Abstract: Trend surface analysis method for spatial interpolation presents a smoothed surface which may be very different from the real spatial distribution, especially for local area with distinct variable data. To solve this problem, a combined global – local interpolation method was proposed, which integrates the trend surface analysis method for global interpolation and the ordinary Kriging method with a post-processing procedure of Yamamoto to correct the smoothing effect for local interpolation. This combined interpolation method was used for the interpolation of regional groundwater level. The spatial distribution results using this combined approach were closer to the observed data. Therefore, the combined interpolation method not only ensured the global optimum but also corrected the smoothing effect of trend surface analysis and increased local estimation accuracy. The interpolation results can depict the distribution of groundwater level more precisely by integrating global and local information of observed data.