South African drought: trend or incident?
Sanne Elfrink
6 July 2018, IJmuiden
Supervisor: Dhr. Dr. Ir. J.H. van Boxel
Abstract
Theewaterskloof Dam near Cape Town on January 17, 2018 (Planet Labs Inc, 2018) Theewaterskloof Dam near Cape Town on February 17, 2015 (Planet Labs Inc, 2018)
Cape Town and other regions in South Africa are suffering from drought. Two explanations are climate change due to increased greenhouse gases or consequences of the occurrence of an El Niño on the Pacific Ocean. The drought is causing economic losses for the South African population and
agriculture. The aim of this research is to look whether the drought in South Africa is induced by climate change, an El Niño event or other processes. A trendline analysis is used to determine climate change and cross correlation is used to check the correlation between precipitation and Oceanic Nino Index (ONI). The trendlines show declines in precipitation rate over most of the regions in South Africa. However, not enough trendlines are significant due to a lack of data. The cross correlation between precipitation and the ONI is in all cases negatively correlated, which indicates that there is a connection between declining precipitation and the resulting weather circumstances during an El Niño. The recent droughts in South Africa are due to the effects of an El Niño, but they are probably influenced and enforced by climate change. The drought in Cape Town as well as the western and southern region of South Africa are more influenced by the effects of El Niño’s than the eastern regions of South Africa.
Over the last century, the amount of greenhouse gases has increased due to human activities. This increased greenhouses gases result in a changing climate with increased temperature as well as changes in precipitation patterns. Additional consequences of climate change are rainfall variability, clustering of wet and dry days and extreme events like droughts (Dennis & Dennis, 2012). South Africa is already experiencing these consequences, but changes in annual precipitation patterns are region dependent in South Africa (Van Wilgen et al., 2016). Since 2015 Cape Town is suffering from a drought and the city was expected to run out of water in 2017 (Kruger & Nxumalo, 2017; Wolski, 2018). According to Wolski (2018), the drought in Cape Town in 2017 was the driest period since 1933. The situation in Cape Town was described as a chronical water shortage and therefore water
restrictions are imposed on the consumers. Besides dry weather conditions, poor management and planning are also enforcing the drought. The water reservoirs are expected to be empty on Day Zero (Wolski, 2018).
Not only the region of Cape Town is suffering from drought, but also other regions are suffering from drought (Van Wilgen et al., 2016). Moreover, Johannesburg is reducing the consumption of water to avoid the city runs out of water (Baudoin et al., 2017). Due to the large surface of South Africa, the country has multiple climates according to the Köppen-Geiger climate classification. For example, the climate of Cape Town is a Mediterranean climate, whereas the climate of Johannesburg is a subtropical highland climate. The different climates mainly exist of B and C climates and a very small part of an Aw climate (Conradie, 2012). The different climates in Africa can be distinguished with help of Figure 1.
Figure 1 shows that most of the arid climates are located in the western, middle and northern region. The warm temperate climates are more located in the eastern and southern region of South Africa.
South Africa is a water-limited country and prone to droughts. Since an increase of 1.5°C is expected along the coast and an increase of 2.0-3.0°C in the inlands in 2050, it is likely that the overall conditions in South Africa become more dry in the future (Dennis & Dennis, 2012). The South African Weather Service (2018) defines a drought as a period with less than 75% of the normal rainfall. The normal rainfall is calculated over the last 30 years of rainfall. Besides the amount of precipitation, other factors such as high temperatures and high wind speeds have to be taken into account.
Drought events are problematic for human health as well as agriculture (Sorensen, 2017). Water restrictions in major cities have consequences on the availability of drinking water and other human usages of water. In addition, sufficient precipitation in arid and semi-arid climates is necessary for agricultural crops to grow. Agriculture is in South Africa one of the most important economic factors. The production of food is not only important for export, but even more important for the self-sufficiency in food, since 40-50% of the population is classified as poor (Machethe, 2004). Grapes grow only in unique climates and the Mediterranean climate in the Western Cape is such a climate. Grapes are affected by droughts and if the drought continues, the consequences are global yield losses and declining economic circumstances in South Africa (Araujo, et al., 2016). Establishing the cause of the drought is important to define a management plan with drought risk.
According to van Boxel (2008) it is notable that the arid and semi-arid areas in the Northern Hemisphere suffered earlier from drought and that the Southern Hemisphere received more precipitation. A possible explanation is the large oceanic surface surrounding South Africa, which means that this circumstances have to be taken in to account. It is possible that the climate change in the Northern Hemisphere is heading in the same direction as the Southern Hemisphere. Since South Africa’s average annual rainfall is only 500 millimetre per year, declining precipitation trends can easily cause dry periods (Dennis & Dennis, 2012). According to future predictions of Engelbrecht & Engelbrecht (2016) are the temperate climate zones decreasing and replaced by hot steppe zones and tropical savannah zones, as implications of increased temperatures and low precipitation.
Climate change and the resulting differences in temperature and precipitation influence the drought in South Africa, but weather conditions are also affected by El Niño. An El Niño situation occurs when the water temperatures of the Pacific are higher than normal. La Niña is the opposite situation with abnormally low water temperatures in the Pacific Ocean (Trenberth, 1997). These differences in sea surface temperature result in changing weather circumstances. During an El Niño event, the precipitation in South Africa is low and climate conditions are dry. On the contrary, during a La Niña is the precipitation in South Africa is high, which results in wet conditions (Bellprat et al., 2015). Most El Niño events take place every 3-6 years and the duration of an El Niño can be 1-2 years (Trenberth, 1997). The drought could also be explained by the event of an El Niño instead of climate
change.
A trend is explained by climate change and an drought incident can be explained by an event like El Niño or by other factors. If the precipitation data show a gradual decline in precipitation, it is more likely that climate change is the cause of the drought. When there is randomness in the dataset for the last couple of years, it is more likely that an event like El Niño is occurring or neither of them.
The aim of this research is to look whether the drought in South Africa is induced by climate change, an El Niño event or other processes. The main question of this research is: “Is the recent drought in South Africa a trend or an incident?”. By means of the following 3 sub questions, the main question will be answered:
How is the recent precipitation pattern related to the past precipitation pattern of South Africa and is there a trend to establish?
What is the influence of an El Niño event on the South African drought?
Is there a relation between climate change and El Niño concerning drought in South Africa?
Methods & Data
To establish if the recent drought in South Africa is a trend or an incident, the precipitation data of South Africa is used. The data is retrieved from the Climate Engine Web Application (Climate Engine Web Application, May 2018). The dataset Modern-Era Retrospective analysis for Research and Applications, Version 2 (MERRA2) is used with a resolution of 0.5 degrees x 0.625 degrees. The time span of the dataset is from January 1, 1980 till March 31, 2018. The Climate Engine Web Application is able to make points or polygons of specific areas. For this research are 18 points established, spread over South Africa’s different climates as well as spread over the country. The datasets are downloaded as daily data because monthly data is convertible to monthly and yearly data and not the other way around. Yearly data will not indicate seasonal variability in the dataset and monthly data can provide a high precipitation value while half of the month was dry.
To conduct a statistical analysis on the precipitation data, Microsoft Excel is used. Graphs are fitted with the yearly data on the x-axis and the precipitation in millimetres on the y-axis. A trendline is added to the graph to establish if there is a significant in the annual precipitation from 1980 till 2017. The coefficients of the regression line can be visualized in Microsoft Excel and they will be explained as well as the significance of the R-squared value with a F-test with a 95% confidence level. The R-squared will determine the variance explained by X and results in the reliability of the
regression line. This statistical analysis is performed for all 18 weather stations. Also the normal values for every month are visualised in a graph. The Excel sheet also contains a column with the monthly precipitation for each year as well as an anomaly for the difference between a specific month of a year and the normal precipitation of that month.
Since there is a difference in precipitation between B-climates and C-climates according to the Köppen-Geiger climate classification, an unpaired two sampled t-test is conducted between a group of 12 B-climates and 6 C-climates. The t-test is conducted on the annual averages of
precipitation in the first t-test and on the trend coefficients in the second t-test. It is likely that the first t-test is significant, since B-climates are classified as more dry climates than C-climates. If the outcome of the second t-test is significant, it means that there is a significant difference between the trends of B-climates and C-climates on drought.
For establishing the relationship between the drought in South Africa and the influence of El Niño, first a dataset of sea surface temperature is required. The National Ocean and Atmospheric Administration (NOAA) monitored El Niño and La Niña as part of the El Niño-Southern Oscillation (ENSO). The strength of an El Niño or La Niña is measured with the Oceanic Nino Index (ONI) of Niño 3.4, that is explained by 3 months of sea surface temperature in the east-central tropical Pacific. The values of the ONI are degrees of Celsius above the 30 years normal of sea surface temperatures in region 3.4 for that month. Values of 0.5°C or higher are related to an El Niño and values with an index of -0.5°C or lower are related to La Niña (Dahlman, 2009).
A monthly dataset of the ONI from 1980 till 2018 is downloaded from the NOAA (NOAA, 2018) and copied to Microsoft Excel. First, a graph of only the ONI data is made to visualize the occurrence of El Niño’s and La Niña’s. To conduct a statistical analysis on the precipitation data and ONI data, a cross correlation spectrum is calculated with time lags between -24 and 24 months. Two types of spectra are produced, a 1 month average and a 12 months average of the correlation between the precipitation anomalies and the ONI. For the 1 month average is the 3 months ONI anomaly of region 3.4 used and for the 12 months average is the average over 12 months of the same anomaly used. The outcomes of the spectra will be explained as well as the correlation between El Niño and precipitation. A correlation with value 1 is a maximum correlation and a correlation with a value of -1 is minimum correlation. Since El Niño circumstances result in dry conditions in South Africa, a minimum correlation is expected between drought in South Africa and El Niño if El Niño is causing the drought.
The program ArcGIS is used to produce maps of South Africa. The first map shows the different provinces in South Africa. This map is downloaded from ArcGIS Online, visible in Appendix 2. Another map layer shows the 18 chosen weather stations of the precipitation data. With the
Georeferencing tool, another layer is added to the map with climate zone data. As a result, the 18 weather stations are visualized in the corresponding climate zone. These maps are made as a support for the visualizing and understanding of the 18 different chosen weather stations and therefore not containing precipitation or ONI data.
The statistical analysis of the data of precipitation and the ONI data will give results concerning a trend or incident. If the trendlines are significant, it is likely that the South African drought is due to climate change. If the outcome of the cross correlation is a minimum correlation, it is likely that El Niño events are the cause of the drought in South Africa. These results will be
described and afterwards a literature study will be conducted. The literature can confirm the
outcome of the statistical analysis as well as contradict the results. This part will be conducted in the discussion of this research.
Results
Before running the statistical analyses, a map was made with ArcGIS to visualize which weather stations are where situated over the country and in what type of climate they are categorised. The map is visible in Figure 2. Figure 2 shows that from the 18 weather stations, 6 are located in zones with a C-climate and 12 are located in zones with a B-climate.
In Figure 3 is a map produced of the provinces in South Africa with the 18 weather stations
Figure 2: Climate map of South Africa including the 18 weather stations
Figure 3: Map of South Africa including the 18 weather stations and 9 provinces
included. The results of the statistical analyses will be shown in a table with the corresponding weather stations and provinces. The province of Northern Cape is the largest province with 5 weather stations, the smaller provinces do only have 1 of 2 weather stations.
As an example is the precipitation of Cape Town visualized in Figure 4. As visible in Figure 4 the trendline of Cape Town is positive. However, the r-squared value is very low and the t-test shows that the r-squared is not significant. In addition, the recent drought in Cape Town that started in 2015 is visualized in the graph as well. Visible is the peak in 2013 of 1000 mm annual precipitation and in 2015 less than 600 mm precipitation. Remarkable is the increase in 2016 and immediately the decrease in 2017. In Figure 5 is the normal monthly precipitations plotted in the graph and shows that Cape Town is summer dry and has precipitation in the winter. The graphs of the other 17 weather stations are visible in Appendix 1.
For all 18 weather stations are graphs made and trendlines fitted, these are summarized in Table 1. The table shows if the region has a B-climate or a C-climate, if there is a positive trend or negative trend and also if this trend is significant or not. Only 5 weather stations have a positive trend, and just one is significant at the 95% confidence level. Of the C-climates 3 have a positive trend and 3 have a negative trend. Out of the 12 stations with B-climates 10 have a negative trend and just 2 have a positive trend. From the 13 decreasing trends are 5 significant, shown in blue in the last columns of Table 1. Another remarkable fact is that the 6 weather stations with a significant trend are located in the south eastern part of South Africa, where 2 regions are located near the ocean. The different climates to distinguish for this 6 regions are: two times BSh, one time BSk, two times Cwb and one time Cfa. The region with Cfa climate is the only significant increasing trend.
It is likely that there is relation between C-climates and increasing trends and B-climates and decreasing trends. After running an unpaired t-test with unequal variances, only the average
precipitation of all years between C-climates and B-climates is significant. However, this is a logical outcome since the climate is classified on precipitation as well as temperature. The B-climates obtain less rainfall than the C-climates, since the B-climates have arid circumstances. The same t-test on the trend coefficients between B-climates and C-climates was not significant.
Provinces Climate B or C Trend mm/year Significant P-value Percentage Significance
Coffee Bay Eastern Cape Cfa 5,14 0,019864 2,0%
Durban KwaZulu-Natal Cfa -3,52 0,38286168 38,3%
Port Elizabeth Eastern Cape Cf 0,30 0,842996 84,3%
Cape Town Western Cape Csb 1,74 0,456989138 45,7%
Johannesburg Gauteng Cwb -5,97 0,007549 0,8%
New Castle KwaZulu-Natal Cwb -8,04 0,001569 0,2%
Komatipoort Mpumalanga BSh -5,14 0,038561551 3,9%
Messina Limpopo BSh 0,84 0,714926888 71,5%
Figure 4: Annual precipitation in mm including trendline and coefficients, Cape Town
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 20 40 60 80 100 120 140 1980 1985 1990 1995 2000 2005 2010 2015 2020 0 200 400 600 800 1000 1200 f(x) = 1.74 x − 2797.84 R² = 0.02
Steenbokpan Limpopo BSh -1,45 0,402243 40,2%
Vryburg North West BSh -3,21 0,096839 9,7%
Bloemfontein Free State BSk -2,96 0,175016 17,5%
Cradock Eastern Cape BSk -5,76 0,015911 1,6%
Matlosana North West BSk -4,46 0,018747 1,9%
Mosselbaai Western Cape BSk -2,35 0,056264 5,6%
Kalahari CBDC Northern Cape BWh -0,36 0,807017 80,7%
Upington Northern Cape BWh 1,60 0,085955 8,6%
Carnarvon Northern Cape BWk -0,68 0,546115 54,6%
Port Nolloth Northern Cape BWk -0,24 0,702366 70,2%
Table 1: Results of the trendline analysis for 18 weather stations
After establishing whether there is a trend of not, the data of the ONI index is visualized in a graph. Figure 6 shows that in 1982, 1997 and 2015 heavy El Niño’s occurred, where 2015 is the largest El Niño since 1980. Other El Niño’s in the period 1980-2017 had lower sea surface temperature anomalies in the Niño3.4. Most El Niño’s and La Niña’s last for a couple of years, especially after big El Niño’s. Remarkable is that shortly after the 2015 El Niño again a small increase in Niño 3.4 was measured. After the 1982 and 1997 El Niño’s, increasing conditions did not occur for 3 till 4 years. 1/1/ 1980 4/1/ 1981 7/1/ 1982 10/1 /198 3 1/1/ 1985 4/1/ 1986 7/1/ 1987 10/1 /198 8 1/1/ 1990 4/1/ 1991 7/1/ 1992 10/1 /199 3 1/1/ 1995 4/1/ 1996 7/1/ 1997 10/1 /199 8 1/1/ 2000 4/1/ 2001 7/1/ 2002 10/1 /200 3 1/1/ 2005 4/1/ 2006 7/1/ 2007 10/1 /200 8 1/1/ 2010 4/1/ 2011 7/1/ 2012 10/1 /201 3 1/1/ 2015 4/1/ 2016 7/1/ 2017 -3 -2 -1 0 1 2 3
ONI 3.4
Figure 6: Oceanic Nino Index region 3.4. Values above 0.5 °C are related with El Niño events and values below -0.5°C are related with La Niña events
The ONI and precipitation data are conducted to a cross correlation and results in spectra with 49 lags, each indicating one month. In case the precipitation is related with an El Niño event, the correlation should be minimal. The ONI exceeds the value of 0.5°C during an El Niño event, where low precipitation is expected. Therefore a negative correlation is anticipated if the two variables are correlated.
Two types of spectra are used, the 1 month average and the 12 months average, because the seasonal variations have to be filtered out. Cape Town is used again to demonstrate the outcome of the statistical analysis. Figure 7 visualizes the 1 month average; here is a correlation of -0,100 with a lag of -8. This means that there is a very small correlation after a lag of 8 months. Figure 8 shows the
12 months average of the cross correlation between the precipitation of Cape Town and the ONI index. There is a correlation of -0,255 with a lag of -12. This means that there is a slightly negative correlation after 12 months. There is less correlation in the one month average than in the 12 months average of Cape Town. Appendix 1 includes the cross correlation spectra of the other weather stations.
The results of the other 17 weather stations are summarized in Table 2. All the correlations are negative values between -0,100 and -0,250 for the one month average and between -0,110 and -0,600 for the 12 months average. The highest correlation is often a lag around 0, but sometimes it is up to even -24 or 24. However, in all regions there is a negative correlation between the ONI and precipitation. Furthermore, the correlation in each case is smaller with a 1 month average instead of a 12 months average. For the lags is this in most of the results the case, but not for all.
Province Climate Cross correlation 1 month average in R Lag in months for 1 month Cross correlation 12 months average in R Lag in months for 12 months
Coffee Bay Eastern Cape Cfa -0,145 -4 -0,410 -6
Durban KwaZulu-Natal Cfa -0,080 -1 -0,200 -2
Port Elizabeth Eastern Cape Cf -0,100 -11 -0,160 -11
Cape Town Western Cape Csb -0,100 -8 -0,255 -12
Johannesburg Gauteng Cwb -0,080 22 -0,225 24
New Castle KwaZulu-Natal Cwb -0,105 -2 -0,150 -2
Komatipoort Mpumalanga BSh -0,200 -2 -0,375 0
Messina Limpopo BSh -0,250 0 -0,600 -2
Steenbokpan Limpopo BSh -0,175 0 -0,300 -1
Vryburg North West BSh -0,075
-0,055
19 1
-0,200 24
Bloemfontein Free State BSk -0,070 -2 -0,110 -1, 24
Cradock Eastern Cape BSk -0,110 -2 -0,250 -4
Matlosana North West BSk -0,090 -3 -0,125
-0,140
-3
24
Mosselbaai Western Cape BSk -0,085 -0,055 21 -9 -0,140 -10 Kalahari CBDC Northern Cape BWh -0,085 -0,075 -0,070 24 -22 0 -0,225 24
Upington Northern Cape BWh -0,240 -2 -0,480 -1
Carnarvon Northern Cape BWk -0,150 -2 -0,320 -3
Port Nolloth Northern Cape BWk -0,070 24 -0,220 24
Table 1: Results of the cross correlation over 18 weather stations
Discussion
From the 18 weather stations spread over South Africa, 13 weather stations show a negative trend in precipitation. However, only for 5 out of this 13 stations the trends are significant. Since just 33,3% of
Figure 8: Cross correlation 12 months average Cape Town Figure 7: Cross correlation 1 month average Cape Town
-24 -18 -12 -6 0 6 12 18 24 -0.300 -0.200 -0.100 0.000 0.100 0.200 0.300 -24 -18 -12 -6 0 6 12 18 24 -0.150 -0.100 -0.050 0.000 0.050 0.100 0.150
the weather stations show a significant trend, it is hard to establish an overall trend. There is
evidence for a negative trend, since most of the stations do have a negative trend. However, there are too less significant trends to establish that climate change is the cause of the drought. It is not excluded that climate change is an influencing factor, but the results of the data do not show this.
According to Ziervogel et al. (2014) the effects of climate changes are already visible. The temperatures in South Africa are increased by 1.5 times the global average increase in temperature. Other effects of climate change are also measured in South Africa, such as an rise in extreme rainfall events, but also an increase in extreme drought events (Hoffman et al., 2009). These drought events are particularly forecasted for the winter rainfall regions, which also can explain the recent drought in Cape Town. The predicted circumstances for the South African precipitation do vary over the region.
Since the precipitation dataset has a length of only 38 years, it is very likely that the statistical analyses are influenced by this length. Since only a small part of the trendlines were significant, it is possible that this is caused by a lack of long records. The South African Weather Service was in the possession of a dataset of approximately 100 years. A Disclosure Statement was filled in and sent to them. However, they were not able to provide the requested data in time. The amount of weather stations with a significant trend can increase by a larger dataset and the certainty of the trend will increase if there is a clear trend. Only 38 years is quite small for statistical analyses on precipitation data. The length of the dataset can change the results.
The results of the cross correlation are given in R values and all values in the table are negative values. Since it is very likely that an increase in sea surface temperatures in ONI 3.4 not directly affect the precipitation in South Africa, a lag of 2 or 3 months is considered as normal. In some cases the seasonal variability has also been considered in relation with the start of El Niño’s in December. Therefore a lag of 6 months or 12 months is also likely. Negative values are expected when the drought is influenced or caused by an El Niño. However, a negative value of -0.600 shows more correlation between less precipitation and an El Niño than a correlation of -0.100. For every region there is a different correlation, as visible in Table 2 Messina and Upington do have a strong
correlation between drought and El Niño and Bloemfontein has a slightly negative correlation. El Niño is affecting drought in South Africa, but not that strong for every single region.
Multiple stations including Cape Town have also a positive correlation between low precipitation and increased ONI 3.4. An explanation could be the seasonal variability in the precipitation data for different regions. Another explanation could be the strength of the El Niño’s and La Niña’s and the fact that the ONI area between 0.5°C and -0.5°C is neither an El Niño nor a La Niña. The correlation is calculated over 38 years and shown in spectra with 49 lags. Years without El Niño’s and La Niña’s are also included in the correlation.
The correlation differences are also divided over the country, therefore no conclusions can be drawn about particular correlated parts of South Africa. The only conclusion about B climates and C climates is that the range of the C climate is quite smaller than the range of B climates. However, there are 12 B climates to distinguish and only 6 C climates. Therefore it is not possible to draw conclusion between the different climates and the degree of correlation.
The results presented in this thesis are in agreement with Rouault et al. (2010). They concluded that significant correlation exists between the rainfall anomalies of South Africa and the sea surface temperatures of the Pacific. Raoualet al. (2010) made a map with these correlations that visualize a strong correlation between El Niño and decreased precipitation on the western and southern coast of South Africa. However, the coastal part in the east situated below Lesotho shows a positive correlation. This can be an explanation for the only significant increase in trendline of Coffee Bay. On the map of Raoualet et al. (2010), Coffee Bay is exactly situated in the positively correlated area, which can explain the increase in precipitation. However, the results of the correlation of Coffee Bay show a negative correlation with a lag of -6 months. This correlation is declining towards 0 with a
lag of -24 months.
Although, the lags that fit the correlation are sometimes unexplainable. A lag of -2 indicates that 2 months after the occurrence of an El Niño, the effects are visible in a declining precipitation. Most El Niño’s start around Christmas, in the South African summer (Trenberth, 1997). Cape Town has for example a dry summer and a wet winter. The effects during an El Niño in a dry season are not different from a normal dry season. That can be an explanation for a larger lag of 6 to 12 months. Within this dataset, there are weather stations where a lag of -24 or 24 is indicated for the strongest correlation. This means that the effects of an El Niño on precipitation are the strongest after 2 years. That does not mean that there is no correlation between El Niño and declining precipitation in an earlier state, only this correlation is less correlated than on the higher lag. The lag issue can be implemented in another research the effects of an El Niño on drought in South Africa.
As visualised in Figure 5 where the ONI is plotted, it is visible that there were large El Niño events during the years of 1982-1983, 1997-1998 and 2015-2016. This types of larger than normal El Niño are called Super El Niño’s and occur approximately only once in two decades (Rao & Ren, 2017). The effects of an super El Niño are much stronger than the effects of an regular El Niño. Weather circumstances are during al El Niño already changed and enforced, but this reinforcement is increased during a super El Niño. The super El Niño’s of Figure 5 are in accordance with Rao & Ren (2017). The recent drought in South Africa can be explained by a super El Niño. Except for super El Niño’s there are types of El Niño’s to distinguish and have to be taken in to account, such as El Niño Modoki. El Niño Modoki consists of a strong warming in the central Pacific, surrounded by colder waters in the east and west (Ashok & Yamagata, 2009).
In most cases the annual precipitations graphs of the 18 weather stations show similar patterns as the ONI. When the ONI is increasing in 2015 for the super El Niño, the precipitation is strongly declining. 2016 is a weak La Niña and in most of the 18 graphs, the precipitation is increasing. There is again a small increase in 2017 in values of ONI region 3.4, afterwards the precipitation is directly declining.
The effects of an El Niño result in an increased level of greenhouse gases, in particular CO2 (Di
Lorenzo et al., 2010; Betts et al., 2016; Chatterjee et al., 2017). Tropical regions are warming and drying during an El Niño, as a result carbon uptake by vegetation is reduced and carbon emissions due to fire are increased. Since carbon is stored in trees and plants, the reduced grow conditions cause a decrease in the uptake of CO2 (Betts et al., 2016; Chatterjee). Since CO2 is an important
greenhouse gas underlying climate change, El Niño events are causing an increase in climate change. However, the effects are working in two ways. Based on future models, it is likely that climate change is causing an increased frequency of extreme El Niño events (Wang et al., 2017; Cai et al., 2014). Today, extreme El Niño’s occur approximately once in two decades a2nd it is anticipated that in the future extreme El Niño’s will occur every 5 till 7 years. In contrary, La Niña events are not increasing in future predictions (Cai et al., 2014). The two types of enforcements are like a vicious circle, they are reinforcing each other.
Conclusion
Since the results of the fitted trendlines are only for a small part significant, the results presented in this thesis provide no statistical evidence to state that the recent drought in South Africa is caused by climate change. However, since 13 out of the 18 weather stations have a declining trend, it is likely that climate change has attributed to the recent drought in South Africa. With a larger dataset of the precipitation in South Africa, it is possible that the results will be significant. The only increasing
significant trend of Coffee Bay is also explained by the high correlation between Niño 3.4 and precipitation anomalies.
A negative correlation between El Niño and declining precipitation is measured over South Africa, which indicates the connection between dry conditions and an El Niño. Furhermore, a similar pattern is visible between low precipitation and positive ONI values, and between high precipitation and low ONI values. In addition, in 2015-2016 a super El Niño did occur. This resulted in increased extreme events, like the drought in South Africa. The western and southern regions of South Africa are more sensitive to the changing conditions of an El Niño than the other regions. Therefore it is very likely that Cape Town is affected by the 2015-2016 El Niño.
The connection between El Niño and climate change seems to work two ways around. Climate change is increasing the frequency of extreme El Niño events and not of extreme La Niña events. However, El Niño events are causing an increase in greenhouse gases such as CO2 and results
in an increasing climate change.
To conclude, according to the results there is not enough significant evidence to establish that a trend like climate change caused the 2015-2018 drought in South Africa. There is enough evidence to confirm that a more incidentally event has occurred; an El Niño. Based on statistical analyses, the recent drought in South Africa is caused by an El Niño event. However, since there are positive feedbacks between El Niño events and climate change it is likely that the recent drought is caused by an El Niño , which is influenced by climate change.
Acknowledgements
References
Araujo, J. A., Abiodun, B. J., & Crespo, O. (2016). Impacts of drought on grape yields in Western Cape, South Africa. Theoretical and applied climatology, 123(1-2), 117-130.
Ashok, K., & Yamagata, T. (2009). Climate change: The El Niño with a difference. Nature, 461(7263), 481.
Baudoin, M. A., Vogel, C., Nortje, K., & Naik, M. (2017). Living with drought in South Africa: Lessons learnt from the recent El Niño drought period. International Journal of Disaster Risk
Reduction, 23, 128-137.
Bellprat, O., Lott, F. C., Gulizia, C., Parker, H. R., Pampuch, L. A., Pinto, I., ... & Stott, P. A. (2015). Unusual past dry and wet rainy seasons over Southern Africa and South America from a climate perspective. Weather and climate extremes, 9, 36-46.
Betts, R.A., Jones, C.D., Knight J.R., Keeling, R.F. & Kennedy J.J. (2016). El Niño and a record CO2 rise.
Nature Climate Change, 6(9), 806-810.
Cai, W., Borlace, S., Lengaigne, M., van Rensch, P., Collins, M., Vecchi, G., … & Jin, F. (2014). Increasing frequency of extreme El Niño events due to greenhouse warming. Nature Climate Change, 4(2), 111-116.
Chatterjee, A., Gierach, M. M., Sutton, A. J., Feely, R. A., Crisp, D., Eldering, A., ... & Schimel, D. S. (2017). Influence of El Niño on atmospheric CO2 over the tropical Pacific Ocean: Findings from NASA’s OCO-2 mission. Science, 358(6360), 1-8.
Climate Engine Web Application (2018). Retrieved from, https://app.climateengine.org/
Conradie, D. C. U. (2012). South Africa’s climatic zones: today, tomorrow. International Green Building Conference and Exhibition, Future Trends and Issues Impacting on the Built Environment, July 25-26, 2012, Sandton, South Africa, 9pp.
Dahlman, L. (2009, August 30). Climate Variability: Oceanic Niño Index., Retrieved from,
https://www.climate.gov/news-features/understanding-climate/climate-variability-oceanic-ni %C3%B1o-index
Dennis, I., & Dennis, R. (2012). Climate change vulnerability index for South African aquifers. Water SA, 38(3), 417-426.
Di Lorenzo, E., Cobb, K.M., Furtado, J.C., Schneider, N., Anderson, B.T., Bracco, A., Alexander, A.M. & Vimont, D.J. (2010). Central Pacific El Niño and decadal climate change in the North Pacific Ocean. Nature geoscience, 3(11), 762-765.
Engelbrecht, C. J., & Engelbrecht, F. A. (2016). Shifts in Köppen-Geiger climate zones over southern Africa in relation to key global temperature goals. Theoretical and applied climatology, 123(1-2), 247-261.
Hoffman, M. T., Carrick, P. J., Gillson, L., & West, A. G. (2009). Drought, climate change and vegetation response in the succulent karoo, South Africa. South African Journal of Science, 105(1-2), 54-60
Machethe, C. L. (2004). Agriculture and poverty in South Africa: Can agriculture reduce poverty. In Paper presented at the Overcoming Underdevelopment Conference held in Pretoria (Vol. 28, p. 29).
NOAA (2018). Analysis Monitoring Detrend Niño 3.4, Retrieved from
http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/detrend.nino34.ascii. txt
Planet Team (2017). Planet Application Program Interface: In Space for life on Earth. San Francisco, CA. http://api.planet.com
Rao, J., & Ren, R. (2017). Parallel comparison of the 1982/83, 1997/98 and 2015/16 super El Niños and their effects on the extratropical stratosphere. Advances in Atmospheric Sciences, 34(9), 1121-1133.
Rouault, M., Pohl, B., & Penven, P. (2010). Coastal oceanic climate change and variability from 1982 to 2009 around South Africa. African Journal of Marine Science, 32(2), 237-246.
Sorensen, P. (2017). The chronic water shortage in Cape Town and survival strategies. International Journal of Environmental Studies, 74(4), 515-527.
South African Weather Service (2018). What is a drought?, Retrieved from
http://www.weathersa.co.za/learning/climate-questions/41-what-is-a-drought
Trenberth, K. E. (1997). The definition of El Niño. Bulletin of the American Meteorological Society, 78(12), 2771-2777.
Van Boxel, J.H. (2008). Signalen van klimaatveranderingen. Van natuurlandschap tot risicomaatschappij: De geografie van de relatie tussen mens en milieu. Amsterdam: Amsterdam University Press, pp 173-179.
Van Wilgen, N. J., Goodall, V., Holness, S., Chown, S. L., & McGeoch, M. A. (2016). Rising temperatures and changing rainfall patterns in South Africa's national parks. International Journal of
Climatology, 36(2), 706-721.
Wang, G., Cai, W., Gan, B., Wu, L., Santoso, A., Lin, X., ... & McPhaden, M. J. (2017). Continued increase of extreme El Niño frequency long after 1.5° C warming stabilization. Nature Climate
Change, 7(8), 568.
Wolski, P. (2018). How severe is Cape Town's “Day Zero” drought?. Significance, 15(2), 24-27.
Ziervogel, G., New, M., Archer van Garderen, E., Midgley, G., Taylor, A., Hamann, R., ... & Warburton, M. (2014). Climate change impacts and adaptation in South Africa. Wiley Interdisciplinary
Appendices
Appendix 1: Excel sheet including data, graphs and statistics
https://drive.google.com/file/d/1cK7LJrQOrcRe6CJFj8hhns7iPe69gnej/view?usp=sharing
Appendix 2: Materials for producing ArcGIS maps
https://drive.google.com/file/d/1vhYabkgsHkdY0lbSi3-xmDhsjBOyr78c/view?usp=sharing https://drive.google.com/file/d/1CRK16psKnDnKSiHhdn9w1KW39S2FImwP/view?usp=sharing https://drive.google.com/open?id=15RBk-OvVsUVLMX3fSLm7CVncH6TTp5TT
