INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. (2016) Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/joc.4847
A global classification of vegetation based on NDVI, rainfall and temperature Xianliang Zhang,a Shuang Wu,b,c Xiaodong Yanb,c* and Zhenju Chena,d* b
a College of Forestry, Shenyang Agriculture University, China Academy of Disaster Reduction and Emergency Management Ministry of Civil Affairs & Ministry of Education, Beijing Normal University, China c State Key Laboratory of Earth Surface Processes and Resource Ecology, Beijing Normal University, China d Qingyuan Forest CERN, Chinese Academy of Sciences, Shenyang, China
ABSTRACT: Vegetation types are generally classified based only on remote sensing vegetation data, and yet they cannot reflect their connection with climate. Climatic vegetation types reflect the regional vegetation characteristics in terms of climate. The distribution of global climatic vegetation types were identified by K-means method based on vegetation and climate data. Fourteen climatic vegetation types were classified based on vegetation and climate data. Every type had distinct climate and vegetation characteristics. The regions with similar normalized difference vegetation index (NDVI) values but with different climate values such as tropical desert and temperate desert could be distinguished. Our new updated climatic vegetation types agreed ‘fair’ with Koeppen climate types and climate type of Zhang and Yan while agreed ‘good’ with vegetation types when checked with Kappa test. KEY WORDS
climate types; K-means method; climatic vegetation; vegetation types
Received 25 November 2015; Revised 24 May 2016; Accepted 2 July 2016
1. Introduction Vegetation and climate are closely correlated with each other (Walter, 1964; Walter and Lieth, 1967; Woodward et al., 2004). On the one hand, climate is the main limitation to vegetation growth (Matthews, 1983; Woodward, 1987), with the distribution of vegetation being highly influenced by climate. On the other hand, the distribution of vegetation determines surface physical parameters such as albedo and leaf area, which in turn influence climate (Charney et al., 1977; Shukla et al., 1990; Zeng and Neelin, 2000; Bartholomé and Belward, 2005). Various climate and vegetation types are distributed across the world, and the distribution of vegetation types is closely coupled with corresponding climate types. Distinguishing the global pattern of climatic vegetation types is crucial for understanding the distribution of vegetation in terms of climate, as well as the relationships between vegetation and climate. In general, there are four ways to classify global vegetation or climate types. The first approach is based on climate data, the most famous of which is Koeppen climate classification (Beck et al., 2005; Baker et al., 2010). Koeppen climate types are defined by thresholds of climate data subjectively determined by referring to their * Correspondence to: X. Yan, Beijing Normal University, 19 Xinjiekouwai Street, Haidian District, Beijing 100875, China. E-mail:
[email protected]; or Z. Chen, Shenyang Agriculture University, 120 Dongling Road, Shenhe District, Shenyang 110866, China. E-mail:
[email protected]
© 2016 Royal Meteorological Society
limitations of regional vegetation, thus to a certain extent enabling Koeppen climate zones to reflect the distribution of vegetation. However, the distribution of Koeppen climate types only reflects the distribution of potential vegetation types. The second approach is based on the correlative relationship between climate and vegetation. In this method, the potential vegetation is forecasted if certain climate conditions are given. The famous models are Holdridge life zones (Holdridge, 1947; Holdridge, 1967), Box models (Box, 1981; Box, 1996), biogeography models [e.g. DOLY (Woodward, 1987) and BIOME3 (Haxeltine and Prentice, 1996)], and dynamic vegetation models [e.g. the Lund Potsdam Jena Dynamic Global Vegetation Model (Sitch et al., 2003; Hickler et al., 2006)]. The third approach is based on data from satellite imagery (e.g. De Fries and Townshend, 1994; Lambin and Ehrlich, 1995; Lu et al., 2003; Alcaraz-Segura et al., 2009; Metzger et al., 2013). With the development of remote sensing technology, information on vegetation dynamics can increasingly be obtained from the measurements of instruments onboard satellites. The vegetation indices such as normalized difference vegetation index (NDVI, Tucker et al., 2004; Tucker et al., 2005) and enhanced vegetation index (EVI, Huete et al., 2002), which are derived from satellite imagery, are widely used to reflect vegetation conditions. Vegetation types can be classified based on NDVI or other vegetation indices (e.g. De Fries et al., 1998; De Fries et al., 2000; Loveland et al., 2000). The fourth group could be the maps based on expert judgement [e.g. World Wild Fund for Nature (WWF) ecoregions by Olson
X. ZHANG et al.
et al. (2001)], which is mainly designed for conserving biodiversity. Among the abovementioned four methods, classifications based on climate data and the correlative relationship between climate and vegetation obtain potential vegetation types. However, climatic vegetation types cannot be fully identified based on only vegetation data because multiple potential vegetation types may be classified into one actual vegetation type due to their similar NDVI variations. For instance, similar vegetation under different climate conditions is probably classified to the same vegetation type in the vegetation classification which was classified based only on vegetation data, but vegetation types are different under different climate conditions (Holdridge, 1967; Box, 1996). Thus, vegetation characteristics can not be well reflected by the vegetation types classified based only on vegetation data. Climate conditions should also be considered when defining vegetation types due to the significant influences of climate on vegetation (de Jong et al., 2012; Buermann et al., 2013; Kim et al., 2014). Multiple potential vegetation types can be clearly classified to their corresponding climatic vegetation types by fusing climate data to remote sensing vegetation data (e.g. NDVI). Therefore, a new objective vegetation classification that depends on both climate and vegetation data was developed in this study. Accordingly, the main goal of this study was to classify climatic vegetation types using the K-means method based on climate and vegetation data.
2. 2.1.
Methods Climate data
Vegetation data
Among vegetation indices, the NDVI is the most commonly used index in vegetation studies (Pan et al., 2003; Breshears et al., 2005; Pettorelli et al., 2005; Tucker et al., 2005). The greenness of vegetation can be indicated using the NDVI (Sellers, 1985; Hansen et al., 2003; Zhou et al., 2007). It is defined as NDVI = (NIR − RED) ∕ (NIR + RED)
(1)
where NIR and RED are the amounts of near-infrared and red light, respectively. NIR and RED should be atmospherically corrected reflectances. NDVI values range from −1 to 1, where negative values correspond to an absence of vegetation and positive values indicate vegetated land. Monthly mean NDVI data at 0.0833∘ × 0.0833∘ © 2016 Royal Meteorological Society
2.3. K-means clustering K-means clustering, a nonhierarchical cluster method, separates multivariable data into several groups according to their distances objectively. Using K-means clustering, the global climate can be classified to 14 climate types effectively (Zhang and Yan, 2014a; Zhang and Yan, 2015). Therefore, the climatic vegetation types were also classified based on the K-means method. Monthly mean temperature, monthly total precipitation and monthly mean NDVI were used as input multivariables in the K-means method. These variables consisted of an n × 36 matrix X: ] [ X=
Global gridded monthly mean temperature and total precipitation data at 0.5∘ × 0.5∘ resolution were obtained from the CRU TS 3.22 data set (Harris et al., 2014). This data set interpolates monthly data from meteorological stations distributed throughout the world to the global land area grid-by-grid for the period 1901–2013. The climate data for the period 1982–2012 were chosen to be used in this study, as this period reflected the availability of vegetation data. 2.2.
(9.2 km × 9.2 km at equator) resolution were retrieved from the advanced very-high-resolution radiometer (AVHRR, Pinzon and Tucker, 2014) Global Inventory Modelling and Mapping Studies (GIMMS) data set (http://ecocast.arc.nasa.gov/data/pub/gimms/3g/) for the period 1982–2013. The vegetation was changing every year. The mean state of vegetation dynamics could not be reflected by a short time of vegetation changes. Thus, 31-year length was used in this study. The resolution of NDVI data is different with that of climate data. In order to unify the resolution of NDVI and climate data, the up-scaling process was carried out for the NDVI data by aggregating the nearest neighbour grids. The arithmetic mean over a six by six window was calculated, then the NDVI data could be up-scaled to 0.5∘ × 0.5∘ resolution.
T11 ⋮ Tn1
··· ⋮ ···
T1m P11 ⋮ ⋮ Tnm Pn1
··· ⋮ ···
P1m NDVI11 ⋮ ⋮ Pnm NDVIn1
··· ⋮ ···
NDVI1m ⋮ NDVInm
(2) where T is monthly mean temperature, P is monthly mean precipitation, NDVI is monthly mean NDVI and n is the number of all the grid cells in the global land area, except the Antarctic. The rows in X represent the monthly attributes, while the columns represent the number of grid cells. Using the K-means method, the columns could be classified to specific clusters based on the monthly attributes (Zhang and Yan, 2014a, 2014b). The characteristics of climate and vegetation in the same latitudes of Northern Hemisphere and Southern Hemisphere are quite similar. However, the seasonal cycle of climate and vegetation in the Southern Hemisphere is opposite to that in the Northern Hemisphere. For instance, the hottest month is July in the Northern Hemisphere, while it is January in the Southern Hemisphere. The growth of vegetation is coupled with the variation in climate. Hence, monthly temperature, precipitation and NDVI in the Southern Hemisphere were adjusted to their corresponding months in the Northern Hemisphere. For instance, January (July) temperature in the Southern Hemisphere and July (January) temperature in the Northern Hemisphere were located in the same column in X after the adjustment. After transformation, vegetation had similar seasonal changes in both the Southern and Northern Hemisphere. Int. J. Climatol. (2016)
IDENTIFY GLOBAL VEGETATION TYPES 90°N EF ET 75°N
Dr Dw
60°N
Ds Cs
45°N
Cf
30°N
Bs 15°N Cw 0° Bw 15°S
Aw
30°S
As
45°S
Am Af
60°S 180°W
150°W
120°W
90°W
60°W
30°W
0°
30°E
60°E
90°E
120°E
150°E
180°E
Figure 1. Geographic distribution of climatic vegetation types classified based on both vegetation and climate data. The name of every type can be found in Table 1.
Table 1. Names of all the climatic vegetation types classified based on climate and vegetation data using K-mean method, and their annual NDVI and climate characteristics. Code
Type
Af Am As Aw Bw Cw Bs Cf Cs Ds Dw Df ET EF
Tropical forest Tropical monsoon forest Tropical grassland Tropical Sahel and semiarid grassland Tropical desert Temperate desert Temperate grassland Temperate maritime climate with Evergreen broadleaf forest Temperate continental climate with deciduous forest Sub-frigid mixed forest Frigid evergreen coniferous forest Frigid deciduous coniferous forest Polar tundra Polar frost
The number of clusters should be assigned prior to do the process of the K-means method. Considering that vegetation is closely coupled with climate, the number of climatic vegetation types should be equal to the number of climate and vegetation types. The number of global climate types is 14 according to previous climate classification (Kottek et al., 2006; Zhang and Yan, 2014a). Moreover, there were around 13 vegetation types in many previous vegetation classifications (e.g. De Fries et al., 2000; Loveland et al., 2000). However, the tropical and temperate deserts which were identified as one type in the previous classifications were classified as two types. Therefore, global climatic vegetation types were classified to 14 types by the K-means method. The monthly temperature and precipitation data were used as input variables in the K-means method. However, © 2016 Royal Meteorological Society
Annual NDVI
Annual temperature (∘ C)
Annual precipitation (mm)
0.74 0.62 0.46 0.29 0.10 0.12 0.30 0.65 0.54 0.42 0.29 0.20 0.09 –
24.95 23.90 22.40 22.33 24.36 9.00 6.33 13.19 4.60 0.32 −5.62 −11.78 −9.64 −17.03
2176.21 1338.14 835.04 362.55 92.73 204.48 414.18 1066.19 721.29 551.05 447.58 315.86 284.93 405.68
the units of temperature and precipitation are different. In order to eliminate the influence of units, standardization should be performed prior to carrying out K-means clustering. However, standardization is not required for NDVI, as NDVI is a dimensionless quantity. The range of NDVI is 0–1 throughout the vegetated land of the world. Several standardization methods were tested and the best method was calculated as Z=
x − min (X) max (X) − min (X)
(3)
where Z is the standard index, with a scale from 0 to 1; X is a variable; x is every value in X; max(X) is the maximum value in X; and min(X) is the minimum value in X. After determining the number of vegetation types and standardizing the monthly temperature and Int. J. Climatol. (2016)
X. ZHANG et al. 27
15
35 30 25 20 15 10 5 0 –5 –10 –15
250
120
Temperature (°C)
25 23 21 19
Af Am As Aw
Precipitation (mm)
17
10 0 –10 Bw Cw Bs Cf Cs
Ds Dw Df ET EF
–20 –30 –40 100
100
200
80
80 60
150 60 100
40
40
50
20
20
0
0
0
0.9
0.9
0.8
0.8
0.7
0.7
0.7
0.6
0.6
0.6
0.5
0.5
0.5
0.4
0.4
0.4
0.3
0.3
0.3
0.2
0.2
0.2
0.1
0.1
0.1
0
0
0
0.8
NDVI
20
r be em r ec be D em ov N ber r o ct be O tem p Se ust ug A ly Ju ne Ju ay M il pr A ch ar y M ruar b Fe ary nu Ja
r be em r ec e D emb ov N ber r o ct e O emb pt Se ust ug A ly Ju ne Ju ay M l i pr A ch ar y M uar br Fe ary nu Ja
r be em r ec be D em ov r N obe er ct b O tem p Se ust ug A ly Ju e n Ju ay M il pr A ch ar ry M rua b Fe ary nu Ja
Month
Figure 2. Monthly NDVI, temperature and precipitation in every vegetation type.
precipitation, vegetation types were classified using the K-means method. Then, the agreement between the results of different classifications was measured using the Kappa test (Cohen, 1960; Brovkin et al., 2003). The kappa value ranges from −1.0 to 1.0. A value of zero indicates that the agreement is no better than that expected by chance, a negative value indicates less agreement than would be expected by chance, and a value of 1.0 represents complete agreement. Agreement is ‘poor to very poor’ when Kappa 0.70 is ‘very good to perfect’.
3.
Results
Fourteen climatic vegetation types were classified using the K-means method based on monthly vegetation and climate data (Figure 1). Every climatic vegetation type had distinct monthly vegetation and climate characteristics to other types (Figure 2). Annual vegetation and climate characteristics were also different for every type (Table 1). © 2016 Royal Meteorological Society
Every type was named by its climate characteristics and its local vegetation (Table 1). The abbreviation of every type refers to the Koeppen classification. Types 1–3 belong to ‘tropical climate’ because they have constant high temperature and precipitation in every month. According to its local vegetation (Figure 1), type 1 is ‘tropical rainforest’ (AF), type 2 is ‘tropical monsoon forest’ (Am) and type 3 is ‘tropical grassland’ (As). Types 4–5 have high temperature but scarce precipitation, so they belong to ‘arid climate’. Type 4 is the transition between grassland and desert, and thus named ‘tropical Sahel and semiarid grassland’ (Aw). Type 5 is named ‘tropical desert’ (Bw) according to its main landscape type. The ‘temperate’ climate includes types 6–9 due to their warm temperatures. Type 6 is ‘temperate desert’ (Cw), Type 7 is ‘temperate grassland’ (Bs), type 8 is ‘temperate maritime climate with evergreen broadleaf forest’ (TeMF) and type 9 is ‘temperate continental climate with deciduous forest’ (Cs). Types 10–12 belong to the ‘cold zone’, with type 10 being ‘sub-frigid mixed forest’ (Ds) and the vegetation of types 11 and 12 being coniferous forest. However, type 12 has a much colder and drier climate than type 11, and they Int. J. Climatol. (2016)
IDENTIFY GLOBAL VEGETATION TYPES 90°N EF ET 75°N
Dr Dw
60°N Ds Cs
45°N
Cf
30°N
Bs
15°N
Cw 0° Bw 15°S
Aw
30°S
As
45°S
Am Af
60°S 180°W
150°W
120°W
90°W
60°W
30°W
0°
30°E
60°E
90°E
120°E
150°E
180°E
Figure 3. Geographic distribution of vegetation types classified based on vegetation-only data. The name of every type can be found in Table 1.
Table 2. For three classifications, the area of regions classified to the same type with the vegetation and climate classification for every type except the frost type. The percentage is calculated based on the consistent area versus the total area. Vegetation-only classification
Af Am As Aw Bw Cw Bs Cf Cs Ds Dw Df ET
Climate classification of Zhang & Yan
Koeppen’s classification
Total area
Consistent area
%
Total area
Consistent area
%
Total area
Consistent area
%
6.63 4.91 3.46 4.75 9.36 2.38 2.50 3.30 3.45 3.42 2.57 2.33 1.24
5.68 3.92 2.62 3.99 6.30 0.00 1.96 1.72 2.98 3.07 1.65 1.40 0.27
86 80 76 84 67 0 78 52 86 90 64 60 22
4.01 6.43 5.42 4.93 6.96 1.76 1.64 5.00 6.09 6.57 5.11 1.79 2.32
2.73 2.96 1.63 2.05 5.33 0.41 0.00 1.36 1.62 2.17 1.72 1.17 0.40
68 46 30 42 77 23 0 27 27 33 34 65 17
5.54 7.42 0.37 6.79 10.11 0.00 2.32 7.15 4.20 0.88 0.43 8.29 3.85
3.64 3.58 0.13 2.12 6.14 0.00 0.11 2.75 1.66 0.39 0.05 1.64 1.02
66 48 34 31 61 0 5 38 40 45 13 20 26
are thus classified as two types. The vegetation in type 12 is deciduous coniferous forest because of the low temperature. Therefore, type 11 is ‘frigid evergreen coniferous forest’ (Dw) and type 12 is ‘frigid deciduous coniferous forest’ (Df). Types 13 and 14 belong to ‘polar climate’, and according to their locations and vegetation (Figure 1), type 13 is ‘polar tundra’ (ET) and type 14 is ‘polar frost’ (EF). In order to compare the classification based on vegetation and climate with the classification based on only vegetation, the vegetation types were also classified based on vegetation-only data (Figure 3). The distribution of vegetation types was similar with former vegetation classifications that were also classified based on vegetation-only data (e.g. De Fries et al., 1998; De Fries et al., 2000; Loveland et al., 2000). Although the distribution of vegetation © 2016 Royal Meteorological Society
types was not totally agree with the former vegetation classification, it is easier to be compared with the climatic vegetation types based on both vegetation and climate data. The distribution of climatic vegetation types were compared with the vegetation types, Koeppen climate types (Kottek et al., 2006) and the climate types classified by Zhang and Yan (2014a). The area of the regions classified to the corresponding types was calculated for every type except the frost type (Table 2). Most vegetation types (12 types) were located in a roughly similar pattern to their corresponding climatic vegetation types, while most climate types (9 types) were located in a quite different pattern to their corresponding climatic vegetation types. In addition, the climatic vegetation types were in fairer agreement with those based on vegetation-only data (Kappa value = 0.68) Int. J. Climatol. (2016)
X. ZHANG et al.
than Koeppen climate types (Kappa value = 0.40) and climate types defined by Zhang and Yan (2014a) (Kappa value = 0.43). Thus, our new updated vegetation types agreed ‘fair’ with Koeppen climate types and climate type of Zhang and Yan (2014a) while agreed ‘good’ with vegetation types.
4.
Discussion
Fourteen climatic vegetation types were classified using the K-means method based on monthly vegetation and climate data. In previous studies, the climate types/vegetation types or were classified based on climate data [e.g. Koeppen climate types and the classification of Zhang and Yan (2014a)]/vegetation data (De Fries and Townshend, 1994; De Fries et al., 1995; De Fries et al., 2000; Bartholomé and Belward, 2005). However, the distribution of climate types could not reflect the distribution of corresponding climatic vegetation types. For instance, the distribution area of tropical rainforest vegetation type was much larger than that of the corresponding rainforest climate type in Africa. Moreover, the vegetation varies with diverse climate conditions, which could not be reflected by the vegetation types. For instance, tropical desert and temperate desert were classified to one type in the previous vegetation classifications [e.g. in Loveland et al. (2000)]. However, their climate characteristics were quite different from each other, which could be seen from the distribution of the climate types (e.g. Zhang and Yan, 2014a). Therefore, the climatic vegetation types could not be satisfactorily classified based only on vegetation or climate data. The vegetation in different climate types was different, while the climate in different vegetation types was different. However, vegetation and climate were correlated closely. The vegetation varied with the change in temperature in the latitudinal direction (tropical, temperate, frigid), while it varied with precipitation in the longitudinal direction (forest, grassland, desert). In order to reflect the link between the corresponding vegetation and climate types, vegetation data should be classified with climate data when defining climatic vegetation types. Also, vegetation, temperature and precipitation should have the same weight when defining vegetation data. The K-means method could classify the multivariable data and give them the same weight. The grid cells that had similar vegetation and climate characteristics could be classified to the same type using K-means method: namely, the corresponding variation between vegetation and climate could be reflected. The advantage of defining vegetation types based on vegetation and climate data would be more obvious when more detailed vegetation types needed to be defined, because similar NDVI types could be distinguished by climate data while similar climate types could be separated by NDVI data. When comparing our vegetation types with the existing vegetation types such as the types defined by De Fries et al. (1995, 1998, 2000) and Loveland et al. (2000), all the vegetation types were roughly located at the same region, © 2016 Royal Meteorological Society
although the detailed comparisons could not be made because their schemes were not available to us. The similar patterns between previous vegetation types and our new updated climatic vegetation types showed that the climate types were well classified by our method. Moreover, the NDVI data reflect the actual vegetation which has been influenced by human activities. The climatic vegetation types were formed by the influences of human activities on potential vegetation types. Therefore, the new updated scheme could reflect the realistic mean state vegetation characteristics of every type. A few disadvantages of the scheme should be noted. First, the fact that it is based on vegetation activity and on climate makes that it is not an independent aggregation layer if one of these parameters is studied. Second, the scale differences between climate observations and vegetation-activity observations enable the scheme works for large-scale (continental to global) applications only, while satellite imagery schemes have a broader scope. In conclusion, global climatic vegetation types were classified by fusing NDVI and climate data, which depicts a distribution pattern of global climatic vegetation types. Other vegetation indices such as EVI, which were used to define vegetation types as NDVI, could be also used to define vegetation types with climate data. Therefore, this work provides an updated map of climatic vegetation types and is important for studying the vegetation in terms of climate.
Acknowledgements This work was funded by the National Natural Science Foundation of China (41271066, 31570632, 41571094 and 31570473). We thank the two anonymous reviewers for their constructive comments.
References Alcaraz-Segura D, Cabello J, Paruelo J. 2009. Baseline characterization of major Iberian vegetation types based on the NDVI dynamics. Plant Ecol. 202: 13–29. Baker B, Diaz H, Hargrove W, Hoffman F. 2010. Use of the Koppen-Trewartha climate classification to evaluate climatic refugia in statistically derived ecoregions for the People’s Republic of China. Clim. Change 98: 113–131. Bartholomé E, Belward AS. 2005. GLC2000: a new approach to global land cover mapping from Earth observation data. Int. J. Remote Sens. 26: 1959–1977. Beck C, Grieser J, Kottek M, Rubel F, Rudolf B (2005) Characterizing global climate change by means of Koppen climate classification, Deutscher Wetterdienst, Offenbach, Germany, 51. Box EO. 1981. Macroclimate and plant form: an introduction to predictive modeling in phytogeography. Dr W. Junk, The Hague. Box EO. 1996. Plant functional types and climate at the global scale. J. Veg. Sci. 7: 309–320. Breshears DD, Cobb NS, Rich PM, Price KP, Allen CD, Balice RG, Romme WH, Kastens JH, Floyd ML, Belnap J. 2005. Regional vegetation die-off in response to global-change-type drought. Proc. Natl. Acad. Sci. USA 102: 15144–15148. Brovkin V, Levis S, Loutre MF, Crucifix M, Claussen M, Ganopolski A, Kubatzki C, Petoukhov V. 2003. Stability analysis of the climate-vegetation system in the northern high latitudes. Clim. Change 57: 119–138. Int. J. Climatol. (2016)
IDENTIFY GLOBAL VEGETATION TYPES Buermann W, Bikash PR, Jung M, Burn DH, Reichstein M. 2013. Earlier springs decrease peak summer productivity in North American boreal forests. Environ. Res. Lett. 8: 24027. Charney J, Quirk WJ, Chow S, Kornfield J. 1977. A comparative study of the effects of albedo change on drought in semi-arid regions. J. Atmos. Sci. 34: 1366–1385. Cohen J. 1960. A coefficient of agreement for nominal scales. Educ. Psychol. Meas. 20: 37–46. De Fries RS, Townshend J. 1994. NDVI-derived land cover classifications at a global scale. Int. J. Remote Sens. 15: 3567–3586. De Fries R, Hansen M, Townshend J. 1995. Global discrimination of land cover types from metrics derived from AVHRR Pathfinder data. Remote Sens. Environ. 54: 209–222. De Fries RS, Hansen M, Townshend J, Sohlberg R. 1998. Global land cover classifications at 8 km spatial resolution: the use of training data derived from Landsat imagery in decision tree classifiers. Int. J. Remote Sens. 19: 3141–3168. De Fries RS, Hansen MC, Townshend J, Janetos AC, Loveland TR. 2000. A new global 1-km dataset of percentage tree cover derived from remote sensing. Glob. Chang. Biol. 6: 247–254. De Jong R, Verbesselt J, Schaepman ME, Bruin S. 2012. Trend changes in global greening and browning: contribution of short-term trends to longer-term change. Glob. Chang. Biol. 18: 642–655. Hansen MC, DeFries RS, Townshend J, Carroll M, Dimiceli C, Sohlberg RA. 2003. Global percent tree cover at a spatial resolution of 500 meters: first results of the MODIS vegetation continuous fields algorithm. Earth Interact. 7: 1–15. Harris I, Jones PD, Osborn TJ, Lister DH. 2014. Updated high-resolution grids of monthly climatic observations – the CRU TS3. 10 Dataset. Int. J. Climatol. 34: 623–642. Haxeltine A, Prentice IC. 1996. BIOME3: an equilibrium terrestrial biosphere model based on ecophysiological constraints, resource availability, and competition among plant functional types. Glob. Biogeochem. Cycles 10: 693–709. Hickler T, Prentice IC, Smith B, Sykes MT, Zaehle S. 2006. Implementing plant hydraulic architecture within the LPJ Dynamic Global Vegetation Model. Glob. Ecol. Biogeogr. 15: 567–577. Holdridge LR. 1947. Determination of world plant formations from simple climatic data. Science 105: 267–268. Holdridge LR. 1967. Life zone ecology. Tropical Science Center: San Jose, Costa Rica. Huete A, Didan K, Miura T, Rodriguez EP, Gao X, Ferreira LG. 2002. Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sens. Environ. 83: 195–213. Kim Y, Kimball JS, Zhang K, Didan K, Velicogna I, McDonald KC. 2014. Attribution of divergent northern vegetation growth responses to lengthening non-frozen seasons using satellite optical-NIR and microwave remote sensing. Int. J. Remote Sens. 35: 3700–3721. Kottek M, Grieser J, Beck C, Rudolf B, Rubel F. 2006. World map of the Köppen-Geiger climate classification updated. Meteorol. Z. 15: 259–263. Lambin E, Ehrlich D. 1995. Combining vegetation indices and surface temperature for land-cover mapping at broad spatial scales. Int. J. Remote Sens. 16: 573–579. Loveland TR, Reed BC, Brown JF, Ohlen DO, Zhu Z, Yang L, Merchant JW. 2000. Development of a global land cover characteristics database and IGBP DISCover from 1 km AVHRR data. Int. J. Remote Sens. 21: 1303–1330. Lu H, Raupach MR, McVicar TR, Barrett DJ. 2003. Decomposition of vegetation cover into woody and herbaceous components using AVHRR NDVI time series. Remote Sens. Environ. 86: 1–18.
© 2016 Royal Meteorological Society
Matthews E. 1983. Global vegetation and land use: new high-resolution data bases for climate studies. J. Clim. Appl. Meteorol. 22: 474–487. Metzger MJ, Bunce RGH, Jongman RHG, Sayre R, Trabucco A, Zomer R. 2013. A high-resolution bioclimate map of the world: a unifying framework for global biodiversity research and monitoring. Glob. Ecol. Biogeogr. 22: 630–638. Olson DM, Dinerstein E, Wikramanayake ED, Burgess ND, Powell GV, Underwood EC, D’Amico JA, Itoua I, Strand HE, Morrison JC. 2001. Terrestrial ecoregions of the world: a new map of life on earth. BioScience 51: 933–938. Pan Y, Li X, Gong P, He C, Shi P, Pu R. 2003. An integrative classification of vegetation in China based on NOAA AVHRR and vegetation-climate indices of the Holdridge life zone. Int. J. Remote. Sens. 24: 1009–1027. Pettorelli N, Vik JO, Mysterud A, Gaillard J, Tucker CJ, Stenseth NC. 2005. Using the satellite-derived NDVI to assess ecological responses to environmental change. Trends Ecol. Evol. 20: 503–510. Pinzon JE, Tucker CJ. 2014. A non-stationary 1981–2012 AVHRR NDVI3g time series. Remote Sens. 6: 6929–6960. Sellers PJ. 1985. Canopy reflectance, photosynthesis and transpiration. Int. J. Remote Sens. 6: 1335–1372. Shukla J, Nobre C, Sellers P. 1990. Amazon deforestation and climate change. Science 247: 1322–1325. Sitch S, Smith B, Prentice IC, Arneth A, Bondeau A, Cramer W, Kaplan JO, Levis S, Lucht W, Sykes MT. 2003. Evaluation of ecosystem dynamics, plant geography and terrestrial carbon cycling in the LPJ dynamic global vegetation model. Glob. Change Biol. 9: 161–185. Tucker CJ, Pinzon JE, Brown ME. 2004. Global inventory modeling and mapping studies. Global Land Cover Facility, University of Maryland, College Park, Maryland. Tucker CJ, Pinzon JE, Brown ME, Slayback DA, Pak EW, Mahoney R, Vermote EF, El Saleous N. 2005. An extended AVHRR 8-km NDVI dataset compatible with MODIS and SPOT vegetation NDVI data. Int. J. Remote Sens. 26: 4485–4498. Walter H. 1964. Vegetation der Erde in oko-physiologischer Betrachtung, Fisher: Jena, German. Walter H, Lieth H. 1967. Climate diagram world atlas. Fischer: Jena, Germany. Woodward FI. 1987. Climate and plant distribution. Cambridge University Press: Cambridge. Woodward FI, Lomas MR, Kelly CK. 2004. Global climate and the distribution of plant biomes. Philos. Trans. R Soc. Lond. B: Biol. Sci. 359: 1465–1476. Zeng N, Neelin JD. 2000. The role of vegetation-climate interaction and interannual variability in shaping the African savanna. J. Clim. 13: 2665–2670. Zhang X, Yan X. 2014a. Spatiotemporal change in geographical distribution of global climate types in the context of climate warming. Clim. Dyn. 43: 595–605. Zhang X, Yan X. 2014b. Temporal change of climate zones in China in the context of climate warming. Theor. Appl. Climatol. 115: 167–175. Zhang X, Yan X. 2015. Deficiencies in the simulation of the geographic distribution of climate types by global climate models. Clim. Dyn. 46: 2749–2757. Zhou L, Dickinson RE, Tian Y, Vose RS, Dai Y. 2007. Impact of vegetation removal and soil aridation on diurnal temperature range in a semiarid region: application to the Sahel. Proc. Natl. Acad. Sci. USA 104: 17937–17942.
Int. J. Climatol. (2016)
