NANO alumni publication: Evaluation of ECMWF and NCEP Reanalysis Wind Fields for Long-Term Historical Analysis and Ocean Wave Modelling in West Africa

NANO Alumni Bennet Atsu K. Foli and colleagues published the following article in the Remote Sensing in Earth Systems Sciences Journal 

Evaluation of ECMWF and NCEP Reanalysis Wind Fields
for Long-Term Historical Analysis and Ocean Wave Modelling in West
Africa

Kwame, B. et al. (2021) Remote Sensing in Earth Systems Sciences Journal. DOI 10.1007/s41976-021-00052-3

Abstract

Ocean wind fields forma significant input to ocean wave models. This study evaluates the accuracy of two major reanalysis wind fields: NCEP-NCAR reanalysis-II (NNR-II) and ECMWF ERA5 wind datasets in the marine domain of West Africa. The objective is to identify the reanalysis data that best represents the wind regimes of the sub-region for use in climate studies and ocean wave modelling. The reanalysis datasets were validated against in situ measurements from PIRATA meteorological buoys in the region. Both reanalysis datasets indicate good agreement with in situ measurements and capture the variability in the wind fields. However, ERA5 wind fields outperform the NNR-II wind fields and better represents the variability in wind fields in the region. They display higher correlation coefficients and R-squared values as well as lower bias and RMSE values for all wind components at all PIRATA buoy locations. Correlation coefficients of 0.92, 0.87, 0.94, and 0.98; R-squared values of 0.83, 0.76, 0.89, and 0.96;mean bias of −0.34±0.75ms−1, 0.25±33.75°, 0.07 ± 0.86 ms−1, and −0.21±0.96ms−1; and RMSE values of 0.82 ms−1, 33.75°, 0.87 ms−1, and 0.98 ms−1 were observed for ERA5 resolved wind speeds, wind directions, and zonal and meridional winds respectively. NNR-II also recorded correlation coefficients of 0.64, 0.7, 0.73, and 0.9; R-squared values of 0.19, 0.39, 0.32, and 0.79; mean bias of 0.12±1.77 ms−1, 8.91±53.43°, 0.55±2.09 ms−1, and −0.31±2.15 ms−1; and RMSE values of 1.77 ms−1, 54.17°, 2.16 ms−1, and 2.17 ms−1 for resolved wind speeds, wind directions, and zonal and meridional winds, respectively. NNR-II winds tend to highly overestimate zonal wind speeds and underestimate meridional wind speeds. Meridional winds are better predicted compared to zonal winds for both NNR-II and ERA5 winds. There was a general overestimation of lower wind speeds and underestimation of higher wind speeds on the part of both reanalysis datasets although this assertion varied with geographical location. To enhance the accuracy of resolved wind velocities and directions in the region, there is the need to improve the estimation of zonal winds in general by both NNR-II and ERA5 winds but with much efforts needed for NNR-II. In effect, ERA5 reanalysis winds better describe the wind regime of West Africa for climate studies and ocean wave modelling.

Keywords:

  • ECMWF reanalysis
  • NCEP reanalysis
  • Ocean wave modelling
  • PIRATA
  • West Africa .
  • Wind fields

Link for the publication here

Would you like to advertise your latest publication/seminar presentation in NANO? Contact us: info_at_nf-pogo-alumni.org