Assessment of Wind Power Potential for the Caribbean region of Colombia: Field Records, Reanalysis and Atmospheric Modeling

Abstract

ABSTRACT: This master’s report presents an assessment of wind power potential for the Caribbean coast of Colombia. The report analyses a large set of field measurements with high temporal resolution, along with data from the reanalysis product ERA5 from the European Center for Medium-Range Predictions (ECMWF) and with high-resolution wind speed fields simulated with the Weather Research and Forecast model (WRF), validated with field measurements. The manuscript consists of two chapters. The first chapter characterizes the wind resource across a wide area that includes the Gulf of Urabá and the San Andrés Islands, with a special focus on La Guajira. The temporal aggregation in the averaging time of some variables shows how finer time intervals (10-minute to hourly) generate more accurate monthly and annual statistics than daily and monthly data, with 10-min average wind speeds in the region in the range of 2.2 m s-1 to 8.3 m s-1. ERA5 performance at reproducing wind speed, direction, and variability compared to in situ measurements can be a valuable information source after identification/treatment of biases for wind characterization. Based on Weibull distributions and hourly Wind Power Density (WPD) calculations, the study finds several locations that offer commercial wind potential, with many having WPD above 800 W m- 2. Furthermore, estimates based on ERA5 show a wide offshore region in the Caribbean with a WPD over 800 W m-2 yearlong. Annual Energy Production (AEP) for the stations with promising potential reach capacity factors around 50%, which are comparable to those of the most productive wind farms worldwide. The second chapter shows the use of the WRF model to estimate the monthly wind regime, the intra diurnal wind variability, and the monthly WPD at resolutions of 9 and 3 km, which are substantially finer than the nearly 30 km equivalent grid size of ERA5. We compared the performance of three planetary boundary layer schemes in WRF (QNSE, YSU and MYJ) at reproducing the monthly average diurnal cycle of wind speeds during the months of higher (June) and lower (October) wind speeds, in four stations with high wind potential, based on metrics from Taylor diagrams. In addition, we compared WRF results with ERA5 estimates. The QNSE scheme outperforms the others in most sites and cases, with averaged fractional bias (FB) of 9.88% for QNSE, 10.83% for YSU, and 9.50% for MYJ, whereas ERA5 reached values up to 14.6%. Using the QNSE scheme, we calculated the WPD for the entire year and compared it with the ERA5 and the observed WPD by means of duration curves, for two of the analyzed points (northernmost locations). The WPD with WRF shows an almost perfect fit to the observed curves, while ERA5 only outperforms WRF in one location, and both ERA5 and WRF underrepresent in one location. Although the fields of WPD estimated with WRF are more detailed than those generated by ERA5, ERA5 information could still be valuable for offshore estimations at larger scales, e.g., the synoptic scale. The estimated WPD values for the month with higher wind speeds are up to 1200 W m-2 in the shoreline and up to 2000 W m-2 offshore, varying between 800 W m-2 and 1600W m-2 inland. WPD values during the month of lower wind speeds vary between 400 W m-2 and 800 W m-2. Due to the high spatial resolution of the WRF model, the obtained results will serve to identify new areas with commercial wind potential in the Caribbean region of Colombia and the north coast of South America.