Aim: Spatial and spectral satellite image segmentation and land cover classification utilising Novel linear regression versus random forest with higher accuracy Linear regression outperform random forest in terms of accuracy Materials and Methods: Multispectral Satellite Image Segmentation using Linear Regression (N=10) and Random Forest (N=10) with the split size of training and testing dataset 60% and 40% using G-power setting parameters: (α=0.05 and power=0.85) respectively Results: Linear Regression with Accuracy 80.04 % is more Accurate than the Random Forest with Accuracy 74.07% and attained the significance value 0.053 (Two tailed, p>0.05) Conclusion: The Linear Regression model is significantly better than the Random Forest for multispectral satellite Novel Image Segmentation.