before normalization of the data. The result of the evaluation shows positive correlation with reflectance of Nred of which values are 0.72 (significant at 5% level). Visualized maps showed that the BLB damage intensity varied in the irrigation block area and individual paddy field.
Our results indicate possibility of developing new damage assessment method which could realize more effective BLB evaluation through integrating drone data into current method and utilizing the integrated data.
