Based on the measured meteorological data，the change of land surface temperature under different weather conditions and the relationship between surface energy and surface temperature are discussed by using the Bowen ratio method to calculate the Splendens ground surface energy. The land surface temperature forecasting models of the oasis-desert ecotone in the southeast of the Dzungaria Basin in Xinjiang is established. The results indicate that: ① The curve of top soil (5 cm，10 cm，15 cm) temperatures，affected by different characteristics of the land surface energy under different weather conditions，shows a "S" type. Rainy days have a curve similar to that of sunny days，only with a weakened amplitude. The cloudy days manifest the feature of linear decrease. The time point of the peak of soil temperature reveals 2.5 h/5cm lagging from the surface downwards. Regardless of the weather，the soil temperature below 40 cm has almost no diurnal variation. ② The correlation coefficient between each energy component and soil temperature decreases continuously with the depth of soils，which reaches the maximum negative correlation at 20 cm. ③ The degree of influence of energy component on soil temperature is as follows: soil heat flux (G) is biggest，followed by the net radiation (Rn) and sensible heat (H)，and latent heat (LE) is minimum. ④ Except for H，the correlation between energy component and the soil temperature under different weather conditions showed that: sunny> the average> cloudy> rainy. The response extent of the soil temperature to the energy change (regression coefficients) almost accords with the same rule. ⑤ Established and certified land surface temperature forecasting models suggest that the land surface temperature is affected only slightly by the hygrothermal energy factor (LE) in the arid oasisdesert ecotone.