It is an efficient approach to estimate winter wheat evapotranspiration (ET) from remote sensing images, but it remains a difficult problem to generate a series of daily or seasonal ETs through temporal interpolation from discrete ET estimations. The SEBAL (surface energy balance algorithm for land) model works well only for areas where sunny days dominate and with a relatively pollution-free atmosphere. This paper presents a modification of the SEBAL model for temporal ET interpolation which considers sunshine duration. This method is demonstrably more suitable for areas such as the Haihe basin in east China. A sensitivity analysis of the sunshine duration parameters was conducted for winter wheat ET in the Shijin irrigation area in Haihe basin for the period starting Oct. 2007 and ending Jun. 2008. Seasonal ET interpolations were based on MODIS products and validated by ground measurements from the Luancheng Agro-Ecosystem Experimental Station. The results show that the seasonal ET interpolations taking account of sunshine duration in the SEBAL model are better than results without considering the sunshine duration. A small change in the sunshine duration parameters results in significant change in the seasonal ET interpolation. Therefore, the SEBAL model is highly sensitive to the sunshine duration parameters and such a modification of SEBAL model improves the results of temporal interpolation.