A simple temporal and spatial analysis of flow in complex terrain in the context of wind energy modelling

A simple temporal and spatial analysis is done on wind speed and direction data from a number of meteorological towers separated by distances between roughly 1 and 100 kilometres. The analysis is done in the context of expected model error in wind energy calculations. The study first uses single point statistics to show the evolution of mean values with time. It is shown that strong seasonal signals are present and that stable means are achieved only after averaging periods of a year or more. The study then uses discrete Fourier transforms to show that significant amounts of spectral energy reside in modes with periods of a few days to less than a day. Frequency dependent cross correlation values are then derived and used to show how correlation between towers diminishes with increasing frequency. The mechanism responsible for this diminished correlation is shown through the comparison of cross-correlation phase as a function of frequency and its relationship to distance between towers. Error in wind energy estimates are shown to be strongly related to correlation and therefore distance over which the prediction is made. In summary, much of the inaccuracy in modelling flow in the context of wind energy calculations is due to a lack of scale separation between the deterministic part of the flow, which is well modelled, and that part of the flow that is stochastic at the length and time scales modelled.