Exploratory Investigation of Extending Rainfall Records in the Western Cape by Means of Dendroclimatology Lauren A. Lester Masters project presented in partial fulfilment of the requirements for the degree of Masters of (Civil) Engineering in the Faculty of Engineering at Stellenbosch University Supervisor: Prof JA du Plessis Department of Civil Engineering December 2019 Stellenbosch University https://scholar.sun.ac.za Declaration By submitting this project electronically, I declare that the entirety of the work contained therein is my own, original work, that I am the sole author thereof (save to the extent explicitly otherwise stated), that reproduction and publication thereof by Stellenbosch University will not infringe any third party rights and that I have not previously in its entirety or in part submitted it for obtaining any qualification. Signature: Date: December 2019 Copyright © 2019 Stellenbosch University All rights reserved ii Stellenbosch University https://scholar.sun.ac.za Abstract Water security is a pressing global issue, although given the recent “Day Zero” crisis experienced in the Western Cape the concern for securing water supplies and sustainable water management is of utmost importance in South Africa. This threat of limited water supplies is only compounded by the growing population and indication of climate change resulting in the country becoming hotter and drier, therefore necessitating more research being done with respect to the effects of climate on South African water surety, and in so doing supporting better, more sustainable, water use and management schemes. Centuries old rainfall records are particularly useful in developing robust climate models, as they may provide insight into any seasonality and cyclicity found within a region’s climate. South Africa, however, has limited reliable rainfall records, with few being in excess of a hundred years therefore leaving much room for improvement in both ascertaining reliable models and in the development of robust, predictive climate models. Successfully used to augment climatic records in parts of Asia, Europe and North America, dendroclimatology offers an opportunity to derive longer rainfall records. Dendroclimatology, a sub- branch of the greater discipline of dendrochronology, is the study and use of trees’ annual growth rings to date and subsequently approximate climatic information, such as rainfall. The ability to extract climatic information from the annual growth ring series is possible since the growth of a tree is not only driven by age or competition but also by climatic factors. To isolate the effect of the climatic factors on the growth, however, the annual growth ring series does need to be detrended through a statistical process known as standardisation and cross-dating. Once the ring series is standardised and cross-dated, a nearby climatic record can be used to establish the correlation between the relevant climatic variable and the tree’s growth thereby allowing the effect of the climate on the growth to be quantified and, if there is a good correlation, projected back or forward to extend the climatic record. Previously, there has not been much success in applying dendroclimatological principles in South Africa – mostly due to the sparse number of indigenous species limiting the sample size and, thus, the accuracy and reliability of the results. The most significant study undertaken in South Africa, at least in terms of prompting greater interest in dendroclimatology, was the “Die Bos” study undertaken by LaMarche and Dunwiddie in the late 1970s. The “Die Bos” study focused on using indigenous Widdringtonia cedarbergensis (wallichii) with the hopes of quantifying a significant correlation between rainfall and the tree growth, however the researchers failed to establish a consistent correlation, noting that the growth and rainfall only correlated over an 18-month period. Other indigenous studies too resulted in non- significant correlations, however all the completed studies do suggest that the potential for a successful correlation remains – going so far as to suggest alternative species, using alien/exotic species, using a greater sample size or assessing other climatic variables, such as temperature. For this research, consideration was only given to the rainfall records in the Western Cape. Four exotic samples from three specific locations were obtained and analysed using both the principles of traditional dendroclimatology and time series-based statistical analyses. The three locations, namely Bishopscourt, Tokai and Wolseley, were selected as complete cross-sectional samples could be obtained. The samples were analysed by means of both conventional dendroclimatology software, namely TSAP-Win, and a more theoretical, statistical, time series centred approach. The TSAP-Win, log-return and ARMA analyses all, however, failed to deliver successful correlations between the rainfall and ring width series with the best iii Stellenbosch University https://scholar.sun.ac.za correlation occurring for the Tokai chronology, at 20.4%. Given that the correlations were statistically insignificant with the limited data sets, it was deemed necessary to attempt to reconstruct a rainfall record with a greater, validated width chronology. The “Die Bos” chronology is an original, validated chronology for the Western Cape and was therefore selected as the final sample that would be analysed through the time series analyses. The results of this, however, were poor with the best correlation of 40.1% therefore negating the possibility of a reliable historic extension. The procedure of how historic rainfall would be determined is, however, included in this research. This research concluded that alternative methods be pursued for the analysis of tree and climate data in the Western Cape, perhaps giving more consideration to machine learning techniques such as random forests or multiple variable regression which allows other climatic variables to be considered. A greater sample set will also be advantageous as the results will inherently bear more statistical significance by meeting the requirements for minimum reliable sample sizes. Additionally, greater sample sizes would also help with generating more accurate chronologies. The possibility for a successful dendroclimatological analysis in the Western Cape is thus still possible. iv Stellenbosch University https://scholar.sun.ac.za Opsomming Watersekerheid is 'n belangrik internasionale aangeleentheid en gegee die onlangse 'Day Zero' -krisis wat in die Wes-Kaap ondervind is, is die versekering van watervoorrade en volhoubare waterbestuur nou uiters belangrik in Suid-Afrika. Die bedreiging van beperkte watervoorrade word vererger deur ‘n groeiende bevolking en 'n aanduiding van klimaatsverandering wat daartoe lei dat die land warmer en droër word. Dit is noodsaaklik dat meer navorsing gedoen word rakende die gevolge van die klimaat op die Suid-Afrikaanse watervoorrade, en sodoende ondersteuning van beter, meer volhoubare watergebruik en bestuur te verseker. Eeue oue reënvalrekords is veral nuttig om robuuste klimaatmodelle te ontwikkel, aangesien dit insig kan bied in die seisoenaliteit en siklusiteit wat in die klimaat van 'n streek voorkom. Suid-Afrika het egter beperkte betroubare reënvalrekords, met slegs 'n paar wat honderd jaar oorskry, wat baie ruimte laat vir verbetering in betroubare modelle sowel as die ontwikkeling van robuuste, voorspellende klimaatmodelle. Dendroklimatologie is suksesvol gebruik om klimaatsrekords in dele van Asië, Europa en Noord-Amerika aan te vul, en bied die geleentheid om langer reënvalrekords af te lei. Dendroklimatologie, 'n vertakking van die groter dissipline van dendrochronologie, is die studie en gebruik van bome se jaarlikse groei ringe vir datering en om sodoende klimaatsinligting, soos reënval, af te lei. Die vermoë om klimaatsinligting uit die jaarlikse groei-ringreeks te verleng, is moontlik omdat die groei van 'n boom nie net deur ouderdom of eksterne faktore soos kompetisie gedryf word nie, maar ook deur klimaatsfaktore. Om die effek van die klimaatfaktore op die groei te isoleer, moet die jaarlikse groei-ringreeks egter ondersoek word deur 'n statistiese proses wat bekend staan as standaardisering en kruisdatering. Sodra die ringreeks gestandaardiseer en gekruisdateer is, kan 'n nabygeleë klimaatrekord gebruik word om die verband tussen die betrokke klimaatveranderlike en die boom se groei te bepaal, waardeur die effek van die klimaat op die groei gekwantifiseer kan word. As daar 'n goeie korrelasie is, kan daar agtertoe óf vorentoe in tyd geprojekteer word om sodoende die klimaatrekord te verleng. Daar was voorheen nie veel sukses met die toepassing van dendroklimatologiese beginsels in Suid-Afrika nie - meestal as gevolg van die geringe aantal inheemse spesies wat die steekproefgrootte beperk het, en dus die akkuraatheid en betroubaarheid van die resultate. Die belangrikste studie wat in Suid-Afrika onderneem is, ten minste ten opsigte van 'n groter belangstelling in dendroklimatologie, was die "Die Bos"-studie wat in die laat 1970's deur LaMarche en Dunwiddie onderneem is. “Die Bos”-studie het gefokus op die gebruik van inheemse Widdringtonia cedarbergensis (wallichii) met die hoop om 'n beduidende korrelasie tussen reënval en die boomgroei te kwantifiseer, maar die navorsers kon nie 'n betroubare korrelasie bewerkstellig nie, met die opmerking dat die groei en reënval slegs oor 'n siklus van 18 maande ooreenstemming getoon het. Ander inheemse