Biography:

Abstract:

Crop yield forecasts a few months before harvest can be of paramount importance for timely initiating food trade secure the national demand and timely organize food transport within countries Forecasting enables planners and decision makers to determine how much to import (in shortfall case) or optionally, to export. The focus of this study is to use remote sensing data to calculate crop yield based on remotely sensed statistical models. The main input parameters of theses models are spectral data either in form of spectral reflectance data that are released from the different SPOT bands or in forms of spectral vegetation indices that are algebraic ratios generated from the spectral reflectance values, the other type of the input factors is leaf area index (LAI) that is a biophysical parameter closely related to crop canopy spectral characteristics and was measured by LAI Plant Canopy Analyzer. The four spectral bands of SPOT4 imagery are: green, red, near infrared, middle infrared, the five vegetation indices that are calculated through different forms of ratios between red and near-infrared bands are : Normalized Difference Vegetation Index (NDVI), Ratio Vegetation Index (RVI), Soil Adjusted Vegetation Index (SAVI), Difference Vegetation Index (DVI) and Infrared Percentage Vegetation Index (IPVI), one vegetation index that is calculated through a ratio between green band and near-infrared band which is called Green Vegetation Index (GVI) and finally Leaf Area Index (LAI)The generated models were validated through two main steps: the first step is the correlation coefficient that is released from the generated models while the second step is the validation through testing the yield that is calculated through the generated models (modeled yield) against the yield that is reported from the technical office of Sakha experimental station (reported yield). Testing modeled yield against reported yield was carried out trough two common statistical tests: standard error of estimate between modeled yield and reported yield and the correlation coefficient for a direct regression analysis between modeled and predicted yield for each generated model. Generally, as shown from the correlation coefficient of the generated models that green and middle infrared bands did not show good accuracy to predict wheat yield while the other spectral bands (red and near infrared) bands showed high accuracy and sufficiency to predict the yield. This was proved through correlation coefficient of the generated models and through the generated models with the two crops for the two seasons Accordingly, the green vegetation index that is generally calculated from green and near infrared bands showed relatively lower accuracy than the rest of the vegetation indices models that are calculated from red and near infrared bands. It is clear that using LAI with other spectral factor increased the accuracy of the generated models as shown from the validation results for all models. The models are applicable after 90 days from sowing date for similar regions with same conditions .