A Technical Guide
for Wine Producers

RECENT ARTICLES   |   WYNBOER HOME

Effect of altitude and distance from the Atlantic Ocean on mean February temperatures in the Western Cape Coastal region


Philip Myburgh, ARC Infruitec-Nietvoorbij, Stellenbosch

Key words: Altitude, distance, ocean, temperature, coastal region

Abstract
Mean February temperature is one of the parameters used to select the most suitable locality for a specific wine grape cultivar. However, temperature data is not always available for all localities or vineyards. Decisions often depend on information gathered at the nearest weather station, which does not necessarily represent the situation at the specific locality. Processing existing data showed that the mean February temperature declines at a rate of ca. 0.5C with a 100 m increase in altitude and increases by ca. 0.6C per 10 kilometre increase in distance away from the ocean. If the altitude and distance from the Atlantic Ocean for a specific locality is known, it could be used in an elementary model to obtain an indication of the mean February temperature. However, factors such as slope and aspect, as well as the effects of topography on air flow and the occurrence of fog, may cause a varying degree of temperature deviations at specific localities.

Introduction

Air temperature is one of the most important atmospheric variables for viticulture. It not only plays a role in growth and development of the grapevine, but also affects juice composition, which in turn determines wine style and quality. Wine aroma in cultivars such as Sauvignon blanc (Marais et al., 1999; Carey et al., 2001) as well as Chardonnay and Cabernet Sauvignon (Carey et al., 2001), is affected by air temperature. Since low air temperature enhances the much sought-after vegetative aroma of Sauvignon blanc, there is an ongoing search for cool localities and, in particular those where mean air temperatures during ripening are lower than 21C. Harvest date (grape maturation rate) can also be influenced by air temperature (Bonnardot, 1997). Hence, the geographic position has to be considered carefully to ensure that vineyards are planted at localities which are suitable for that particular cultivar.


FIGURE 1: Study area and localities of weather stations. Numbers designate the weather stations listed in Table 2.

Various thermal variables, including temperature and cumulative heat units, are used to demarcate specific areas. There appears to be a good relationship between mean temperature and cumulative heat units (De Villiers et al., 1996 and references therein). Based on this, mean February temperature (MFT), i.e. during the ripening period for vineyards in the Western Cape Coastal region, has been suggested as a reliable parameter to distinguish between the thermal regimes of different localities (De Villiers et al., 1996). Subsequently, a map was produced which divided the Western Cape into 2C increment classes to demarcate five thermal regimes with respect to MFT. In order to compare thermal regimes between vineyards, or to select the ideal localities for new vineyards, air temperatures are increasingly being measured on a meso scale by means of automatic logging systems (J. Joubert & F. Knight, personal communications). Since at least ten years of data are considered necessary to provide accurate long-term atmospheric variables (Anonymous, 1989), whereas measuring air temperatures over only three or four seasons may be misleading. A possible way to solve this problem could be to determine the correlation between temperatures measured over the short term at a particular vineyard and those measured over the same period at the nearest weather station. If this correlation is within statistically acceptable limits, it could be used to estimate long-term mean temperature values at the particular vineyard or farm.

On average, air temperature in South Africa decreases at a rate of 0.3C per 100 m increase in altitude compared to 0.6C for Europe (Saayman, 1981 and references therein). The presence of large water bodies will also influence air temperature over the adjacent land via land-sea breeze circulation (Bonnardot et al., 2001). Compared to the stronger sea breezes occurring along the Namib desert, those in the Cape Peninsula are generally regarded as light westerly breezes that are not usually felt beyond five miles from the shore (Jackson, 1954). More recently it was shown that the land-sea breeze circulation during February on the northern side of False Bay in the Western Cape had an effect on air temperature above the adjacent land (Bonnardot et al., 2001). It was also concluded that the sea breeze could penetrate as far as 100 km inland from False Bay. A more detailed study revealed that the cooling effect of the sea breeze during February was prominent near the coastline, but that it declined rapidly with an increase in distance from False Bay (Bonnardot et al., 2003). In contrast, results presented by Conradie et al. (2003) indicated that there was no relationship between the distance from False Bay and MFT measured at five vineyards in the Durbanville/Stellenbosch districts over a period of seven years. In fact, their data showed a significant increase in MFT with an increase in the distance to Table Bay on the Atlantic seaboard. As expected, MFT decreased with an increase in altitude (Conradie et al., 2003). This suggested that the Atlantic Ocean might play a more prominent role in MFT in the Coastal region of the Western Cape than the land-sea breeze off False Bay.

The objective of this study was to determine whether MFT in the entire Coastal region of the Western Cape could be related to distance from the Atlantic Ocean, and to altitude.

Materials and methods

The study area stretched over ca. 265 km from Stellenbosch in the south to Lutzville in the north. In the east it is bordered by an almost continuous series of mountain ranges and in the west by the Atlantic Ocean (Fig. 1). The greatest east-west distance, i.e. from the mountain ranges to the Atlantic Ocean, is ca. 100 km. Data in Table 1, as presented by Conradie et al. (2003), were used to obtain a multiple linear regression mathematical model that could be used to estimate MFT at a specific locality as a function of altitude and distance to the Atlantic Ocean.


TABLE 1: Geographic details and mean February temperature for five weather stations in the Durbanville/Stellenbosch region (after Conradie et al., 2002).

To validate the model, predicted MFT values were compared to long-term ones measured at weather stations in the study area. For this purpose, altitudes and MFT for these weather stations were obtained from the long-term climate statistics for the Winter Rainfall Region (Anonymous, 1989). Of the 47 weather stations listed in the study area, only those with more than ten years' MFT data available were considered for validation of the model (Table 2). More recent data that were collected over the same period as the data used to calculate the above-mentioned model were not included in the validation data set. In doing so, an independent data set was used for the validation. The locations of the 25 selected weather stations are indicated in Figure 1. The distance to the Atlantic Ocean was measured on a road map of the Western Cape (Map Studio Productions, Cape Town). Since the weather stations in Table 1 were more or less in an easterly direction from Table Bay, distances between the Atlantic Ocean and the 25 weather stations were measured in a similar direction. Predicted MFT values were compared to actual MFT values by means of simple linear regression.


(Click image to enlarge)
TABLE 2. Geographic details, distance from the Atlantic Ocean (D), mean February temperature (MFT) and daily wind-run (W) for 25 weather stations in the Western Cape coastal region (n is number of years for which data were available).

Results and discussion

The equation obtained to estimate MFT from the data set for the five weather stations in the Durbanville/Stellenbosch districts as presented by Conradie et al. (2003), was as follows:

MFT = 20.85 - 0.0052341 x A + 0.06369 x D (R2 = 0.9983; se = 0.04C)

(Eq. 1)

where A is altitude (m) and D is distance from the Atlantic Ocean (km). The R-squared statistic indicated that A and D in equation 1 explained 99.8% of the variability in MFT at the 95% confidence level. According to the model, MFT at the coastline was estimated as ca. 20.9C. Furthermore, the model estimated that mean air temperature declined at a rate of ca. 0.5C with a 100 m increase in A. This indicated that air temperature in the Durbanville/Stellenbosch districts tended to decline more rapidly with altitude compared to the average of 0.3C per 100 m for South Africa, but tended to be slightly lower than the value for Europe (Saayman, 1981 and references therein). Air temperature was estimated to increase by ca. 0.6C per 10 kilometre increase in D away from the ocean. This increase was lower than the ca. 1.0C increase per 10 kilometre measured during February in a northerly direction from False Bay (Bonnardot et al., 2001). It was assumed that the relationship between MFT and D was linear. However, due to the strong cooling effect of the land-sea breeze near the coast (Bonnardot et al., 2003), this may actually not be the case. More weather stations near the coast would be necessary to determine a more realistic non-linear relationship between MFT and D.


TABLE 3: Mean February wind direction distribution measured at five weather stations in the Durbanville/Stellenbosch region from 1994 to 2003. Data supplied by ARC Institute for Soil, Climate and Water, Pretoria.

During February, the average wind direction at the five weather stations was westerly, i.e. from North-West to South-West, for 53% of the time (Table 3). With the exception of the weather station at Kuilsrivier, where the wind often tended to blow from the south, wind direction varied mostly between NW and SW at the other stations. The strong westerly component at Durbanville, i.e. the weather station closest to the Atlantic Ocean, was probably caused by land-sea breeze circulation. The observed trends in wind direction probably explain why there was a stronger relationship between MFT and distance to Table Bay compared to the distance to False Bay as in the data presented by Conradie et al. (2003).

Altitudes of the 25 weather stations in the study area varied between 31 m for Lutzville, and 755 m for Heldervue which is located on the Piketberg mountain (Fig. 1, Table 2). The mean altitude of all the weather stations was 190 m. February was the warmest month for all the weather stations, except for the weather station at Langgewens, where January was the warmest month in terms of mean temperature (data not shown). Mean February temperatures ranged between extremes of 18.9C and 25.0C as measured at the Nortier and Ideal Hill weather stations, respectively. The average MFT was 22.8C for the 25 weather stations, and below 21C at three of them. The actual MFT values correlated reasonably well with maximum February temperature (R2 = 0.925; s.e. = 0.59C) and to a lesser extent with minimum February temperature (R2 = 0.796; s.e. = 0.55C) (data not shown). However, MFT correlated poorly with the diurnal temperature fluctuation, i.e. the difference between maximum and minimum temperatures (R2 = 0.432; s.e. = 1.14C).

At the Nortier and Vredenburg Co-op weather stations, the low actual MFT was probably caused by the close proximity of the Atlantic Ocean (Bonnardot et al., 2003), whereas the effect of high altitude on air temperature was illustrated at the Heldervue weather station (Table 2). The MFT values did not follow a normal distribution, which suggested that the lower temperatures seemed to be the exception rather than the rule for the studied area. Due to historic land use patterns, most of the weather stations had been erected further away from the Atlantic Ocean and at altitudes below 200 m, i.e. in the areas where cereals, fruit and grapes are produced. This is probably a more realistic reason for the skew distribution of the MFT values in the study area.


FIGURE 2: Relationship between measured and predicted mean February temperature (MFT) for 25 weather stations in the Western Cape Coastal Region. Numbers designate the weather stations listed in Table 2.

When the predicted MFT values were compared to the actual long-term MFT measured at the 25 weather stations, the R-squared statistic indicated that equation 1 explained ca. 76% of the variability in MFT at the 95% confidence level (Fig. 2). The standard error of the relationship between actual and predicted MFT was ca. 0.8C. On a meso scale, i.e. in the vicinity of the weather stations, a number of factors could have contributed to the standard error of MFT estimation. In summer, horizontal, north-facing slopes receive considerably more incoming solar radiation than steep, south-facing ones (Wooldridge & Beukes, 2005). Other possible factors which might have caused error, are interference of airflow by the topography (Saayman, 1981), and the occurrence of fog that reduced incoming solar radiation (Williams et al., 1994)

An over estimation of MFT occurred in the case of the Porterville Co-op weather station, i.e. the one furthest inland from the Atlantic Ocean, if D is measured in a westerly direction (Fig. 1). The MFT measured over four years at Saron, i.e. 10 km south of Porterville, was only 0.5C higher, which confirmed that this was a relatively cool area. When D was measured in a north westerly direction, which is the shortest distance to the Atlantic Ocean, a more realistic estimate of MFT was obtained (data not shown). This suggested that the mountain ranges could have diverted cool air from the Atlantic Ocean towards the Porterville area. Although D was only three kilometres for the Nortier weather station, MFT was also over estimated (Fig. 2). Land-sea breeze circulation probably lowered the temperature at such close proximity to the Atlantic Ocean (Bonnardot et al., 2003). The occurrence of fog, which is common along the west coast of southern Africa (Olivier & Stockton, 1989), could also have decreased radiation, and consequently reduced air temperature at this particular weather station (Williams et al., 1994). However, sunshine hours per day were not measured at Nortier to allow comparison to other weather stations. Under estimation of MFT occurred at Klawer, Graafwater and Philadelphia (Fig. 2). Daily wind-run for these weather stations tended to be lower than the mean value of 192 km per day for the 25 weather stations (Table 2). Hence, it could be possible that the topography in the vicinity of these weather stations diverted or reduced airflow, thereby increasing the air temperature. In the case of the Klawer weather station, its location on a western aspect could also have caused MFT to be warmer than the predicted value.

These results showed that not only altitude, but also distance to the Atlantic Ocean in excess of 60 km had a significant effect on MFT in the study area. This suggested that significant air flow or land-sea breeze circulation occurred in a westerly direction. Westerly or south-westerly sea breezes during the afternoon are not uncommon in the Stellenbosch wine producing area (Bonnardot et al., 2001). Furthermore, the proximity of the Atlantic Ocean seemed to have had an effect on the MFT over longer distances compared to the 35 km reported for sea breezes around False Bay (Bonnardot et al., 2003).

Conclusions

The study showed that existing climatic data could be used to generate more information than that which is presently used to determine the best locality for a particular grapevine cultivar. Since altitude and distance from the Atlantic Ocean contributed significantly to mean February temperature in the Western Cape coastal region, these two geographical variables could be used in an elementary multiple linear regression model to estimate MFT. Hence, this model could be regarded as a first approach to quantify the thermal regime at a specific locality. Accurate geographical variables can easily be obtained by means of a road map and modern technology such as handheld global positioning systems. The model should be refined by future research to improve the accuracy of MFT prediction so that it can be used in combination with a geographical information system for demarcation of sites best suited for specific grapevine cultivars. The extent of the land-sea breeze circulation associated with the Atlantic Ocean should also be investigated by future research to refine the prediction of air temperature in the Western Cape coastal region. Since MFT does not necessarily provide a reliable indication of diurnal temperature fluctuation, it would also be useful if this parameter could be estimated accurately for a specific locality.

For more information contact Philip Myburgh at myburghp@arc.agric.za

References

Anonymous, 1989. Climate statistics for the Winter Rainfall Region. Section Agrometeorology, Elsenburg, Soil and Irrigation Research Institute, Dept. Agric. Water Supply. Private Bag X1, Elsenburg.

Bonnardot, V., 1997. Some climatic indices for Pinot noir maturation at a meteorological station in Burgundy. S. Afr. J. Enol. Vitic. 18 (1), 19-23.

Bonnardot, V., Carey, V.A., Planchon, O. & Cautenet, S., 2001. Sea breeze mechanism and observations of its effects in the Stellenbosch wine producing area. Wynboer, October 2001, 107-113.

Bonnardot, V., Planchon, O., Carey, V.A. & Cautenet, S., 2003. Diurnal wind, relative humidity and temperature variation in the Stellenbosch-Groot Drakenstein wine-growing area. S. Afr. J. Enol. Vitic. 23 (2), 62-71.

Carey, V.A., Bonnardot, V.M.F., Schmidt, A. & Theron, J.C.D., 2001. The interaction between vintage, vineyard site (mesoclimate) and wine aroma of Vitis vinifera L. cvs. Sauvignon blanc, Chardonnay and Cabernet Sauvignon in the Stellenbosch-Klein Drakenstein wine growing area, South Africa (1996-2000). In: Proc. 26th O.I.V. World Vine & Wine Congress, October 2001, Adelaide, Australia. pp. 139-152.

Conradie, W.J., Carey, V.A., Bonnardot, V., Saayman, D., & Van Schoor, L.H., 2002. Effect of different environmental factors on the performance of Sauvignon blanc grapevines in the Stellenbosch/Durbanville districts of South Africa. S. Afr. J. Enol. Vitic. 23 (2), 78-91.

De Villiers, F.S., Schmidt, A., Theron, J.C.D. & Taljaart, R., 1996. Onderverdeling van die Wes-Kaapse wynbougebiede volgens bestaande klimaatskriteria. Wynboer Tegnies, January 1996, 10-12.

Jackson, S.P., 1954. Sea breezes in South Africa. S. Afr. Geogr. J., 13-23.

Marais, J., Hunter, J.J. & Haasbroek, P.D., 1999. Effect of canopy micro-climate, season and region on Sauvignon blanc grape composition and wine quality. S. Afr. J. Enol. Vitic. 20 (1), 19-30.

Olivier, J. & Stockton, P.L., 1989. The influence of upwelling extent upon fog incidence at Luderitz, Southern Africa. Int. J. Climatol., 69-75.

Saayman, D., 1981. Klimaat, grond, en wingerdbougebiede. In: Burger, J. & Deist, J. (eds) Wingerdbou in Suid-Afrika, 48-66.

Williams, L.E., Dokoozlian, N.K. & Wample, R., 1994. Grape. In: Schaffer, B. & Anderson, P.C. (eds). Handbook of environmental physiology of fruit crops, Vol. I, Temperate crops. CRC Press, Boca Raton, 85-133.

Wooldridge. J. & Beukes, H., 2005. Radiant solar energy interception in the Western Cape. Wynboer, February 2005, 9-11.


Wynboer is incorporated in WineLand, magazine of the SA wine producers.

Subscribe to WineLand

Visit our sister sites:


South African wine farmers' representative organisation


Facts, figures, contact details and much more in the 2009/10 Directory

UP COPYRIGHT (C) 2000 WineLand