The essences of traditional grey forecasting theory are background value and class ratio. In traditional GM(1,1) model, the background value is an average one which is restricted on a point and class ratio is too limited to be accepted that may only be agreeable to monotone increasing or decreasing cases. Therefore, a linear assumption and an optimal alpha are introduced for background value. Besides, base on spatial perspective, an error analysis will be constructed to improve comprehension of this model. A comparison of example indicates that the modified approach is probably to reduce forecasting error by RMSE evaluation. Besides, this is the first part of series paper and gradual modifications will also be proposed to enhance applications of GM(1,1) in the future.