I have run a regression (OLS) to estimate the strength of the relationship between imperviousness (dependent variable) and population density, and land cover using a model impervious surface extracted from high resolution images from two different periods at least 10years apart. i am getting adjusted sq R values between 0.7  and 0,955095 for various scenarios at 5% probability. Now i want to develop a model that will use these variables to predict how much imperviousness is in an area of a given population density and land cover type. I hope to further use this to establish a link with run-off generation. I will be grateful for your suggestions. (I am using works by Bauer et al 2005, 2007 and Chabaeva 2004 as a guide)