Correlation analysis on soil nutrients and wavelet decompositions of satellite imagery

Understanding soil nutrients using remote sensing imagery are highly insisted by researchers for a nation with commendatory agricultural landscapes such as India. A robust prediction model is always useful to perform cost efficient prediction using remote sensing imagery. This paper reports a preliminary analysis on wavelet transformations and their correlations with various soil nutrient contents to support and improve the prediction model. Firstly, we provide the chemical analysis report of our study regions, which are Baggi, Ibrahimpur, Mogara and Wai villages from Maharahstra province of India. The report reveals pH level, electrical conductivity, carbon, phosphorous and potassium contents of soil samples. Subsequently, we acquired satellite imagery and perform wavelet decompositions using first order Daubechies, Symlet, Biorthogonal and Coilflet wavelet transformations. Analysis is performed in MATLAB to determine the correlation between the principle wavelet components and the soil nutrients. The analytical outcomes are believed to be a cornerstone for deriving wavelet decomposition based prediction models.