Temporal and spatial analyses of temperature in a French wine-producing area: the Loire Valley


C. Bonnefoy, COSTEL Laboratory, UMR6554 LETG CNRS, University of Rennes 2, place du Recteur Henri le Moal 35043 Rennes Cedex, France. E-mail: cyril.bonnefoy@univ-rennes2.fr


Global warming is causing earlier phenological dates for vines and changes in the quality of wines all around the world. To understand how vines are liable to react to future climate evolution, the climate of viticultural areas needs to be known as accurately as possible. In this article, temporal evolution and spatial variability of temperature in the wine producing area of the Loire Valley were analysed with a multi-scalar approach. First, the regional evolution of temperature and bioclimatic indices were studied over the past 60 years in the wine-producing area of the Loire Valley and showed a general warming at all locations in the regional weather station network (Météo-France). Secondly, temperature data from weather stations located within the vineyards between Angers and Saumur were studied for the year 2010 and underlined spatial variability of temperature between the different plots according to the different topoclimates. Finally, particular attention was paid to the Coteaux du Layon vineyards (sweet wine producing appellation) where 21 temperature data loggers were set up in the vine rows to study climate at local scale. The study showed, in particular, that the spatial variability of temperature and bioclimatic indices in these vineyards was greater than those observed at larger scales. Copyright © 2012 Royal Meteorological Society

1. Introduction

Global climate change impacts agriculture and our ecosystems (Rosenzweig and Hillel, 1998; Reddy and Hodges, 2000; Soussana, 2001). In Europe, foresight studies predict drought to become more frequent in the future, frost events to become rarer and the warm season to become longer (Parry, 2000; Schultz, 2000; Bolle 2003; Stock et al., 2003; IPCC, 2007). Extreme heat, as experienced by Western Europe in 2003, is expected to occur more frequently (Beniston, 2004; Meehl and Tebaldi, 2004; IPCC, 2007). Some consequences have already been observed in different regions worldwide. Frost decrease has been associated with an increase in yields and better crop quality (Harrison et al., 2000; Nemani et al., 2001). Sirotenko et al. (1997) observed greater yield variability in Eastern Europe.

Vineyards are particularly sensitive to climate and especially temperature (Pouget, 1968; Buttrose and Hale, 1973; Gerbier and Remois, 1977). That is why phenological vine data has been used to reconstruct temperature series over long past periods (Le Roy Ladurie, 1971; Chuine et al., 2004). García de Cortázar Atauri et al. (2005) created a predictive model of budbreak based on temperature. Many temperature-related bioclimatic indices have been developed for viticulture (Amerine and Winkler, 1944; Gladstones, 1992, Due et al., 1993; Riou, 1994; Huglin and Schneider, 1998).

Due to climate change, the migration of suitable wine-growing areas to the north of Europe has already been observed, especially in England where surfaces covered with vineyards are increasing (Kenny and Shao, 1992; Reilly et al. 1996; Palutikof, 2000). The phenological cycle of vines has also been affected by climate change. In Europe, the different phenological events occur 6–25 d earlier than 30–50 years ago and the duration between the different phenological events has shortened by 14–17 d (Jones, 2006; Ramos et al., 2008).

More specifically in France, Delécolle et al. (1999) and Seguin (2002) showed a significant earliness from three weeks (Bordeaux) to one month (Colmar) in the flowering date of the Chasselas grapevine variety. Phenological observations in Montreuil-Bellay (Loire Valley wine-producing area) have shown earlier harvest dates by 10–20 d since 1976 (Barbeau, 2007). In Alsace (Eastern France), phenological stages occur 15–23 d earlier than in the 1940s (Duchêne and Schneider, 2004). At the same time, the duration between flowering and ripening shortened by 8 d. In Beaune (Burgundy), the shortening of the period from flowering to harvest was significantly associated with an increase in the maximum temperature in August from 1990 (Bonnardot, 1996) and harvest dates were two weeks earlier at the end of the 20th century than in the years 1960–1970 (Madelin et al., 2008). In Bordeaux, the growing season shortened by 13 d over the period 1952–1997 (Jones and Davis, 2000). Temperature increases have also been observed in the Languedoc, Bordeaux and Provence wine producing regions (Tondut et al., 2006; Bois and Van Leeuwen, 2008; Bridier and Quénol, 2010).

Climate change is not homogenous (Pielke et al., 2002) and viticulture is not affected in the same way everywhere. Grapevine growers will have to adapt their cultural practices according to their regions' climate. Many studies stressed climate variability within a wine-growing region and showed probable future migrations of vineyards to cooler coastal climates like in Australia (Webb et al., 2007, Hall and Jones, 2010), New Zealand (Salinger, 1987), South Africa (Bonnardot and Carey, 2008; Bonnardot et al., 2011) or North America (Lobell et al., 2006; White et al., 2006; Jones et al., 2010). However regional scale climate studies may not be sufficient. As vineyards are sensitive to their environment (slopes, aspects, proximity of a river …), additional local climate investigation should provide practical information for decision making at plot level. As yet, few studies have dealt with fine scale analysis of climate in vineyards. Using the 1971–2000 PRISM 400 meters resolution climate grids, Jones (2010) spatialized bioclimatic indices throughout the American Viticultural Areas in the Western United States. South African Wine of Origin districts are also well endowed with weather stations. Various terroir studies undertaken in the Stellenbosch district have shown significant mesoclimatic differences over short distances due to the proximity of the Atlantic Ocean and complex topography (Carey, 2001; Conradie et al., 2002; Hunter and Bonnardot, 2002). In France and thanks to a very fine network of weather stations, Madelin (2004) emphasized some frost risk areas in the Champagne vineyards. As methods for the spatialization of temperature using fine climatic networks, Geographic Information Systems have been developed (Carrega, 1994; Fury and Joly, 1995) and applied to various wine growing regions (Hurnard, 1982; Quénol et al. 2004; Madelin and Beltrando, 2005; Sturman and Tapper, 2006; Bois, 2007; Bonnardot et al., 2012). Spatialization of temperature by means of meso-scale atmospheric modelling is also used over wine regions (Bonnardot et al., 2002, 2005; Bonnefoy et al., 2009).

In this article, temperature in the Loire Valley wine producing area was analysed with a multi-scalar approach. The Loire Valley is the 3rd largest French wine producing area (InterLoire, 2011) and stretches from Nantes in the West to Sancerre in the East. Scale interlocking is necessary to conduct such a study. Knowledge of the regional climate helps us to understand the general climatic context of wine producing areas but finer climatic networks are also essential to improve our understanding of different local climates in vineyards. The temperature evolution in the Loire Valley has therefore been analysed since the middle of the 20th century, using the regional weather station network (Météo France). Other weather stations were then used to refine our analysis thanks to the TERVICLIM program (ANR-JC 07–194103) and the National Institute of Agronomic Research (INRA) network. Data from these stations installed in the mid Loire Valley vineyards between Angers and Saumur were analysed for the year 2010. The climate of this area is described as a temperate oceanic climate with slight differences between minimum and maximum temperature. Rainfall is relatively low with less than 600 mm year−1. Cabernet Franc, Chenin Blanc and Gamay grape varieties are mainly cultivated and the global volume of wine production is around 700,000 hl with 25% exported (Vaudour, 2003). Finally, a local scale study of the Coteaux du Layon vineyards, located south-west of Angers, was undertaken. This appellation area covers around 2000 ha and vineyards are located on the slopes surrounding the Layon River running from the South of Maine-et-Loire towards the Loire River in the north (Barbeau et al., 1999). Chenin Blanc is used in this area to produce sweet wines after over-maturation of grapes. This grape variety is cultivated on fine soils (40–50 cm) with a poor hydric balance. Several temperature sensors (TERVICLIM program) were set up in one of the most famous appellations (Quarts-de-Chaume) in order to improve our understanding of the local climate, which allows the production of these renowned sweet wines. Increased knowledge of this local scale climate will also help in understanding the future impacts of global climate change over this wine-appellation region and which terroirs that could be initially less impacted by climate change.

2. Data and methods

2.1. Regional analysis of temperature

Data from the regional Météo France network were used to characterize the temperature evolution in the North West Central area of France and especially in the Loire Valley (Figure 1) since the mid 20th century. Annual data were calculated from monthly data then analysed for the 1946–2010 period for Nantes, Tours, Chateauroux, Bourges, Poitiers, Le Mans and Orleans, 1947–2010 for Angers, 1951–2010 for Saumur and 1953–2010 for Romorantin. According to Meteo-France metadata there was no change in the measurement locations during these periods, except for in Tours, where the measurement location was moved from an urban to a suburban area in 1964. This move resulted in a slight cooling from the end of the sixties onwards. The statistical Pettitt test was applied to these datasets over the chosen period of 1953–2010, in order to detect possible ruptures in the series (Pettitt, 1979). This test has been largely used to study changes of stable conditions in rainfall and temperature datasets (Demarée, 1990; Sutherland et al., 1991; Vannitsmen and Demarée, 1991; Dikbas et al., 2010). The test allows the detection of possible ruptures in datasets at each moment (t) from 1 to n, by comparing the two distributions on both sides. Therefore, this test is really sensitive to any change in both distributions means and more particularly in the middle of a time-serie (Costa and Soares, 2009). The robustness of this test has notably been proved by Lubès et al. (1994) and Lubès-Niel et al. (1998) through hydrological applications. In this article and in case of ruptures, pre- and post rupture trends were calculated for different weather stations by subtracting the means of the both distributions. The same number of years was conserved for the two periods in order to correctly assess the evolution of temperature. The significance of these trends was tested with the Mann–Whitney test that calculates if the means of two distributions are significantly different or not.

Figure 1.

Studied area: map of the Loire Valley showing the locations of vineyards and weather stations

Daily data for Nantes, Angers and Saumur is available for between January 1950 and September 2010 and between January 1960 and September 2010 for Tours. The evolution of different bioclimatic indices was studied (Table 1). Growing degree days (GDD) were calculated for each year during the phenological season (April to October): GDD = ∑Tmdaily − 10 with Tmdaily (daily mean temperature) > 10 (Winkler et al., 1974). This index, also named Winker index, is widely studied because correlations between this index and the different phenological stages are statistically significant. Huglin index (Huglin and Schneider, 1998) was also calculated for each year during the April-September period: IH = ∑[(Tmdaily − 10) + (Txdaily − 10)/2] × k (with k day length coefficient and Txdaily maximum temperature). This index is usually used as an indicator of sugar content in grapes. Both indices help in classifying the stations according to their climates types for viticulture (Table 1). Variability and trends were analysed over the 60 year-period and the Pettitt test was applied to underline eventual ruptures in the datasets. The cool night index (Tonietto, 1999; Tonietto and Carbonneau, 2004) was also studied. Normally, this index represents the mean minimum temperature in September (ripening period of grapes) in the Northern Hemisphere, but in this article the mean minimum temperature between mid-August and mid-September was used, since global warming causes precocity in phenological stages. Indeed, Barbeau (2007) showed the average harvest date in the Loire Valley was henceforth around September 15th. The cool night index classifies nights into different categories from the coldest to the warmest (Tab 1). Cool nighttime temperatures during ripening promote the biosynthesis of anthocyanins and abscisic acids (Yamane et al., 2006; Koshita et al., 2007), giving the grapes their skin color and the wine its quality.

Table 1. Bioclimatic indices for vine
Winkler indexHuglin indexCool night index
Region °C °FCharacteristicRankRangeCharacteristicsRankRange
  1. Source: adapted from Vaudour (2003) and Tonietto and Carbonneau (2003) ©UMR 6554 LETG.

 Very warmIH + 3IH > 3000 
V> 2205≥ 4000WarmIH + 22400≤ IH ≤3000Warm nightsIF1IF > 18 °C
IV1927 to 22053501 to 4000Warm temperateIH + 12100≤ IH ≤2400Temperate nightsIF216 °C ≤ IF < 18 °C
III1650 to 19263001 to 3500TemperateIH-11800≤ IH ≤2100Cool nightsIF314 °C ≤ IF < 16 °C
II1371 to 16492501 to 3000CoolIH-21500≤ IH ≤1800Very cool nightsIF412 °C ≤ IF < 14 °C
I< 1371< 2500Very coolIH-3IH ≤1500Cold nightsIF5IF ≤12 °C

2.2. Topoclimatic analysis of temperature

A weather stations network was set up in the Terroirs d'Anjou (between Angers and Saumur) in 2008 through the Terviclim program and the INRA. This network (Figure 2) enables the local climate of these terroirs to be studied. These automated weather systems (Campbell Sci. Inc., Logan Utah) record temperature, humidity, wind direction and speed, global radiation and rainfall every 15 min. The temperature probes (CS215) are all located in natural ventilated shelters (T351 RS) 1.5 m above the soil surface. In this article, the minimum and maximum daily temperature data during the 2010 growing season (April 1st–October 31st) was studied. A hierarchical ascendant classification (Ward method) has been applied to these datasets to possibly distinguish statistically different behaviors of temperature between the stations. Huglin and Winkler indices were also calculated for the nine stations over this growing season. The cool night index for the same stations was calculated over the period from mid-August to mid-September.

Figure 2.

Terviclim/INRA weather stations network in Anjou (a) and data loggers network in the Coteaux du Layon vineyards (b)

2.3. Temperature analysis in the Coteaux du Layon vineyards

During April 2009, twenty one data loggers located in a natural ventilated shelter (RS3 solar radiation shield, Onset®, MA, USA) were installed on vine rows in the Coteaux du Layon vineyards (Figure 2) in order to assess the spatial variability of temperature. These sensors are named Tinytag Talk 2 (Gemini Data Loggers Ltd., UK) and made up with a temperature logger and a thermistor probe (10K NTC). This probe records temperature from − 40 °C to 125 °C with an accuracy of 0.4 °C and a resolution of 0.05 °C or better. They have been programmed to record inside canopy temperature every 15 min, 1 m above the soil surface. For these reasons and compared with a traditional weather station, temperature can be slightly underestimated during nighttime and in some cases overestimated during daytime. Sensors were located according to different slopes, elevations, aspects and basic terroir units (Morlat and Bodin, 2006) in one of the most famous appellations (Quarts-de-Chaume) where sweet wines are produced after the development of Botrytis cinerea (Barbeau et al., 1999). Thermal conditions which promote the development of Botrytis cinerea were notably studied.

Minimum and maximum temperature data were analysed for the year 2009, as a result of too many missing values at several data loggers in 2010. Temperature evolution for the growing season (April to October) was presented for the five most contrasted loggers selected thanks to a hierarchical ascendant classification. The growing season, usually from April 1st to October 31st, was shortened by 17 d at the beginning of the 2009 season due to the last logger being installed on 18th April 2009. Temperature has also been analysed during one frost event in March 2010 and one heat wave in August 2009. Indeed, spring frost could become a real issue for winegrowers. Winter temperature would be milder and as a consequence budbreak would be earlier in spring. Therefore buds would be more vulnerable to a frost event given their advanced developing state. During hot days, temperatures above the threshold of 35 °C can also be hazardous for vine, and cause berries warming and drying (Crespy, 1992; Dokoolzian and Bergqvist, 2001). Winker and Huglin indices were calculated over the same period for 19 data loggers. Indeed, some daily data (less than 10 d) were missing for the sensors ROC_tt6 and BEA_tt9. These indices were not compared to those calculated at the regional scale but they show interesting micro-variability of climate at fine scale. The cool night index was also calculated using these data over the period from mid-August to mid-September.

3. Regional aspects of global warming on temperature and bioclimatic indices

3.1. A general temperature increase since the 1960s

The Loire Valley and generally Western-Central France has a temperate oceanic climate (‘Cfb’ type of climate according to the Köppen's classification). The studied area is entirely subject to Northern Atlantic weather systems, but at a regional scale, the continentality causes changes in space-time distribution of climatic features between coastal and inland areas (Bonnefoy et al., 2010). Thus, the annual and daily increase in temperature range from the coast inland and the weather station of Nantes close to the Loire estuary records the lowest temperature ranges among the selected weather stations in Western-Central France.

Temperature has been increasing since the middle of the twentieth century and especially since the end of the eighties (Figure 3). Moreover, the statistical Pettitt test reveals some ruptures (at 99% significance level) in the temperature series. The year 1987 was determined as a rupture in the maximum temperature series at all stations with an increase of 0.8 °C in Nantes and 1.3 °C in Saumur (Table 2). On the other hand, the rupture varies in the minimum temperature series at the stations from 1980 in Nantes and Angers to 1993 in Poitiers. The increase in minimum temperature varies between 0.6 °C in Poitiers and 1.2 °C in Le Mans. All of these trends are significant (at 99% significance level) according to the Mann–Whitney test. Despite its relatively flat topography (low plateaus), the lower and middle Loire River basin is characterized by internal climatic differences (Pédelaborde, 1957; Pagney, 1988; Planchon, 1997). The combination of geographical features causing these differences may affect vine growing. Indeed, the proximity of a river, a lake or the position of a station in a valley or in a plateau directly impacts on climatic parameters. As vine responds pretty well to climatic conditions, plant growing is affected by earlier or later phenological stages, frost vulnerability, diseases, etc. That is also why wine makers have adapted their cultural practices and chosen the right varieties of vine according to the environment (soils, climate, slopes aspect …).

Figure 3.

Evolution of Temperatures in the North-West Central France between 1948 and 2010

Table 2. Climatic ruptures in North-West Central France and temperature trends (° C) between pre- and post-rupture periods (Tn: minimum temperature/Tx: maximum temperature/Tm: mean temperature). Lowest and highest trends are indicated in bold
 Tm ruptureTm trendTn ruptureTn trendTx ruptureTx trend
  1. Data: Météo France ©UMR 6554 LETG

Le Mans19871.219801.219871.2

All the regional weather stations used are located in the main cities of this region. If they do not match exactly with the climate under where vineyards grow up, they reflect the general evolution of climate in this wine-making region. Besides, all the stations used respond to the World Meteorological Organization norms and as a consequence, all data sets are protected from an eventual Urban Heat Island effect. Nevertheless, we checked the differences in temperature for the season 2010 between the two synoptic stations of Angers and Saumur and respectively two other stations (Epiré and Souzay) from the Terviclim and INRA network, which are their rural homologous. According to the Mann–Whitney test, there are no significant differences between the ‘urban’ and ‘rural’ weather stations regarding the minimum temperatures. If we look at the maximum temperatures, there are no significant differences between Saumur and its rural homologous (Souzay) but differences are significant between Angers and Epiré. These significant differences could be due to the south aspect of the slopes where is located Epiré which explains higher maximum temperatures at this station.

3.2. Bioclimatic indices: a shift to upper levels of each class

Concerning the Winkler index and according to the statistical Pettitt test, statistical ruptures were found in the datasets and fall in 1986 for Angers and Saumur but in 1981 for Nantes and Tours. Consequently, the Winkler index increased significantly between the pre- and post-rupture periods (Figure 4). The index rose by between 167 GDD in Tours and 239 GDD in Saumur. As a consequence, Saumur is now classified in the Viticultural Region II (1371–1649 GDD) according to the Winkler classification (Table 1). This corresponds initially to climates such as those of Napa (USA), Budapest (Hungary) or Santiago (Chile) (Vaudour, 2003). The other stations fall into Region I (< 1371 GDD) where Dijon (France), Bordeaux (France) or Coonawara (Australia) can be found. The Huglin index also increased during the same period (Figure 5). The statistical Pettitt test shows a rupture (significant rate > 99%) in 1981 in Nantes, 1986 in Saumur and 1988 in Angers and Tours. These stations used to be classified as cool climate areas before the rupture but now they tend towards a temperate climate, as previously observed in Toulouse or Bordeaux in France. The climate of Saumur was temperate before the rupture and can now be considered as a warm temperate climate, as was the case in Montpellier (France), Santiago (Chile), or Madrid (Spain), 30 years ago (Vaudour, 2003). The increase of these indices may have implications for grapes composition and phenology as they are strongly linked with. Each grape variety needs a specific amount of heat to bud, bloom or ripen (Van Leeuwen et al., 2008). With global change, these amounts could be reached earlier in the season and it would mean earlier phenological stages for vine. Taking into account the climatic projections (IPCC, 2007), some varieties will need to be substituted by more adapted ones (later grape varieties).

Figure 4.

Evolution of GDD at four weather stations (a: Nantes, b: Angers, c:Saumur, d: Tours) in the Loire Valley (France) over two periods, before and after the climatic ruptures

Figure 5.

Evolution of the Huglin index at four weather stations (a: Nantes, b: Angers, c: Saumur, d: Tours) in the Loire Valley (France) over 1950/1960 and 2010

With respect to the cool night index, Figure 6 shows the percentage of each class for the thirty-year periods of 1961–1990, 1971–2000 and 1981–2010. Based on the 50-year long-term mean minimum temperatures (1961–2010), the mildest nocturnal conditions were observed in Nantes (13 °C) and the coolest in Tours (12.4 °C), although both values fall into the same category (Table 1). The percentage with cold conditions at night has decreased since the 1961–1990 period. The greatest decrease was observed in Nantes (1961–1990: 30%, 1971–2000: 20%, 1981–2010: 6.7%). Meanwhile, the percentage with very cool conditions at night increased everywhere, except in Saumur where minimum temperature seems to have risen faster. As a consequence, the cool nights category (1961–1990: 10%, 1971–2000: 20%, 1981–2010: 33.3%) in Saumur rose more than the very cool nights category. These cool nights became more common in the other stations, except in Tours (continental influences) where the percentage remained stable. In general, fewer cold nights (< 12 °C) are observed during this ripening period. The cool night indices remain in colder categories in the Loire Valley but if the temperature continues to increase, night thermal conditions during the ripening period of grapes would be warmer which would modify abscisic acids and anthocyanins synthesis. Cooler conditions at night and higher differences between daytime and nighttime temperature are beneficial for this synthesis. As a consequence if these conditions would change, characteristics of Loire Valley wines would change.

Figure 6.

Evolution of the cool night index at four weather stations in the Loire Valley over different periods

However, since all bioclimatic indices involve thresholds, all these results have to be taken with precaution.

4. Analysis of temperature in the Terroirs d'Anjou

4.1. Growing season temperature analysis

Slight differences in temperature are observed between the nine weather stations set up in the vineyards of Anjou; however, these differences emphasize their geographic situation. The dendrogram (Figure 7) of the hierarchical ascendant classification, applied to the 2010 vegetative season datasets, shows 53% of the total variance of minimum temperatures and 77% of the total variance of maximum temperatures are explained respectively by four and three classes. Concerning minimum temperature, the weather stations of Faye d'Anjou, Beaulieu and Cléré located on plateaus (elevation > 70 m and slope < 3%) fall in the first class. This class is characterized (Table 3) by a moderate range of temperature and a seasonal mean minimum temperature of 10.5 °C. The second class is composed of all the stations located in south of Angers, close to and influenced by the Loire River (Haute-Perche/La Marre Lalande/Brissac). This class is also characterized by a low temperature range and one of the highest seasonal mean minimum temperatures (11 °C), all these stations benefitting from the mild breezes coming from the Loire River. Chaume shows up as a single third class, the only one at a valley floor with the lowest temperatures through the area. The weather stations of St Cyr-en-Bourg and Souzay are classified in the fourth class since they are located in the area of Saumur, which is usually a warm part of the region. Concerning maximum temperature, Cléré and Beaulieu are in the first class but the station of Faye d'Anjou falls in the second class with the stations of Haute Perche, La Marre Lalande and Brissac. Table reveals that those two first classes are relatively close with moderate temperatures influenced by higher elevations for some stations and by the Loire River proximity for others. The third class is composed of the warmest locations: Chaume, according to the position of the weather station at the floor of a valley, St Cyr-en-Bourg and Souzay, corresponding to the warmest area by Saumur. This analysis shows differences in temperatures behavior according to the environment of the station and implies potential differences for vine-growing during the phenological season.

Figure 7.

Dendrogram of the hierarchical ascendant classification applied on the 2010 vegetative season temperature datasets (a: minimum temperature, b: maximum temperature) of the Terviclim/INRA weather stations network in Anjou

Table 3. Classes description of the hierarchical ascendant classification of temperature applied on nine weather stations datasets from the Terviclim/INRA network in Anjou (season 2010)
Minimum temperatures (° C)Maximum temperatures (° C)
 1st Class2nd Class3rd Class4th Class 1st Class2nd Class3rd Class
  1. Data: Terviclim/INRA ©UMR 6554 LETG.

Minimum0.51.3− 1.21.1Minimum10.910.511.6
1st quartile7. quartile18.918.920.0
3rd quartile13.714.113.014.13rd quartile26.326.227.4

4.2. Bioclimatic indices analysis: two climatic zones represented

Concerning the Winkler index, all the stations can be classified in region I (Table 4) except for Souzay and St Cyr-en-Bourg which are classified in region II (region of Saumur). According to the Huglin index, most of the stations experience a temperate climate, except for Beaulieu, Brissac and Faye d'Anjou which experience a cool climate. As a result of this index taking maximum temperature into account, the station of Chaume in the Coteaux du Layon area recording the highest maximum temperatures (foot of the valley) has one of the highest indices. Calculation of the cool night index reveals conditions above 12 °C everywhere, except in Chaume where night conditions are cooler (11.2 °C). On the contrary, this index is higher in Beaulieu (12 °C) situated on the plateau, although this station is only 1.5 kilometers away from the station of Chaume at the bottom of the Coteaux du Layon area close to the river. Differences concerning temperature and indices are observed between the different stations according to their position in relation to the Loire River, on a plateau, at the bottom or top of a slope … Temperature seems to be milder and show a reduced range between minimum and maximum temperature close to the Loire River. In the Coteaux du Layon area, a large difference between the station of Beaulieu and Chaume is recorded with a wide range of temperature in Chaume and less contrasted temperatures in Beaulieu. As a consequence, the climate of this area where sweet wines are produced deserves to be analysed in greater detail.

Table 4. Bioclimatic indices in nine weather stations from the Terviclim and INRA network in Anjou during the phenological season of 2010Thumbnail image of
  • Data: Terviclim/INRA network ©UMR 6554 LETG.

  • 5. Fine scale temperature analysis in the Coteaux du Layon area

    5.1. High spatial variability of temperature

    The 21 data loggers installed throughout the vineyards show important spatial temperature variability. Figure 8 illustrates the contrasts during the growing season of 2009 between five locations, representing the most different loggers according to the results of the hierarchical ascendant classification (results are not shown in this article). It appears that these loggers are located in the main different configurations which correspond to miscellaneous aspects, elevations and slopes. SLA_tt2 is located at the lowest elevation of the network and with a north aspect. ROC_tt6 has a south aspect and is at the bottom of a slope whereas BEA_tt1 and ROC_tt1 are at the top (plateaus). Lastly BEA_tt8 is at a mid-slope position with a south aspect. Many nights with thermal inversion are observed, especially with clear skies and radiative conditions. As a consequence, the lowest mean minimum temperature of the season (9.6 °C) is observed at the sensor SLA_tt2 (less elevated sensor, bottom of a slope). On the opposite, the sensors BEA_tt1 and ROC_tt1 record the highest mean minimum temperatures, respectively 10.7 °C and 11.1 °C. Thus, when the sky is overcast lowest minimum temperatures can also be recorded at high elevations (plateau, highest points of plots …). The mean seasonal maximum temperature varies from 23.6 °C at the highest point (BEA_tt1) to 24.2 °C at the lowest point (SLA_tt2) and at the south-facing mid-slope position of the sensor BEA_tt8. As a result, the maximum range of temperature is observed at the bottom of the plots and the minimum range at the top.

    Figure 8.

    Minimum (a) and maximum (b) temperature evolution (15 d moving average) for five representative data loggers in the Coteaux du Layon vineyards during the phenological season of 2009

    5.2. Spatial variability of temperature during two extreme events

    These differences of mean temperature between the sensors can be more marked during ‘extreme events’. In this study, two climatic events have been investigated: one frost event and one heat wave. The frost event was on 15 March 2010 and even if the budbreak did not begin at this point, it is interesting to understand how temperature reacts in that configuration. The heat-wave began on 15 August in 2009 and lasted 5 d with a slight cooling on 17 August. Temperature often exceeded the threshold of 35 °C in many stations.

    The map of 15 March 2010 with the ground level analysis over Europe (Figure 9) shows a high pressure system off the Brittany coast, conveying a cold north-western flow over France. The result was cooler temperatures than usual and even frost in most of the regions. In the Loire Valley, the synoptic stations of Angers and Saumur recorded respectively − 0.1 °C and − 1.2 °C. The other stations of the Terviclim/INRA network showed minimum temperatures between − 2.1 °C in St-Cyr-en-Bourg and 0.6 °C in Haute-Perche. This same night the spatial variability of temperature was even stronger at fine scale in the Coteaux du Layon (Figure 10) where temperature dropped from 1.3 °C to − 3.1 °C. The severest frost was observed at the bottom of the plots close to the Layon River and temperature remained positive at the top of the slopes and on the plateau. This situation is typical when the sky is really clear which can be really hurtful for vine, especially at the end of March and the beginning of April, during budbreak.

    Figure 9.

    Ground level analyses over Europe during one frost event (15 March 2010) and one hot day (15 August 2009)

    Figure 10.

    Temperatures ( °C) recorded by data loggers in the Coteaux du Layon vineyards during the frost event of 15 March 2010 (a) and the hot day of 15 August 2009 (b)

    The map of 15 August 2009 with the ground level analysis (Figure 9) shows a ridge of high pressure crossing Europe but also a low pressure system arriving from Spain and conveying tropical air all over the country. The Loire Valley is particularly kept under this advection of hot air coming directly from North Africa. This type of situation may occur every year and then, being this one a particularly intense event. The temperature reached 34.9 °C in Angers and 36.7 °C in Saumur. Once more and due to the contrasted topography, different aspects of the plots and characteristics of soils, the spatial variability of temperature was really marked at the scale of the Coteau du Layon with maximum temperatures between 35.9 °C and 40 °C. The temperature repartition was more complex but the highest temperatures were generally recorded at the bottom of plots with a south aspect, while lowest temperatures were recorded at some highest elevations on well-ventilated plots. The maximum temperature exceeded everywhere the threshold of 35 °C which can cause in some cases damages on grapes.

    5.3. Contrasting bioclimatic indices

    Initially used at a regional scale, Winkler and Huglin indices were calculated within the Coteaux du Layon vineyards in order to assess differences at local scale. Figure 11 shows both indices calculated for the 19 sensors where no daily data are missing. Within these vineyards, the GDD vary between 1184 and 1481. According to the classification of the regional Winkler index, most of locations would fall into Region I, which constitutes the coolest viticultural area in the world (850–1389 GDD). However, six of the 19 locations have values which would correspond to Region II (1389–1667 GDD). It shows how climate may vary at local scale. The highest indices are calculated in the middle slopes and/or south-facing slopes, while the lowest indices are calculated at high or low elevation where slopes are less marked and/or north-facing. Concerning the other index based on the Huglin calculation, it varies from 1953 to 2191. According to the regional Huglin index, almost all locations would fall into the ‘temperate climate’ group, while one would fall into the ‘warm temperate climate’ group (2100–2400). A difference of more than 200 units between the highest and the lowest value is considerable for this small area. The lowest indices are calculated for the highest locations and the highest indices for the lowest and/or south-facing locations. Indeed, the highest maximum temperatures are regularly observed in the lowest locations and the Huglin index calculation takes greater account of the maximum temperature than the Winkler index, which is only based on mean temperature.

    Figure 11.

    Bioclimatic indices for 19 data loggers in the Coteaux du Layon vineyards during the phenological season of 2009

    With respect to the cool night index, this index is correlated (R2 = 0.72) with elevation (Figure 12), due to frequent thermal inversions (Bonnefoy et al., 2009). As a consequence, the lowest index (9.6 °C) is recorded at the lowest location (at 28 m on a northern slope). On the contrary, nights are warmer at the top of the plots with an index above 12 °C beyond an altitude of 65 meters. As a consequence, nocturnal conditions can be classified into two different categories in the Coteaux du Layon vineyards: cold nights at the lowest points and very cool nights at the highest points. This coolness is very beneficial for vines and especially for grape maturation (Calame et al., 1977).

    Figure 12.

    Cool night index for 21 data loggers in the Coteaux du Layon vineyards in 2009 (calculated from mid-August to mid September) R2 = 0.72/Curve equation: y = 0.19 × exp(0.48x)

    6. Discussion and conclusion

    Evolution and spatial variability of temperature were studied within the wine-growing area of the Loire Valley. The evolution of temperature over the past 60 years in the Loire Valley revealed a general warming in this region. The calculation of bioclimatic indices for viticulture from Nantes to Tours showed a shift to upper levels, causing vineyards to grow in warmer conditions. As a consequence, vine phenology has already been affected and earlier harvest dates for many varieties have been observed in the Loire Valley (Barbeau, 2007). The maturity index has also risen and the average temperature of the growing season has almost reached the optimum temperature (Jones et al., 2004). A questionnaire of the perception of climate change distributed to many winegrowers from France, Germany and Italy showed professionals are aware of climate change and they have already observed change in quality of their wines (Battaglini et al., 2009). In France in the Bordeaux region, Loire Valley and Rhone Valley, most of winemakers have noticed an improvement in quality of wines for the past 10–20 years. To illustrate this quality enhancement, recently, the appellation Quarts-de-Chaume became the first Grand-Cru appellation of the Loire Valley.

    Regional scale analysis gave us the general climatic context in this area but was insufficient for such a study. Indeed, spatial variability can sometimes be as significant at fine scale as at larger scales. The same differences with bioclimatic indices were observed between the 21 data loggers within the 600 ha of vineyards (Coteaux du Layon) as between the 9 weather stations located in the region of Angers and Saumur. This observation showed influences of different local factors between vineyards or even inside a vineyard (elevation, aspect, position on the slope, proximity of a river…). Further study will aim at specifying which factors prevail in this variability. A multi-criteria model will be created in order to spatialise temperature in the Coteaux du Layon area during different events (spring frost, heat wave…), the final aim being to automate this model for the entire phenological season.

    This study helped to increase knowledge of the climate of the Loire Valley wine-growing area, especially those of the Terroirs d'Anjou and the Coteaux du Layon area at fine scale. The general climatic context is a warming for all the regional stations. This new insight into local scale climate in vineyards will enable a better understanding of how climate change is liable to affect vines and the phenological cycle. Climatic modeling will be carried out at fine scale (200 m resolution) and for different dates with the Regional Atmospheric Modeling System (RAMS). Results will be compared with our multi-criteria model to reveal the contributions and limits of both models. The final aim will be to run future modeling according to two scenarios (A2/B1) so as to fully understand impacts on viticulture.