The off-line and fixed annual average failure probability model is used in conventional reliability assessment, it is difficult to adapt to the current needs of power system operation. Power grid faults have great temporal relationship with disastrous weather, and meteorological disasters have distinct seasonal characteristic, thus power grid risk will fluctuate with time change. When descripting power component failure rate, it is necessary to expand from chronological time to historical period and taking consideration of the reasons especially external meteorological environments which cause faults, in order to describe the temporal and spatial environment dependent failure rate. The component failure rate model is established and applied to power grid reliability assessment, based on conventional statistical model in different timescale and under different meteorological disasters. The proposed models are utilized to analyze the 500 kV transmission lines failure data during nearly 3 years, meanwhile, the proposed method is applied to assess the reliability of a provincial power grid in southwest China. The results indicate that the reliability indexes in different time intervals have major difference with annual average index, and it proves the necessity of temporal and spatial environments dependent reliability assessment.