Water consumption of South African vineyards : a modelling approach based on the quantified combined effects of selected viticultural, soil and meteorological parameters

Myburgh, P. A. (1998-12)

Thesis (PhD)--Stellenbosch University, 1998.

Thesis

AFRIKAANSE OPSOMMING: Normaalweg word slegs een of twee stelle gewasfaktore in kombinasie met 'n verwysingwaterverbuik (ET) vir die beraming van gewas waterverbruik (ET) van wingerde gebruik. As gevolg van variasie tussen wingerde beperk hierdie gewasfaktore die akkuraatheid waarmee produsente besproeiing kan bestuur om produksie en gehalte te optimiseer. Die doel van hierdie studie was om 'n waterverbruikmodel, wat die variasie tussen wingerde in ag neem, te ontwikkel. Die hittepuls tegniek is gekalibreer om sapvloei oor kort periodes in wingerdstamme te meet. 'n Kalibrasiekurwe vir sapvloei teenoor tyd is ontwikkel. Om vir variasie van sapvloei in xileem voorsiening te maak, is vier sensors per stam gebruik. Sapvloei is in geselekteerde wingerde onder verskillende toestande gemeet. Daaglikse sapvloei per stok het toegeneem met blaaroppervlak. Uurlikse sapvloei het nie reglynig met straling toegeneem nie, wat daarop gedui het dat maksimum huidmondjie-opening net 'n vaste hoeveelheid transpirasie toegelaat het. In sommige gevalle het sapvloei gedurende die dag tydelik afgeneem wat op 'n waterbesparingsmeganisme as gevolg van huidmondjie-sluiting onder toestande van hoe ligintensiteit gedui het. In vergelyking met droëlandtoestande, het besproeiing slegs aanvanklik hoë sapvloeipieke geïnduseer. Dit het impliseer dat, benewens verhoogde transpirasie, turgiditeit in selle ook herstel is. In vergelyking met loweroppervlak-oriëntasie en meteorologiese toestande, het oeslading en besproeiing beperkte effekte op daaglikse sapvloei gehad. Tagtig persent van variasie in sapvloei kan op grand van blaaroppervlak en ET voorspel word. As gevolg van verskille in die hoeveelheid blare aan straling blootgestel, was die voorspelling meer akkuraat wanneer tussen horisontale en vertikale lowers onderskei is. Aangesien toename in beskaduwing met 'n toename in lowerdigtheid, asook kultivareienskappe en waterspanning, nie in ag geneem is nie, word hierdie modelle as 'n eerste benadering beskou. 'n Li-Cor LAl -2000 Plant Canopy Analyzer (PCA) is in geselekteerde wingerde gekalibreer om blaaroppervlakindeks (BOIpca) te meet. Alhoewel die PCA werklike blaaroppervlakindeks (BOI) onderskat het, is die nou korrelasie tussen BOl pca en BOI gebruik om blaaroppervlakontwikkeling te meet. Blaaroppervlakontwikkeling is in agt wingerde, wat ten opsigte van kultivar, plantafstand en prieelstelsel verskil het, in vyf wingerdbougebiede gemeet. Seisoenale verandering in blaaroppervlak kon met derde orde polinomiese vergelykings, met dag van groeiseisoen as enigste veranderlike, voorspel word. Afsonderlike potensiële groeikurwes is vir die Winterreën en Somerreën gebiede ontwikkel. Die gebruik van lootmassa om maksimum blaaroppervlak te voorspel, was minder akkuraat as wanneer die vars blaarmassa gebruik is. Horisontale priële het meer blare per eenheid lootmassa as vertikale priële geproduseer. Water wat daagliks in bogrondse dele van die wingerdstok gestoor word, het slegs breukdele van 'n millimeter beloop en kon dus in die voorspelling van waterverbruik geïgnoreer word. Verdamping vanaf die grondoppervlak (E) is met mini-lisimeters in wingerde gemeet om die Boesten & Stroosnijder model te evalueer en aan te pas. Onder nat grondtoestande was daar 'n neiging tot hoër E in die middelste gedeelte van die werksry. Verdamping vanaf onbewerkte grond sonder 'n deklaag is met aanvaarbare akkuraatheid deur die model voorspel. Die model moes egter aangepas word om interaksie tussen die lower en grond in ag te neem. Daar was 'n neiging tot hoër E onder horisontale priële as by vertikale priële gedurende fase twee van verdamping. Verdamping het ook tussen die ses grondtipes wat in hierdie studie gebruik is, verskil. 'n Strooideklaag het E betekenisvol onder relatiewe nat toestande beperk. Vir die meeste gronde het die deklaag na tien dae geen verskil in vergelyking met grond sonder 'n deklaag gehad nie. Daar was 'n reglynige verband tussen kumulatiewe E en ET vasgestel. Kumulatiewe E was gemiddeld ongeveer 30 % van kumulatiewe ET. Die transpirasie en oppervlakverdamping modelle is gekombineer om as basis vir 'n voorlopige voorspellingsmodel vir evapotranspirasie te dien. Gesimuleerde ET is teenoor werklike ET, soos gemeet in agt wingerde onder verkillende toestande, vergelyk. Hierdie wingerde het 'n reeks veranderlikes soos besproeiingstelsel, grondtipe, grondwateronttrekkingspeile, prieelstelsel en groeikrag aangespreek. Slegs eenvoudige parameters soos lootmassa, plantafstand, loweroppervlak-orientasie, verwysingswaterverbruik en 'n konstante waarde wat die verdampingsverliese vanaf 'n spesifieke grondtipe bepaal, is as invoere gebruik. Gesimuleerde ET het bevredigend met werklike ET vergelyk, en dus die akkuraatheid van die model bevestig.

ENGLISH ABSTRACT: Generally only one or two sets of crop coefficients are used in combination with a reference crop evapotranspiration (ET) to estimate evapotranspiration (ET) of vineyards. Due to the variation among vineyards, these crop coefficients restrict the accuracy of estimating ET required by producers to manage irrigation to optimize yield as well as grape and wine quality. The aim of this study was to develop a water consumption model that could account for variation among vineyards. The heat pulse velocity technique was calibrated for measuring sap flow over short periods of time in grapevine trunks. A calibration curve of sap flux against time was developed. At least four probes were used per trunk to account for sap flow variability in xylem. Hourly sap flow was measured in selected vineyards under different conditions. Diurnal sap flow increased with leaf area per vine. Hourly sap flow did not increase linearly with net radiation, which suggested that maximum stomatal opening only allowed a fixed amount of transpiration. In some cases, sap flow also showed a temporary decrease during the day, which indicated a possible water saving mechanism resulting from stomatal closure at high light intensities. In comparison to non-irrigated grapevines, irrigation induced initial high sap flow peaks which indicated that, in addition to increased transpiration, turgidity was also regained. Eighty percent of variation in total diurnal sap flow could be explained by means of linear regression when only leaf area and ET were considered. Due to differences in amount of leaves exposed to direct net radiation, variation in sap flow was predicted more accurately by linear models for horizontal and vertical canopies, respectively. Since increase in shading with increase in leaf layers, cultivar characteristics and water stress effects were not accounted for, these models are regarded as a first approach. A Li-Car LAl-2000 Plant Canopy Analyzer (PCA) was calibrated to measure leaf area index (LAlpca) in selected vineyards. Although the PCA underestimated actual leaf area index (LAI) , the close correlation between LAIpca and actual LAI was used to measure leaf area development. Leaf area development was measured in eight vineyards varying in cultivar, vine spacing and trellising system in five grape growing regions. Seasonal leaf area development could be predicted by means of a third order polynomial equation using day of season as the independent variable. Potential growth curves were developed for the Summer and Winter Rainfall regions, respectively. Using cane mass to predict maximum leaf area was not as accurate as using leaf fresh mass. Horizontal canopies tended to produce more leaf area per unit cane mass in comparison to vertical canopies. Water stored daily in the above-ground parts of the grapevine only amounted to fractions of a millimetre, suggesting that water used for maintaining cell turgidity and physiological processes other than transpiration could be ignored in modelling water consumption. Evaporation losses from the soil surface (E) were measured under grapevine canopies by means of mini-lysimeters to evaluate and adapt the Boesten & Stroosnijder evaporation model. Under wet soil conditions, E tended to be higher in the middle section of the work row. Evaporation from unmulched, untilled soil was estimated with acceptable accuracy by the Boesten & Stroosnijder model. Since this model was initially developed for bare, fallow soils, some adaptations were necessary to account for canopy shading effects. During stage two evaporation, E tended to be higher under a horizontal trellis in comparison to a vertical trellis. Furthermore, E differed between the six soil types used in this study. Mulching reduced E significantly under relatively wet soil conditions. For most soils there were no difference between E for unmulched and mulched soils ten days after irrigation. Cumulative E from mulched soil correlated linearly with cumulative ET and generally amounted to approximately 30 % of cumulative ET. A combination of the transpiration and evaporation models were used as basis to design a draft water consumption model to estimate evapotranspiration of individual vineyards. Simulated ET was compared to actual ET measured for eight vineyards under different conditions. These vineyards represented various sets of variables which included irrigation system, soil type, soil water depletion level, trellising system and vigour. Only inputs such as cane mass, vine spacing, canopy surface orientation, reference crop evapotranspiration data as well as a constant value, which determines the amount of evaporation from a specific soil type, were required for the estimation of ET. The accuracy of the model was verified satisfactorily by simulation of measured ET.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/56049
This item appears in the following collections: