Estimating Leaf Area Index of a corn silage field Using a Modified Commercial Digital Camera

Leaf area index (LAI) is an important indicator of plant growth and yield. Therefore, monitoring the spatial and temporal distribution of LAI at agricultural farms could be a significant predictor of how well the various elements of farm management strategies such as irrigation scheduling and uniformity have been implemented. The purpose of this study is to outline how pictures taken by a modified digital camera can be used for estimating the LAI of a corn silage field. It focuses on how to utilize a combination of a simple digital photographic cameras with spectral filters, designed to provide multispectral images in the visible and near-infrared domains, to estimate LAI. In order to remove the sources of errors in the resulting images, procedures to perform image vignetting corrections, geometric distortions corrections, and elimination of radiometric bidirectional effects are suggested. Due to high spatial resolution of this imaging system (at the level of a few centimeters), separation of surfaces with and without plant cover was accomplished well. This separation process was also useful in determination of percentage of vegetation cover (crop density). The leaf area index had the highest correlation with the vegetation cover percentage (R2 = 0/919), and the NIR spectral band (R2 = 0/741). There was a high correlation between the two spectra of red and green with the NIR spectral band. This correlation indicates that with the presence of the NIR spectral band, the effect of red and green spectral bands on estimating leaf area index is insignificant. Therefore, a multivariable regression model was generated to estimate leaf area index as a function of only two parameters, namely vegetation cover percentage and spectral band NIR. The performance of the developed model was evaluated by comparing its predicted values of LAI with corresponding measured values. The adjusted coefficient of determination of this comparison was 96.6%, which indicates that 96.6% of the variation in the estimated leaf area index values is explained by the two variables (vegetation cover percentage and NIR spectral band) incorporated into the model.