Regional Atmospheric Modeling System (RAMS) was applied to complete
a series of numerical simulations of convective plumes over ice leads.
Two RAMS configurations were used: (1) the LES (Large Eddy Simulation)
formulation for the horizontal scale up to several hundreds meters, and (2) the mesoscale formulation for the horizontal scale up to several tens of kilometers. The simulations were performed for different combinations of input parameters characterizing a single idealized ice lead (width, surface heat flux) as well as different wind velocity. Finally, source-receptor relationships in a flow over ice leads were investigated with the aid of Lagrangian tracer particles. This approach allowed us to study not only vertical heat flux as distributed from the source but also the vertical heat flux as seen by a sensor at given location in reference to the ice lead, i.e., in terms of upwind flux footprints. The concept of flux footprint has been applied in many studies but so far to homogenous surface only. Our paper presents its extension to nonhomogeneous terrain.