Description
One of the weaknesses of the renewable energies is the variability associated to the natural processes that are involved in the power generation. The main proposal of this project is the generation of new EPS members by applying statistical strategies to obtain initial (IC) and boundary conditions (BC) involved in the system. The application of this methodology will identify similar atmospheric conditions to provide the IC and BC to characterize the forcing and uncertainties of weather-like nature that contribute to errors in predictions. The application of this new methodology could be a very powerful tool in the wind power industry, reducing on one hand the computational costs associated to the EPS forecasts and avoiding on the other hand the observational data to calibrate the new model. The experience in statistical and numerical modeling provided by the researcher will be available for the donor partner on the other hand the knowledge of the specific requirements of the offshore wind power industry will be provided by the donor partner.
Summary of project results
One of the main weaknesses of ensemble prediction systems (EPS) is the high computational load involved. This project addressed this weakness by proposing an innovative statistical methodology to produce perturbations on initial and boundary conditions of meteorological models. The methodology is based on finding similar situations to an atmospheric patter in a historical database. The results obtained by the proposed methodologies has allowed to perform an EPS model with low computational requirements which is able to represent the atmospheric uncertainties. The interest of the project lies not only in studying the predictability of weather surface variables but also, and not least, in generating high-resolution probabilistic predictions of extreme weather event in the Iberian Peninsula from the results of integration of several models in a domain covering the North Atlantic and Mediterranean Area. The project has developed and validated a meteorological EPS. After a first stage of proposal of the reference model some methodologies for perturbing initial and contour conditions have been proposed, implemented and validated. Finally, long databases created in previous stages have been applied to make a statistical analysis along-term predictability. The evaluation of the analogs modelling and atmospheric patter recognition as alternate techniques to produce perturbed initial and boundary conditions has shown encouraging results, all the analysed statistics on this project (bias, RMSE, correlation, rank histograms, RME and BSS) have shown an improvement of results after 18 forecasting hours. Such improved results have a higher effect forecasting extreme weather events. The results obtained for simulations up to 18 hours can be explained by the differences introduced in the initial conditions by analog patterns. Despite the decreasing of skill is small on this first simulation hours, this becomes a leading issue to be studied and corrected in future research projects. The statistical analysis of long-term uncertainties has revealed a climatological structure of predictability, such structure must be studied opening a new climatic variable to be analysed. In the context of predictability analysis, the results obtained in this project have led to the participation of the research group in the European project NEWA (New European Wind Energy Atlas).
Summary of bilateral results
The partnership has established a scientific bound between research groups in Geophisical Institute of Bergen and Complutense University of Madrid. Future European research projects such as NEWA or COPERNICUS (The European Earth Observation Programme) will be the field where incipient international cooperation will be sustained and developed. Beyond the scientific interest of the project, the obtained results found innovative methodologies that lead to a significant reduction of computational resources required by EPS. The international cooperation of the project is considered and essential value as it allows testing the model in different geographic locations such as Spain and Norway. It provides information about the skills of the model under different climatic conditions. The cooperation also provides a workspace where knowledge of both institutions in the field of wind energy can be transferred. The experience in statistical and numerical modelling provided by the Spanish researcher is available for the host institution; on the other hand, the knowledge of the specific requirements of the offshore wind power industry is provided by the host institution. The proposed framework therefore presented different approaches to the same question mixing the meteorological modelling interest of onshore and offshore wind power generation.