INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. 26: 303–321 (2006) Published online 13 February 2006 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/joc.1244

INFLUENCE OF SEASONAL PRESSURE PATTERNS ON TEMPORAL VARIABILITY OF VEGETATION ACTIVITY IN CENTRAL SIBERIA

2

SERGIO M. VICENTE-SERRANO,1, *† MANUELA GRIPPA,1 NICOLAS DELBART,1,2 THUY LE TOAN1 and LAURENT KERGOAT1 1 Centre d’Etudes Spatiales de la Biosph` ere (CESBIO), 18 avenue, Edouard Belin, bpi 2801, 31401 Toulouse cedex 9, France Frontier Research Center for Global Change, Ecosystems Change Research Group 3173-25 Showa machi, Kanazawa-ku, Yokohama, Kanagawa, 236-0001, Japan Received 16 August 2005 Revised 28 July 2005 Accepted 28 July 2005

ABSTRACT This paper analyses the spatial distribution of the inter-annual variability of vegetation activity in central Siberia and its relationship with atmospheric circulation variability. We used NOAA-AVHRR NDVI series from Pathfinder Land Data Set at 1° of spatial resolution, and we calculated the annual vegetation activity in each pixel (aNDVI) from 1982 to 2001. Principal component analysis (PCA) was used to determine the general spatial patterns of inter-annual variability of vegetation activity. We identified three main modes, which explain more than 50% of the total variance, each corresponding to a large region. By means of surface pressure grids, we analysed the main patterns of the seasonal atmospheric circulation in the study area: its variability was summarised by means of a few circulation modes and the patterns differ significantly between winter, spring and summer. However, a pattern with a North–South dipole structure represents the general spatial pattern of atmospheric circulation. We investigated the effect of seasonal atmospheric circulation patterns on the inter-annual variation of vegetation activity. In general, the strongest relationships between the atmospheric circulation variability, climate and the aNDVI variability were found in areas where the climatic characteristics are more limiting for the vegetation development, such as the northern regions. This may be explained by the fact that in these areas the variability of atmospheric circulation modes determines summer temperatures, which have a direct impact on vegetation activity. Copyright  2006 Royal Meteorological Society. KEY WORDS:

vegetation activity; NDVI; atmospheric circulation; teleconnections; principal component analysis; climate–vegetation relationships; Siberia

1. INTRODUCTION Several studies have provided evidence that climate is an important constraint to vegetation activity and its temporal variability (Schultz and Halpert, 1993; Nemani et al., 2003). In cold regions and in areas where moisture availability is sufficient for vegetation to develop, the main constraint is temperature (Braswell et al., 1997; Suzuki et al., 2000, 2001; Gong and Ho, 2003). At present, some serious questions arise about the possible consequences of climate change for the distribution of vegetation types and biomass production. The influence of climate on vegetation needs to be better observed and understood in order to develop possible strategies to mitigate the negative consequences of climate change. On a global scale, there is evidence of recent climate changes in terms of temperature increase (Houghton et al., 2001), increase of extreme events (Easterling et al., 2000) and changes in the main atmospheric circulation modes (Hurrell, 1995; Trenberth, 1997). The effects of these changes on vegetation * Correspondence to: Sergio M. Vicente-Serrano, Instituto Pirenaico de Ecolog´ıa, CSIC (Spanish Research Council), Campus de Aula Dei, P.O. Box 202, Zaragoza 50080, Spain; e-mail: [email protected] † Current address: Instituto Pirenaico de Ecolog´ıa, CSIC (Spanish Research Council), Campus de Aula Dei, P.O. Box 202, Zaragoza 50080, Spain

Copyright  2006 Royal Meteorological Society

304

S. M. VICENTE-SERRANO ET AL.

have been observed in large areas of the Northern Hemisphere, mainly as an increase of vegetation activity (Lucht et al., 2002; Slayback et al., 2003; Zhou et al., 2003) and an earlier onset of greening (Myneni et al., 1998). Different studies have analysed the influence of precipitation, snow cover and temperature on vegetation dynamics and biomass at high latitudes (Dye and Tucker, 2003; Nemani et al., 2003). Nevertheless, few papers have investigated the influence of atmospheric circulation on vegetation (i.e. Buermann et al., 2003; Gong and Ho, 2003). This may be important in order to better understand the effects of climate change on ecosystems. Most analyses of the influence of atmospheric circulation on vegetation activity have been focused on the impact of El Ni˜no Southern Oscillation (ENSO) (e.g. Eastman and Fulk, 1993; Kogan, 2000; SalinasZavala et al., 2002), which is one of the main sources of atmospheric circulation variability on a global scale (Philander, 1990; New et al., 2001). Kogan and Wei (2000) analysed the response of global vegetation to ENSO and showed that this atmospheric mode mainly affects tropical areas and some regions of the Southern Hemisphere. However, Buermann et al. (2003) also indicated some ENSO influence in some areas of the Northern Hemisphere. In the Northern Hemisphere, although ENSO affects atmospheric circulation, precipitation and temperature (Rogers, 1984; Ropelewski and Halpert, 1987; Fraedrich, 1994), other atmospheric modes dominate climate variability (Hurrell and Van Loon, 1997; Thompson et al., 2002). These modes are summarised by means of teleconnection indices such as the Arctic Oscillation (AO) (Thompson and Wallace, 1998), oHa 19OscillNation

variab

SEASONAL PRESSURE PATTERNS IN CENTRAL SIBERIA

Water Evergreen Needleleaf Forests Deciduous Needleleaf Forests Deciduous Broadleaf Forests Mixed Forests

305

Woodlands Grasslands (Steppe) Croplands Bare Mosses and Lichens

Figure 1. Location of the study area and land-cover map. Obtained from University of Maryland (http://www.geog.umd.edu/landcover/ 8km-map.html), DeFries et al. (1998)

as snow, the inter-annual variability of which is significantly influenced by atmospheric circulation and the ocean (Clark et al., 1999; Ye, 2000). The north-eastern areas are not only characterised by the presence of permafrost conditions (Malevsky-Malevich et al., 2001) but also by an important inter-annual variability of precipitation (Frauenfeld et al., 2004). Figure 1 shows the spatial distribution of land cover (DeFries et al., 1998). There is a clear organisation of vegetation in latitudinal belts created by the climatic characteristics (Suzuki et al., 2000; Zhang et al., 2004). Steppe and crops are dominant in the south-western areas characterised by higher temperatures and low precipitation. Different bands of forests are observed at latitudes of about 55–67 ° N. The forest types differ, depending on climatic characteristics, mainly temperatures, which determine their phenological cycles (Zhang et al., 2004). North of the forest belts are tundra areas, where the activity of vegetation, composed of mosses and lichens, is limited to the summer months, when snow cover disappears.

3. METHODS 3.1. Estimation of the spatial and temporal variability of vegetation from remote sensing Optical remote sensing has been widely used to analyse vegetation activity (Nicholson et al., 1990; Ringrose and Matheson, 1991; Kogan, 2001). Different satellite sensors provide spectral data of the terrestrial surface in different spatial and temporal resolutions. The principles for vegetation monitoring using this kind of sensor are based on the response of vegetation to radiation in the visible and near-infrared regions of the electromagnetic spectrum. Visible radiation is mainly absorbed by vegetation, while near-infrared radiation is primarily reflected (Knipling, 1970). These properties have made it possible to create vegetation indexes that have been used to derive the vegetation type and cover and to monitor the dynamics of vegetation activity (Kogan, 2001). The most widely used among these indexes is the normalised difference vegetation index (NDVI) (Tucker, 1979), which is calculated as NDVI = (infrared − red)/(infrared + red). Different NDVI global data series are available from AVHRR data (PAL and GIMMS NDVI) (Justice and Townshend, 1994, James and Kalluri, 1994). The PAL-NDVI database (http://daac.gsfc.nasa.gov) provides Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)

306

S. M. VICENTE-SERRANO ET AL.

data for every 10 days from 1981 to 2000. The homogenisation of the series has been checked with good results (Smith et al., 1997, Kaufmann et al., 2000). Although some errors in the computation of solar zenith angles (SZA) have been found (NASA, 2004), Kaufmann et al. (2000) concluded that the PAL-NDVI data set temporal homogeneity is not affected by SZA changes. Shabanov et al. (2002) also provided evidence that the NDVI trends recorded in some regions of the world are not related to errors in the data process, and Gong and Ho (2003) indicated that the NDVI-climate coupling modes are stable and detectable even when there are errors in the original NDVI data set. Here, we used the PAL-NDVI data set at 1° of spatial resolution and at temporal resolution of 10 days from 1982 to 2001. From this data set, we calculated the vegetation activity for each year, adapting the method used by Zhang et al. (2003) to determine phenological dates from the MODIS vegetation index. This method fits logistic functions y(t) to the PAL-NDVI annual time series in each pixel: y(t) =

c 1 + e(a+bt)

+d

(1)

where c + d is the maximum NDVI value and c the minimum value each year. One such function fits the ascending part of the NDVI cycle, and another one fits the descending part. This method permits the elimination of perturbations due to noise. To avoid snow effect at the early and late stages of the NDVI cycle, the cumulated aNDVI is taken as the sum of the two functions between the dates of maximum curvature on the fitted functions. Consequently, the NDVI cumulated in this way relates the two quantities that are important for this study: the duration of the activity season and the intensity of this activity, denoted by the maximum NDVI value. For each pixel of 1° of spatial resolution, we obtained a time series of 20 years that we used to investigate the inter-annual variability of vegetation activity in central Siberia. 3.2. Study of the spatial and temporal variability of vegetation activity by principal component analysis To summarise the inter-annual vegetation activity and identify the main modes of vegetation evolution, we used principal component analysis (PCA). The application of PCA to different time series in a specific area can be performed in spatial (S) or temporal (T) modes (Jollife, 1990; Serrano et al., 1999). The T-mode, where the time observations are the variables and the pixels the cases, permits the identification of the general spatial patterns of climatic variables. In contrast, the S-mode is used to determine the main patterns of the time series evolution. In remote sensing, PCA has usually been applied to summarise spectral data (Richards, 1993). However, it has also been used in the T-mode to classify land uses and land cover (Townshend, 1984) and in the S-mode to analyse vegetation variability or land use changes (Townshend et al., 1985; Fung and LeDrew, 1987; Eastman and Fulk, 1993). In this paper, we used PCA in the S-mode from the aNDVI time series of each pixel to identify the main modes of the variability of vegetation activity. Since in the study area the vegetation cover is highly variable in space, we standardised the original aNDVI values to avoid differences caused by environmental conditions or by vegetation type and coverage. The standardised series of NDVI have been widely used to detect spatio-temporal anomalies in vegetation activity (Maseli et al., 1992; Gutman and Ignatov, 1995; Hartmam et al., 2003). Peters et al. (2002) showed that long series of NDVI are close to a normal distribution and that the standardisation procedure is a good approach with which to normalise vegetation series in time and space. Consequently, previous to PCA, each series of 1° was standardised according to its mean and standard deviation. Finally, the components obtained from the PCA were rotated to redistribute the final explained variance. The rotation procedure makes it possible to obtain a clearer separation of components that maintain their orthogonality (Hair et al., 1998). We used the Varimax rotation (Kaiser, 1958), which is the most widely applied option because it produces more stable and physically robust patterns (Richman, 1986; Ye, 2002; Kadioglu, 2000; Serrano et al., 1999). The land-cover map by DeFries et al. (1998) (Figure 1) was used to calculate the percentage of each land cover that corresponds to the representative areas of each component according to significant rotated empirical orthogonal functions (REOF). Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)

SEASONAL PRESSURE PATTERNS IN CENTRAL SIBERIA

307

3.3. Analysis of the seasonal atmospheric circulation variability in Siberia using surface pressure data The atmospheric circulation was characterised using a database of monthly grids of surface pressure provided by NCEP–NCAR (http://dss.ucar.edu/datasets/ds010.1/) between 1982 and 2001 at a resolution of 5° . We also considered pressure fields located east and west of the study area to obtain non-local circulation patterns, because atmospheric circulation in central Russia usually has a large spatial scale (Clark et al., 1999; Aizen et al., 2001). The chosen database has been widely tested (Trenberth and Paolino, 1980; Basnett and Parker, 1997) and used to obtain atmospheric circulation patterns in the Northern Hemisphere (Linderson, 2001; Spellman, 2000). We applied PCA analysis in T-mode to calculate the general circulation modes in Russia. Such an analysis had been done by Barnston and Livezey (1987), who identified the main teleconnection patterns in the Northern Hemisphere. However, the atmospheric modes obtained for the whole Northern Hemisphere may be too general to identify the influence of atmospheric circulation on vegetation activity in northern areas. For this reason, we used an intermediate spatial scale (between synoptic weather types and general hemispheric patterns). We worked with standardised seasonal series of surface pressure values to identify the principal modes of atmospheric variability (Trigo and Palutikof, 2001). In the study area, the climate has a high seasonality (Fukutomi et al., 2003) and the atmospheric patterns that control it may differ significantly from season to season (Aizen et al., 2001; Ye, 2001; 2002). As a consequence, we adopted a seasonal timescale, separating winter (December, January and February), spring (March, April and May) and summer (June, July and August). We did not take into account the autumn because,

308

S. M. VICENTE-SERRANO ET AL.

dynamics and their climate impact over the last few decades (i.e. Mo and Livezey, 1986; Hurrell, 1995; Hurrell and Van Loon, 1997). We also took into account two global indices of atmospheric circulation: The AO (http://www.cpc.ncep.noaa.gov/products/precip/Cwlink/daily ao index/ao index.html, Thompson and Wallace, 1998) and the Southern Oscillation Index (obtained from the Climate Research Unit of the University of East Anglia, http://www.cru.uea.ac.uk/cru/data/soi.htm, Ropelewski and Jones, 1987).

4. RESULTS 4.1. Spatio-temporal variability of inter-annual vegetation activity in central Siberia Figure 2 shows the 11 rotated temporal components obtained from PCA applied to the aNDVI series. Figure 3 shows the spatial patterns of REOFs. The first component represents the temporal evolution of aNDVI values in south-eastern areas, the second individuates a large area in the central-western region and

3

3

2

2

COMPONENT 2

COMPONENT 1

SEASONAL PRESSURE PATTERNS IN CENTRAL SIBERIA

1 0 -1 -2 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

COMPONENT 4

COMPONENT 3

1 0 -1 -2 -4 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

-2

1 0 -1 -2 -3 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

2

4

1

COMPONENT 6

COMPONENT 5

-1

2

2

-3

0 -1 -2

3 2 1 0 -1 -2 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

-3 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 3

3

2

2

COMPONENT 8

COMPONENT 7

1 0

-3 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

3

1 0 -1 -2 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

1 0 -1 -2 -3 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 3

COMPONENT 10

2 COMPONENT 9

309

1 0 -1 -2 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

2 1 0 -1 -2 -3 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

COMPONENT 11

3 2 1 0 -1 -2 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

Figure 2. Temporal evolution of rotated components

Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)

310

S. M. VICENTE-SERRANO ET AL. REOF 1

REOF 2

REOF 3

REOF 4

REOF 5

REOF 6

REOF 7

REOF 8

REOF 9

REOF 10

REOF 11 1000

0

1000 2000 3000 Kilometers

-0.8 -0.6 -0.45 -0.2 0.0 0.2 0.45 0.6 0.8

Figure 3. Spatial distribution of the 11 REOF obtained from principal component analysis. Values higher than 0.45 (p < 0.05) are statistically significant. This figure is available in colour online at www.interscience.wiley.com/ijoc 20 18 16 Limit of selected components

% OF VARIANCE

14 12 10 8 6 4 2 0 1

2

3

4

5

6

7

8 9 10 11 12 13 14 15 16 17 18 19 COMPONENTS

Figure 4. Total variance explained by the different components

Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)

311

SEASONAL PRESSURE PATTERNS IN CENTRAL SIBERIA

Table I. Trend analysis of the series of the first three components Components

Rho

Slope (increase/year)

1 2 3

0.29 0.13 0.37

0.04 0.02 0.04

Table II. Proportion of land cover types in the different components (areas selected according to REOF > 0.45, p < 0.05). 1: Evergreen needleleaf forests, 2: Deciduous needleleaf forests, 3: Deciduous broadleaf forests, 4: Mixed forests, 5: Woodlands, 6: Grasses (Steppe), 7: Croplands, 8: Bare, 9: Mosses and lichens Land cover

Component 1 Component 2 Component 3

1

2

3

4

5

6

7

8

9

0.26 0.32 0.01

0.19 0.10 0.13

0.06 0.11 0.00

0.17 0.14 0.00

0.06 0.08 0.14

0.10 0.05 0.01

0.10 0.07 0.00

0.00 0.00 0.02

0.06 0.12 0.69

Table III. Seasonal atmospheric patterns obtained from PCA

Component 1 2 3 4 5 6

Winter % of variance 21.4 15.0 14.5 12.4 11.1 10.1

% accumulated 21.4 36.3 50.8 63.2 74.4 84.4

Component 1 2 3 4 5 6 7

Spring % of variance 20.2 15.3 11.8 10.3 9.8 8.5 7.2

% accumulated 20.2 35.4 47.2 57.5 67.2 75.7 82.9

Component 1 2 3 4 5 6 7

Summer % of variance 18.2 11.8 11.8 10.3 9.5 9.4 8.7

% accumulated 18.2 30.0 41.9 52.2 61.7 71.1 79.8

Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)

312

S. M. VICENTE-SERRANO ET AL.

< -2

-1.2

-0.4

0.4

1.2

>2

Figure 5. Main atmospheric circulation modes during winter season (1982–2001). This figure is available in colour online at www.interscience.wiley.com/ijoc

Figure 6 shows the spring atmospheric circulation modes. Spring mode 1 also shows a dominant zonal circulation, but, unlike winter mode 1, low pressures in the north and high pressures in the south are observed, which implies dominant flows from the west. Patterns 2, 4, 6 and 7 indicate dominant meridian circulation and cyclonic, north- or west-dominant advective flows in central Siberia. Modes 3 and 5 have a quasi-zonal circulation with dominant low pressures in the south and high pressures in the north. Figure 7 shows the most important circulation modes observed in the summer. Mode 1 indicates high pressures in Arctic areas and low pressures in the south, similar to winter mode 1, although, in summer mode, the pressure gradient is more important. The other atmospheric configurations show a higher spatial complexity with different centres of low or high pressures. 4.4. Influence of seasonal atmospheric circulation modes on the inter-annual variability of aNDVI in central Siberia Table IV reports the results of the empirical models created to explain the inter-annual variability of aNDVI using atmospheric circulation modes in the areas represented by the three main components. Table V reports the variables included in each model, the standardised beta coefficients and the partial correlation. For component 1, only 24% of the total variance of aNDVI inter-annual variability is explained by atmospheric circulation. Only one atmospheric circulation mode is included in the model by means of the stepwise regression procedure (summer mode 5). Nevertheless, the correlation of this atmospheric pattern and the temporal evolution of aNDVI values in the areas represented by component 1 is significant (R = −0.49, p < 0.05). The model for component 2 (west-central areas) includes three variables: two atmospheric summer modes (2 and 7) and the winter mode 5. In this case, a high percentage of the variance of the inter-annual variability of vegetation activity is explained by these three modes of atmospheric circulation (63%). Summer mode 7 has the main role in the model (partial correlation = 0.65). The model for component 3 also explains the high percentage of the total variance (53%) and it includes two atmospheric modes (summer mode 1 and winter mode 4). Summer mode 1 is the most important with a partial correlation of 0.68. Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)

313

SEASONAL PRESSURE PATTERNS IN CENTRAL SIBERIA

< -2

-1.2

-0.4

0.4

1.2

>2

Figure 6. Main atmospheric circulation modes during spring season (1982–2001). This figure is available in colour online at www.interscience.wiley.com/ijoc

< -2

-1.2

-0.4

0.4

1.2

>2

Figure 7. Main atmospheric circulation modes during summer season (1982–2001). This figure is available in colour online at www.interscience.wiley.com/ijoc

Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)

314

S. M. VICENTE-SERRANO ET AL.

Table IV. Results of the regression models used to explain temporal evolution of aNDVI by atmospheric circulation

Component 1 Component 2 Component 3

R

R2

R2 Corrected

0.49 0.8 0.73

0.24 0.63 0.53

0.2 0.57 0.48

Table V. Standardised beta coefficient and partial correlation of the atmospheric circulation variables for each regression model Beta Component 1 −0.49 Component 2 Mode 2 (summer) 0.39 Mode 7 (summer) 0.52 Mode 5 (winter) −0.39 Component 3 Mode 1 (summer) 0.63 Mode 4 (winter) 0.46 Mode 5 (summer)

-1 -2 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

COMPONENT 2

COMPONENT 1

1 0

−0.49 0.53 0.65 −0.53 0.68 0.55

3

3 2

Partial

2 1 0 -1 -2 -3 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

3 COMPONENT 3

2 1 0 -1 -2 -3 -4 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

Figure 8. Comparison among time series of the different components and the models obtained from atmospheric circulation (dashed lines). This figure is available online at www.interscience.wiley.com/ijoc

Figure 8 shows the temporal evolution of the three main components of vegetation activity (continuous lines) and the models obtained from atmospheric circulation (dashed lines). Model 1 shows high uncertainty, but the generally positive trend of this component is also reproduced by the model, with less variability than the aNDVI data. For components 2 and 3, it is clear that the temporal evolution of the models obtained from atmospheric circulation has a strong similarity to the real aNDVI evolution and only in 1989 does model 3 show a significant difference from the aNDVI data. Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)

SEASONAL PRESSURE PATTERNS IN CENTRAL SIBERIA

315

The influence on temperature of the seasonal atmospheric circulation patterns, included in the explanatory models discussed above, is illustrated in Figures 9, 10 and 11. Figure 9 shows the spatial distribution of the correlation (R) between REOF series of summer atmospheric circulation mode 5 and the summer mean temperature in the study area. This mode produces positive thermal anomalies in the central region. However, there is a negative correlation with the summer temperature in the south-eastern sector due to cold flows from the north. The aNDVI evolution in this area coincides with the evolution of component 1. In these high-altitude areas, the main climatic factor limiting vegetation development is temperature. Summer mode 5 produces a temperature decrease in southern areas that might limit vegetation activity and explain the negative correlation between this atmospheric mode and aNDVI evolution. For component 2, three atmospheric circulation variables are included in the model (summer modes 2 and 7 and winter mode 5). Figure 10 shows the correlation between these atmospheric modes and the climatic parameters. Summer mode 2 has a negative impact on the summer temperature over large regions in the centre of the study area, where the temporal evolution of vegetation is represented by component 2. Summer mode 7 also has a positive effect on the temperature in areas represented by component 2. The influence of winter mode 5 on the climate parameters is less clear, and it does not clearly affect temperature in the western areas of central Siberia. Nevertheless, we have found a positive correlation between this atmospheric mode and winter precipitation, hence snowfall. In areas in which the aNDVI temporal evolution is represented by component 3, the model created by atmospheric circulation patterns included summer mode 1 and winter mode 4. The spatial distribution of the correlation between these atmospheric modes and the climatic variables is shown in Figure 11. Summer mode 1 affects temperature positively in the northern areas: a very high correlation is found in areas represented by component 3 of aNDVI evolution. Since temperature is the main constraint to vegetation activity in this region, higher values of this climatic parameter enhance vegetation development. Moreover, although summer mode 1 plays the main role in explaining the aNDVI variability in northern areas, winter atmospheric circulation also has a significant impact. We have found a negative correlation between winter mode 4 and the winter mean temperature in these areas (Figures 11.2). It is difficult to explain the relationship between this winter atmospheric mode and vegetation activity based on temperature. However, winter mode 4 affects the precipitation variability in the northern regions (Figures 11.3). The correlation between this atmospheric mode and precipitation in these areas is negative and highly significant. 4.5. Relationship between seasonal atmospheric circulation modes in Siberia and the main hemispherical and global teleconnection indexes Finally, we analysed the relationships between the evolution of global and hemispherical teleconnection modes and regional circulation patterns obtained from PCA in Russia between 1982 and 2001. The purpose was to understand whether aNDVI evolution is linked to general atmospheric circulation in the Northern Hemisphere. Table VI shows the correlation between the regional seasonal patterns obtained

-0.8

-0.6

-0.4

-0.2

0.4

0.6

0.8

Figure 9. Spatial distribution of correlation between summer atmospheric mode 5 and summer mean temperature. This figure is available in colour online at www.interscience.wiley.com/ijoc Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)

316

S. M. VICENTE-SERRANO ET AL.

-0.8

-0.6

-0.4

-0.2

0.4

0.6

0.8

Figure 10. Spatial distribution of correlation between 1- summer atmospheric mode 2 and summer mean temperature, 2- summer atmospheric mode 7 and winter mean temperature, 3- winter atmospheric mode 5 and winter precipitation. This figure is available in colour online at www.interscience.wiley.com/ijoc

Table VI. Correlation between the seasonal atmospheric circulation patterns obtained from pressure grids and seasonal teleconnection patterns (average of monthly values) Teleconnection

Correlation

Winter Component Component Component Component Component Component

1 2 3 4 5 6

POL – NAO WP SOI PNA

−0.82∗∗ – 0.76∗∗ −0.47∗ −0.61∗∗ −0.57∗∗

Spring Component Component Component Component Component Component Component

1 2 3 4 5 6 7

AO EA/WR NP NP – – SCA

0.70∗∗ 0.49∗ −0.55∗ −0.58∗∗ – – 0.60∗∗

Summer Component Component Component Component Component Component

1 2 3 4 5 6

Component 7

Teleconnection

Correlation

AO – AO – – –

−0.50∗ – −0.61∗∗ – – –

–

–

in the study area and the general atmospheric circulation modes in the Northern Hemisphere (average of monthly teleconnection indexes over the season). In winter, the atmospheric circulation modes are closely related to the main atmospheric circulation patterns of the Northern Hemisphere. Winter Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)

SEASONAL PRESSURE PATTERNS IN CENTRAL SIBERIA

-0.8

-0.6

-0.4

-0.2

0.4

0.6

317

0.8

Figure 11. Spatial distribution of correlation between 1- summer atmospheric mode 1 and summer mean temperature, 2- winter atmospheric mode 4 and winter mean temperature, 3- winter atmospheric mode 4 and winter precipitation. This figure is available in colour online at www.interscience.wiley.com/ijoc

mode 1 has high correlation with the Polar (POL) teleconnection (r = −0.82) and mode 3 is related to the NAO (r = 0.76). Winter mode 5, included in the aNDVI model for component 2, is significantly related to ENSO and winter mode 4 (included in the aNDVI model for component 3) is related to WP pattern. In spring, there is also a high level of correlation with the atmospheric circulation modes of the Northern Hemisphere. In summer, the season in which the number of atmospheric circulation modes affecting aNDVI variability is higher, the regional patterns correlate poorly to the teleconnection indexes: only mode 1 and 3 are significantly related to the AO index and the correlation values are not very high. In summary, the relationship between the atmospheric circulation modes that significantly influence the inter-annual variability of vegetation activity in central Siberia and the general teleconnection patterns of the Northern Hemisphere is weak.

5. DISCUSSION AND CONCLUSIONS In this paper, we have analysed the influence of the inter-annual variability of seasonal atmospheric circulation patterns on vegetation activity in central Siberia. We found that the spatial patterns of vegetation activity variability are quite complex. Three main temporal patterns were identified, which represent the temporal Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)

318

S. M. VICENTE-SERRANO ET AL.

evolution in three large regions (SE, W and N). The temporal evolution of aNDVI in the first 2 PC series is not related to differences in land cover since components 1 and 2 represent forests of the same type (mainly evergreen needleleaf forest). Only in tundra areas representing component 3 does land cover determine aNDVI evolution. The temporal series of aNDVI, representative of the three main areas, show a positive increase in vegetation activity between 1982 and 2001, in agreement with other results on the NDVI trends in the northern Eurasian areas (Bogaert et al., 2002; Shabanov et al., 2002; Slayback et al., 2003). We have found the highest increase in the northern areas of tundra. It is difficult to establish the connections between climatic and vegetation dynamics. Nevertheless, we have observed a significant impact of atmospheric circulation on inter-annual variations of aNDVI. We have summarised the seasonal atmospheric variability in Siberia by means of a few circulation modes. It is interesting to note that the patterns are significantly different for winter, spring and summer, although the general surface pressure mode during the three seasons is indicative of the northern annular mode structure reported by Thompson and Wallace (1998). The effect of these atmospheric circulation patterns on the variability of vegetation activity differs spatially. To explain the aNDVI evolution by atmospheric circulation in areas represented by component 1, we have shown that summer mode 5 affects temperatures negatively in these areas, and this can have a negative influence on vegetation activity. These regions are mainly mountainous, and temperature is the main constraint to vegetation development. However, a high percentage of the total variance of vegetation activity in these areas is not explained by atmospheric circulation. In fact, the model for component 1 explains only 24% of the total variance. Nevertheless, this model shows a positive trend similar to that observed in the aNDVI series. The explanation of component 2 (western areas) by means of atmospheric circulation modes is more reliable. However, the influence of summer modes 2 and 7 and winter mode 5 on this component is difficult to establish in terms of precipitation and temperature. In these central regions, an important connection between winter and summer precipitation and Atlantic and Pacific Ocean surface temperatures has been identified (Ye, 2000; 2001; 2002). For this reason, it is difficult to exclusively associate the dynamics of climate to regional atmospheric circulation. Moisture and temperature of air mass may play a more important role than the spatial distribution and intensity of pressure patterns. In tundra areas, characterised by more extreme climatic conditions, aNDVI variability is closely related to atmospheric circulation. We have shown that summer mode 1 significantly affects the inter-annual variability of vegetation activity in these regions. We found a high positive correlation between summer mode 1 and the summer temperatures. There is also a negative correlation between winter mode 5 and precipitation. Since in these areas winter precipitation is solid, the relationship of this atmospheric circulation mode with vegetation activity is indirect, through the winter snowpack. We have provided evidence that shows that atmospheric circulation in Siberia has a certain relationship with the general teleconnection patterns of the Northern Hemisphere; nevertheless, the seasonal variations are important. In winter and spring, the atmospheric circulation modes observed are significantly related to general teleconnection indexes, but, in summer, the relationship is not significant. This confirms that there is more local atmospheric circulation during summer and has important implications for identifying the relations between climate and vegetation activity. It may explain why other studies using teleconnection indexes in relation to vegetation activity have not found significant correlations for the northern Siberian areas (Buermann et al., 2003). The vegetation activity of these regions is recorded mainly in summer and, during this season, the climate (mainly temperature), which determines vegetation activity, is mainly related to regional circulation. In conclusion, the relationship between the variability of atmospheric circulation, climatic conditions and vegetation activity is clearer in areas where climatic conditions are more limiting, as in the northern regions. In more southern areas, the climatic conditions are less restrictive and other factors and feedback mechanisms (such as climate, soil, human activities and vegetation) can affect vegetation activity variability. ACKNOWLEDGEMENT

The authors wish to acknowledge financial support from the Siberia II project (5th Framework Program of the European Commission). Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)

SEASONAL PRESSURE PATTERNS IN CENTRAL SIBERIA

319

REFERENCES Aizen EM, Aizen VB, Melack JM, Nakamura T, Ohta T. 2001. Precipitation and atmospheric circulation patterns at mid-latitudes of Asia. International Journal of Climatology 21

320

S. M. VICENTE-SERRANO ET AL.

Kogan FN, Wei G. 2000. Using AVHRR data for detection of El Ni˜no/La Ni˜na impacts on land ecosystems. Advances in Space Research 26: 1165–1168. Linderson M. 2001. Objective classification of atmospheric circulation over southern Scandinavia. International Journal of Climatology 21: 155–169. Lucht W, Prentice IC, Myneni RB, Sitch S, Friedlingstein P, Cramer W, Bousquet P, Buermann W, Smith B. 2002. Climatic control of the high-latitude vegetation greening trend and Pinatubo effect. Science 296: 1687–1689. Malevsky-Malevich SP, Molkentin EK, Nadyozhina ED, Shklyarevich OB. 2001. Numerical simulation of permafrost parameters distribution in Russia. Cold Regions Science and Technology 32: 1–11. Maseli F, Conese C, Petkov L, Gilabert MA. 1992. Use of NOAA-AVHRR NDVI data for environmental monitoring and crop forecasting in the Sahel. Preliminary results. International Journal of Remote Sensing 13: 2743–2749. Mitchell TD, Jones PD. 2005. An improved method of constructing a database of monthly climate observations and associated highresolution grids. International Journal of Climatology. 25: 693–712. Mo KC, Livezey RE. 1986. Tropical-Extratropical Geopotential Height teleconnections during the northern hemisphere winter. Monthly Weather Review 114: 2488–2515. Myneni RB, Tucker CJ, Asrar G, Keeling CD. 1998. Inter-annual variations in satellite-sensed vegetation index data from 1981 to 1991. Journal of Geophysical Research 103: 6145–6160. NASA. 2004. Pathfinder AVHRR Land Data. http://daac.gsfc.nasa.gov. Nemani RR, Keeling CD, Hashimoto H, Jolly WM, Piper SC, Tucker CJ, Myneni RB, Running SW. 2003. Climate-driven increases in global terrestrial net primary production from 1982 to 1999. Science 300: 1560–1563. Nicholson SE, Davenport ML, Malo AR. 1990. A comparison of the vegetation response to rainfall in the Sahel and east Africa, using normalized difference vegetation index from NOAA-AVHRR. Climatic Change 17: 209–241. New M, Todd M, Hulme M, Jones P. 2001. Precipitation measurements and trends in the twentieth century. International Journal of Climatology 21: 1899–1922. Peters AJ, Walter-Shea EA, Vi˜na A, Hayes M, Svoboda MD. 2002. Drought monitoring with NDVI-based standardized vegetation index. Photogrammetric Engineering and Remote Sensing 68: 71–75. Philander SG. 1990. El Ni˜no, La Ni˜na and the Southern Oscillation. Academic Press: San Diego, CA. Richards JA. 1993. Remote Sensing Digital Analysis: An Introduction. Springer-Verlag: New York. Richman MB. 1986. Rotation of principal components. Journal of Climatology 6: 29–35. Rogers JC. 1984. The association between the north Atlantic oscillation and the southern oscillation in the northern hemisphere. Monthly Weather Review 112: 1199–2015. Ropelewski CF, Jones PD. 1987. An extension of the Tahiti-Darwin Southern oscillation index. Monthly Weather Review 115: 2161–2165. Ropelewski CF, Halpert MS. 1987. Precipitation patterns associated with the high phase of the Southern oscillation. Journal of Climate 2: 268–284. Ringrose S, Matheson W. 1991. A Landsat analysis of range conditions in the Botswana Kalahari drought. International Journal of Remote Sensing 12: 1023–1051. Salinas-Zavala CA, Douglas AV, D´ıaz MF. 2002. Inter-annual variability of NDVI in northwest Mexico. Associated climatic mechanisms and ecological implications. Remote Sensing of Environment 82: 417–430. Schultz PA, Halpert MS. 1993. Global correlation of temperature, NDVI and precipitation. Advances in Space Research 13: 277–280. Serrano A, Garc´ıa JA, Mateos VL, Cancillo ML, Garrido J. 1999. Monthly modes of variation of precipitation over the Iberian peninsula. Journal of Climate 12: 2894–2919. Siegel S, Castelan NJ. 1988. Nonparametric Statistics for the Behavioral Sciences. Mc-Graw-Hill: New York. Shabanov NV, Zhou L, Knyazikhin Y, Myneni RB, Tucker CJ. 2002. Analysis of inter-annual changes in northern vegetation activity observed in AVHRR data from 1981 to 1994. IEEE Transactions on Geoscience and Remote Sensing 40: 115–130. Slayback DA, Pinzon JE, Los SO, Tucker CJ. 2003. Northern hemisphere photosynthetic trends 1982–99. Global Change Biology 9: 1–15. Smith PM, Kalluri SNV, Prince SD, DeFries RS. 1997. The NOAA/NASA pathfinder AVHRR 8-km land data set. Photogrammetric Engineering and Remote Sensing 63: 12–31. Spellman G. 2000. The use of an index-based regression model for precipitation analysis on the Iberian peninsula. Theoretical and Applied Climatology 66: 229–239. Suzuki R, Tanaka S, Yasunari T. 2000. Relationships between meridional profiles of satellite-derived vegetation index (NDVI) and climate over Siberia. International Journal of Climatology 20: 955–967. Suzuki R, Nomaki T, Yamasuri T. 2001. Spatial distribution and its seasonality of satellite-derived vegetation index (NDVI) and climate in Siberia. International Journal of Climatology 21: 1321–1335. Thompson DWJ, Wallace JM. 1998. The Artic oscillation signature in the wintertime geopotential height and temperature fields. Geophysical Research Letters 25: 1297–1300. Thompson DW, Baldwin MP, Wallace JM. 2002. Stratospheric connection to Northern Hemisphere wintertime weather: Implications for prediction. Journal of Climate 15: 1421–1428. Townshend JRG. 1984. Agricultural land-cover discrimination using thematic mapper spectral bands. International Journal of Remote Sensing 5: 681–698. Townshend JRG, Goff TE, Tucker CJ. 1985. Multitemporal dimensionality of images of normalized difference vegetation index at continental scales. IEEE Transactions on Geoscience and Remote Sensing 6: 888–895. Trenberth KE. 1997. The definition of El Ni˜no. Bulletin of the American Meteorological Society 78: 2771–2777. Trenberth KE, Paolino DA. 1980. The northern hemisphere sea-level pressure data set: trends, errors and discontinuities. Monthly Weather Review 108: 855–872. Trigo RM, Palutikof JP. 2001. Precipitation scenarios over Iberia. A comparison between direct GCM output and different downscaling techniques. Journal of Climate 14: 4422–4446. Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)

SEASONAL PRESSURE PATTERNS IN CENTRAL SIBERIA

321

Tucker CJ. 1979. Red and photographic infrared linear combinations for monitoring vegetation. Remote Sensing of Environment 8: 127–150. Vicente-Serrano SM, Heredia A. 2004. NAO influence on NDVI trends in the Iberian Peninsula (1982–2000). International Journal of Remote Sensing 25: 2871–2879. Wallace JM, Gutzler DS. 1981. Teleconnections in the geopotential height field during the Northern Hemisphere winter. Monthly Weather Review 109: 784–812. Wang GL. 2003. Reassessing the impact of North Atlantic oscillation in the sub-Saharan vegetation productivity. Global Change Biology 9: 493–499. Wang JXL, Cho HR. 1997. Spatial-temporal structures of trend and oscillatory variabilities of precipitation over Northern Eurasia. Journal of Climate 10: 2285–2298. Wang G, You L. 2004. Delayed impact of the North Atlantic oscillation on biosphere productivity in Asia. Geophysical Research Letters 31: L12210, DOI: 10.1029/2004GL019766. Ye H. 2000. Decadal variability of Russian winter snow accumulation and its associations with Atlantic sea surface temperature anomalies. International Journal of Climatology 20: 1709–1728. Ye H. 2001. Characteristic of winter precipitation variation over northern central Eurasia and their connections to sea surface temperatures over the Atlantic and Pacific oceans. Journal of Climate 14: 3140–3155. Ye H. 2002. Observed regional and climatological associations between spring and summer precipitation over northern central Eurasia. Water Resources Research 32: 1317, DOI: 10.1029/2001WR001060. Zhang X, Friedi MA, Schaaf CB, Strahler AH. 2004. Climate controls on vegetation phenological patterns in northern mid- and high latitudes inferred from MODIS data. Global Change Biology 10: 1133–1145. Zhang X, Friedl MA, Schaaf CB, Strahler AH, Hodges JCF, Gao F, Reed BC, Huete A. 2003. Monitoring vegetation phenology using MODIS. Remote Sensing of Environment 84: 471–475. Zhou L, Kaufmann RK, Tian Y, Myneni RB, Tucker CJ. 2003. Relation between inter-annual variations in satellite measures of northern forest greenness and climate between 1982 and 1999. Journal of Geophysical Research. 108(D1): 4004, DOI: 10.1029/2002JD002510.

Copyright  2006 Royal Meteorological Society

Int. J. Climatol. 26: 303–321 (2006)