<![CDATA[Research on the development of a paddy field classification model based on Landsat remote sensing images aims to obtain a rapid classification of paddy field models. This study uses input multitemporal Landsat images (path/row 122/064) in 2017. The research was conducted in Subang regency, which is one of the center of West Java rice production. The method used in this study is the threshold method for the multi-temporal Landsat image index. As a reference, detailed scale spatial information on paddy fields base is used which is supplemented with data from field surveys using drones. First, an atmospheric correction of Landsat images was carried out using DOS (Dark Object Subtraction) Method, then transformation image to several indices: Enhance vegetation Index (EVI), Normal Difference Water Index (NDWI), and Normal Difference bare Index (NDBI) was carried out. For cloudy images, the index is filled with interpolation techniques from the index value before and after. The next step is smoothing index and statistical analysis to obtain minimum, maximum, mean, median, range (maximum - minimum), EVI_planting, EVI_harvesting, mean_planting-harvesting, mean_vegetative, mean_generative, NDWI_planting, NDWI_harvesting, NDBI_planting, and NDBI_harvesting. Classification accuracy is calculated by using the confusion matrix technique using detailed scale spatial information references. Based on the analysis and test of accuracy that has been done on several models, the highest accuracy is generated by the 1.5 stdev threshold model four index parameters (EVI_min, EVI_Max, EVI_range, and EVI_mean) with an accuracy of 86.56% and a kappa value of 0.716.]]>
Copyrights © 2019
