The Association for Rural Advancement (AFRA) has run a Drought Project since October 1992. The project's primary focus was to monitor drought conditions in the AFRA's operational area and to assist communities to access relief through providing them with relevant information and helping them to submit proposals to relief agencies. In attempting to do this, it became clear that the effects of drought on rural black communities was not a well understood phenomenon. We also found that relief strategies were informed more by economic, agricultural and meteorological criteria than social ones. This Special Report aims to contribute to the development of a more appropriate drought management strategy, especially with regard to rural communities. In attempting to make this contribution, the Report examines the factors involved in redefining drought and drought relief perspectives in relation to rural communities. However, the range of factors involved and the scarcity of useful information on them, dictate that this is merely an introduction. The complexities and implications arising from the introduction of these factors into drought management in South Africa should be explored on an on-going basis. Because of the scarcity of relevant information about the effects of drought on black rural communities in South Africa, this Report has relied on AFRA's limited monitoring of such communities. Some information was also gathered through Participatory Rural Appraisal (PRA) methods. There were also difficulties in getting information about relief schemes and their operations as most relief agencies were reluctant to release this information. As a result, almost all the relief related information in this Special Report was obtained from publications and reports of the National Consultative Forum on Drought.

The main objective of the study was to assess the extent to which resource-poor households in selected villages of Nkonkobe Local Municipality in the Eastern Cape Province of South Africa are vulnerable to drought by using an improvised remote sensing and Geographic Information System (GIS)-based mapping approach. The research methodology was comprised of 1) assessment of vulnerability levels and 2) the calculation of established drought assessment indices comprising the Normalized Difference Vegetation Index (NDVI) and the Normalized Difference Water Index (NDWI) from wet-season Landsat images covering a period of 29 years from 1985 to 2014 in order to objectively determine the temporal recurrence of drought in Nkonkobe Local Municipality. Vulnerability of households to drought was determined by using a multi-step GIS-based mapping approach in which 3 components comprising exposure, sensitivity and adaptive capacity were simultaneously analysed and averaged to determine the magnitude of vulnerability. Thereafter, the Analytical Hierarchy Process (AHP) was used to establish weighted contributions of these components to vulnerability. The weights applied to the AHP were obtained from the 2012 - 2017 Nkonkobe Integrated Development Plan (IDP) and perceptions that were solicited from key informants who were judged to be knowledgeable about the subject. A Kruskal-Wallis H test on demographic data for water access revealed that the demographic results are independent of choice of data acquired from different data providers (χ2(2) = 1.26, p = 0.533, with a mean ranked population scores of 7.4 for ECSECC, 6.8 for Quantec and 9.8 for StatsSA). Simple linear regression analysis revealed strong positive correlations between NDWI and NDVI ((r = 0.99609375, R2 = 1, for 1985), 1995 (r = 0.99609375, R2 = 1 for 1995), (r = 0.99609375, R2 = 1 for 2005) and (r = 0.99609375, R2 = 1 for 2014). The regression analysis proved that vegetation condition depends on surface water arising from rainfall. The results indicate that the whole of Nkonkobe Local Municipality is susceptible to drought with villages in south eastern part being most vulnerable to droughts due to high sensitivity and low adaptive capacity.

Summaries in English, Afrikaans and Sepedi / The Limpopo Province is a disaster-prone province, with drought being the most common natural disaster. From the year 2012 onwards, the province experienced extremely dry conditions that culminated in a severe drought in 2016. This negatively impacted the livelihoods of smallholder livestock farmers and the welfare of their livestock. The study investigated the coping strategies that were adopted by smallholder livestock farmers during drought conditions and the factors that were associated with animal loss during the drought that affected the region between 2014 and 2016. A randomly selected sample of 281 smallholder livestock farmers aged 18 years and older from the Greater Letaba Local Municipality participated in the study. Structured interviews, aided by questionnaires were used to collect the primary data. Proportions of categorical variables and the mean and standard deviation for continuous variables were computed and presented as tables and figures. A Poisson regression model was fitted to the data to identify factors that were significantly associated with loss of animals during the drought. More than half (55.50%; n=116) of the participants were made aware of the 2014–2016 drought through the agricultural extension officers, followed by 19.14% (n=40) who got to know about it through radio channels. More than half of the participants (58.29%; n=123) were aware of the impending drought. The most common support received from government agencies to help cope with the drought was in the form of animal feed (80%; n=124). Although most of the farmers (73.55%; n=114) benefitted from the support they received, slightly more than half (53.74%; n=151) did not cope well with the drought conditions. While Bellevue (B=-0.199; 95% CI: -0.380 -0.019) was negatively associated with loss of animals, Mokwakwaila (B=0.568; 95% CI: 0.405 0.731) had a strong positive association with loss of animals. Being married (B=-0.060; 95% CI: -0.305 0.183) or divorced (B= -0.035; 95% CI: -0.316 0.246) was negatively associated with loss of animals. Years of experience in farming (B=0.022; 95% CI: 0.010 0.033) and not receiving support during were strongly positively associated with loss of animals (B=0.324; 95% CI: 0.189 0.459). The low number of farmers who were aware of the impending drought and the large number of farmers who did not cope well suggests that many farmers in the area were not prepared for the drought. Groups such as widows, widowers and farmers who have many years of farming experience are high-risk groups and should be targeted for interventions in the event of a drought. More measures are needed to ensure that all agricultural centres are prepared and supported in event of a drought so as to minimise the impact of drought on local communities. / Die provinsie Limpopo is 'n rampgevoelige provinsie, met droogte as die mees algemene natuurramp. Vanaf 2012 het die provinsie uiters droë toestande beleef wat in 2016 op 'n ernstige droogte uitgeloop het. Dit het 'n negatiewe uitwerking gehad op die lewensonderhoud van kleinboere en die welstand van hul vee. Die studie het ondersoek ingestel na die hanteringstrategieë wat deur veeboere in kleinvee tydens droogtetoestande aangeneem is, en die faktore wat verband hou met diereverlies tydens die droogte wat die streek tussen 2014 en 2016 geraak het. 'N Lukraak geselekteerde steekproef van 281 kleinboere van 18 jaar en ouer van die Greater Letaba Local Munisipaliteit het aan die studie deelgeneem. Gestruktureerde onderhoude, gehelp deur vraelyste, is gebruik om die primêre data in te samel. Verhoudings van kategoriese veranderlikes en die gemiddelde en standaardafwyking vir deurlopende veranderlikes is bereken en as tabelle en figure aangebied. 'N Poisson-regressiemodel is op die data toegepas om faktore te identifiseer wat beduidend verband hou met die verlies aan diere tydens die droogte. Meer as die helfte (55,50%; n=116) van die deelnemers is bewus gemaak van die droogte 2014–2016 deur die landbouvoorligtingsbeamptes, gevolg deur 19,14% (n =40) wat dit via radiokanale leer ken het. Meer as die helfte van die deelnemers (58,29%; n=123) was bewus van die dreigende droogte. Die mees algemene steun van regeringsinstansies om die droogte die hoof te bied, was in die vorm van veevoer (80%; n=124). Alhoewel die meerderheid van die boere (73,55%; n=114) voordeel getrek het uit die steun wat hulle gekry het, het die droogtetoestande nie goed hanteer nie (53,74%; n=151). Terwyl Bellevue (B= -0.199; 95% CI -0.380 -0.019) negatief geassosieer is met verlies aan diere, is Mokwakwaila (B= 0,568; 95% CI 0,405 0,731) sterk positief geassosieer met verlies aan diere. Om getroud te wees (B= -0.060; 95% CI -0.305 0.183) of geskei (B= -0.035; 95% CI 0.316 0.246) was negatief geassosieer met verlies aan diere. Jare se ondervinding in die boerdery (B=0,022; 95% CI 0,010 0,033) en om nie ondersteuning gedurende te ontvang nie, was sterk positief geassosieer met die verlies van diere (B=0.324; 95% CI 0.189 0.459). Die lae aantal boere (ongeveer die helfte) wat bewus was van die dreigende droogte, dui daarop dat baie boere in die omgewing nie voorbereid was op die droogte nie. Groepe soos weduwees, wewenaars en boere met baie jare se boerdery-ervaring, is hoërisikogroepe en moet geteiken word vir ingrypings in die geval van 'n droogte. Meer maatreëls is nodig om te verseker dat alle landbousentrums voorberei en ondersteun word in geval van 'n droogte om die impak van droogte op plaaslike gemeenskappe te verminder. / Komelelo ke bothata bjo bogolo kudu go tsa temo/bolemi, segolothata re lebeletse balemirui ba bannyane. Limpopo province e na le kgatelelego ye kgolo kudu ka komelelo gagolo ge re lebeletse tsa bolemi. Nako le nako komelelo e tsea karolo ye kgolo moo e feleletsago e gatelela tsa temo. Tabakgolo ya rena kego lebelela ditsela le mehuta ye e fapanego yeo e shomishitshwego ke balemi go lwantshana le bothata bja komelelo gareng ga ngwaga wa 2014- 2016. Thuto ye e kgobokantshitshwe gotwsa go masepala wa motse selegae wa Greater Letaba Local Municipality, karolo ya Mopani, profenseng ya Leboa mo Afrika Borwa. Tshedimosho ye e hweditshwe ka mokgwa wa poledisano le balemi bao ba fapanego ba go lekana nomoro ya 281 ya balemi go dinaga tsa go fapanafapana. Poledisano ebile ka mokgwa wa peakanyanyo ya dopotsisho. Tshedimosho ye e kopantshitshwe le go hlathollwa ka mokgwa wa go ikgetha wa Strata Version 14. Hlathollo ya go ikgetha e berekishitshwe ka mokgwa wa ditiragalo yoya ka nako le dipalopalo tsa go kopantswa fao tahlegelo e sa tsebjego go ka lekanywa.le bokae la diperesente. Mabakakgolo ao a hlolago tahlegelo ya diruiwa a nyakishitswe ka mokgwa wa Poisson Regression Model. Bogolo bja (64.77%) go bao ba arabilego ebile banna le bogolo bja (74.38%) bja balemi ba be ba tseba ka komelelo ye e batamelago. Go feta halofo (55.50%) ya bao ba arabilegoba tsebishitshwe ka komelelo yeo e batamelang go tswa go bagakolodi, gomme gwa latela thelebisheni (8.13%) le dikuranta (1.44%). Go feta bogolo bja (58.29%) bja bao ba ikarabetsego ba laeditswe gore ba be ba tseba ka komelelo yeo e batamelago ke fao ba bego ba ikemiseditse. Bontshi bja thekgo (80%) yeo balemi ba e hweditsego ebile ka mokgwa wa dijo tsa diphoofolo, gwa latela latela thekgo yeo e filwego ka mokgwa wa meets le dithibela malwetsi go diphoofolo. Bogolo bjago makatsa (73.55%) bo laeditse gore thekgo yeo ba e hweditsego e ba tshwetsi molemo le diphoofolo tsa bona, go feta bonnyane (26.45%) bjoo bo rilego thekgo yeo ba e hweditsegobka nako ya komelelo ga se ya ba tshwela mohola goba gaya ba hola ka selo. Go ba modudi wa Bellevue (B= -0.199; 95% CI: -0.380 -0.019, p= 0.031) le Mokwakwaila (B=0.568; 95% CI: 0.405 0.731, p= 0.0001) ebile nthla ye bohlokwa go tahlegelo ya dirui. Go oketsa seo, go nyadiwa (B=0.942; 95% CI: 0.737 1.201), go hladiwa (B= 0.966 (95% CI: 0.729 1.279), le palo ya mengwaga ya botsebingo tsa temo (B=1.022; 95% CI: 1.010 1.034, p= 0.0001), le ge eba molemi ga se a hwetse thusho ka nako ya komelelo (B=0.324; 95% CI: 0.189 0.459, p= 0.0001) ebile dinhla tsa bohlokwa go tahlegelo ya dirui ka nako ya komelelo. Palo ya fase ba balemi bao ba bego bas a tsebe ka komelelo yeo e bego e batamela, e kgathile tema ye kgolo mo tabeng ya go paledisha balemi go ipeakanyela kgahlanong le komelelo yeo e bego e batamela. Ka fao seo se ba amile gampe. Mo nakong ye e tlang, matsapa a mantsi a swanetswi go tseiwa go kgonthishisha gore tsa temo ka moka di itokishetsa ka go lekana. Mmusho o swanetse go beeletsa kudu go lefapha la bagakolodi ka go oketsa palo ya mafapha a bagakolodi. / Agriculture and Environmental Sciences / M.Sc. (Agriculture)

Thesis (M. Sc. (Agricultural Economics)) -- University of Limpopo, 2021 / Agriculture is not only the backbone, but also an important sector of the South African economy. It provides food and employment to a majority of people in the country, especially in the rural areas. Smallholder farmers play an important role in livelihood creation and the alleviation of poverty among the population of the Limpopo Province. However, despite their significant contribution, smallholder farmers’ production is still low. Climate change has bought increasing frequencies and severity of drought conditions and uncertainties in the length and quality-growing season. Drought threatens the production of maize as a staple food and without measures to counter climate change, food security will be a major problem in South Africa. This study therefore examined factors determining the adoption of drought tolerant maize among smallholder farmers in the Lepelle-Nkumpi Municipality. Primary data was collected using semi-structured questionnaires to achieve the objectives of the study. Multistage sampling was used for the study because larger clusters were subdivided into smaller and more targeted groupings for surveying. Descriptive Statistics and the Binary Probit Model were used to analyse the data. The results of the Probit Regression analysis indicated that farm size, hired labour and maize produced per hectare had positive significant influence on the probability of farmers adopting drought tolerant maize varieties. Farm size and maize produced per hectare were statistically significant at 1% and hired labour was statistically significant at 5%. Based on the sample of this study,74% of the households grew non-drought tolerant maize varieties, while 26% of the smallholder farmers grew drought tolerant maize varieties. The results indicate that 24,4% of the farmers were not affected by any constraints in terms of their adoption of drought tolerant maize varieties whilst 76,6% said they are affected by those constraints in Lepelle-Nkumpi Municipality. Based on the study’s findings, it is recommended that extension officers should make it a priority to provide smallholder farmers with timely and accurate information. Extension officers should effectively disseminate information about the adoption of drought tolerant maize through a combination of different pathways.

Thesis (M.Sc.(Agriculture (Agricultural Economics)) -- University of Limpopo, 2022 / Understanding the factors that influence farmers' decisions to take preventive measures against natural hazards provides insight that can be used to develop user-specific interventions to support their adaptation processes. The use of Protection Motivation Theory in analysing climate risk adaptation behaviour is driven by the increase in climate change, which is projected to increase the frequency and severity of climate related risks such as heatwaves, floods, and droughts. Given the importance of livestock in rural communities, information about their adaptation must be prioritised; yet, this is not the case, as most climate change adaptation research focus on crop production. The main aim of the study was to analyse the drought risk preparedness of smallholder livestock farmers in the Limpopo Province's Blouberg Local Municipality. The study's specific objectives were to identify and describe the socioeconomic characteristics of smallholder livestock farmers in the Blouberg Local Municipality, as well as to determine the drought coping and adaptation strategies used by them and to evaluate the protection motivation theory components influencing that coping and adaptation behaviour. The study collected primary cross-sectional data from 130 smallholder livestock farmers in the Blouberg Local Municipality using a semi-structured questionnaire. The farmers' drought risk coping and adaptation strategies were described using descriptive statistics, while multiple linear regression was used to test whether protection motivation theory variables influence the adaptation and coping choices of smallholder livestock farmers in Blouberg Local Municipality. According to the findings, smallholder livestock farmers in Blouberg Local Municipality use four measures on average to protect their livestock against drought. With an R2 adjusted of 0.70, protection motivation theory variables explain 70% of the variation in farmer protection motivation. Perceived risk probability, perceived severity, perceived self-efficacy, and perceived costs were significant variables associated with farmers' protection motive. It is recommended that interventions meant to increase drought risk resilience of the farmers should prioritise early warning signals to increase perceived probability of the farmers, create platforms for information exchange to increase perceived severity, teach farmers methods practically to increase perceived self-efficacy and keep the price of utilising measures low to decrease perceived cost / Risk and Vulnerability Science Centre (RVSC)

A case study on drought coping mechanisms was conducted among small-scale farmers in the Motheo District of the Free State Province in Republic of South Africa, to determine how farmers cope with drought effects with or without external influence in terms of drought relief packages from the government and non-governmental organizations. Data was collected by administering a semi-structured questionnaire to 200 farmers. The data were captured and analysed using SPSS to obtain frequency, cross-tab, univariate ANOVA as well as logistic regression analysis. Findings of the study revealed that only 12.5 percent of the respondents were aware of drought, while a larger percentage of 87.5 of the respondents were not aware of a drought incidence before its onset, which made them more vulnerable to the drought disaster; 8.5 percent of them protected water sources for livestock while 91.5 percent of the farmers did not protect water sources for their livestock because they farm on a communal land; 42.5 percent provided supplementary feeds to livestock during the drought, but 57.5 percent did not provide supplementary feed for their animals for lack of funds. Ninety-nine (99.0) percent of the respondents shared grazing lands while only 1 per cent did not because most farmers operates on a communal system of farming; 35.5 percent changed cropping systems; 50.5 percent had alternative water sources for crops which included mini and hand irrigation systems while 49.5 percent of the respondents depended solely on streams and rivers available in the villages; 19.3 percent sold or pledged assets in order to be able to cope with drought effects while most farmers did not pledge or sell assets not because they did not want to, but because they did not have assets to sell. / Envornmental Science / Thesis (M.Sc. (Environmental Management))

A case study on drought coping mechanisms was conducted among small-scale farmers in the Motheo District of the Free State Province in Republic of South Africa, to determine how farmers cope with drought effects with or without external influence in terms of drought relief packages from the government and non-governmental organizations. Data was collected by administering a semi-structured questionnaire to 200 farmers. The data were captured and analysed using SPSS to obtain frequency, cross-tab, univariate ANOVA as well as logistic regression analysis. Findings of the study revealed that only 12.5 percent of the respondents were aware of drought, while a larger percentage of 87.5 of the respondents were not aware of a drought incidence before its onset, which made them more vulnerable to the drought disaster; 8.5 percent of them protected water sources for livestock while 91.5 percent of the farmers did not protect water sources for their livestock because they farm on a communal land; 42.5 percent provided supplementary feeds to livestock during the drought, but 57.5 percent did not provide supplementary feed for their animals for lack of funds. Ninety-nine (99.0) percent of the respondents shared grazing lands while only 1 per cent did not because most farmers operates on a communal system of farming; 35.5 percent changed cropping systems; 50.5 percent had alternative water sources for crops which included mini and hand irrigation systems while 49.5 percent of the respondents depended solely on streams and rivers available in the villages; 19.3 percent sold or pledged assets in order to be able to cope with drought effects while most farmers did not pledge or sell assets not because they did not want to, but because they did not have assets to sell. / Envornmental Science / Thesis (M.Sc. (Environmental Management))

MENVSC / Department of Geography and Geo-Information Sciences / South Africa was badly affected by the recent 2015/16 severe drought. Water levels in dams declined drastically resulting in decimation of livestock herds and widespread crop failure. Mopani District Municipality is comprised of many agricultural activities that contribute to the economy and social development of the country. The study evaluated the nature of the drought hazard - its impacts, vulnerability and adaptation strategies employed by rural communities of Mopani District. The study used a mixed method approach with both quantitative and qualitative datasets. The district was divided into two distinct climatic areas, the eastern lowveld which includes the Greater-Giyani, Ba-Phalaborwa and Maruleng Local Municipalities and the western highveld which includes Greater- Tzaneen and Greater- Letaba Local Municipalities. Questionnaires were administered among community members whilst key informant interviews were conducted among relevant government and municipal officials. Anomalies in long term climate data were analysed to determine the frequency and intensity of drought in the district. Drought characterisation was done using a Standardised Precipitation and Evapotranspiration Index whilst vegetation anomaly maps, maize yields and dam level data were used to analyse the impacts of drought across the district. Levels of vulnerability to drought were determined using the Household Vulnerability Index. Spatially distinct patterns of drought conditions across the district were remarkable with wet conditions on the western highveld along the escarpment and harsh dry conditions towards the eastern lowveld. It was found that nearly half the time there is some form of drought or another in the district which may be linked to the remote El Nino phenomenon. Community vulnerabilities have a direct impact on human welfare and different strategies are employed to adapt to drought hazards both at community and district levels. The study showed a link between drought hazard extent and vulnerability. Community members are adapting using conservation agriculture, selling fire-wood, accessing boreholes and rearing chickens, amongst other means to survive in these harsh climatic conditions. Local government intervention strategies include supply of seeds and fertilisers, selling fodder at a cheaper price and supplying water using trucks. The findings of this study contribute to disaster risk reduction efforts in Mopani District Municipality / NRF

PhDH / Department of Hydrology and Water Resources / Demand for water resources has been on the increase and is compounded by population growth and related development demands. Thus, numerous sectors are affected by water scarcity and therefore effective management of drought-induced water deficit is of importance. Luvuvhu River Catchment (LRC), a tributary of the Limpopo River Basin in South Africa has witnessed an increasing frequency of drought events over the recent decades. Drought impacts negatively on communities’ livelihoods, development, economy, water resources, and agricultural yields. Drought assessment in terms of frequency and severity using Drought Indices (DI) in different parts of the world has been reported. However, the forecasting and prediction component which is significant in drought preparedness and setting up early warning systems is still inadequate in several regions of the world. This study aimed at characterising, assessing, and predicting drought conditions using DI as a drought quantifying parameter in the LRC. This was achieved through the application of hybrid statistical and machine learning models including predictions via a combination of hybrid models. Rainfall and temperature data were obtained from South African Weather Service, evapotranspiration, streamflow, and reservoir storage data were obtained from the Department of Water and Sanitation while root zone soil moisture data was derived from the NASA earth data Giovanni repository. The Standardised Precipitation Index (SPI), Standardised Precipitation Evapotranspiration Index (SPEI), Standardised Streamflow Index (SSI), and Nonlinear Aggregated Drought Index (NADI) were selected to assess and characterise drought conditions in the LRC. SPI is precipitation based, SPEI is precipitation and evapotranspiration based, SSI is based on streamflow while NADI is a multivariate index based on rainfall, potential evapotranspiration, streamflow, and storage reservoir volume. All indices detected major historical drought events that have occurred and reported over the study area, although the precipitation based indices were the only indices that categorised the 1991/1992 drought as extreme at 6- and 12- month timescales while the streamflow index and multivariate NADI underestimated the event. The most recent 2014/16 drought was also categorised to be extreme by the standardised indices. The study found that the multivariate index underestimates most historical drought events in the catchment. The indices further showed that the most prevalent drought events in the LRC were mild drought. Extreme drought events were the least found at 6.42%, 1.08%, 1.56%, and 4.4% for SPI, SPEI, SSI, and NADI, respectively. Standardised indices and NADI showed negative trends and positive upward trends, respectively. The positive trend showed by NADI depicts a decreased drought severity over the study period. Drought events were characterised based on duration, intensity, severity, and frequency of drought events for each decade of the 30 years considered in this study i.e. between 1986 – 1996, 1996 – 2006, 2006 – 2016. This was done to get finer details of how drought characteristics behaved at a 10-year interval over the study period. An increased drought duration was observed between 1986 - 1996 while the shortest duration was observed between 1996 - 2006 followed by 2006 - 2016. NADI showed an overall lowest catchment duration at 1- month timescale compared to the standardised indices. The relationship between drought severity and duration revealed a strong linear relationship across all indices at all timescales (i.e. an R2 of between 0.6353 and 0.9714, 0.6353 and 0.973, 0.2725 and 0.976 at 1-, 6- and 12- month timescales, respectively). In assessing the overall utilisation of an index, the five decision criteria (robustness, tractability, transparency, sophistication, and extendibility) were assigned a raw score of between one and five. The sum of the weighted scores (i.e. raw scores multiplied by the relative importance factor) was the total for each index. SPEI ranked the highest with a total weight score of 129 followed by the SSI with a score of 125 and then the SPI with a score of 106 while NADI scored the lowest with a weight of 84. Since SPEI ranked the highest of all the four indices evaluated, it is regarded as an index that best describes drought conditions in the LRC and was therefore used in drought prediction. Statistical (GAM-Generalised Additive Models) and machine learning (LSTM-Long Short Term Memory) based techniques were used for drought prediction. The dependent variables were decomposed using Ensemble Empirical Mode Decomposition (EEMD). Model inputs were determined using the gradient boosting, and all variables showing some relative off importance were considered to influence the target values. Rain, temperature, non-linear trend, SPEI at lag1, and 2 were found to be important in predicting SPEI and the IMFs (Intrinsic Mode Functions) at 1, 6- and 12- month timescales. Seven models were applied based on the different learning techniques using the SPEI time series at all timescales. Prediction combinations of GAM performed better at 1- and 6- month timescales while at 12- month, an undecomposed GAM outperformed the decomposition and the combination of predictions with a correlation coefficient of 0.9591. The study also found that the correlation between target values, LSTM, and LSTM-fQRA was the same at 0.9997 at 1- month and 0.9996 at 6- and 12- month timescales. Further statistical evaluations showed that LSTM-fQRA was the better model compared to an undecomposed LSTM (i.e. RMSE of 0.0199 for LSTM-fQRA over 0.0241 for LSTM). The best performing GAM and LSTM based models were used to conduct uncertainty analysis, which was based on the prediction interval. The PICP and PINAW results indicated that LSTM-fQRA was the best model to predict SPEI timeseries at all timescales. The conclusions drawn from drought predictions conducted in this study are that machine learning neural networks are better suited to predict drought conditions in the LRC, while for improved model accuracy, time series decomposition and prediction combinations are also implementable. The applied hybrid machine learning models can be used for operational drought forecasting and further be incorporated into existing early warning systems for drought risk assessment and management in the LRC for better water resources management. Keywords: Decomposition, drought, drought indices, early warning system, frequency, machine learning, prediction intervals, severity, water resources. / NRF

Thesis (MSc)--Stellenbosch University, 2015. / ENGLISH ABSTRACT: Understanding the effects of climate change is one of the cardinal issues within the natural resource management circles. Increased droughts are part of these changes. Afrotemperate forests, as well as their drier Afromontane counterparts suffer from periodic and seasonal droughts respectively. To better understand the effect of droughts on these forests, three key species namely Olea capensis (Iron wood), Podocarpus latifolius (Common Yellow wood) and Pterocelastrus tricuspidatus (Candle wood), were analysed using dendroecologic techniques. Two sites in the Southern Cape were selected according to a West-to-East moisture gradient, with the drier site being close to George and the medium moist site at the Diepwalle estate in the vicinity of Knysna. Growth ring measurements from each of the species were used to calculate basal area and basal area increment during the lifetime of the trees. Drought years for the sites were then selected based on the Standardised Precipitation and Evapotranspiration Index (SPEI), also indicated by the growth during the drought periods. Differences in growth patterns for all three species were observed. An event analysis was then used to quantify the difference in the resistance (Rt), recovery (Rc), resilience (Rs) and relative resilience (RRs). With values standardised around one (Rt, Rc and Rs) and zero (RRs), it was seen that the Candle wood had the highest (~0.92) resistance and the Yellow wood had the highest (~1.3) recovery after the drought. Iron wood stood apart from the other two species in the sense that it only reacted negatively towards the drought one year after the event in most cases. It was concluded that each of the species were significantly different in their reactions towards drought. This specific difference in drought reaction can give way to the possibility that the species together adapted to relieve the stress of a short drought by splitting the available resources over a longer period. / AFRIKAANSE OPSOMMING: Dit is van uiterse belang vir bestuur doeleindes om die veranderende klimaat oor die wêreld te verstaan, insluitend die droogtes wat daarmee gepaard gaan. Die Afrotemperate woud-tipe, asook sy droeër teenstaander, die Afromontane, lei gereeld aan sporadiese en seisonale droogtes. Om hierdie woud-tipe se reaksie tot droogtes beter te verstaan, was drie boom spesies naamlik Ysterhout (Olea capensis), Kershout (Pterocelastrus tricuspidatus) en gewone Geelhout (Podocarpus latifolius), gekies vir die gebruik in ‘n dendro-ekologiese studie. Twee areas was gekies van ‘n wes-tot-oos droogte gradient, met die droeër blok in die George omgewing en die meer vogtige een naby aan Knysna. Die jaarring metings van elke boom was gebruik om beide die basale oppervlakte en die basale oppervlak groei per jaar aan te teken. ‘n Gestandardiseerde reenval en evapotranspirasie indeks (SPEI) was gebruik om vas te stel jare waarin matige tot sterk droogtes gebeur het. Hierdie gekose jare het aanduiding gegee dat daar wel ‘n verskil waargeneem was in die groei patrone van elke spesie gedurende die droogtes. ‘n Gebeurtenis analise is gebruik om ‘n kwantitatiewe verskil te kon sien in die weerstand (Rt), herstel (Rc), weerstandbiedendheid (Rs) en relatiewe weerstandbiedendheid (RRs). Die was waargeneem dat Kerhout die hoogste weerstand (0.92) toon, terwyl die Geelhout ‘n hoër herstel waarde (1.3) gehad het. Ysterhout het apart van die ander twee spesies gestaan in dìe dat dit eers een jaar na die droogte ‘n reaksie getoon het teenoor die droogte. Dit was dus gevind dat daar spesifieke verskil is tussen al drie van die spesies teen opsigte van stres reaksies was. Hierdie verskil kan dan wel ook moontlik aandui dat hierdie spesies en woud-tipe op so ‘n anier aangepas is dat dit die stress gedurende ‘n kort droogte versprei oor ‘n langer tydperk.