Abstract: Traditional methods of requirements engineering has not been able to really capture the elasticity and all the information contained in requirements as expressed by clients who are only domain experienced or even experts. The system analyst which is the first interface of requirements capture, traditionally a person recognizes the requirements from his perspective of possible computer aided solutions. From their digital background the decisions normally are to ignore or workaround requirements that seem imprecise or conflicting. A lot of information contained in the informal human communication is lost. In this study, we describes a novel process using Fuzzy Approximation Theorem. First, we discuss the current methodology and why it has not been able to represent the information contained in the inexact, imprecise and conflicting human language of the client or domain expert. We then examine the problems faced in requirement engineering in Pollution forecasting and present a process using Fuzzy logic.