Integrating climate and satellite remote sensing to assess the reaction of Vitis vinifera L. cv. Cabernet Sauvignon to a changing environment

Southey, T. O (2017-03)

Thesis (PhDAgric)--Stellenbosch University, 2017.

Thesis

ENGLISH ABSTRACT: Context - In the context of climate change, factors such as seasonal variability and limitations of available water resources, have increased pressures on the production of table wines, and could continue to do so without effective adaptive strategies. Problem formulation and aims (objective) - The classification of the climate in the Western Cape as well as long term and in-season monitoring are complicated due to the difficulties in accessing climate data, quality of the data is not assured, and limitations due to the sparse spatial distribution of currently logging weather stations. Reliable climate data can be costly, and currently requires intensive data validation; hence the study aimed to find an alternative resource to quantify the climate over the spatial extent of the Western Cape, for possible semi-real time applications. Temperature is one of the main climatic variables driving grapevine response; hence this study validated the use of freely available remote sensing land surface temperature images - a collection of daily maximum and minimum temperature layers. The aims of the study were to quantify climate change and seasonal variability in the Western Cape with the best station network possible within the limits of the spatial and temporal resolution availability. Study design - The study included extreme climatic conditions by selecting sites over a climatic band, and multiple factor analysis was used to evaluate the interaction of climate with grapevine phenology, growth, ripening and wine attributes. This was done to highlight the possible driving factors that can be used in future climatic modelling. Findings and impact - For the estimation of mean weather station temperature, the results from this study can be considered promising, given the simplicity of the statistical models employed, the robustness of the resampling techniques and the high accuracy achieved under these limitations. The weather station mean temperature can therefore be supplemented/predicted using the mean land surface temperatures, based on the regression equations attained from the final analysis with a calibration error of 2.4°C and a prediction error of 2.6°C. The daily mean land surface temperature and weather station temperature data exhibited a strong linear relationship (r = 0.86, p < 0.001, and N = 29), and good prediction accuracy. These land surface temperature maps are intrinsically spatialised, providing daily temperature values that in the past would have only been possible by spatial interpolation of sparse weather station networks, which could only be as accurate as the input data. Seasonal variability was prominent in driving grapevine response. The variability was seemingly intensified by extreme climate events such as extreme wind, rainfall or higher temperatures earlier in growing season as well as during the ripening period. This confirmed the difficulty to predict seasonal conditions in the context of climate change. Seasonal variability and grapevine responses could better be described with finer scale analysis of the climate profile, considering and accounting for the amount of hours in specific temperature, wind and relative humidity ranges. This was a novel approach, as it allowed analysis of the climate data without preconceived ideas around specified thresholds. This could possibly explain grapevine responses outside of a conventional scientific frame of reference. Grapevine phenological stages seemed to be more affected by temperatures outside of the growing season, which highlighted a need for reviewing the current climatic indices used to describe the growing season. Temperatures throughout the year, with the exception of August and September, seemed to affect flowering. The summer months (December, January and February) with more observed hours between 30-35°C and 35-40°C, had a negative correlation with flowering date as days after 1 September (the flowering and vintage precocity indices were earlier). In this study, the date of flowering was most affected by temperature and tended to “set the pace” for phenology in the season. Flowering date as days after budburst could also potentially be used to predict harvest date for Cabernet Sauvignon over sites with an accuracy of only a few days. The study proved that within a general warming trend, the climate in the Western Cape could be both warming and cooling, depending on the area or months in the context of the long term mean. This emphasises the need for continuous semi-real time climate data, such as the daily land surface temperature layers that can be used as an alternative to supplement weather station networks using a regression model to account for the remote sensing imaging. The acquisition and processing of the land surface temperature layers can be automated and extended to other crops cultivated in South Africa. This study has provided some insights into the understanding of cultivating Cabernet Sauvignon in the context of warmer and cooler climatic conditions. Almost every response of the grapevine was affected by climate. It was shown that grapevine growth tempo and final shoot length were sensitive to seasonal variability and water constraints and indirectly affected leaf area per vine that in turn could alter ripening tempo and final wine quality. The seasonal climatic conditions could be masked with more detailed viticultural management practices to ensure a balanced grapevine through the selection of trellis system, pruning and canopy management. In the context of climate change, the aim is to match the cultivar growth and ripening response to the climatic conditions of a site. Finer scale analysis of the climate using hourly frequency data approaches may aid in improving adaptive strategies for the future. This is especially relevant in the context of climate change and the complex terrain of the Western Cape affecting the diurnal shifts of climate over seasons and short distances. The study showed that warm to moderate climatic conditions where grapevines experience moderate water constraints tended to be more balanced in terms of growth and ripening, yielding complex wines with favourable attributes for Cabernet Sauvignon. Prospectives for future work would include the integration of climate maps in the context of grapevine modelling along with a spatial view of the seasonal shifts. This could help with improved management and adaptation. This further emphasises the need for within season monitoring of temperature, plant water status as well as other environmental parameters. It can also be further unlocked by remote sensing products, also in the context of water management in the future. The study provided some foundations for a larger database of climate, land surface temperature and phenology for the identification of cultivar distributions compared to more ideal cultivar distribution in the context of a warmer future with possibly more limited water resources.

AFRIKAANSE OPSOMMING: Konteks - In die konteks van klimaatsverandering het faktore soos die seisoenale wisselvalligheid en beperkings in beskikbare waterhulpbronne die druk op die produksie van tafelwyn verhoog en kan dit so voortduur as daar nie doeltreffende aanpassingstrategieë is nie. Probleemformulering en doelwitte - Die klassifikasie van die klimaat in die Wes-Kaap, sowel as langtermyn en binne-seisoen monitering, word belemmer deur moeilik bekombare klimaatsdata, die kwaliteit van die data wat nie verseker is nie, sowel as beperkings met betrekking tot die yl verspreide weerstasies wat tans funksioneer. Die doelwitte van hierdie studie was om klimaatsverandering en seisoenale wisselvalligheid in die Wes-Kaap te kwantifiseer met die beste netwerk van stasies moontlik binne die perke van die beskikbare ruimtelike en tydsresolusie. Betroubare klimaatsdata kan duur wees en vereis tans intensiewe data validasie; die studie het dus gemik om ’n alternatiewe hulpbron te vind om die klimaat oor die ruimtelike omvang van die Wes-Kaap te kwantifiseer vir moontlike semi-intydse toepassings. Temperatuur is een van die vernaamste klimaatsveranderlikes wat die wingerdrespons dryf; hierdie studie het dus die gebruik van vrylik beskikbare afstandswaarnemingsbeelde van grondoppervlaktemperatuur gestaaf in die vorm van ’n versameling van daaglikse maksimum- en minimum temperatuurlae. Studie ontwerp – Die studie het ekstreme klimaatstoestande ingesluit deur liggings oor ’n klimaatsband te kies, en veelvuldige faktoranalise is gebruik om die interaksie van klimaat met wingerdfenologie, groei, rypwording en wyneienskappe te evalueer. Dit is gedoen om die moontlike dryfvere uit te lig wat in toekomstige klimaatsmodellering gebruik kan word. Bevindings en impak - Vir die skatting van die gemiddelde weerstasietemperatuur kan die uitslae van hierdie studie as belowend beskou word, gegewe die eenvoudigheid van die statistiese modelle wat gebruik is, die robuustheid van die tegnieke vir beeldverwerking en die hoë akkuraatheid wat onder hierdie beperkings verkry is. Die gemiddelde temperatuur van die weerstasie kan dus aangevul/voorspel word deur die gemiddelde temperature van die grondoppervlak te gebruik, gebaseer op die regressievergelykings wat vanaf die finale analise met ’n kalibrasiefout van 2.4°C en ’n voorspellingsfout van 2.6°C verkry is. Die daaglikse gemiddelde grondoppervlaktemperatuur en die temperatuurdata van die weerstasie het ’n sterk liniêre verhouding gewys (r = 0.86, p < 0.001, en N = 29), en goeie voorspellingsakkuraatheid. Hierdie kaarte van grondoppervlaktemperatuur is opsigself ruimtelik, en verskaf dus daaglikse temperatuurwaardes wat in die verlede slegs moontlik sou gewees het deur die ruimtelike interpolasie van yl weerstasienetwerke, wat net so akkuraat soos die insetdata sou wees. Seisoenale wisselvalligheid is ’n prominente aspek wat die wingerdstokrespons dryf. Die wisselvalligheid word skynbaar vergroot deur uiterste klimaatsgebeure, soos uiterste wind, reënval of hoër temperature vroeër in die groeiseisoen, asook tydens rypwording. Hierdie het die moeilikheid om seisoenale toestande in die konteks van klimaatsverandering te voorspel, bevestig. Seisoenale wisselvalligheid en wingerdstokresponse kon beter beskryf word binne die fyner skaalanalise van die klimaatsprofiel, wat die aantal ure binne spesifieke temperatuur-, wind- en relatiewe humiditeitsreekse beter weerspieël. Hierdie was ’n nuwe benadering, aangesien dit voorsiening gemaak het vir die analise van klimaatsdata sonder enige vooroordele as gevolg van spesifieke drempelwaardes. Dit kan moontlik wingerdstokresponse buite ’n konvensionele wetenskaplike verwysingsraamwerk verklaar. Die fenologiese stadiums van die wingerdstok het geblyk om meer deur temperature buite die groeiseisoen geaffekteer te word, wat die behoefte na vore gebring het om die huidige klimaatsindekse wat gebruik word om die groeiseisoen te beskryf, te hersien. Temperature dwarsdeur die jaar, buiten in Augustus en September, blyk om blom te beïnvloed. Die somermaande (Desember, Januarie en Februarie), met meer waargenome ure tussen 30-35°C en 35-40°C, het ’n negatiewe korrelasie gehad met blomdatum as dae ná 1 September (die blom- en oes-vroegheidsindekse het vroeër waardes getoon). In hierdie studie was blomdatum die meeste geaffekteer deur temperatuur en het dit geneig om die toon aan te gee vir fenologie in die seisoen. Blomdatum as dae ná bot kan ook potensieel gebruik word om die oesdatum van Cabernet Sauvignon oor liggings te voorspel met ’n akkuraatheid van slegs ’n paar dae. Hierdie studie het bewys dat, binne ’n algemene neiging tot warmer toestande, die klimaat in die Wes-Kaap beide kan opwarm of afkoel, afhangend van die gebied of maande in die konteks van die langtermyn gemiddelde. Dít benadruk die behoefte aan volgehoue semi-intydse klimaatsdata, soos die daaglikse grondoppervlak- temperatuurlae wat gebruik kan word as ’n alternatief om weerstasienetwerke aan te vul deur middel van ’n regressiemodel om rekenskap te gee van die afstandswaarnemingsbeelde. Die verkryging en prosessering van die grondoppervlak- temperatuurlae kan geoutomatiseer word en aan ander gewasse wat in Suid-Afrika gekweek word, uitgebrei word. Hierdie studie verskaf sekere insigte wat ons begrip van Cabernet Sauvignon in die konteks van warmer en koeler klimaatstoestande verbeter. Feitlik elke reaksie van die wingerdstok word deur klimaat beïnvloed. Daar is getoon dat die groeitempo en finale lootlengte van die wingerdstok gevoelig was vir seisoenale wisselvalligheid en waterstremming en indirek beïnvloed is deur blaaroppervlak per stok, wat op sy beurt weer die rypwordingstempo en finale wynkwaliteit kon verander. Die seisoenale klimaatstoestande kon gemaskeer word deur aangepaste wingerdkundige bestuurspraktyke om ’n gebalanseerde wingerdstok te verseker deur die keuse van opleistelsel, snoei en lowerbestuur. In die konteks van klimaatsverandering, is die doel om die kultivar se groei- en rypwordingsreaksie by die klimaatstoestande van ’n ligging te pas. Fyner skaalanalises van die klimaat deur gebruik te maak van uurlikse frekwensie- databenaderings kan moontlik help om aanpassingstrategieë vir die toekoms te verbeter. Hierdie is veral relevant in die konteks van klimaatsverandering en die kompleksie Wes-Kaapse terrein wat daaglikse verskuiwings in weer/klimaat oor seisoene en kort afstande beïnvloed. Die studie het getoon dat warm tot matige klimaatstoestande waarbinne wingerdstokke matige waterspanning ervaar het, meer gebalanseerd was in terme van groei en rypheid en komplekse wyne met gunstige eienskappe vir Cabernet Sauvignon is gelewer. Moontlikhede vir toekomstige werk kan insluit die integrasie van klimaatskaarte in die konteks van wingerdstokmodellering, tesame met ’n ruimtelike ontleding van die seisoenale verskuiwings. Dit sal help om bestuur en aanpassing te verbeter. Dit benadruk ook verder die behoefte aan binne-seisoen monitering van temperatuur, plantwaterstatus asook ander omgewingsparameters. Hierdie kan ook verder ontsluit word deur afstandswaarnemingsprodukte te gebruik, asook in die toekoms in die konteks van waterbestuur. Hierdie studie verskaf etlike grondslae vir ’n groter databasis van klimaat, grondoppervlaktemperatuur en fenologie, vir die identifisering van huidige kultivarverspreiding in vergelyking met meer ideale verspreiding in die konteks van ’n warmer toekoms met moontlik meer beperkte waterhulpbronne.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/101124
This item appears in the following collections: