Problems and methods of numerical and experimental studies of wind flow in the surface layer in the coastal zone "sea-land" and its impact on the urban area with high-rise constructions in view of a discrete roughness terrain and its topography are discussed. Some environmental conditions associated with the transport and dispersion of contaminants are under consideration. Among the problems to be solved, there are several interconnected tasks: prediction of unsteady wind loads on high-rise buildings in urban area taking into account discrete roughness (including forests and topography) in the coastal zone; rational arrangement of buildings in the city area on the basis of the minimum drag criterion; experimental study of mean and fluctuation characteristics in turbulent flow over isolated body, tandem and packages of bodies. Application of numerical simulation in the framework of multiblock computational technologies to make valid prediction of wind loads in urban area with high-rise buildings using an approach based on solving URANS equation system jointly with SAS-MSST models of turbulence can be used. The special attention is paid to development of the coordinated conditions on entrance border of calculating area on the basis of the solution of a three-dimensional task on evolution of atmospheric boundary layer on a rough surface. Experimental study of turbulent characteristics in shear areas of a flow around building models is performed by means of the intrusive diagnostics which are capable of measuring instantaneous positive and negative velocities with high time resolution, including the hot-wire technique developed at ITAM SB RAS and by split-fiber film technique developed at Department of Astronautics and Aeronautics (DAA) of National Chen Kung University (NCKU). Accuracies of the employed intrusive diagnostics are examined by comparing with the data obtained with the non-intrusive PIV technique. At studying of non-stationary processes the application of methods of the Wavelet-analysis and empirical decomposition of fluctuations, developed in DAA on time-frequency modes with Hilbert transformation is employed.