276] The Harmonoise engineering method is developed for computation of long-term noise indicators Lden and Lnight. Since sound propagation effects can strongly depend on meteorological conditions, the method must be able to handle different meteorological conditions that occur during a year. In order to obtain a more accurate description of the sources, a subdivision into subsources at different heights is necessary, distinguishing different physical mechanisms: rolling noise, traction noise and, for high speed trains, aerodynamic noise. Combining such comprehensive source and propagation models into an integrated method for use in complex 3-dimensional noise maps, a balance had to be found between the accuracy of computation versus computation speed, requiring optimization and simplification of algorithms. On the other hand, for continuity of the model, additional algorithms had to be developed to improve methods that have been developed merely for ‘academic cases’. The Fresnel-zone weighting principle, adopted from the Nord2000 method, can even be extended to reflecting and diffracting obstacles with finite length. Algorithms have been developed for transformation of meteorological data into an equivalent ray curvature, depending on the direction of propagation and local geometry. The resulting engineering method is flexible in such a way that it can be used both for detailed computations in case of noise assessment and for noise mapping, requiring a higher speed of computations. This computation time is saved by reduction of input data, using the same calculation engine. 1 STRUCTURE OF THE HARMONOISE PROJECT The HARMONOISE projects consists of 7 work packages. Work package 1 has been subdivided into WP1.1 for road sources and WP1.2 for railway sources. These work packages provide physical models for an acoustical description of individual, moving vehicles. In work package 2, a reference model for sound propagation is being developed, used for validation of point-to-point computations. Work package 4 provides empirical data from different measurement sites for validation of the engineering method in practical situations.