Implementation of a Model Output Statistics based on meteorological variable screening for short‐term wind power forecast

Abstract

A combination of physical and statistical treatments to post-process numerical weather predictions (NWP) outputs is needed for successful short-term wind power forecasts. One of the most promising and effective approaches for statistical treatment is the Model Output Statistics (MOS) technique. In this study, a MOS based on multiple linear regression is proposed: the model screens the most relevant NWP forecast variables and selects the best predictors to fit a regression equation that minimizes the forecast errors, utilizing wind farm power output measurements as input. The performance of the method is evaluated in two wind farms, located in different topographical areas and with different NWP grid spacing. Because of the high seasonal variability of NWP forecasts, it was considered appropriate to implement monthly stratified MOS. In both wind farms, the first predictors were always wind speeds (at different heights) or friction velocity. When friction velocity is the first predictor, the proposed MOS forecasts resulted to be highly dependent on the friction velocity-wind speed correlation. Negligible improvements were encountered when including more than two predictors in the regression equation. The proposed MOS performed well in both wind farms, and its forecasts compare positively with an actual operative model in use at Risø DTU and other MOS types, showing minimum BIAS and improving NWP power forecast of around 15% in terms of root mean square error. Further improvements could be obtained by the implementation of a more refined MOS stratification, e.g. fitting specific equations in different synoptic situations.

An improved random forest model of short‐term wind‐power forecasting to enhance accuracy, efficiency, and robustness

“…Although the time scale in short-term wind power forecasting is only 1 to 3 days and there are hundreds of data points (assuming a sampling interval is 15 min), the time scale for historical data of each physical variable (including temperature, air pressure, humidity, wind direction, wind speed at different heights, and output power of wind farms) is at least a year with nearly 40 000 data points. 30 In the massive historical data chains with multiple variables, only a few characteristic variables and 1 part of the data samples are strongly related to the forecasting day. The matrix of characteristics correlation between the forecasting and historical samples is sparse and can hardly meet the minimal-redundancy and maximal-relevance principles.…”

High-resolution modelling of atmospheric dispersion of dense gas using TWODEE-2.1: application to the 1986 Lake Nyos limnic eruption

“…On the one hand, statistical downscaling methodologies (e.g. Ranaboldo et al, 2013;Devis et al, 2013;Kirchmeier et al, 2014) build on finding correlations between global/regional model simulations and observations during long periods of time in order to identify patterns used to forecast in time and space by extrapolation. This has proven to be effective for some applications (e.g.…”

Wind speed forecast correction models using polynomial neural networks

“…Purely statistical models, on the other hand, are excellent at forecasting idiosyncrasies in local weather but are usually worthless beyond about 6 h [2]. Model output statistics (MOS) can combine both the complex numerical forecasts, based on the physics of the atmosphere to forecast the large-scale weather patterns, with the use of regression equations in the statistical post-processing to clarify the surface weather details [3]. Several further correction methods were developed to estimate and eliminate global weather system model errors, induced due to uncertain initial conditions, data and computational limitations [4].…”

“…n À number of input variables xðx 1 ; x 2 ; …; x n Þ aða 1 ; a 2 ; …; a n Þ; … À vectors ðmatrixesÞ of parameters (1) polynomial PDE terms (3). It replaces mathematical operators and symbols in a PDE by the ratio of the corresponding variables.…”