جستجوی مقالات مرتبط با کلیدواژه « مدل سازی عملکرد » در نشریات گروه « مکانیزاسیون کشاورزی »

Prediction of sugarcane yield is very important for a wide range of applications like, sugarcane production management, preparation of sugarcane refineries and pre-sales and warehouse industrial products. In this study, a model based on time-series processing of vegetation indices, Normalized Difference Vegetation Index (NDVI), Green Normalized Differnce Vegetation Index (GNDVI) and Enhanced Vegetation Index (EVI) extracted from satellite images, were used to estimate sugarcane yield. Overall 474 Landsat 7 satellite images from January 2001 to December 2017 obtained from USGS (U.S. Geological Survey) were processed. At first the DN (Digital Number) of pixels were converted to TOA (Top of Atmosphere) reflectance and then the distorted pixels due to not clear sky such as cloud, shadow, snow and ice were eliminated. Consequently, the average of the vegetation indices values of study region for every images were computed. Then the weekly time-series of vegetation indices were calculated via interpolation. The accumulated vegetation indices values from 15th to 44 th week of year and average observed yields efficiency were evaluated by regression model. The result showed the NDVI and GNDVI vegetation indices with R2=0.63, RMSE=4.71 ton/ha and R2=0.60, RMSE 4.93 ton/ha, respectively, have good relations with sugarcane stem yield efficiency. The 2017 sugarcane yield of MianAB Sugarcane Agro-Industry Company efficiency was predicted as 86.35 ton/ha using the NDVI model which was 4.16 ton/ha less than observed value.