Modern optical transport networks are built on the basis of spectral densification technologies with a flexible frequency grid and have the ability to programmatically control optical channels. When designing and operating these networks, routing and spectrum assignment problems for composite optical channels, called light paths, are solved. Existing light path models are mostly descriptive or limited to routing. The set of acceptable light paths is proposed to be formed based on the ratio of optical signal power to noise power in the light path bandwidth for a particular type of signal. A number of optimality criteria for solving the problems of search and distribution of light paths is presented. A mathematical model of a light path is developed, its strict definition of a light path is given and its main properties are formalized to improve the efficiency of spectral resource management in optical transport networks. On the basis of the proposed model it is possible to solve the subproblems of light path routes search taking into account the requirements for non-intersectivity of spectra, contiguity of elementary frequency intervals of each light path, continuity of the spectrum on the whole route.