Clim Dyn (2016) 46:1131–1150

DOI 10.1007/s00382-015-2636-8

Global semi‑arid climate change over last 60 years

Jianping Huang1 · Mingxia Ji1 · Yongkun Xie1 ·
Shanshan Wang1 · Yongli He1 · Jinjiang Ran1

Received: 19 November 2014 / Accepted: 29 April 2015 / Published online: 13 May 2015
© The Author(s) 2015. This article is published with open access at Springerlink.com

Abstract This study analyzes areal changes and regional Keywords Semi-arid · Climate change · Expansion ·
climate variations in global semi-arid regions over 61 years Aridity index
(1948–2008) and investigates the dynamics of global semi-
arid climate change. The results reveal that the largest
expansion of drylands has occurred in semi-arid regions 1 Introduction
since the early 1960s. This expansion of semi-arid regions
accounts for more than half of the total dryland expansion. Semi-arid regions cover ~15 % of the Earth’s land sur-
The area of semi-arid regions in the most recent 15 years face and supported 14.4 % of the global population in the
studied (1990–2004) is 7 % larger than that during the first year 2000 (Safriel and Adeel 2005). The ecosystems in
15 years (1948–1962) of the study period; this expansion such regions are fragile and sensitive to strong interactions
totaled 0.4 × 106 and 1.2 × 106 km2 within the American between human activities and climate changes (Charney
continents and in the Eastern Hemisphere, respectively. 1975; Huang et al. 2010; Rotenberg and Yakir 2010; Xue
Although semi-arid expansion occurred in both regions, the 1996; Zeng et al. 1999). Most human effects on semi-arid
shifting patterns of the expansion are different. Across the climates are related to agricultural exploitation of lands.
American continents, the newly formed semi-arid regions Introduction of agriculture into such areas, whose natural
developed from arid regions, in which the climate became condition is already agriculturally marginal, might be moti-
wetter. Conversely, in the continental Eastern Hemisphere, vated by a number of economic or social factors (Liu et al.
semi-arid regions replaced sub-humid/humid regions, in 2008). These factors include the need to provide food for
which the climate became drier. The climate change in dry- a larger population, the decreasing productivity of existing
ing semi-arid regions over East Asia is primarily dominated agricultural land, and the profit motive. Semi-arid popula-
by a weakened East Asian summer monsoon, while the tions rely primarily on rain-fed agriculture for their liveli-
wetting of semi-arid regions over North America is primar- hoods, and fluctuations in precipitation greatly affect the
ily controlled by enhanced westerlies. availability of surface water resources that are necessary to
sustain agricultural irrigation (Liu and Xia 2004; Schwin-
ning et al. 2004). For instance, a slight shift in rainfall pat-
terns or a temperature change could be disastrous. In addi-
tion, much of the arid and semi-arid land has undergone
Electronic supplementary material The online version of this substantial anthropogenic change via urbanization and
article (doi:10.1007/s00382-015-2636-8) contains supplementary
material, which is available to authorized users.
land-use changes (Gao et al. 2010; Li et al. 2012; Xu et al.
2011; Bounoua et al. 2009; Georgescu et al. 2009; Scan-
* Jianping Huang lon et al. 2006; Fricke et al. 2009). Thus, semi-arid climate
[email protected] change is important to investigate for the purposes of pol-
1 icy making.
Key Laboratory for Semi‑Arid Climate Change
of the Ministry of Education, College of Atmospheric The long-term global warming trend is particularly
Sciences, Lanzhou University, Lanzhou, Gansu, China enhanced over semi-arid regions (Huang et al. 2012; Ji

13
1132 J. Huang et al.

et al. 2014). Both observations and modeling studies indi- extending from 1948 to the present. The PREC/L is derived
cate that in a warmer climate, droughts may become longer from gauge observations taken at over 17,000 stations, and it
in duration and more severe in currently drought-prone was collected from two large individual datasets: the Global
regions because of enhanced evaporation and reduced pre- Historical Climatology Network version 2 (GHCN2), and
cipitation (Dai 2013a). Drying trends may occur most sig- the Climate Anomaly Monitoring System (CAMS). There-
nificantly in semi-arid and arid regions as a result of global fore, the PREC/L can be regularly updated in near real time
warming, leading to an intensification of the hydrological with a sufficient number of stations and unique interpola-
cycle, with climatologically wet regions becoming wetter tion methods. The definition of drylands by precipitation is
and dry regions becoming drier (Chou et al. 2009; Held as follows: areas that experience annual precipitation (P) of
and Soden 2006; Seager et al. 2010). Recent climate model less than 500 mm; drylands are further classified into hyper-
simulation shows that the area of global drylands is pro- arid (P < 25 mm), arid (25 mm ≤ P < 250 mm), and semi-
jected to expand by ~10 % by 2100, except in India and arid (250 mm ≤ P < 500 mm) regions (Thomas 2011).
northern tropical Africa, which will become wetter (Feng In addition to the PREC/L, we used the corresponding
and Fu 2013). Decadal-scale climate anomalies and land surface air temperature (SAT) dataset of CPC in this study,
surface degradation have also occurred in these regions which is labelled as the GHCN_CAMS Gridded 2 m tem-
(Dregne 2002), and the risk of land degradation is great- perature (Fan and Van den Dool 2008). This dataset has the
est in the middle of the aridity gradient, which is mostly same temporal coverage and spatial resolution as the
located in semi-arid drylands (Safriel and Adeel 2005). PREC/L. To justify the uncertainty of the dataset, it was
Although historical climate changes in semi-arid regions compared with several observation-based SAT datasets,
have been discussed in previous studies (Bader and Latif such as the Parameter-Elevation Regressions on Independ-
2003; Giannini et al. 2003; Hoerling et al. 2006; Lu 2009; ent Slopes Model (PRISM) dataset,1 the National Climate
Lu and Delworth 2005; McCabe et al. 2008; Méndez and Data Center (NCDC) Climate Division Data, and the Cli-
Magaña 2010; Zeng et al. 1999), most of these studies mate Research Unit (CRU) dataset. The preliminary results
are regionally limited by focusing on a particular region show that the quality of GHCN_CAMS land SAT analysis
or nation, i.e., Sahelian semi-arid regions (Hoerling et al. is reasonably high and that it can capture most of the com-
2006; Lu and Delworth 2005); few studies emphasize mon temporal-spatial features in observed climatology as
global semi-arid areas. In this work, we study semi-arid cli- well as the anomaly fields over both the regional and global
mate change from a global perspective. domains (Fan and Van den Dool 2008).
Three major questions are addressed in this study. Has
expansion occurred in all semi-arid regions over the last 2.2 Estimation of potential evapotranspiration (PET)
61 years? If so, the shifts of the semi-arid regions and other
climate categories are important to understand. How has The methods commonly used to estimate PET in climate
the aridity changed in these regions? What are the dynam- studies are typically classified into three types: temper-
ics of global climate change in semi-arid regions? To ature-based methods (Thornthwaite 1948; Hamon 1963;
understand the regional climate variability, this study con- Hargreaves and Samani 1985), radiation-based methods
ducts a systematic examination of areal changes in Sect. 4 (Makkink 1957; Priestley and Taylor 1972; Turc 1961) and
and an assessment of climate change in semi-arid regions combination methods (Penman 1948; Monteith 1981).
in Sect. 5. Section 6 illustrates the dynamics of semi-arid Because air temperature is correlated with net radia-
climate change globally, and the conclusions and discus- tion and humidity, temperature-based PET methods
sion are presented in Sect. 7. appear to perform relatively well in climatological appli-
cations (Shaw and Riha 2011). However, recent studies
indicate that simple temperature-based PET estimates are
2 Datasets and methodology not appropriate for determining trends such as long-term
evaporative demands and drought projections (Sheffield
2.1 Observation data et al. 2012; McAfee 2013; Greve et al. 2014) because tem-
perature is not always the primary factor controlling PET
The precipitation dataset that we mainly used in this study is (Donohue et al. 2010; Shaw and Riha 2011; Hobbins et al.
obtained from the National Centers for Environmental Pre- 2012). In fact, PET is not only a function of air tempera-
diction (NCEP) Climate Prediction Center (CPC) (Chen ture but also of radiative and aerodynamic controls (Pen-
et al. 2002). It is typically labelled as the PRECipitation man 1948).
REConstruction over Land (PREC/L) dataset, and it is an
observation station-based dataset with a globally covered 1
PRISM group, Oregon State University, see http://www.prismcli-
0.5° × 0.5° latitude–longitude resolution for the period mate.org.

13
Global semi-arid climate change over last 60 years 1133

Regarding radiation-based PET methods, the Priestley National Center for Atmospheric Research (NCAR) rea-
and Taylor (1972) method is a simplification of the Penman­– nalysis, not from surface observations.
Monteith method. The former only retains the radiation
term, and it explicitly ignores the impact of the aerody- 2.3 Aridity index (AI)
namic control term, which is the effect of the water vapour
pressure deficit (VPD), wind speed and canopy resistance The aridity index (AI) represented the degree of cli-
(Wang and Dickinson 2012). It has been noted that the fail- matic dryness is defined as the ratio of annual precipita-
ure to explicitly account for changes in available energy, air tion (P) to annual potential evapotranspiration (PET). For
humidity, and wind speed can cause some drought indices the purposes of the AI, drylands are defined as regions
to identify an artificial trend towards more droughts under a with AI of less than 0.65 and are further classified into
warming climate (Sheffield et al. 2012). hyper-arid (AI < 0.05), arid (0.05 ≤ AI < 0.2), semi-arid
The Penman–Monteith algorithm is derived from physi- (0.2 ≤ AI < 0.5), and dry sub-humid (0.5 ≤ AI < 0.65)
cal principles, and it is superior to empirically based for- subtypes (Middleton and Thomas 1997). The hyper-arid
mulations, which usually only consider the effect of tem- regions are the driest, and followed by arid, semi-arid, and
perature and radiation. The Penman–Monteith algorithm dry sub-humid regions in all four types of drylands. The
used in this study considers many meteorological param- AI is provided by Feng and Fu (2013) and calculated from
eters related to the evapotranspiration process, such as net 1948 to 2008, and the spatial resolution of the data is 0.5°
radiation, relative humidity, wind speed and temperature by 0.5°.
data, as well as information on vegetation characteristics In addition, Feng and Fu (2013) verified the uncer-
(Allen et al. 1998). tainty of different datasets when calculating PET and AI.
900
Two temperature and precipitation datasets were used: the
0.408�(Rn −G)+γ T +273 u2 (es −ea )
PET = a CPC PREC/L and University of Delaware (UDel) (Legates
∆+γ (1+0.34u2 )
and Willmott 1990a, b) datasets. The solar radiation, spe-
where Rn is the net radiation at the crop surface (MJ m−2 cific humidity and wind speed used to calculate the PET
d−1), G is the soil heat flux density at the soil surface were also selected from two datasets: the GLDAS and the
(MJ m−2 d−1), Ta is the mean daily air temperature at Twentieth Century Reanalysis (20CR) datasets (Compo
a 2-m height (°C), u2 is the wind speed at a 2-m height et al. 2011). Therefore, four combinations (i.e., CPC and
(m/s), es is the saturation vapor pressure at a 2-m height GLDAS, CPC and 20CR, UDel and GLADS, UDel and
(kPa), ea is the actual vapor pressure at a 2-m height (kPa), 20CR) were obtained and compared. The results showed
Δ is the slope of the vapor pressure–temperature curve a good agreement in the global averaged temporal varia-
(kPa/°C), and γ is the psychometric constant (kPa/°C). tions. However, we note that the trends may be slight dif-
The employed data of net surface radiation (include short- ferent in specific regions by using individual precipitation
wave and longwave net radiaton), specific humidity, wind datasets, especially in Central Asia (see Fig. S3–S6 in Sup-
speed, and soil heat flux density are from the data prod- plementary). As indicated by Schiemann et al. (2008), the
uct of Global Land Data Assimilation Systems (GLDAS) precipitation pattern is most realistically represented using
at a 0.5° resolution; the data begin in1948 (Rodell et al. UDel in central Asia by comparing a wide range of obser-
2004). The solar radiation data are corrected for 1983– vational precipitation data sets, and the large biases of the
2000 based on two global radiation products (Sheffield observed precipitation may be leaded by the foothills of the
et al. 2006). Specific humidity and wind speed data are large mountain systems such as the central Tien Shan and
adjusted according to the observationally based climato- the eastern Pamirs regions of very little precipitations. For
logical methods developed by the CRU at the University similar precipitation trends in the semi-arid regions from
of East Anglia (Mitchell and Jones 2005). The regionally UDel and PREC/L in central Asia, we note that the data set
averaged wind speed trends from GLDAS are −0.0009 of PREC/L precipitation seems also reasonable to present
and −0.0013 m s−1 a−1 in global land surface and North- the changes of precipitation trend in central Asia.
ern Hemisphere (see Fig. S1 & Fig. S2 in Supplementary)
respectively, which are consistent with the declining trend 2.4 Köppen–Geiger climate classification
in observed near-surface wind speed (−0.014 m s−1 a−1)
as indicated by McVicar et al. (2012). However, the mag- Since the Köppen climate classification was first pro-
nitude of trend in GLDAS is much less than that of the posed by Köppen (1884), various classifications based on
observations over global land, so we should note that using Köppen’s original method have been developed, such as
the GLDAS wind speed dataset may lead to biases toward Thornthwaite-type method (Thornthwaite 1948; Feddema
increasing PET and drying climate in some areas because 2005), Köppen-Trewartha (Trewartha and Horn 1980),
GLDAS gets its wind speed and humidity from NCEP/ Köppen–Geiger method (Köppen 1936; Geiger 1954).

13
1134 J. Huang et al.

Generally, the Köppen-Trewartha is a simplified method 2.5 Circulation data

used in many studies, which is only developed by using
temperature and precipitation, and the Thornthwaite-type The East Asia summer monsoon (EASM) index, which is
method is infrequently used because it is too complex archived at http://ljp.gcess.cn/dct/page/65544, is used here.
(Thornthwaite 1948; Feddema 2005). Comparing with the The EASM index is defined using the area-averaged, JJA-
above methods, the Köppen–Geiger method is based on averaged, dynamically-normalized winds at 850 hPa within
the vegetation types of a region, together with temperature the East Asian monsoon domain (10°–40°N, 110°–140°E).
and precipitation, and it is the most common climate clas- For details of this normalization, please see Li and Zeng
sification in use. As a result, we choose Köppen–Geiger (2002). In addition, the NCEP/NCAR reanalyses including
classification in this paper. Recently, updates to the Köp- the geopotential heights (HGT) and winds at 850 hPa are
pen–Geiger climate classification were published in sev- also used here (Kalnay et al. 1996).
eral studies based on observed temperature and precipita-
tion (Kottek et al. 2006; Peel et al. 2007). In this work,
we adopt the Köppen–Geiger climate classification map 3 Comparison of dryland definitions
for the period of 1951–2000 with 0.5° resolution; this map
was published by Kottek et al. (2006). Both the map and The spatial distributions of drylands (as identified by AI or
the underlying digital data are available online at http:// annual precipitation from 1961 to 1990 and the Köppen–
koeppen-geiger.vu-wien.ac.at/present.htm. These data pro- Geiger climate classification for 1951–2000) are shown
duce a total of 31 climate classes described by three let- in Fig. 1a–c. Although there are no fixed or precise limits
ters. The first letter describes the main classes, namely, for the biomes, the major dryland biomes include savanna,
equatorial climates (A), arid climates (B), warm temperate grassland, and desert (Safriel and Adeel 2005). Figure 1d
climates (C), snow climates (D) and polar climates (E). In shows the major dryland surface vegetation types from
this work, we describe only the four arid classes of type B, MODIS. As shown in Fig. 1a, global drylands defined by
i.e., hot desert climate (BWh), cool desert climate (BWk), AI is primarily distributed in the middle and low latitudes
hot steppe climate (BSh), and cool steppe climate (BSk). and closely match the map from UNEP (Middleton and
The second letter accounts for precipitation, where “W” Thomas 1997). Major hyper-arid areas occur over the cen-
represents desert and “S” represents steppe. The third let- tral and northern Sahara Desert, the Rub al Khali Desert in
ter denotes the temperature, where “h” and “k” are warm the eastern Arabian Peninsula, the Iran/Afghanistan/Paki-
and cool, respectively. stan border area, the Taklimakan Desert in northwestern

Fig. 1  The global distribution of drylands over 1961–1990 according resolution imaging spectroradiometer (MODIS) (ISLSCP II MODIS
to three definitions: a aridity, b precipitation, and c Köppen–Geiger (Collection 4) IGBP Land Cover), available online at http://daac.ornl.
climate classification; d surface vegetation types from the moderate- gov/cgi-bin/dsviewer.pl?ds_id=968

13
Global semi-arid climate change over last 60 years 1135

China, the Gobi Desert in Mongolia/China border area and classification may mistakenly classify regions with dis-
the Atacama Desert area (Fig. 1a), where the MODIS land tinctively different weather patterns and vegetation as
cover types are barren (Fig. 1d). Generally, arid areas are belonging to the same climate type (Trewartha and Horn
located over the southern Sahara Desert, southern Africa, 1980). For example, western Washington and Oregon in
western Arabia Peninsula, Central Asia, areas of Mongolia the northwest US are classified as the same climate type as
and northern China, and much of Australia (Fig. 1a), where southern California (in the southwest US), although the two
the land cover types are barren or sparsely vegetated, i.e., regions have strikingly different weather and vegetation
shrubland and grassland (Fig. 1d). Semi-arid and dry sub- types (Feng et al. 2014). Furthermore, particular drylands
humid regions are located in the western United States, the may be classified as other climate types, i.e., Mediterra-
west coast of South America, Central and East Asia and nean semi-arid regions, which are classified as Csb or Csa
a large portion of Australia outside of the central desert by the Köppen–Geiger climate classification, where the “s”
regions, where the land cover type is primarily grassland indicates “summer dry” and “b” and “a” indicate “warm
(Fig. 1d). summer” and “hot summer”, respectively. Thus, the arid
Compared with Fig. 1a, the dryland area defined by climate (type B) defined by the Köppen–Geiger climate
annual precipitation (Fig. 1b) is significantly smaller than classification does not represent the total drylands, and it is
that defined by AI everywhere except for the latitudes significantly smaller than the global drylands because sev-
around 50°–60°N. The dryland area defined by precipita- eral climate subtypes are classified in each major climate
tion is larger in Asia and North America at approximately type based on the seasonal variations in temperature and
60°N, such as in Siberia, while the AI classifies these areas precipitation. Compared with Fig. 1a, the drylands defined
as humid, due to the low PET. In Fig. 1c, the Köppen–Gei- by the Köppen–Geiger climate classification are signifi-
ger BWh subtype is located in the subtropical latitudes cantly smaller in most areas, i.e., East Asia, Central Asia,
between 30°S and 30°N. Such areas include the entire and North America. As indicated by Kottek et al. (2006), a
Sahara Desert, the Arabian Peninsula, large regions of Iran, total of 29.1 % of the global land area is covered by Köp-
southern Pakistan and Afghanistan, southern Africa, and pen–Geiger arid climates, but the areal coverage of the total
much of Australia. The BWk areas are mainly located in drylands according to the AI definition is 6.1 × 107 km2, or
East and Central Asia, as well as small areas in the Great 41 % account for the global land (Safriel and Adeel 2005).
Basin Desert of the western United States, northern Africa Thus, the drylands in the following sections are based on
and southern Australia. The BSh regions tend to be located the definition of AI.
in the tropics and subtropics, which are most commonly
found along the fringes of subtropical deserts. This type
of area is primarily located in regions such as the Sahel in 4 Expansion of semi‑arid regions
northern Africa, southern Africa, small portions of India
and Mexico, and a large portion of Australia outside of the The dryland subtypes of hyper-arid, arid, semi-arid and dry
central desert regions. This type of region also can be found sub-humid cover 1.1 × 107 km2, 1.9 × 107 km2, 2.3 × 107
on the poleward side of the northern Africa deserts, which km2, and 0.9 × 107 km2, accounting for 7, 13, 15, and 6 %
typically feature a Mediterranean climate. The BSk subtype of the land surface, respectively. For all of these subtypes
tends to be located in temperate zones, and the majority of of global drylands, the semi-arid region is the largest, and it
these areas are located in East Asia, Central Asia and North accounts for more than one-third of the total dryland area.
America in the Northern Hemisphere and sections of South Figure 2 shows the temporal variations in the areal cov-
America and interior southern Australia in the Southern erage of drylands and its subtypes from 1948 to 2008. The
Hemisphere. global drylands have greatly expanded because the dryland
Of the three definitions of drylands, the AI definition area in the more recent 15 years (1990–2004) is 2.6 × 106
is most reasonable and reliable, and the dryland cover- km2 (or 4 %) larger than that during 1948–1962 (Fig. 2a).
age agrees well with the surface vegetation types. Obvi- Feng and Fu (2013) also obtained similar results by analyz-
ously, precipitation alone is insufficient for defining cli- ing the observations for 1948–2008. The expansions of the
matic boundaries (Safriel and Adeel 2005). In Australia, dryland subtypes are 0.6 × 106, −0.1 × 106, 1.6 × 106, and
for example, the generally accepted limit for an arid zone 0.5 × 106 km2 for the hyper-arid, arid, semi-arid and dry
is 250 mm, but particular areas of the tropical northwest are sub-humid categories, respectively (Fig. 2b–e). The largest
clearly arid even though the mean annual rainfall exceeds dryland expansion has occurred in semi-arid regions, where
500 mm (Mabbutt 1979). In addition, the annual precipi- the area has expanded since the early 1960s, accounting
tation in Siberia is less than 500 mm, but the major land for more than half of the total dryland expansion (Fig. 2c).
cover type is forest, and the climate is humid. As inves- The area of dry sub-humid regions experienced few fluc-
tigated in previous studies, the Köppen–Geiger climate tuations prior to the 1980s, decreased in the early 1980s,

13
1136 J. Huang et al.

Fig. 2  Temporal variations in

the area (units: 106 km2) for a (a)
total drylands, b dry sub-humid
regions, c semi-arid regions,
d arid regions, and e hyper-
arid regions for the period of
1948–2008. The 15-year run-
ning mean (thick blue curves)
is applied to emphasize climate
change
(b)

(c)

(d)

(e)

and significantly expanded after the late 1980s. The areal Table 1  The area transitions (106 km2) between semi-arid regions
changes in arid regions show strong decadal variations, and other classes for 1990–2004 relative to 1948–1962
with large areas before the 1970s, reduced areas in the Semi-arid to arid Sub-humid to semi-arid Humid to semi-arid
1970s, and resumed expansion in the early 1980s; the area 1.6 2.84 0.11
for the most recent 15 years is comparable with that present Arid to semi-arid Semi-arid to sub-humid Semi-arid to humid
before the 1970s. The arid regions had the smallest areas 1.1 0.87 0.01
during the 1970s as a result of large positive precipitation
anomalies in the Southern Hemisphere, which led to a tran-
sition from arid to semi-arid regions (Feng and Fu 2013).
The AI of a semi-arid region is within the range of humid to semi-arid regions (3.0 × 106 km2). The amount
0.2–0.5, according to the definition. Transitions from of the decrease from semi-arid regions to other subtypes
semi-arid regions to other dryland subtypes will occur if is 2.5 × 106 km2, which also includes two transitions from
the AI falls outside the range of 0.2–0.5. To understand semi-arid to arid regions (1.6 × 106 km2) and semi-arid
the changes in semi-arid regions during the last 61 years, to sub-humid/humid regions (0.9 × 106 km2). Therefore,
two periods are chosen: the first 15 years (from 1948 to the net change in the semi-arid region is 1.6 × 106 km2 by
1962) and the most recent 15 years (from 1990 to 2004) subtracting the reduced areas from the newly formed areas.
of the monitoring period. Table 1 shows all of the transi- Comparing the two types of shifts, the semi-arid area that
tions in the semi-arid regions, which include arid to semi- transited from sub-humid/humid regions is nearly 3 times
arid, sub-humid/humid to semi-arid, semi-arid to arid, and larger than that from arid regions. This result indicates
semi-arid to sub-humid/humid regions. The newly formed that the newly formed semi-arid regions primarily trans-
semi-arid regions that transited from other subtypes are ited from sub-humid and humid regions, which account for
4.1 × 106 km2 and they are formed by two transitions from 73 % of the newly formed semi-arid regions. Comparing
arid to semi-arid regions (1.1 × 106 km2) and sub-humid/ the transitions between semi-arid (sub-humid/humid) and

13
Global semi-arid climate change over last 60 years 1137

Fig. 3  Global distribution of semi-arid regions and their transi- arid regions (dark red), arid to semi-arid regions (dark green), and
tions to other climate classifications, including old semi-arid regions semi-arid to sub-humid/humid regions (blue) for 1990–2004 relative
(grey), sub-humid/humid to semi-arid regions (orange), semi-arid to to 1948–1962

arid (semi-arid) regions, the area transiting from semi-arid Table 2  Latitude/longitude information for eight regions
(sub-humid/humid) to arid (semi-arid) is much larger than Region Longitude (E) Latitude (N)
that transiting from arid (semi-arid) to semi-arid (sub-
humid/humid); therefore, more semi-arid (sub-humid/ 1 East Asia 80° to 140° 30° to 60°
humid) regions become drier. 2 Central Asia 40° to 80° 30° to 60°
Figure 3 demonstrates the global distribution of semi- 3 Northern Africa −20° to 40° 5° to 30°
arid regions and their transitions to other subtypes. As 4 North America −130° to −89° 30° to 60°
shown in Fig. 3, the transition from sub-humid/humid 5 South America −77° to −50° −60° to −30°
(semi-arid) to semi-arid (arid) primarily occurs in East 6 Southern Africa 9° to 43° −40° to −10°
Asia, east Australia, and northern and southern Africa. 7 Central/West Australia 110° to 139° −40° to −10°
The transition from semi-arid (arid) to sub-humid/humid 8 East Australia 139° to 156° −40° to −10°
(semi-arid) occurs in central/west Australia, the central US,
and southern South America. The results indicate that the
transitions to drier types primarily occurred in the Eastern The shifts from sub-humid/humid to semi-arid, semi-
Hemisphere, except for in central/west Australia, whereas arid to arid, semi-arid to sub-humid/humid, and arid to
the shifts to wetter types occurred in the mid-latitudes in semi-arid for eight typical semi-arid regions are shown
North and South America. Notably, two temperate semi- in Fig. 4. The results show that, except for Central Asia
arid regions in North America and East Asia show dif- and northern Africa, 6 semi-arid areas in the most recent
ferent changes, in which more arid and semi-arid regions 15 years (1990–2004) are larger than those during 1948–
became wetter in North America but semi-arid and dry sub- 1962. The largest semi-arid expansion occurred in East
humid regions became drier in East Asia. To investigate Asia, which accounts for nearly 50 % of the global semi-
the regional climate changes in semi-arid regions globally, arid expansion. In comparing the four semi-arid regions
eight semi-arid regions are selected, i.e., East Asia, Cen- in the Northern Hemisphere, except for North America,
tral Asia, northern Africa, North America, South America, the area that shifted to a drier class is much larger than the
southern Africa, central/west Australia, and east Australia. area that shifted to a wetter class, particularly in East Asia
The latitude/longitude information is listed in Table 2 and and northern Africa (Fig. 4a, b). For example, the area that
labeled in Fig. 3. shifted to drier classes is four times larger than the area that

13
1138 J. Huang et al.

Fig. 4  The transitions from (a) (b)

sub-humid/humid to semi-arid,
semi-arid to arid, semi-arid to
sub-humid/humid, and arid to
semi-arid regions over the eight
semi-arid regions of
a East Asia, Central Asia,
northern Africa (in the Eastern
Hemisphere); b North America
(in the Northern Hemisphere);
c southern Africa, central/west
Australia, and east Australia
(in the Eastern Hemisphere);
and d South America (in the
Southern Hemisphere)
(c) (d)

Table 3  The areal changes (106 km2) in semi-arid regions over the both areas, the shifting patterns of the expansion are dif-
American continents and the continents in the Eastern hemisphere ferent. The newly formed semi-arid regions shift from arid
Change Eastern America regions, in which the climate becomes wetter within the
hemisphere American continents. Conversely, in the continents of the
Sub-humid/humid to semi-arid 2.6 0.4
Eastern Hemisphere, the expansion of semi-arid regions
replaces the sub-humid/humid regions in which the cli-
Semi-arid to arid 1.5 0.1
mate becomes drier. For example, the transition from sub-
Semi-arid to sub-humid/humid 0.4 0.5
humid/humid to semi-arid classes occurs over 2.6 × 106
Arid to semi-arid 0.5 0.6
km2 of the continental Eastern Hemisphere; this area is
five times larger than the area transited from arid to semi-
arid (0.5 × 106 km2). Within the American continents, the
shifted to wetter classes in East Asia. However, in North areal change in the sub-humid/humid to semi-arid areas
America, the area that shifted to wetter types is nearly ten (0.4 × 106 km2) is less than the areal change from arid to
times larger than the area that shifted to drier categories. semi-arid (0.6 × 106 km2). The results indicate that wet-
In the Southern Hemisphere, the climate primarily shifted ting of arid regions plays an important role in the expan-
to drier categories in southern Africa and east Australia but sion of semi-arid areas in the American continents versus
shifted to wetter categories over central/west Australia and the drying climate trend in the continents of the Eastern
South America (Fig. 4c, d). In all eight semi-arid regions Hemisphere. The causes of the different behaviors between
considered, two of the three regions that shifted to wetter the American continents and the Eastern Hemisphere are
types are within the American continents. addressed in Sect. 6.
The area changes in the semi-arid regions between the
American continents and the Eastern Hemisphere are
shown in Table 3. The semi-arid areas in the more recent 5 Semi‑arid climate change
15 years (1990–2004) are 0.4 × 106 km2 and 1.2 × 106
km2 larger than those during 1948–1962 within the Ameri- Climate change associated with changes in precipitation
can continents and in the Eastern Hemisphere, respec- and PET would lead to changes in AI and thus in the areal
tively. Although expansions of semi-arid regions occur in extent of drylands and its subtypes. The linear trends in

13
Global semi-arid climate change over last 60 years 1139

(a)

(b)

(c)

(d)

Fig. 5  The distributions of the linear trends in the AI, precipitation, PET, and SAT in semi-arid regions globally from 1948 to 2008. a P/PET
trend (/decade), b P trend (mm/decade), c PET trend (mm/decade), d temprature trend (°C/decade)

AI, precipitation, and PET in semi-arid regions from 1948 including northeastern China, northern Mongolia, east Aus-
to 2008 are shown in Fig. 5. The AI decreases, suggest- tralia, the Sahel in northern Africa, southern Africa, and
ing drying trends during the last 61 years over East Asia, Mediterranean regions, while wetter trends are apparent in

13
1140 J. Huang et al.

(a) (d) (g)

(b) (e) (h)

(c) (f) (i)

Fig. 6  The variations in the regionally averaged precipitation, PET, regions, and the black curve denotes the old semi-arid regions. The
and AI in the semi-arid regions over East Asia (left column), Cen- thick blue (black) line illustrates the linear trend for the curve, and a
tral Asia (middle column), and North America (right column) from asterisk indicates a trend that exceeds the 99 % confidence level. a, d,
1948 to 2008. The blue curve indicates the newly formed semi-arid g Annual precipitation. b, e, h Annual PET. c, f, i Annual P/PET

central/west Australia, South America, North America, and trend is particularly pronounced in the eastern Sahel and
other small areas, such as northern Central Asia and west- higher latitudes over the Northern Hemisphere, i.e., East
ern China (Fig. 5a). Precipitation decreases in the Sahel, and Central Asia (Fig. 5d). The different trends between
southern Africa, east Australia, and East Asia but increases the PET and temperature in particular regions, e.g., North
in central/west Australia and within the American conti- America and southern Africa, imply that temperature is not
nents (Fig. 5b). The patterns of precipitation are consistent always the primary factor controlling PET (Donohue et al.
with those for AI, with only particular regional differences. 2010; Shaw and Riha 2011; Hobbins et al. 2012), which is
This result suggests that precipitation is a key factor that also influenced by other factors, such as, winds, net radia-
affects AI. The PET decreases over the American conti- tion, and humidity.
nent, but PET increases over other Northern Hemisphere Figure 6 shows the variations in the regionally averaged
continents. However, this pattern is reversed in the South- precipitation, PET, and AI in three mid-latitude semi-arid
ern Hemisphere, and the PET increases in South America regions over East Asia, Central Asia, and North America
but decreases in southern Africa and northern Australia in the Northern Hemisphere. Considering that the newly
(Fig. 5c). During the same period, the surface air temper- formed semi-arid regions transited from other climate
ature (SAT) warms in most of the semi-arid regions; this types, they are different from the semi-arid regions of

13
Global semi-arid climate change over last 60 years 1141

the same domain already in place; thus, each of the semi- formed semi-arid regions are much larger than those in the
arid regions is divided into old and newly formed semi- old semi-arid regions in the drying semi-arid regions over
arid regions. In East Asia, precipitation decreases in both East Asia and Central Asia. In contrast, the variations in the
areas, but the decrease in the precipitation is 57 mm in precipitation, PET and AI show little difference between
the newly formed semi-arid regions; this value is 3 times the newly formed and old semi-arid regions in the wetting
higher than that in the old semi-arid regions (Fig. 6a). region over North America.
The PET increases by 132 mm in the newly formed Northern Africa is also an important semi-arid region,
semi-arid regions; this value is two times higher than which has received more attention because of the scarcity
that in the old semi-arid regions (Fig. 6b). The decrease of rainfall and the long periods of drought that occurred in
in precipitation and increase in PET lead to a significant 1970s and 1980s (Dai et al. 2004; Hulme et al. 2001). The
decrease in AI, which indicates that two types of semi-arid PET increases in both the old and newly formed semi-arid
regions became drier over the past 61 years, particularly regions, and it has remained at a stable level since the 1980s
in the newly formed semi-arid regions (Fig. 6c). The PET (Fig. 7b). The precipitation and AI both decrease signifi-
increases significantly in both types of semi-arid regions cantly in the old and newly formed semi-arid regions, and
in Central Asia (Fig. 6e). The rates of change in precipita- the time series indicates strong decadal variations (Fig. 7a,
tion between the old and newly formed semi-arid regions c). The maximum values occur in the early 1950s; then, the
are remarkably different (Fig. 6d). The regionally aver- values decrease and remain steady with reduced precipita-
aged precipitation shows a sharp decrease of 81 mm in tion and AI until the mid-1980s. These values have partially
the newly formed semi-arid regions; this value is nearly 9 recovered since the driest period in the mid-1980s. This
times higher than that in old semi-arid regions. Thus, the decadal variation in precipitation was also demonstrated in
decrease in the AI in the newly formed semi-arid regions
is 4 times larger than that in the old semi-arid regions
(Fig. 6f). The time series of regionally averaged precipita- (a)
tion, PET, and AI are consistent between the old and newly
formed semi-arid regions, without significant changes in
North America (Fig. 6g–i). The PET shows strong dec-
adal variations, with the largest value occurring during the
1950s, observed decreases in the mid-1960s, and minimum
values in the mid-1980s before increasing again in the
early 1990s. The upward trend in precipitation and down-
ward trend in PET leads to an increasing trend in the AI
in both the old and newly formed semi-arid regions; thus (b)
indicates the semi-arid regions became wetter over the past
61 years. The annual variability in the AI is completely
consistent with that of precipitation, which indicates that
the precipitation is a dominating factor in the wetting trend
of North America. Although historical records reveal that
droughts occurred in the late 1980s and at the beginning
of the 21st century, the higher AI and precipitation in the
most recent 20 years provide evidence that the current cli-
mate is indeed becoming wetter compared with the climate (c)
of the 1950s in North America.
Comparing these regions in Fig. 6, the regionally aver-
aged precipitation, PET, and AI show consistent upward
or downward trends between the newly formed and old
semi-arid regions. Precipitation decreases in East Asia
and Central Asia but increases in North America, and the
PET shows an increasing trend in East Asia and Central
Asia, in contrast to the decreasing trend in North America.
As a result, the AI is consistent with precipitation, which
reveals that the climate becomes drier in East Asia and
Central Asia but becomes wetter in North America. Nota- Fig. 7  Same as Fig. 6 but for northern Africa semi-arid regions. a
bly, the changes of precipitation, PET and AI in the newly Annual precipitation. b Annual PET. c AnnualP/PET

13
1142 J. Huang et al.

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

Fig. 8  Same as Fig. 6 but for southern Africa (left column), central/west Australia (middle column), and east Australia (right column) semi-arid
regions. a–c Annual precipitation. d–f Annual PET. g–i Annual P/PET

previous studies (Dai et al. 2004; Hulme et al. 2001; Zeng Africa; thus, the drying trend in the newly formed semi-
et al. 1999). arid regions in East Asia is more sensitive to the decrease in
For two severely drying semi-arid regions in the North- precipitation than the trend in northern Africa because the
ern Hemisphere over East Asia and northern Africa, the climatological PET of the newly formed semi-arid regions
decrease in the precipitation in the old semi-arid regions in northern Africa (1785 mm) is much larger than that in
over northern Africa is 8 times larger than the changes in East Asia (902 mm).
East Asia; the rate of change in the decreasing PET also Figure 8 shows the time series of the regionally averaged
shows a rate twice as large as that in East Asia. As a result, precipitation, PET, and AI in semi-arid regions over south-
the drying trend in the old semi-arid regions over north- ern Africa, central/west Australia, and east Australia in the
ern Africa is more severe than that in East Asia. In con- low latitudes of the Southern Hemisphere. The decreasing
trast, during the same period, the precipitation decreases trends in the AI in southern Africa and east Australia reveal
by 173 mm in the newly formed semi-arid regions over that the semi-arid regions became drier (Fig. 8g, i). Fur-
northern Africa, which is 3 times higher than that in East thermore, the rates of change in the AI in the newly formed
Asia (57 mm); only a slight difference is observed in the areas are much larger than the changes in the old semi-arid
variability in the PET between East Asia and northern regions. In central/west Australia, the annual variations and
Africa. However, the AI changes are comparable in the linear trends in the regionally averaged precipitation, PET,
newly formed semi-arid regions in East Asia and northern and AI are all consistent between the old and new semi-arid

13
Global semi-arid climate change over last 60 years 1143

Fig. 9  Linear trends in the mean geopotential height (1 gpm/decade, interval = 2 gpm/decade) at 850 hPa using the NCEP/NCAR re-analyses
for a DJF and b JJA

regions (Fig. 8d–f). The increasing trend in the precipita- However, the effects of these factors on climate change in
tion and significant decreasing trend in the PET contribute semi-arid regions might vary by region. In this section, we
to an upward trend in the AI, which indicates that the cli- discuss the causes of the different behaviors in semi-arid
mate became wetter in central/west Australia, even though climate changes over the eight typical semi-arid regions via
the wetting trend is not significant. Similar results for the the effect of planetary-scale internal dynamical processes
regionally averaged precipitation and PET show significant and atmospheric-ocean interactions.
increases in the old and newly formed semi-arid regions First, the linear trend in the atmospheric circulations
over South America. The AI shows upward trends of 0.06 from the NCEP/NCAR re-analyses during the period of
and 0.04, which indicate that the semi-arid regions became 1948–2008, including the geopotential heights (HGT) and
wetter over the last 61 years in South America (figure not winds at 850 hPa are examined, as indicated in Figs. 9 and
shown). 10. Comparing the influences of the westerly circulation
and monsoon circulation on semi-arid areas in North Amer-
ica and East Asia, North America is more strongly influ-
6 Dynamics of climate change in semi‑arid regions enced by the westerly circulation, but East Asia is more
strongly influenced by the monsoon circulation (Trenberth
Semi-arid climate change may be caused by both global- et al. 2000; Wang and Ding 2006). Under global warm-
and regional-scale factors. The global-scale factors include ing, the temperature rise has not been evenly distributed
planetary-scale internal dynamical processes, SST anoma- over the globe; the North Pacific has cooled along 40°N
lies, well-mixed greenhouse effects and external forcing. (figure not shown; IPCC 2007), and the Aleutian Low has

13
1144 J. Huang et al.

Fig. 10  Linear trends in the wind (vectors, 1 m/s decade) and mean zonal wind (shading, 1 m/s decade) at 850 hPa using the NCEP/NCAR re-
analyses in a DJF and b JJA

been deepening, particularly in winter (Fig. 9). This long- pressure center accompanied by anomalously strong north-
term strengthening of the Aleutian Low causes an increase erlies along its eastern flank (Fig. 10b), implying the East
in Western US cool-season precipitation (e.g. Huang et al. Asian summer monsoon (EASM) is weakening (Fig. 11).
2005; Dai 2013b), highly owing to the accelerated west- This result is consistent with the previous studies (Xu et al.
erly along the southern flank of the deepened Aleutian Low 2006; Zhu et al. 2012; Zuo et al. 2012). This is in related
(Fig. 10). This enhanced westerly favors substantially more to a reduction in the meridional thermal gradients between
moisture-rich air and storm tracks from Pacific to the west Asian continent and adjacent oceans due to pronounced
North America, thereby favoring precipitation and contrib- warming over the tropics (IPCC 2007; Ueda et al. 2006; He
uting to the wetting trend in this region. Additionally, warm et al. 2014). Because the EASM is weakening, the moisture
Atlantic SSTs also influence the precipitation over the advected by the monsoon cannot reach the margin of semi-
western US, particularly in summer (Kushnir et al. 2010; arid area; thus, drying is induced over these regions. Over-
Feng et al. 2011), primarily by modulating the impact of all, the two different circulation systems, i.e., the wester-
the El Niño–Southern Oscillation (ENSO)-like SST forc- lies and the EASM, play decisive roles in the wetting of the
ing in the Pacific, which is comparatively weak (Mo et al. semi-arid climate in North America and the drying in East
2009). However, for the semi-arid region in East Asia, the Asia, respectively.
climate is largely controlled by the Asia monsoon system, In contrast to East Asia, the moisture of Central Asia is
particularly the summer monsoon. As shown in Fig. 9b, primarily delivered by the prevailing westerlies in the mid-
almost all of East Asia is controlled by an anomalous high latitudes (Chen et al. 2010). Because of the anomalous

13
Global semi-arid climate change over last 60 years 1145

Fig. 11  The annual mean time

series of the East Asia summer
monsoon index. The black
curves indicate the 11-year
Gaussian-type filtered values,
and the dashed lines denote the
linear trend in the curves

high pressure center over most of Asia (Fig. 9), anomalous accompanied with broad rising pressure trend over south-
easterlies exist to the south and anomalous westerlies exist ern Africa. These changes in the SST and regional circula-
to the north of Central Asia, implying that the prevailing tion inhibit convection and induce a drying trend over the
westerlies are becoming stronger in the north and weaker semi-arid region in southern Africa. In addition, the rainfall
in the south (Fig. 10). Similar circulation trend happens in in southern Africa is also affected by ENSO via the influ-
boreal spring and autumn as well. These changes induce ence on the SST fluctuations in the western Indian Oceans,
the wetting in the northernmost regions of Central Asia and with drier than normal conditions during El Niño and wet-
the drying over most of the southern area, as indicated in ter conditions during La Niña (Nicholson and Kim 1997;
Fig. 5b, whereas the semi-arid climate is drier, on average, Richard et al. 2000; Mason 2001; Reason and Rouault
across all of Central Asia, particularly under the influence 2002; Misra 2003; Kane 2009). During the last half of the
of the pronounced warming in this region. 20th century, El Niño events strengthened, occurred more
The climate over the semi-arid region in northern Africa frequently, and had a more stable relationship with South
is inextricably linked to the West African monsoon (WAM) Africa because of the increasing trend in the Pacific Dec-
system, which blows from the tropical Atlantic Ocean adal Oscillation (PDO) index, thereby worsening the semi-
across the Sahel and brings rain to this region, primarily arid climate over southern Africa.
during the boreal summer (Nicholson 2013; Nicholson and The annual rainfall has changed significantly in recent
Kim 1997). Many results with various atmospheric mod- decades in the semi-arid region of Australia. Precipita-
els forced by the long-term global sea surface temperature tion has declined in east Australia but has shown a marked
(SST) reveal that the oceans control the change in the pre- increasing trend in central/west Australia, mainly during
cipitation in the African Sahel (Bader and Latif 2003; Gian- the austral summer season (Wardle and Smith 2004; Shi
nini et al. 2003; Hagos and Cook 2008; Lu and Delworth et al. 2008). The marked changes over the Indian Ocean,
2005). Although warming of the tropical Indian Oceans is particularly since 1970, could trigger the overlying atmos-
a non-negligible component of historical Sahelian drought pheric circulation that generates anomalous ascending
(Bader and Latif 2003; Hagos and Cook 2008; Lu and Del- motion west of 140°E and anomalous subsidence east of
worth 2005), the Atlantic Ocean SSTs played an impor- 140°E; these changes are responsible for the wetting over
tant role in the recovering rainfall in the 1990s (Hagos central/west Australia (promoting convection) and dry-
and Cook 2008; Hoerling et al. 2006; Ting et al. 2009). ing observed over east Australia (inhibiting convection)
As shown in Fig. 9b, an anomalously high pressure center (Taschetto and England 2008; Li et al. 2012). In addition,
occurs around the Sahel in the long-term trend, and the an anomalous anticyclonic circulation at low levels in the
land-sea low-level pressure gradient is weakened, which east tends to weaken the easterlies from the tropical west-
drives a weaker monsoon flow and induces a drying trend ern Pacific, thus diminishing the transport of moisture and
in this semi-arid region. exacerbating the drying over east Australia. An anomalous
The semi-arid region in southern Africa receives the cyclonic circulation over central/west Australia brings
most rainfall in the austral summer season, primarily due moist air to the northwest, leading to an increase in rainfall
to the tropical-extratropical influences and associated con- over the semi-arid region in central/west Australia (Fig. 9a).
vection, which are strongly influenced by the surround- The drying trend over east Australia could be associated
ing oceans, particularly the Indian Ocean. The Indian with ENSO (Taschetto and England 2008; Shi et al. 2008).
Ocean has warmed markedly since 1970 (IOCI 2002), In approximately 1976, the PDO entered a warm phase,

13
1146 J. Huang et al.

with an increasing frequency of El Niño events and a dry- and AI in the newly formed semi-arid regions are much
ing trend over east Australia. larger than those in the old semi-arid regions in the drying
The semi-arid region in South America is located lee- semi-arid regions. However, the precipitation, PET and AI
ward of the Andes Mountains, which are oriented north– show similar trends, with only slight differences between
south, and is strongly influenced by the typical westerly the newly formed and old areas in the wetting semi-arid
flow (Garreaud et al. 2013; Garreaud 2007; Parker et al. regions that transitioned from drier climate types.
2007). Most of the humid air masses transported by the Notably, two temperate semi-arid regions in North
westerlies from the Pacific are blocked by the Andes, and America and East Asia show different climatic patterns.
downslope subsidence dries the area east of the Andes via The distinct behaviors of the old and newly established
adiabatic warming (Paruelo et al. 1998; Garreaud et al. semi-arid regions in North America and East Asia reveal
2013). Therefore, the precipitation over the semi-arid the different responses of semi-arid areas to the decadal
region east of the Andes is strongly and negatively corre- variability in the westerlies and monsoon systems under
lated with the 850-hPa zonal wind (Garreaud 2007; Gar- global warming. The climate change in the drying semi-
reaud et al. 2013). During the last six decades, re-analysis arid region over East Asia is mainly dominated by a weaker
data suggested that an anomalously high pressure center East Asian summer monsoon, whereas the wetting semi-
existed over Patagonia (Fig. 9a, b), accompanied by anom- arid regions over North America are mainly controlled
alous easterlies (westerlies) along the northern (southern) by enhanced westerlies, which are related to the reduced
flank (Fig. 10a, b). This high pressure trend is similar to the ocean-land temperature gradient under global warming (He
result shown by Thompson and Wallace (2000), who indi- et al. 2014).
cated that these results to be at least qualitatively reliable Furthermore, the decadal variations and long-term trends
despite concerns about the reliability of the NCEP–NCAR in sea surface temperatures (SSTs) of the world oceans also
reanalysis over this region (e.g. Randel and Wu 1999; play a crucial role in the widespread drying over semi-arid
Marshall 2003). Accordingly, the westerly flow tended to regions by providing boundary forcing, as well as initiat-
decrease in the mid-latitudes (approximately 45°S); there- ing and sustaining particular circulation anomalies in the
fore, the anomalous onshore winds bring additional mois- atmosphere at the corresponding time scales (Kushnir et al.
ture and favoring a wetting trend in the semi-arid region to 2002; Alexander 2013). Evidence indicates that the wide-
the east of the Andes. spread drying that has occurred over the decades in semi-
arid regions is also controlled by atmospheric-ocean inter-
actions (Huang et al. 1998; Wang et al. 2014). Studies using
7 Conclusions and discussion atmospheric models forced by long-term SSTs support the
claim of a controlling role of the oceans on the climate of
Of all subtypes of global drylands, semi-arid regions have the semi-arid region in the Sahel (Bader and Latif 2003;
the largest area, which accounts for more than one-third of Giannini et al. 2003; Hoerling et al. 2006; Lu 2009; Lu and
the total dryland area. The largest expansion of drylands Delworth 2005) and other extra-tropical semi-arid regions
also occurs in semi-arid regions, which account for half (McCabe et al. 2008; Méndez and Magaña 2010). Warm-
of the total dryland expansion. Globally, semi-arid regions ing of the tropical oceans is a non-negligible component of
have expanded over the last 61 years (1948–2008), and the historical Sahelian droughts (Hagos and Cook 2008; Ting
area during 1990–2004 is 1.6 × 106 km2 (or 7 %) larger et al. 2009), and the current recovery can be attributed to
than that during 1948–1962. The expansion of semi-arid the variability in the Atlantic Ocean (Hoerling et al. 2006;
regions across the continents in the Eastern Hemisphere Ting et al. 2009). Warming in the tropical Pacific could
accounts for 75 % of the total expanded semi-arid regions weaken summer monsoons and thus drier conditions in
(1.6 × 106 km2) than that of 25 % across the American con- eastern China (Li et al. 2010; Zhou et al. 2009).
tinents. Although expansions of semi-arid regions occur in In addition to the effect of the oceans, previous stud-
both areas, the shifting patterns of the expansion are dif- ies also emphasized that the interaction between land
ferent. The newly formed semi-arid regions are shifted surfaces (land use and land cover change) and the atmos-
from arid regions, where the climate becomes wetter, over phere is critical for drying trends in semi-arid regions
the American continents. Conversely, in the continents (Huang et al. 2008; Guan et al. 2009; Wang et al. 2010).
of the Eastern Hemisphere, the expansion of semi-arid A positive feedback process was proposed by Otterman
regions replaces the sub-humid/humid regions, where the (1974) and explained by (Charney 1975) as follows: bare
climate becomes drier. In all eight semi-arid regions con- soil reduces the net radiation at the surface, thus creating
sidered, five of them become drier, but another three, as evaporation and moist static energy declines, which induce
North America, South America and central/west Australia, drier and warmer conditions (Nicholson 2000; Xue 1996;
become wetter. The rates of change in precipitation, PET Xue and Shukla 1993; Zeng et al. 1999). Increased local

13
Global semi-arid climate change over last 60 years 1147

anthropogenic dust aerosols associated with human activi- millennium in arid central Asia: a review, synthesis and com-
ties, such as agriculture and industrial activities might parison with monsoon region. Quat Sci Rev 29(7–8):1055–1068
Chou C, Neelin J, Chen C, Tu J (2009) Evaluating the “rich-get-
play an important role in the drying trends over the semi- richer” mechanism in tropical precipitation change under global
arid regions (Cook et al. 2009; Su et al. 2008; Huang et al. warming. J Clim 22(8):1982–2004
2010, 2011, 2014; Mulitza et al. 2010; Wang et al. 2013). Compo GP, Whitaker JS, Sardeshmukh PD, Matsui N, Allan RJ, Yin
Thus, human activities can also lead to significant change X, Gleason BE, Vose RS, Rutledge G, Bessemoulin P, Brönni-
mann S, Brunet M, Crouthamel RI, Grant AN, Groisman PY,
over semi-arid regions. An attribution of human-induced Jones PD, Kruk MC, Kruger AC, Marshall GJ, Maugeri M,
land use change is needed to examine the anthropogenic Mok HY, Nordli Ø, Ross TF, Trigo RM, Wang XL, Woodruff
contribution. Effort has been made to discriminate between SD, Worley SJ (2011) The twentieth century reanalysis project.
climate and human-induced dryland changes (Evans and Q J Roy Meteor Soc 137(654):1–28. doi:10.1002/qj.776
Cook BI, Miller RL, Seager R (2009) Amplification of the North
Geerken 2004; Omuto et al. 2010; Wessels et al. 2007). American “Dust Bowl” drought through human-induced land
However, these studies are mainly case studies that require degradation. Proc Natl Acad Sci USA 106(13):4997–5001
high-quality satellite observations (e.g., normalized differ- Dai A (2013a) Increasing drought under global warming in observa-
ence vegetation index data (NDVI)). Therefore, the change tions and models. Nat Clim Change 3(1):52–58
Dai A (2013b) The influence of the inter-decadal Pacific oscil-
in global human-induced semi-arid regions requires further lation on US precipitation during 1923–2010. Clim Dyn
research. 41(3–4):633–646
All of the factors discussed contribute to the climate Dai A, Lamb PJ, Trenberth KE, Hulme M, Jones PD, Xie P (2004) The
change observed over semi-arid regions. The drying or wet- recent Sahel drought is real. Int J Climatol 24(11):1323–1331
Donohue RJ, McVicar TR, Roderick ML (2010) Assessing the abil-
ting trends in semi-arid regions might be dominated by SST ity of potential evaporation formulations to capture the dynam-
anomalies or by local land surface processes due to human ics in evaporative demand within a changing climate. J Hydrol
activities and global warming. Further studies are needed to 386(1):186–197
clarify the contributions of these factors. Dregne H (2002) Land degradation in the drylands. Arid Land Res
Manag 16(2):99–132
Evans J, Geerken R (2004) Discrimination between climate and human-
Acknowledgments This work was jointly supported by the National induced dryland degradation. J Arid Environ 57(4):535–554
Basic Research Program of China (2012CB955301), National Sci- Fan Y, Van den Dool H (2008) A global monthly land surface air
ence Foundation of China (41175134 and 41305060), the Fundamen- temperature analysis for 1948–present. J Geophys Res Atmos
tal Research Funds for the Central Universities (lzujbky-2015-ct03), (1984–2012) 113(D1):D01103. doi:10.1029/2007JD008470
and China 111 project (B13045). Feddema J (2005) A revised Thornthwaite-type global climate clas-
sification. Phys Geogr 26(6):442–466
Open Access This article is distributed under the terms of the Crea- Feng S, Fu Q (2013) Expansion of global drylands under a warming
tive Commons Attribution 4.0 International License (http://creativecom- climate. Atmos Chem Phys 13(19):10081–10094
mons.org/licenses/by/4.0/), which permits unrestricted use, distribution, Feng S, Hu Q, Oglesby R (2011) Influence of Atlantic sea surface
and reproduction in any medium, provided you give appropriate credit temperatures on persistent drought in North America. Clim Dyn
to the original author(s) and the source, provide a link to the Creative 37:569–586
Commons license, and indicate if changes were made. Feng S, Hu Q, Huang W, Ho C, Li R, Tang Z (2014) Projected climate
regime shift under future global warming from multi-model, multi-
scenario CMIP5 simulations. Global Planet Change 112:41–52
References Fricke K, Sterr T, Bubenzer O, Eitel B (2009) The oasis as a megac-
ity: Urumqi’s fast urbanisation in a semiarid environment. Erde
Allen R, Pereira L, Raes D, Smith M (1998) Crop evapotranspiration 140(4):449–463
—guidelines for computing crop water requirements. FAO Irri- Gao Q, Kang M, Xu H, Jiang Y, Yang J (2010) Optimization of land
gation and Drainage Paper 56. FAO, Rome use structure and spatial pattern for the semi-arid loess hilly–
Alexander M (2013) Extratropical air–sea interaction, sea surface gully region in China. Catena 81(3):196–202
temperature variability, and the Pacific decadal oscillation. In: Garreaud R (2007) Precipitation and circulation covariability in the
Sun D, Bryan F (eds) Climate dynamics: why does climate extratropics. J Clim 20(18):4789–4797. doi:10.1175/JCLI4257.1
vary? American Geophysical Union, Washington, pp 123–148 Garreaud R, Lopez P, Minvielle M, Rojas M (2013) Large-scale
Bader J, Latif M (2003) The impact of decadal-scale Indian Ocean control on the Patagonian climate. J Clim 26(1):215–230.
sea surface temperature anomalies on Sahelian rainfall and the doi:10.1175/jcli-d-12-00001.1
North Atlantic Oscillation. Geophys Res Lett 30(22):2169. doi: Geiger R (1954) Klassifikation der Klimate nach W. Köppen. In:
10.1029/2003GL018426 Landolt-Börnstein–Zahlenwerte und Funktionen aus Physik,
Bounoua L, Safia A, Masek J, Peters-Lidard C, Imhoff M (2009) Chemie, Astronomie, Geophysik und Technik, vol 3. Springer,
Impact of urban growth on surface climate: a case study in Berlin, pp 603–607
Oran, Algeria. J Appl Meteorol 48(2):217–231 Georgescu M, Miguez-Macho G, Steyaert L, Weaver C (2009) Cli-
Charney J (1975) Dynamics of deserts and drought in the Sahel. Q J R matic effects of 30 years of landscape change over the Greater
Meteorol Soc 101(428):193–202 Phoenix, Arizona, region: 1. Surface energy budget changes. J
Chen M, Xie P, Janowiak JE, Arkin PA (2002) Global land precipita- Geophys Res Atmos (1984–2012) 114(D5):D05110. doi:10.10
tion: a 50-yr monthly analysis based on gauge observations. J 29/2008JD010745
Hydrometeorol 3:249–266 Giannini A, Saravanan R, Chang P (2003) Oceanic forcing of Sahel
Chen F, Chen J, Holmes J, Boomer I, Austin P, Gates J, Wang N, rainfall on interannual to interdecadal time scales. Science
Brooks S, Zhang J (2010) Moisture changes over the last 302(5647):1027–1030

13
1148 J. Huang et al.

Greve P, Orlowsky B, Mueller B, Sheffield J, Reichstein M, Senevi- Kane R (2009) Periodicities, ENSO effects and trends of some
ratne SI (2014) Global assessment of trends in wetting and dry- South African rainfall series—an update. S Afr J Sci
ing over land. Nat Geosci 7(10):716–721 105(5–6):199–207
Guan X, Huang J, Guo N, Bi J, Wang G (2009) Variability of soil Köppen W (1884): Die Wärmezonen der Erde, nach der Dauer der
moisture and its relationship with surface albedo and soil heissen, gemässigten und kalten Zeit und nach der Wirkung der
thermal parameters over the Loess Plateau. Adv Atmos Sci Wärme auf die organische Welt betrachtet (The thermal zones
26(4):692–700. doi:10.1007/s00376-009-8198-0 of the earth according to the duration of hot, moderate and cold
Hagos SM, Cook KH (2008) Ocean warming and late-twentieth-cen- periods and to the impact of heat on the organic world). –Mete-
tury Sahel drought and recovery. J Clim 21(15):3797–3814 orol. Z. 1, 215–226. (trans, ed: Volken E, Brönnimann S. Mete-
Hamon WR (1963) Computation of direct runoff amounts from storm orol. Z. 20 (2011), 351–360)
rainfall. Int Assoc Sci Hydrol Pub 63:52–62 Köppen W (1936) Das geographische System der Klimate. In: Kop-
Hargreaves G, Samani Z (1985) Reference crop evapotranspiration pen W, Geiger R (eds) Handbuch der Klimatologie, vol 1, Gebr-
from temperature. Appl Eng Agric 1:96–99 Borntrager, Berlin, pp 1–44
He Y, Huang J, Ji M (2014) Impact of land–sea thermal contrast on Kottek M, Grieser J, Beck C, Rudolf B, Rubel F (2006) World map of
interdecadal variation in circulation and blocking. Clim Dyn. the Koppen–Geiger climate classification updated. Meteorol Z
doi:10.1007/s00382-00014-02103-y 15(3):259–263
Held IM, Soden BJ (2006) Robust responses of the hydrological cycle Kushnir Y, Robinson W, Bladé I, Hall N, Peng S, Sutton R (2002)
to global warming. J Clim 19(21):5686–5699 Atmospheric GCM response to extratropical SST anomalies:
Hobbins M, Wood A, Struebel D, Werner K (2012) What drives the synthesis and evaluation. J Clim 15:2233–2256
variability of evaporative demand across the conterminous Kushnir Y, Seager R, Ting M, Naik N, Nakamura J (2010) Mecha-
United States? J Hydrometeorol 13:1195–1214 nisms of tropical Atlantic SST influence on North American
Hoerling M, Hurrell J, Eischeid J, Phillips A (2006) Detection and precipitation variability. J Clim 23:5610–5628
attribution of twentieth-century northern and southern African Legates DR, Willmott CJ (1990a) Mean seasonal and spatial vari-
rainfall change. J Clim 19:3989–4008 ability in gauge-corrected, global precipitation. Int J Climatol
Huang J, Higuchi K, Shabbar A (1998) The relationship between the 10(2):111–127
North Atlantic oscillation and El Niño-Southern oscillation. Legates DR, Willmott CJ (1990b) Mean seasonal and spatial vari-
Geophys Res Lett 25:2707–2710 ability in global surface air temperature. Theor Appl Climatol
Huang H, Seager R, Kushnir Y (2005) The 1976/77 transition in pre- 41(1–2):11–21
cipitation over the Americas and the influence of tropical sea Li J, Zeng Q (2002) A unified monsoon index. Geophys Res Lett
surface temperature. Clim Dyn 24(7–8):721–740 29(8):1274. doi:10.1029/2001GL013874
Huang J, Zhang W, Zuo J, Bi J, Shi J, Wang X, Chang Z, Huang Z, Li H, Dai A, Zhou T, Lu J (2010) Responses of East Asian summer
Yang S, Zhang B, Wang G, Feng G, Yuan J, Zhang L, Zuo H, monsoon to historical SST and atmospheric forcing during
Wang S, Fu C, Chou J (2008) An overview of the semi-arid cli- 1950–2000. Clim Dyn 34(4):501–514
mate and environment research observatory over the Loess Pla- Li A, Wu J, Huang J (2012) Distinguishing between human-induced
teau. Adv Atmos Sci 25(6):1–16 and climate-driven vegetation changes: a critical application of
Huang J, Minnis P, Yan H, Yi Y, Chen B, Zhang L, Ayers J (2010) RESTREND in inner Mongolia. Landsc Ecol 27(7):969–982
Dust aerosol effect on semi-arid climate over Northwest China Liu C, Xia J (2004) Water problems and hydrological research in
detected from A-Train satellite measurements. Atmos Chem the Yellow River and the Huai and Hai River basins of China.
Phys 10(14):6863–6872 Hydrol Process 18(12):2197–2210
Huang J, Fu Q, Zhang W, Wang X, Zhang R, Ye H, Warren S (2011) Dust Liu H, Yin Y, Tian Y, Ren J, Wang H (2008) Climatic and anthropo-
and black carbon in seasonal snow across northern China. Bull Am genic controls of topsoil features in the semi-arid East Asian
Meteorol Soc 92(2):175–181. doi:10.1175/2010BAMS3064.1 steppe. Geophys Res Lett 35(4):L04401. doi:10.1029/200
Huang J, Guan X, Ji F (2012) Enhanced cold-season warming in 7GL032980
semi-arid regions. Atmos Chem Phys 12(12):5391–5398 Lu J (2009) The dynamics of the Indian Ocean sea surface tempera-
Huang J, Wang T, Wang W, Li Z, Yan H (2014) Climate effects of dust ture forcing of Sahel drought. Clim Dyn 33(4):445–460
aerosols over East Asian arid and semiarid regions. J Geophys Lu J, Delworth TL (2005) Oceanic forcing of the late 20th century
Res Atmos 119:11398–11416. doi:10.1002/2014JD021796 Sahel drought. Geophys Res Lett 32(22):L22706. doi:10.
Hulme M, Doherty R, Ngara T, New M, Lister D (2001) African cli- 1029/2005GL023316
mate change: 1900–2100. Clim Res 17(2):145–168 Mabbutt JA (1979) Desert landforms. MIT, Cambridge, p 340
IOCI (2002) Climate variability and change in southwest Western Makkink G (1957) Testing the Penman formula by means of Lysim-
Australia. Technical report, Indian Ocean Climate Initiative eters. J Inst of Water Eng 11:277–288
Panel, Perth, p 34 Marshall G (2003) Trends in the Southern Annular Mode from Obser-
IPCC (2007) Climate change 2007: the physical science basis. In: vations and Reanalyses. J Clim 16:4134–4143
Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt K, Mason S (2001) El Niño, climate change, and Southern African cli-
Tignor M, Miller H (eds) Contribution of Working Group I to mate. Environmetrics 12:327–345
the fourth assessment report of the Intergovernmental Panel on McAfee S (2013) Methodological differences in projected potential
Climate Change. Cambridge University Press, Cambridge evapotranspiration. Clim Change 120:915–930
Ji F, Wu Z, Huang J, Eric P (2014) Evolution of land surface air tem- McCabe GJ, Betancourt JL, Gray ST, Palecki MA, Hidalgo HG
perature trend. Nat Clim Change. doi:10.1038/nclimate2223 (2008) Associations of multi-decadal sea-surface temperature
Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D, Gandin variability with US drought. Quat Int 188(1):31–40
L, Iredell M, Saha S, White G, Woollen J, Zhu Y, Leetmaa A, McVicar TR, Roderick ML, Donohue RJ, Li LT, Van Niel TG,
Reynolds R, Chelliah M, Ebisuzaki W, Higgins W, Janowiak J, Thomas A, Grieser J, Jhajharia D, Himri Y, Mahowald NM
Mo K, Ropelewski C, Wang J, Jenne R, Joseph D (1996) The (2012) Global review and synthesis of trends in observed ter-
NCEP/NCAR 40-Year Reanalysis Project. Bull Am Meteorol restrial near-surface wind speeds: implications for evaporation.
Soc 77:437–471 J Hydrol 416:182–205

13
Global semi-arid climate change over last 60 years 1149

Méndez M, Magaña V (2010) Regional aspects of prolonged mete- Schiemann R, Luthi D, Vidale P, Schar C (2008) The precipitation
orological droughts over Mexico and Central America. J Clim climate of Central Asia—intercomparison of observational and
23(5):1175–1188 numerical data sources in a remote semiarid region. Int J Clima-
Middleton N, Thomas D (1997) World atlas of desertification, 2nd tol 28:295–314
edn. Arnold, a member of the Hodder Headline Group, London Schwinning S, Sala OE, Loik ME, Ehleringer JR (2004) Thresholds,
Misra V (2003) The influence of Pacific SST variability on the pre- memory, and seasonality: understanding pulse dynamics in arid/
cipitation over southern Africa. J Clim 16:2408–2418 semi-arid ecosystems. Oecologia 141(2):191–193
Mitchell T, Jones PD (2005) An improved method of constructing a Seager R, Naik N, Vecchi GA (2010) Thermodynamic and dynamic
database of monthly climate observations and associated high- mechanisms for large-scale changes in the hydrological cycle in
resolution grids. Int J Climatol 25(6):693–712 response to global warming. J Clim 23(17):4651–4668
Mo K, Schemm J, Yoo S (2009) Influence of ENSO and the Atlan- Shaw S, Riha S (2011) Assessing temperature-based PET equations
tic multidecadal oscillation on drought over the United States. J under a changing climate in temperate, deciduous forests.
Clima 22:5962–5982 Hydrol Process 25:1466–1478
Monteith J (1981) Evaporation and surface temperature. Q J Roy Sheffield J, Goteti G, Wood E (2006) Development of a 50-year high-
Meteor Soc 107(451):1–27 resolution global dataset of meteorological forcings for land
Mulitza S, Heslop D, Pittauerova D, Fischer HW, Meyer I, Stuut surface modeling. J Clim 19(13):3088–3111
J-B, Zabel M, Mollenhauer G, Collins JA, Kuhnert H (2010) Sheffield J, Wood EF, Roderick ML (2012) Little change in global
Increase in African dust flux at the onset of commercial agricul- drought over the past 60 years. Nature 491(7424):435–438
ture in the Sahel region. Nature 466(7303):226–228 Shi G, Cai W, Cowan T, Ribbe J, Rotstayn L, Dix M (2008) Variability
Nicholson S (2000) Land surface processes and Sahel climate. Rev and trend over the northwest Western Australian rainfall: obser-
Geophys 38(1):117–139 vations and coupled climate modeling. J Clim 21:2938–2959
Nicholson S (2013) The West African Sahel: a review of recent stud- Su J, Huang J, Fu Q, Minnis P, Ge J, Bi J (2008) Estimation of Asian
ies on the rainfall regime and its interannual variability. ISRN dust aerosol effect on cloud radiation forcing using Fu–Liou
Meteorol 2013:1–32 radiative model and CERES measurements. Atmos Chem Phys
Nicholson S, Kim J (1997) The relationship of the El Niño Southern 8(10):2763–2771
oscillation to African rainfall. Int J Climatol 17:117–135 Taschetto A, England M (2008) An analysis of late 20th century
Omuto C, Vargas R, Alim M, Paron P (2010) Mixed-effects modelling trends in Australian rainfall. Int J Climatol 29:791–807
of time series NDVI–rainfall relationship for detecting human- Thomas D (2011) Arid zone geomorphology: process, form and
induced loss of vegetation cover in drylands. J Arid Environ change in drylands, 3rd edn. Wiley-Blackwell, xxiv
74(11):1552–1563 Thompson D, Wallace JM (2000) Annular modes in the extratropical cir-
Otterman J (1974) Baring high-albedo soils by overgrazing—a hypoth- culation. Part I: month-to-month variability. J Clim 13:1000–1016
esized desertification mechanism. Science 186(4163):531–533 Thornthwaite CW (1948) An approach toward a rational classification
Parker D, Folland C, Scaife A, Knight J, Colman A, Baines P, Dong of climate. Geogr Rev 38(1):55–94
B (2007) Decadal to multidecadal variability and the climate Ting M, Kushnir Y, Seager R, Li C (2009) Forced and internal
change background. J Geophys Res. doi:10.1029/2007jd008411 twentieth-century SST trends in the North Atlantic. J Clim
Paruelo JM, Beltrán Adriana, Jobbágy Esteban, Sala Osvaldo E, Gol- 22(6):1469–1481
luscio RA (1998) The climate of Patagonia: general patterns Trenberth KE, Stepaniak DP, Caron JM (2000) The global monsoon
and controls on biotic processes. Ecol Austral 8:85–101 as seen through the divergent atmospheric circulation. J Climate
Peel M, Finlayson B, McMahon T (2007) Updated world map of the 13(22):3969–3993
Köppen–Geiger climate classification. Hydrol Earth Syst Sci Trewartha G, Horn L (1980) An introduction to climate, 5th edn.
11:1633–1644 McGraw-Hill, New York, p 437
Penman HL (1948) Natural evaporation from open water, bare soil Turc L (1961) Evaluation de besoins en eau d’irrigation, ET poten-
and grass. Proc R Soc Lond A 193(1032):120–145 tielle. Ann Agron 12:13–49
Priestley C, Taylor R (1972) On the assessment of surface heat flux Ueda H, Iwai A, Kuwako K, Hori M (2006) Impact of anthropogenic forc-
and evaporation using large scale parameters. Mon Weath Rev ing on the Asian summer monsoon as simulated by eight GCMs.
100:81–92 Geophys Res Lett 33:L06703. doi:10.1029/2005GL025336
Randel W, Wu F (1999) Cooling of the Arctic and Antarctic polar Wang B, Ding Q (2006) Changes in global monsoon precipitation
stratospheres due to ozone depletion. J Clim 12:1467–1479 over the past 56 years. Geophys Res Lett 33(6):L06711. doi:10.
Reason C, Rouault C (2002) ENSO-like decadal variability and South 1029/2005GL025347
African rainfall. Geophys Res Lett 29:1638. doi:10.1029/200 Wang K, Dickinson R (2012) A review of global terrestrial evapotran-
2GL014663 spiration: observation, modeling, climatology, and climatic var-
Richard Y, Trzaska S, Roucou P, Rouault M (2000) Modification of iability. Rev Geophys 50:RG2005. doi:10.1029/2011RG000373
the Southern African rainfall variability/El Niño southern oscil- Wang G, Huang J, Guo W, Zuo J, Wang J, Bi J, Huang Z, Shi J (2010)
lation relationship. Clim Dyn 16:883–895 Observation analysis of land–atmosphere interactions over the
Rodell M, Houser P, Jambor U, Gottschalck J, Mitchell K, Meng C, Loess Plateau of northwest China. J Geophys Res 115:D00K17.
Arsenault K, Cosgrove B, Radakovich J, Bosilovich M (2004) doi:10.1029/2009JD013372
The global land data assimilation system. B Am Meteorol Soc Wang X, Doherty S, Huang J (2013) Black carbon and other light-
85(3):381–394 absorbing impurities in snow across Northern China. J Geophys
Rotenberg E, Yakir D (2010) Contribution of semi-arid forests to the Res Atmos 118:1471–1492
climate system. Science 327(5964):451–454 Wang S, Huang J, He Y, Guan Y (2014) Combined effects of the
Safriel U, Adeel Z (2005) Dryland systems. In: Hassan R, Scholes R, Pacific decadal oscillation and El Nino-southern oscillation on
Ash N (eds) Ecosystems and human well-being, current state global land dry–wet changes. Sci Rep 4:6651. doi:10.1038/
and trends, vol 1. Island Press, Washington, pp 625–658 srep06651
Scanlon BR, Keese KE, Flint AL, Flint LE, Gaye CB, Edmunds WM, Wardle R, Smith I (2004) Modeled response of the Australian mon-
Simmers I (2006) Global synthesis of groundwater recharge in soon to changes in land surface temperatures. Geophys Res Lett
semiarid and arid regions. Hydrol Process 20(15):3335–3370 31:L16205. doi:10.1029/2004GL020157

13
1150 J. Huang et al.

Wessels K, Prince S, Malherbe J, Small J, Frost P, VanZyl D (2007) Zeng N, Neelin JD, Lau K-M, Tucker CJ (1999) Enhancement of
Can human-induced land degradation be distinguished from the interdecadal climate variability in the Sahel by vegetation inter-
effects of rainfall variability? A case study in South Africa. J action. Science 286(5444):1537–1540
Arid Environ 68(2):271–297 Zhou T, Gong D, Li J, Li B (2009) Detecting and understanding the
Xu M, Chang C, Fu C, Qi Y, Robock A, Robinson D, Zhang H (2006) multi-decadal variability of the East Asian summer monsoon-
Steady decline of east Asian monsoon winds, 1969–2000: evi- recent progress and state of affairs. Meteorol Z 18(4):455–467
dence from direct ground measurements of wind speed. J Geo- Zhu C, Wang B, Qian W, Zhang B (2012) Recent weakening of north-
phys Res Atmos. doi:10.1029/2006JD007337 ern East Asian summer monsoon: a possible response to global
Xu Y, B-s Tang, Chan EH (2011) State-led land requisition and trans- warming. Geophys Res Lett. doi:10.1029/2012GL051155
formation of rural villages in transitional China. Habitat Inter- Zuo Z, Yang S, Kumar A, Zhang R, Xue Y, Jha B (2012) Role of
national 35(1):57–65 thermal condition over Asia in the weakening Asian sum-
Xue Y (1996) The impact of desertification in the Mongolian and mer monsoon under global warming background. J Clim
the Inner Mongolian grassland on the regional climate. J Clim 25(9):3431–3436
9(9):2173–2189
Xue Y, Shukla J (1993) The influence of land surface properties on
Sahel climate. Part 1: desertification. J Clim 6(12):2232–2245

13