On Optimal Prediction of Atmospheric Ekman Layer Fluctuations over Short Lead Times

Abstract

Wind and temperature time series, sampled at 10 min intervals 250 and 600 m above the ground, are analyzed. For horizontal wind components and temperature forecasting, autoregressive integrated models of the type (p, 1, 0); p ≈ 4 provide the best alternatives among autoregressive integrated moving average models. No single model is substantially better than others. For the vertical velocity component, no model provides a substantially better alternative than to predict it equal to zero.