A Model for Estimation of Agricultural Soil Fertilizer Using Simulated Data

摘要

Precision agriculture needs continuous monitoring of soil and its fertilizer contents qualitatively as well as quantitatively. The most used fertilizers are Calcium Nitrate, Potassium Sulphate, Ammonium nitrate, Ammonium Sulphate, Potassium nitrate, and NPK .Usually different proportions of these fertilizers are added to the soil for different crops. These fertilizers affect the reflectance of the soil proportional to their shares and consequently produce a unique reflectance for the mixed soil. Knowing the reflectance of these fertilizers as well as the reflectance of the soil and their proportions in the mixed soil, one can simulate the reflectance of mixed soil. Then extracting the reflectance of the pixel from satellite image and comparing it with the simulated data, one can estimate the proportions of each fertilizer in the mixed soil. A set of simulated data for different proportions of different fertilizers was made. For this, the reflectance curves of fertilizer as well as the reflectance curve of the base soil were used. The method consists of a regression between the field reflectance with those of simulated data. The results show a R~2 > 0.80 between predicted and measured data for different Fertilizers. The method is implemented for the Hyperion images where the results were promising.