Due to the greenhouse effect, the air temperature is now rising, and the changes (drop or rise) of the water level of salt lakes in arid-semi-arid salt lake areas of western China is speeding up. Though there are many lakes in northern Tibetan Plateau, there are no lake level records, except at salt Lake Zabuye, where a long-term field observatory station was established in 1990 by Open Laboratory of Saline Lake Resources and Environment, CAGS, and lake level was recorded monthly. How to forecast the short-term lake-level change accurately and quantitatively according to the historical lake-level records is a major issue concerning the fate of salt lake resource development. In this paper, according to the lake-level records observed during 1991-2003 in salt Lake Zabuye in Tibetan Plateau, we discussed the applications of two forecasting methods of time-serial analysis, the model of Winters linear and seasonal exponent smoothness and the model of ARIMA seasonal multiplication, in forecasting the dynamic lake-level change of salt lakes.