The wind energy industry in Mexico is developing rapidly with the capacity projected to increase by a factor of 5 by 2030. However, the increased penetration of wind energy presents a series of challenges for energy system stakeholders including generators, system operators and policy makers. The aims of this collaborative project is to; (1) characterise the long term statistics of wind generation in Mexico, (2) understand the large-scale meteorological drivers of wind generation in Mexico and (3) determine the skill of Numerical Weather Prediction (NWP) models at estimating wind generation in Mexico for a range of lead times. This poster will present the initial results on the characterisation and predictability of wind power in Mexico.
State-of-the-art reanalysis datasets have been used extensively to determine the long-term characteristics of wind generation for several nations/regions [1,2]. This study examines the applicability of such methods for Mexico using the MERRA2 dataset. In addition, 1 year of WRF simulations have been performed to estimate the wind field at a range of high spatial resolutions (1, 3, 15, 75 and 120 km). To assess the performance of the models; (1) the wind fields from both have been compared to observational data at a number of sites across the country and (2) the estimated wind power generation at a location in Arriaga, Chiapas has been compared to metered data.
The preliminary results show that MERRA2 provides a good representation of the low frequency variability (daily to annual) of wind generation in Mexico. However, in some regions the model does not capture the variability on shorter time scales. The WRF simulations do provide an improved representation at these time scales. However, further analysis is required to determine the benefit of this to wind resource assessment.
The predictability of wind power in Mexico has been assessed using the operational global ensemble forecasts archived by the International Grand Global Ensemble (TIGGE) from ten forecasting centres. The wind fields from the models are used in conjunction with turbine power curves to estimate the power output of several wind farms in Mexico. The skill of the forecasts at estimating the generation will be assessed as a function of lead time using several metrics including the Continuous Rank Probability Skill Score (CRPSS). In particular, the analysis will determine the predictability of key wind generation events, such as extreme ramping and periods of persistently low or high generation.