Improved Inverse Distance Weighted method based on regionalized variable theory

Because there are some defects in theory of Inverse Distance Weighted (IDW), this method does not consider the direction effect of samples around the point to be evaluated and the gathering effect of samples distribution. And some parameters of IDW, such as effective distance, exponential, are decided by experience and the precision of result is affected. This paper introduces regionalized variable theory into IDW to get the effective distance and the parameters of anisotropy ellipsoid by smooth continuity of variogram, and find the value of exponential by Neural Network or Genetic Algorithm. In the end of this paper, an example of improvement IDW is given to compare with Ordinary Kriging, and the result of examination proves precision and reliability of Improvement IDW.