Spatial interpolation-based mapping of the spring frost hazard in the Champagne vineyards

Transcription

1 Meteorol. Appl. 12, (2005) doi: /s Spatial interpolation-based mapping of the spring frost hazard in the Champagne vineyards M. Madelin 1,2 & G. Beltrando 1,3 1 Université Paris VII, 2 place Jussieu, Paris Cedex 05, France 2 UMR Géographie-cités Equipe PARIS, 13 rue du Four, Paris, France 3 UMR PRODIG Equipe DYNMIRIS (CC7001), 2 place Jussieu, Paris Cedex 05, France malika.madelin@parisgeo.cnrs.fr The Champagne vineyards, famous worldwide, are located at the northern limit of viticulture. Consequently, they are very sensitive to spring frosts that can occur after bud burst. These spring frosts occur mainly in radiative atmospheric situations (low speed wind, clear sky) where spatial variations in minimum temperatures are very important. The latter depend on several factors at different scales: from the macroclimatic scale, where the geographical location (latitude/longitude) has a strong influence, to the topoclimatic scale, where environmental and topographical factors play an important role. After a brief recall of the definition of these different factors extracted from a Digital Elevation Model and a land-use database, we propose a spatial interpolation of minimum temperatures in the Champagne vineyards based on multiple regression. From this interpolation, we can map the frost hazard for the whole vineyard from data measured over five spring seasons ( ) and for about 20 weather stations. In this paper, we present this mapping technique and compare it with the data for spring 2003 when particularly strong frosts occurred. 1. Introduction Located at the northern limit of traditional winegrowing areas, the Champagne vineyards (Figure 1) are particularly sensitive to frost events that occur during spring after the breaking of the first buds. In some years, frost damage can affect large areas (during spring 2003, for example, more than 50% of the Champagne vineyards were destroyed by severe frosts). Owing to the great economic importance of these vineyards, a forecasting system that aims to predict minimum temperatures has been in use since 1980 at the Centre Météo-France Marne. Based on a statistical analysis of average temperature differences, it provides a short time estimate of temperatures at four points in the Marne vineyards (largest subset of the Champagne vineyards) from the forecasted temperature at Reims (reference station). Recently, thanks to the siting of new meteorological stations in the vineyard, this statistical forecasting system has been significantly improved (Madelin et al. 2002). However, during nights with clear skies and cool winds, it has been shown (Geiger, 1965; Mahrt 1986; Carrega 1995; Daly et al. 1994; Lindkvist et al. 2000; Halley et al. 2003) that important differences can be observed in the temperatures of stations just a few kilometres apart (up to 8 C in Champagne; Beltrando 1998). Such large differences are not taken into account by the present forecasting system. This spatial variability of temperatures reflects the importance of the local environment, especially the topography (Bouchet 1965; Kalma et al. 1992). In this paper, we present a spatial analysis of climatic data for the Marne vineyards that highlights the role of topographic factors in the spatial variability of minimum temperatures. First, we describe the climatic and topographic data used in this study. Then, we present the methods used to analyse these data, methods which mainly rely on a multiple regression-based interpolation. This allows the estimation of average minimum temperatures for any point in the Marne vineyards from the temperatures measured at 23 meteorological stations. Our results are presented in the form of a map which represents the average minimum temperatures for the whole of the Marne vineyards and could be implemented in a GIS geared to the needs of winegrowers (Burrough & McDonnell 2000). In the final section, we propose a mapping of frost risk estimates and compare it with the damage observed during spring

2 M. Madelin & G. Beltrando The climatic data used in this study are daily minimum temperatures measured during the springs of 1998 to 2003 (spring is defined as 15 March to 31 May) at 23 meteorological stations homogeneously dispersed through the Marne vineyards (Figure 2a). This network of measurement stations was set up in the 1990s in order to improve climatic knowledge of these vineyards, particularly the frost events. These stations belong to two institutions: the French national meteorological agency Météo-France ( and the Champagne wine growers association Comité Interprofessionnel du Vin de Champagne (CIVC, In order to study phenomena such as thermal inversions and cold air gravity flows, the stations are located at different heights on the hillsides (Figure 2a). Their precise geographic coordinates were measured by means of a GPS (Global Positioning System). Figure 1. Location of the Marne vineyard the main part of the Champagne vineyard, among the French vineyards (Source: CORINE Land Cover). 2. Data 2.1. Climatic data In addition to this temperature data, wind speed and cloud cover data at the Reims-Courcy weather station have also been used to define the atmospheric conditions each night Digital Elevation Model (DEM) To define the relevant topographic parameters, we used a Digital Elevation Model (DEM) provided by the Institut National Géographique (IGN). This model provides the altitude, with an altimetric resolution of 2 5 m, of each point in an area spreading over 40 km from west to east and over 60 km from north to south, with a planimetric resolution of 50 m. Figure 2b shows a map of the relief of the studied area based on an interpolation (triangulation method) of the DEM. The relief is relatively flat (maximum elevation: 287 m; minimum: 60 m). It is characterised by the valleys of the Marne and the Vesle (altitude < 80 m) and the plateau of Montagne de Reims (altitude > 220 m). The Figure 2. From left to right: (a) location of the meteorological stations, (b) land-use distribution (c) elevation map. 52

3 Mapping of spring frost hazard in Champagne vineyard clear sky: when three out of the four nightly observations (21h, 00h, 03h, 06h UTC) in Reims show cloud cover of less than 2 octats. cool wind: when three out of the four nightly windspeed observations (21h, 00h, 03h, 06h UTC) in Reims are less than 2 m s 1. Figure 3. The computation of the slope according to spatial windows of increasing sizes. south-eastern parts of the studied area is characterised by the Champagne Pouilleuse plain (altitude < 120 m) Land-use data In order to locate precisely the wine-growing areas included in our study area, we used vector land-use data from the BD Carto database provided by IGN. This database uses 11 land-use categories. Figure 2c shows the main land-use types in the studied area: forest (550 km 2 ) mainly on plateaus, vines (200 km 2 ) on the hillside above the plains, and other cultivation (1300 km 2 ). 3. Methods In order to establish a relationship between the daily minimum temperatures and the geographic (coordinates) and topographic (elevation, slope, aspect) data described in the previous section, we performed several statistical analyses. For each night, we computed a model for the variation of the weather station minimum temperature in two steps: polynomial regression with the geographic coordinates; stepwise linear multiple regression on the residuals of the polynomial regression with the topographic parameters. In order to take into account phenomena at different scales, the topographic parameters were computed according to several spatial windows (Joly et al. 2003; seefigure3). These models were then used to estimate the minimum temperature at each point of the DEM grid (according to its geographic coordinates and its topographic characteristics). Since the spatial variability of the minimum temperatures is particularly important for the nights with clear skies and cool wind conditions, we decided to limit this study to the set of those nights that satisfied this requirement. The criteria used to define clear sky and cool wind conditions (Madelin et al. 2002) are as follows: 4. Results and discussion We present in this section the results of our modelling. First, we show the mapping of the studied area in terms of the average estimated minimum temperatures over the whole set of selected nights. Then we focus on a particular night for which severe frosts were reported and compare the map of the minimum temperatures estimated using our model with a map representing the frost damage in the vineyards Mapping of the average estimated minimum temperature over the whole vineyard Figure 4 is a map of the average estimated minimum temperatures. It shows very schematically a thermal gradient from NW to SE. Looking closer, we observe a decrease in temperatures along the hillside, which reflects the significant factors of longitude, slope, etc.. However, it also shows the limits of the present model owing to the characteristics of the weather station network, namely: a spatial heterogeneity which induces border effects; location of the weather stations mainly on the vineyard hillsides: no information is therefore available for the eastern plain areas and for the forests above the Montagne de Reims. Owing to these limitations, it is necessary to restrict the estimation of minimum temperature to the winegrowing areas. In order to do so, we have thus created a mask hiding the non wine-growing areas (using the digital land-use database BD Carto in Figure 2b). Figure 5 shows a close-up of the area to the north-east of Montagne de Reims where, thanks to this mask, only the wine-growing areas are visible. This close-up shows that hillside bottoms are colder than tops owing to the flow of cold winds down the slope (with gravity) (Bouchet 1965; Mahrt 1986). The different orientations also seem to play a role in the repartition of the estimated minimum temperatures. Using this map, the wine growers can identify the most frost-sensitive areas under clear-sky and cool wind conditions, and decide whether or not to install frost protection systems. Note that similar maps can be produced for other meteorological conditions (for example, easterly advection of air masses). 53

4 M. Madelin & G. Beltrando Figure 4. Average minimum temperature repartition during nights with clear sky and cool wind conditions. a comparison (Figure 6b). The atmospheric conditions of this particularly cold night were characterised by an advection with airmasses coming from the east, after several days of clear sky and cool wind conditions (typical radiative conditions). Figure 5. Average minimum temperature in the NE area of the Montagne de Reims (clear sky, cool wind) Towards a comparison of meteorological and agricultural frosts We now focus on a particular night in the spring of 2003 (11 April) when severe frosts occurred in the vineyards, and compare a map of the observed frost damage with the map showing the miminum temperatures estimated using our model. Figure 6a shows the repartition of frost damage for spring 2003 as a per cent of the total wine growing area, based on the vineyard communes. Because this damage was observed mainly during the night of 11 April 2003, we used our model to draw a map of the minimum temperatures estimated for that same night as 54 The repartitions observed on these two maps are quite similar: the hillsides located to the east of the vineyard (Montagne de Reims; Côte des Blancs) encountered very low temperatures, unlike the Vallée de la Marne. However, if we look more carefully, we observe that the vineyard areas located to the north of Montagne de Reims are weakly affected by frosts (less than 32% of the area) even though the estimated temperatures are very low. This can be explained mainly by the type of vine repartition: these weakly damaged areas contain mostly Pinot Meunier which starts growing later than Chardonnay and Pinot Noir (the average start of the growing season for Chardonnay was 4 April 2003, for Pinot Noir 9 April 2003 and for Pinot Meunier 13 April 2003). Thus Pinot Meunier would have been less susceptible to this early frost event on 11 April. 5. Conclusion In this paper, we have established relationships between the minimum temperatures in the Marne vineyards and the geographic and topographic factors which play a role in their variation. This allowed us to build a map of the average estimated minimum temperatures across the whole vineyard area which can be used by wine growers to identify the frost sensitive areas. A comparison of the model estimations and the observed frost damage during spring 2003 shows a fairly good agreement between observed and predicted frost events. Nevertheless, the

5 Mapping of spring frost hazard in Champagne vineyard Figure 6. Comparison of meteorogical and agricultural frosts during spring discrepancy observed in the northern areas illustrates the difference between the agronomic frost (i.e. that affecting certain types of vines according to the start of their growing seasons) and meteorological frost (temperatures below a given threshold). Acknowledgements The authors gratefully acknowledge the data support of Comité Interprofessionnel du Vin de Champagne and Météo-France. References Beltrando, G. (1998) Les gelées printanières en Champagne viticole. Quelques résultats à partir d un nouveau réseau de stations automatiques. La Météorologie 8(21): Bouchet, R. J. (1965) Problèmes des gelées de printemps. Agricultural Meteorology 2(3): Burrough, P. A. & McDonnell, R. (2000) Principles of Geographical Information Systems. Oxford: Oxford University Press, 333pp. Carrega, P. (1995) A method for the reconstruction of mountain air temperatures with automatic cartographic applications. Theor. & Applied Climatol. 52: Daly, C., Neilson, R. P. & Phillips, D. L. (1994) A statistical topographic model for mapping climatological precipitation over mountainous terrain. J. Appl. Meteorol. 33(2): Geiger, R. (1965) The Climate Near the Ground. Cambridge, Mass.: Harvard University Press, 161pp. Halley, V., Eriksson, M. & Numez, M. (2003) Frost prevention and prediction of temperatures and cooling rates using GIS. Austr. Geogr. Stud. 41: 3: Joly, D., Nilsen, L., Fury, R., Elvebakk, A. & Brossard, T. (2003) Temperature interpolation at a large scale: test on a small area in Svalbard. Int. J. Climatol. 23(3):

6 M. Madelin & G. Beltrando Kalma,J.D.,Laughlin,G.P.,Caprio,J.M.&Hamer,P.J.C. (1992) Advances in Bioclimatology. Vol. 2: The Bioclimatology of Frost. Berlin: Springer-Verlag, 144pp. Lindkvist, L., Gustavsson, T. & Bogren, J. (2000) A frost assessment method for mountainous areas. Agric. & Forest Meteorol. 102: Madelin, M., Beltrando, G., Camblan, M. & Langellier, F. (2002) Amélioration de la prévision des gelées printanières dans le vignoble de la Marne en Champagne. Hommes et Terres du Nord 1: Mahrt, L. (1986) Nocturnal Topoclimatology, WCP-117, WMO/TD-No. 132, 76pp. 56