An exploration of South Africa's wind climate using station records and reanalysis data

Doctoral Thesis

2016

Permanent link to this Item
Authors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher

University of Cape Town

License
Series
Abstract
Sparse information about the wind climate of South Africa behooves an exploration of the drivers of surface wind speed, especially in the context of wind resource assessment. This work quantifies the coupling between the synoptic circulation states and station-scale flows to develop a process-based regionalisation of wind regimes over the country .A thorough inspection of available South African Weather Service (SAWS) wind records is conducted and a quality control procedure is applied. The procedure reveals a large proportion of the data are missing and existing data contain numerous errors such that only107 of the original 960 stations passed the quality control criteria. However, data from these107 stations only overlap temporally 2% of the time, which makes the data inappropriate fora regionalisation procedure. To ameliorate this, a method for incorporating bias-corrected time series data from a reanalysis data set is developed. Data from the 0.3◦ resolution hourly Climate Forecast System Reanalysis (CFSR) be-tween 1989-2010 is selected to improve the temporal coverage of the station data. The raw CFSR data overestimates wind speeds and underestimates the temporal variability and long-term trends. A bias correction method based on the wind speed and direction, time of day and month of the year is developed which successfully removes the mean error on wind speed and direction and improves the correlation with station records. This is achieved without disrupting spatial correlation patterns. Corrected and extended wind time series from each station site are used for the regionalisation. The regionalisation uses a self-organising map (SOM) to define the archetypal synoptic circulation patterns in the reanalysis data set and the influence of these on the local wind climate is quantified. 12 representative atmospheric states are defined by the SOM that are consistent with the existing literature and capture the major synoptic circulation states. A hierarchical clustering is then used to define wind climate regions based on the coupling between these circulation states and the extended station data. Six relatively cohesive spatial wind-climate groupings are identified that are physically consistent with the driving synoptic environment and are characteristic in terms of terrain and response to synoptic drivers. This process-based regionalisation facilitates a future assessment of potential changes in the wind climate of South Africa as a result of a warming world.
Description

Reference:

Collections