Significant reductions in weather data size without loss of any useful information can be accomplished thorough the use of data compression and rounding to the least significant digit. The source of the weather data's compression can be traced to the data's smoothness and rounding. Rounding off the insignificant digits and extracting the smoothness results in increased symbol frequency. Common applications such as gzip and bzip2 encode frequent symbols with less bits. Hence, more data may be sent over a fixed bandwidth channel if it is first compressed. This paper compares several compression techniques with respect to twelve kilometer weather data from the ETA model.