Current practices in mineral resource estimation challenge resource engineers to analyse locally varying anisotropies more effectively. These anisotropies, driven by local geological conditions, necessitate either sub-domaining of a geological or mineral domain or the use of locally aligned search ellipsoids. This paper introduces a new algorithm designed to enhance estimation accuracy by systematically addressing and incorporating local anisotropies as part of the estimation process. The algorithm begins by generating directional experimental variograms within a domain at set increments to define anisotropy directions. Once identified, it proceeds by automatically fitting authorized variogram models (Spherical, Exponential or Gaussian) to the experimental variograms, minimising Root Mean Squared Errors (RMSE) to determine the longest axis of global anisotropy. This axis serves as the search radius for filtering local data within a geological neighbourhood. Subsequently, only the filtered local data are used to generate local experimental directional variograms for each estimation locations, followed by auto-fitting and grade estimation using a locally adapted anisotropic variogram. To test the algorithm, a case study was generated using Walker Lake data set. The case study aimed to estimate unknown values at 725 grid locations from the 470 samples covering the Walker Lake area. The case study with a small sample set showed that more accurate estimation of unknown values at the grid locations is possible for the Walker Lake area if the local variogram anisotropy is determined and used before the estimation of the value at each grid location.