Climate variables covered
The climate variables presented here are primarily based on temperature and rainfall and some of their derivative variables. We included several derivative variables that are particularly important in pome and stone fruit production, namely, climatic conditions conducive to frost, chill accumulation, heat accumulation, sunburn, red colour development, and insect pest life cycles.
Users may question why agriculturally significant climatic variables such as hail and strong winds are not covered in this document. The reason is that current Global Climate Models (GCMs) remain weak at modelling these variables in space and time, and the climate database for these variables is poor.
Spatial resolution of mapping: Quinary Catchments
Maps shown here have been prepared at a spatial resolution of so-called Quinary Catchments (QCs, or Quinaries; Schulze and Horan, 2010), which are relatively homogeneous agricultural and hydrological spatial units regarding climate, topography and soils. The Western Cape Province is comprised of 1 401 such Quinaries.
Historical climate of South Africa and the Western Cape
The historical climate of the Western Cape for the 50-year period 1950-1999 provides the reference, or baseline, against which projected impacts of climate change can be evaluated.
Rainfall and rainfall derivatives
In South Africa as a whole, rainfall is considered the most important input into any agricultural and hydrological assessment model. A comprehensive database (1950-1999) of quality-controlled rainfall data in southern Africa was compiled by Lynch (2004). From that database, a rainfall station, termed the ‘driver’ station, was selected for each QC, with that station’s data considered representative of the daily rainfall of that Quinary (Schulze et al., 2010). Each driver station contained a 50-year record of daily rainfall from 1950 to 1999. The selection of driver stations was followed by the determination of multiplicative month-by-month rainfall adjustment factors (from the 1 arc minute raster of median monthly rainfalls created by Lynch, 2004) for each Quinary and these were then applied to the driver station’s daily records in order to render its daily rainfall more representative of that of the Quinary. This resulted in a unique 50-year daily rainfall record for each of the Quinaries covering the region.
Temperature and temperature derivatives
Daily maximum and minimum temperature values facilitate estimations to be made, either directly or indirectly, of solar radiation, vapour pressure deficit and potential evaporation (Schulze, 2008). Procedures outlined by Schulze and Maharaj (2004) enabled the generation of a 50-year quality-controlled historical time series (1950-1999) of daily maximum and minimum temperatures at any unmeasured location in the Western Cape at a spatial resolution of one arc minute of latitude / longitude (~1.7 x 1.7 km). At each of these grid points the maximum and minimum temperatures were computed for each day of the 50-year data period by methods described Schulze and Maharaj (2004), who also achieved excellent verifications of results from this methodology. From the study of Schulze and Maharaj (2004) representative grid points were determined for each of the Quinaries covering the Western Cape, using techniques outlined in Schulze et al. (2010). The resulting 50-year series of daily maximum and minimum temperatures for each Quinary was then used to generate daily estimates of solar radiation and vapour pressure deficit and from these, daily values of reference potential evaporation were computed.
General Circulation Models (GCMs) and scenarios used in this study
Future climate projections (which are NOT forecasts nor predictions) are scenario descriptions of possible future conditions based on the current understanding of the physics of the atmosphere, on assumptions about changing greenhouse gas emissions and their atmospheric concentrations, as well as on assumptions of future technological, economic and demographic trends. The skill of projections (i.e. their accuracy) depends strongly on how far into the future projections are made, which of a number of possible future greenhouse gas emissions pathways is considered (the thicker lines in Fig. 9), and on the climate variable considered (e.g. temperature projections are generally more skilful than rainfall projections). Deriving key regional messages about future potential change thus requires assessing multiple lines of evidence. Climate projections are therefore assessed here from a range of climate models generically termed GCMs, i.e. General Circulation Models, as it is not possible to identify a ‘best’ model for all relevant climate variables for South Africa. This range of outcomes from different GCMs for a specific future pathway is shown by the different thin coloured lines in Fig. 9 for each of the thicker coloured lines of an emissions pathway.
Projections of impacts on the agricultural sector in South Africa (and other sectors as well) are often complicated by different scientists applying different sets of climate scenarios and using different modelling approaches. The various climate projections used in the impact studies presented here have been based on certain case studies of the Intergovernmental Panel on Climate Change’s (IPCC) Special Report on Emission Scenarios (SRES) so-called A2 emission scenario. This is essentially a “business as usual” scenario representing CO2 equivalent levels of above 500 ppm by 2050. Other case studies have used outputs from GCMs driven by the various so-called RCPs, or Representative Concentration Pathways.
The two sets of GCMs that were used are listed below, and on the relevant maps the set of multiple GCMs which were used in a specific analysis are referred to. All the GCMs used were accredited by the South African Long-Term Adaptation Scenarios initiative of the South African National Department of Environmental Affairs.
The first set of GCMs used were CMIP3 GCMs which were downscaled / distributed by the Climate Systems Analysis Group (CSAG) of the University of Cape Town and derived from global scenarios produced by five IPCC AR4 approved GCMs. All these GCMs were statistically downscaled to over 2 000 climate stations in South Africa and then further bias corrected for the 1 401 Quinary Catchments covering the Western Cape Province by techniques described in Schulze et al. (2010).
In all cases the future scenario A2 was used together with the following GCMs:
- CGCM3.1(T47)
- CNRM-CM3
- ECHAM/MPI-OM
- GISS-ER and
- IPSL-CM4
In each case, daily values of rainfall, maximum and minimum temperatures, and computed daily values of solar radiation, maximum and minimum relative humidity and reference potential evaporation (using methods given in Schulze, 2008) were generated for three 20-year time periods:
- the present (1971-1990),
- the intermediate future (2046-2065), and
- the more distant future (2081-2100).
The first two time periods were used here.
The second set of climate scenarios used were from the World Climate Research Programme sponsored Coordinated Regional Climate Downscaling Experiment CORDEX, in each case with daily rainfall and maximum / minimum temperature (again with derived daily values of solar radiation, relative humidity and potential evaporation). For these climate scenarios two 30-year periods were used:
- 1976-2005 (termed the historical, or present, climate), and
- 2016-2045 (termed the immediate future).
The modelling was conducted for the immediate future scenario, together with the RCP8.5 scenario. Again, the GCMs were downscaled to the 1 401 Quinary Catchments and then bias corrected for local topography by methods described in Schulze et al. (2014).
The GCMs used were:
- CCCma-CanESM2_historical_RCA5_1976
- CCCma-CanESM2_rcp85_RCA5_2016
- CNRM-CERFACS-CNRM-CM5_historical_RCA5_1976
- CNRM-CERFACS-CNRM-CM5_rcp85_RCA5_2016
- ICHEC-EC-EARTH_historical_RCA5_1976
- ICHEC-EC-EARTH_rcp85_RCA5_2016
- NCC-NorESM1-M_historical_RCA5_1976
- NCC-NorESM1-M_rcp85_RCA5_2016
- NOAA-GFDL-GFDL-ESM2M_historical_RCA5_1976
- NOAA-GFDL-GFDL-ESM2M_rcp85_RCA5_2016
These climate models are referred to on the maps as the CMIP5 GCMs. It is important to remember that all maps for the projected future, and change from present to future, are means of four or five different GCMs, except for red colouring where only one GCM was used. The variance between the model outcomes is not shown; rather, the mean or median of the different outcomes is shown.
A new set of GCMs (CMIP6) is now becoming available. Over the last two decades of model development, there has been very little change in the projections for temperature and temperature-derived variables. We thus have high confidence that the GCMs used in this study provide a reliable picture of the future that will not be altered substantially as the new GCMs begin to be used. However, projections for rainfall and rainfall-derived variables remain challenging. As the model developers address the weaknesses and inconsistencies in the older GCMs, the new set of GCMs can alter the picture somewhat. We thus have lower confidence in the rainfall and rainfall-derived variables, and have decided to present only the maps for historical climate. However, we have included future climate results for dry and wet spells since the science shows better agreement around increasing variability of rainfall. We recommend that the analyses are updated once the new GCMs have become available.