An original approach is proposed to the problem of reconstructing mesometeorological fields in the territories uncovered with observational data, which is based on the optimum combination of alternative techniques for spatial prediction. Some grounding in theory of appropriate mathematical methods, such as polynomial fitting, optimal extrapolation, and modified method of clustering of arguments (MMCA) are examined along with an integrated algorithm for spatial prediction of vertical profiles of temperature and wind velocity.