I'm looking at a site in Africa, where three 80m-ish masts have been measuring for nearly a year. The wind rose for all is predominantly easterly. The masts lie near the edge of a 2000m plateau which is west of the masts. Two of the masts are roughly the same distance from the "edge", where terrain drops ~500m. They predict themselves and each other well (within 0.1%). One mast lies nearer the edge, on similar altitude to the other two, however it seems to benefit from higher windspeeds as the wind "falls" down the escarpment to the west. To predict this mast from the others leads to a ~15-20% under-prediction, to predict the others from this one leads to a ~15-20% over-prediction. The masts are measuring at multiple heights and have temp, rel hum and pressure measurements at the base. Is there anything I can do by way of parameter adjustment to improve mast to mast predictions for the "edge" mast to account for the acceleration in windspeed caused by the drop in terrain?