We used the ﬂux tower GPP to evaluate the SOS derived from the NDVI data sets and used the ﬁeld pheno- logical observations to examine and evaluate the trends in SOS. It should be noted that there is a scale mis- match between remote sensing pixels and footprints of ﬁeld sites (e.g., eddy covariance ﬂux towers). The pixel size of the NDVI data sets, particularly GIMMS and GIMMS (8 km), is typically larger than the footprint of an eddy covariance ﬂux tower. Moreover, a pixel is of a ﬁxed size for a speciﬁc sensor, while the footprint of an eddy covariance ﬂux tower varies with wind speed, wind direction, tower height, and surface roughness (Kljun et al., 2015). To examine the representativeness of each ﬂux site and assess the scale mismatch between its footprint and the NDVI pixels, we calculated the fractional cover of grassland within the 1 km × 1 km and 8 km × 8 km windows surrounding the tower. The fractional cover of grassland was close to 1.0 for all sites except Maqu. This indicates that using ﬂux tower GPP to evaluate the SOS derived from the NDVI data sets was reasonable to some extent despite the scale mismatch. The ﬁeld phenological obser- vations for a given site were based on a speci ﬁc species, while the pixel in which the site is located consists of multiple vegetation types and/or multiple species. Although in situ measurements have been widely used to evaluate phenology derived from satellite data (Coops et al., 2007; Turner et al., 2003), the scale mismatch can result in signi ﬁcant discrepancies in springphenology between in situ measurements and satellite observa- tions (Chen et al., 2008). In addition, the smoothing of daily GPP using the 15 day moving window could also introduce uncertainty to the derived SOS.
Relationships Between Tree-Ring Phenology and Climate Factors. The April through June minimum temperature appears to have had the highest influence on SOS on the TP during the period 1960– 2014, as also demonstrated for other temperate, boreal, and tim- berline ecosystems in the Northern Hemisphere (24). Because our study region is characterized by high altitudes, it is reasonable that temperature has a significant effect on the starting date of the growing season. As a result, significant correlations between temperature and SOS were obtained by the principle of limiting factor, i.e., that just one factor can have an impact on a biosystem at any particular moment of time in the VS model simulation. Compared with the period 1960–1981, the 1982–2014 phenologi- cal data showed a stronger temperature signal, consistent with the observed significant (P < 0.01) spring warming after 1982 ( SI Appendix, Fig. S10 ). Clearly, increases in April through June minimum temperatures were correlated with the earlier SOS across the whole region (Fig. 2) (see refs. 31−33 for supporting evidence). An increase of 1 °C in April through June minimum temperature caused an advance of 6.94 d in the SOS of our study region (Fig. 2). The results predicted by our model (∼7 d per 1 °C) closely matched in situ observations of SOS derived from den- drometer data of Qilian juniper [from an elevational gradient in the northeastern TP (33)] and from microcoring data of European larch from two elevational transects in Switzerland (34). Both studies were based on the so-called “space-for-time/warming ex- periments” approach (34), whereby long-term changes in the timing and duration of tree growth per shift in degree Celsius (i.e., time) are substituted by changes along altitudinal transects (i.e., space). Monitoring results from 1,321 trees belonging to 10 conifer species located at 39 sites in North America, Europe, and Asia (29.62°N to 66.2°N, 72.87°W to 94.7°E, 60 m a.s.l. to 3,850 m a.s.l
Bershaw et al. (2012), Hren et al. (2009) and Caves et al. (2015) (Fig. 8a, b). In general, the model shows a good agreement with precipitation and VSMOW-weighted modern surface waters δ 18 O, including stream, lake and spring waters (data from Bershaw et al., 2012; Hren et al., 2009; Quade et al., 2011), as testified by a Pearson coefficient of 0.86 be- tween modelled and observed precipitation δ 18 O (Fi. 8c). This comparison shows the ability of LMDZ-iso to repro- duce the decrease in δ 18 O from the Indian subcontinent to the Himalayan foothills and with minimum values over the Himalayas. Simulated increase in δ 18 O over the TP with the distance from the Himalayas is also consistent with data sam- pled along a southwest–northeast transect across the Plateau (Bershaw et al., 2012). However, over the northern margins of the TP, LMDZ-iso underestimates simulated δ 18 O in pre- cipitation (Fig. 8a). This model data mismatch may result from two types of uncertainties. First despite the high res- olution obtained with a zoomed grid, restricted topographic features could not be well-captured over some parts of the TP, which could lead our simulations to miss local processes affecting δ 18 O in rainfall. Second, overestimating the wester- lies fluxes (see the comparison with the ERA moisture trans- port on Fig. 5a) could lead to underestimate δ 18 O over the northern part of the TP, through advection of depleted air masses. Nevertheless, despite our model not capturing the absolute maximal values well, the regional latitudinal gra- dient is correctly represented, and most observed values are
underestimate the greening trend on the TP to different extents (the details can be found in Text S1 in the Supporting Information), largely due to the impacts of sensor shifts or degradation [ 14 , 16 , 17 , 20 ].
GNDVI increased in the eastern and northern TP but decreased in the south-central and southwest TP during 2000–2015 (Figure 2 ), and most ecosystem zones experienced major shifts in GNDVI between 2009 and 2011 (Figure 4 ). The main explanation for this is that spatially heterogeneous temporal trends in major climatic factors induced complex responses in vegetation growth on the TP [ 59 – 61 ]. This study shows that the role of temperature, precipitation, and solar radiation played in vegetation dynamics varied depending on ecosystem zones, where the vegetation composition and humidity are different. Vegetation growth in the southwestern and northeastern TP requires more water as these are mostly arid and semi-arid areas of alpine meadow and alpine steppe. These areas are consequently greatly affected by water availability. We show that vegetation greening was mainly the result of increasing precipitation on the northeastern TP, which is in accordance with the findings of other similar studies [ 24 ]. Geographically, the northeastern and southwestern TP are characterized by distinct water vapor sources and different thermal heating systems [ 62 ]. In contrast, within the context of the weakening Indian monsoon [ 62 ], vegetation browning in the southwestern TP was caused by a water deficit (Figure S2), which induced by interactions among changes in temperature, solar radiation, and precipitation [ 60 , 63 ]. In this respect, the results of this study contradict the previous study that increasing temperature drove vegetation greening within the Yarlung Zangbo River Basin [ 26 ], perhaps because the greening trend was over-estimated on the basis of SPOT-VGT-NDVI (Figure S3). Furthermore, a two-month lag effect response of vegetation to temperature, precipitation and radiation appeared in IIC2 zone (Figure 5 ), which was consistent with the previous result that the browning trend in this region was associated with a delayed vegetation green-up date, likely affected by pre-monsoonal droughts [ 23 , 64 , 65 ].
Here, multi-temporal landscape characterisation using VIs have been applied within an epidemiological context, with results showing that the MODIS 16-day composite VI data product has suf ﬁ cient sensitivity to track the seasonal vegetationphenology at this study site. Although the 250 m resolution of this imagery means multiple land cover types could potentially be contained within a single pixel, the relatively homogenous grassland dominated areas within this study site suggest that the use of MODIS time-series VI datasets is appropriate for capturing the landscape and vegetation dynamics at an ecologically meaningful scale. By using satellite VI data as a surrogate measure of biophysical characteristics of the landscape and capturing the dynamic vegetation and phenological characteristics driving the biological systems of that site (as previously performed by Lunetta, Knight, Ediriwickrema, Lyon, & Worthy, 2006), a better description of the fundamental variability of the landscape can be achieved, and the links between seasonal landscape dynamic and small mammal parasite transmission reservoirs better understood. The potential in ﬂ uence of small mammal population cycles must also be acknowledged (evidence has been provided that O. curzoniae populations on the eastern Tibetanplateau can be cyclic, with cycle duration N 6 – 7 years (Giraudoux et al., 2006)), meaning that Ochotona spp. habitats de ﬁ ned by the signal captured by MODIS EVI 16-day may correspond to periods when Ochotona spp. are locally scarce during the low density phase of a cycle (Giraudoux et al., 2007). However, while Ochotona spp. densities may vary due to multi-annual population cycles, their colonies are geographically very stable over several years (Giraudoux, personal communication, May 1, 2015), meaning that the methodological approach employed at this site is both appropriate and highly informa- tive in this context.
Located in a transition zone between the continental cli- mate of Central Asia and the Indian Monsoon system, the Nam Co drainage basin including the western Nyainqentan- glha Mountains has been pointed out as a key research area in Tibet, and is also investigated in this study. The recent rise of the lake level of Nam Co, one of the largest and highest lakes on the TiP (year 2000: 1980 km 2 area, 4724 m a.s.l. lake level altitude), has been attributed to glacier retreat as well as to an increase of precipitation in recent decades (e.g., Wu and Zhu, 2008; Krause et al., 2010). Precipitation increase in central TiP during this period has also been reported by Liu et al. (2009). However, the TiP remains a sparsely ob- served region, and there is limited availability of meteoro- logical data. This is especially true for long-term weather records necessary for reliable climatological studies (Frauen- feld et al., 2005; Kang et al., 2010). In particular, no long- term data from weather stations are existing at elevations above 4800 m a.s.l. Generally, the geographical distribution of weather stations is biased towards lower altitudes, flat ar- eas and specific land-cover types excluding high-mountain regions covered by glaciers. The question of the respective contributions of glacier retreat and precipitation increase to rising lake levels on the TiP remains unanswered, so far, for these reasons.
2018 ). AMO was obtained from NOAA’s Earth System Research Laboratory (Earth System Research Labo- ratory 2018a ). The MEI is based on six variables observed over the Pacific Ocean. This time series was also downloaded from NOAA’s Earth System Research Laboratory ( 2018b ). Each climate index, except MEI, was provided as a monthly index, which we summa- rized into seasonal indices by calculating the average for winter (DJF), spring (MAM) and summer (JJA). MEI was provided as a bimonthly index (e.g. DEC- JAN), which we summarized in similar seasons as the other indices. We are not presenting the results for the fall season, because we are interested in the potential predictive capability of the climate indices on the peak of the growing season, which for this area occurs in the late spring or summer. An overview of the spring indices (MAM) since 2001 can be found in figure 1 . Table 1 provides the Spearman correlation between these individual indices. A significant negative cor- relation is revealed between spring AMO, NAO and EAWR indices. There is a significant positive correla-
same time, but oscillate throughout the Holocene, with the highest values centred around ~ 2 cal ky BP (Figs. 4a and 4b). Although MAs ratios cannot precisely point to the type of past burnt vegetation, they can classify general vegetation categories. According to their published ranges (Fabbri et al., 2009), our data suggest that grasses dominated the area for the oldest section of the Paru Co core and that softwood began to grow in the region after ~ 10.74 cal ky BP. Grasses, softwood and hardwood may have oscillated until 8.6 cal ky BP. Hardwood generally dominated the vegetation between 8.6 to 7.7 cal ky BP, followed 30
The phenological dynamics of terrestrial ecosystems reflect response of the Earth’s biosphere to inter- and intra- annual dynamics of the Earth’s climate and hydrologic regimes (Myneni, Keeling, Tucker, Asrar, & Nemani, 1997; Schwartz, 1999; White, Thornton, & Runnings, 1997) . Because of the synoptic coverage and repeated temporal sampling that satellite observations afford, remotely sensed data possess significant potential for monitoring vegetation dynamics at regional to global scales (e.g., Myneni et al., 1997 ). In the last decade, a number of different methods have been developed to determine the timing of vegetation greenup and senescence (i.e., the start and end of growing season) using time series of normalized difference vegetation index (NDVI) data from the Advanced Very High Resolution Radiometer (AVHRR). These methods have employed a variety of different approaches including the use of specific
Based on 38 years (1959-1996) of climate observations and statistical analysis, the annual mean temperature increased during this period ranged from 0.4˚C to 0.6˚C in the area of Haibei Alpine Tundra Ecosystem Re- search Station (Li et al., 2004), that is located on northeastern part of Qinghai-TibetanPlateau (37˚N, 101˚E). In order to study alpine tundra vegetation changes at the regional scale, we model alpine tundra vegetation spatial and temporal dynamics in response to global warming by integrating a raster-based cellular automata and a Geographic Information System (Zhang et al., 2008). Temperature changes across the study area are not only due to elevation, but also to aspect and distance from the nearest stream channel. The liner regression model provided a temperature spatial distribution based on elevation alone, which is the primary step. The normalized temperature surface created by the Multi-Criteria Evaluation (MCE) is highly representative of the potential temperature distribution in a normalized fuzzy format. Assuming each vegetation type in the raster cell unit re- acts as homogeneous entity, we conduct a spatial and temporal simulation by combining cellular automata and MCE provided in the IDRISI software (Eastman, 2003).
Our measured mean N g of each functional type was higher than that reported in humid alpine meadow of east Tibet (Jiang et al., 2012; Liang et al., 2015) and the global average (Reich, Oleksyn, & Tilman, 2004; Vergutz et al., 2012), while mean P g of each functional type was much lower than the global average (Reich et al., 2004). This result indicated alpine plants were more limited by phosphorus than nitrogen, and were likely to better adapt to alpine and infer- tile environment. Except for legume, both N and P resorption effi- ciencies of other functional types showed obviously higher values than the global- scale average compiled by Aerts (1996) and Yuan and Chen (2009), suggesting very efficient nutrient conservation of alpine plants on the Plateau. Leaf nutrient resorption is consid- ered highly proficient if N s and P s are below 7 and 0.5 g/kg, respec- tively, and as ultimate potential resorption if N s and P s as low as 3 and 0.1 g/kg, respectively (Killingbeck, 1996). Accordingly, alpine species had very high resorption of N and P, and sedge and grass showed almost ultimate resorption. The higher resorption efficiency and proficiency suggested that alpine plants were likely well adapted to nutrient- limited environment through high internal N and P recy- cling (Freschet, Cornelissen, van Logtestijn, & Aerts, 2010; Norris & Reich, 2009). Moreover, the PRE of different functional types was higher than the NRE, indicated that P is more limited than N on the Changtang Plateau.
Earthquake prediction thus far has proven to be a very difficult task, but changes in situ stress appear to offer a viable approach for forecasting large earthquakes in Tibet and perhaps other continental regions. High stress anomalies formed along active faults before large earthquakes and disappeared soon after the earthquakes oc- curred in the TibetanPlateau. Principle stress increased up to ~2 - 5 times higher than back- ground stress to form high stress anomalies along causative faults before the Ms 8.1 West Kunlun Pass earthquake in November 2001, Ms 8.0 Wenchuan earthquake in May 2008, Ms 6.6 Nimu earthquake in October 2009, Ms 7.1 Yushu earthquake in April 2010 and the Ms 7.0 Lushan earthquake in April 2013. Stress near the epi- centers rapidly increased 0.10 - 0.12 MPa over 45 days, ~8 months before the Ms 6.6 Nimu earth- quake occurred. The high principle stress ano- malies decreased quickly to the normal stress state in ~8 - 12 months after the Ms 8.1 West Kunlun Pass and the Ms 8.0 Wenchuan earth- quakes. These high stress anomalies and their demise appear directly related to the immediate stress rise along a fault prior to the earthquakes and the release during the event. Thus, the stress rise appears to be a viable precursor in prediction of large continental earthquakes as in the TibetanPlateau.
Permafrost boundaries in the earlier maps have often been plotted on topographic maps by hand using conventional cartographic techniques (Tong and Li, 1983; Shi and Mi, 1988; Li and Cheng, 1996). The standard and most widely used map is the Map of Permafrost on the Qinghai-TibetanPlateau (Li and Cheng, 1996). In this map, the permafrost boundaries were mainly determined using air temperature isotherms combined with field data, satellite images, and many relevant maps. After 2000, GIS techniques have been applied to permafrost mapping of the TP. Simple empirical models with minimal data requirements were established to consider the permafrost characteristics on the TP. These in- clude the elevation model (Li and Cheng, 1999) and mean an- nual ground temperature (MAGT) (Nan et al., 2002). Mean- while, some models with simplified physical processes appli- cable to high-latitude permafrost were transferred to simulate the permafrost distribution on the TP. Examples are the frost index (Nelson and Outcalt, 1983) and the temperature at the top of permafrost (TTOP) (Smith and Riseborough, 1996; Wu et al., 2002a). These models link permafrost tempera- ture with surface temperature using seasonal surface transfer functions and subsurface thermal properties, which can pro- vide reasonable assessments of permafrost distribution when the permafrost upper boundary conditions and regional soil thermal properties are satisfied. Recently, a global permafrost zonation index (PZI) was established based on the relation- ships between the air temperature and the occurrence of per- mafrost (Gruber, 2012). The PZI can represent broad spatial patterns but does not provide the actual extent of the per- mafrost.
al., 2010; Mischke, 2012). Investigations were mainly conducted in the more easily accessible northern and eastern part of the plateau, and only a few studies (e.g. Li et al., 1991; Zhu et al., 2010) improved the available knowledge of the local ostracod fauna of its central and western part. Furthermore, ostracod studies in the central and southern parts of the TibetanPlateau focussed on Holocene and Late Pleistocene faunas whereas the precursors of these partly endemic species are not known. Investigations on Plio-Pleistocene ostracods from the TibetanPlateau are restricted to the Qaidam Basin so far (e.g. Sun et al., 1988; Yang et al., 1997; Mischke et al., 2006, 2010) where they are a valuable tool for biostratigraphy in hydrocarbon exploration. This work focuses on Plio-Pleistocene ostracods of the Zhada Basin located in the western TibetanPlateau. Previous works carried out in this area concern its tectonic origins (Wang et al., 2004; 2008; Saylor et al., 2010b) and palaeoenvironmental reconstruction (Saylor et al., 2010a) using mostly pollen records and sedimentological analyses. Kempf et al. (2009), who investigated petrographic and sedimentological properties, were the first to describe also elements of the ostracod fauna in this area. They found some typical endemic taxa like Leucocytherella sinensis and several not identified species. In this work we present the Plio-Pleistocene ostracod assemblage recovered from 105 sub-samples of Joel Saylor’s stratigraphic “South Zhada” (“SZ”) section, localised in the southern part of the Zhada Basin and already sedimentologically analysed and dated by Saylor (2008) in order to improve the taxonomic data base on ostracods of this area for future palaeoecological and potentially stratigraphical studies.
Synoptic-scale conditions in IOP2 show that, as soon as there is convective activity in the BoB in late May, PT increased signiﬁcantly over the pla- teau. However, this is not expected to be due to adiabatic heating induced by the Hadley-type circu- lation accompanied by convection in South Asia, but latent heat release of convection at the BoB. At the same time, a cooler air mass related to syn- optic-scale meandering ﬂow cools the plateau even in the period of strong convection. With the cool air mass passing and enhancing the convective activity, the warmer air suddenly spread while cen- tered on the eastern plateau. This is adiabatic heat- ing due to the downward motion of the Hadley- type circulation. Tamura et al. (2010) showed climatological process of the heating over the pla- teau and summarized as an adiabatic heating in- duced by the Hadley-type circulation is a dominant mechanism. However, in the seasonal progression of a certain year, the variation in atmospheric tem- perature over the plateau is caused by a combina- tion of several synoptic-scale processes.
Seventy-seven healthy native Tibetans (34 males and 43 females, 14 –18 years of age) living in the southeast Tibet Autonomous Region (southeast Nagqu, Xigaze, Lasa, Nyingchi, and Chamdo) (altitude, 2300 –5300 m) were recruited. Their ancestors were all native Tibetans living on the Qinghai-TibetanPlateau. They were without any prior descent to the lowlands or ascent to higher altitudes. All had a normal body mass index. They were all 10th grade students and had been enrolled in a high school at Chengdu (altitude, ⬍ 400 m) for half a month. The control subjects were 80 (34 males and 46 females, 14 –18 years of age) adolescents living at sea level (altitude, ⬍ 50 m), matched with the native Tibetans by sex, age, and education. Subjects were excluded if they had a his- tory of mountain sickness, neurologic disorder, or head injury. The experimental protocol was approved by the Research Ethics Review Board of Xiamen University. Procedures were fully ex- plained, and all subjects provided written informed consent be- fore participating in the study.
The risk increases up to an elevation of 3177 m and then declines as elevation increase, because in the lower height, forest and thickly grass would be harmful for marmot making doggishness, while in the higher height, there would be lack of food. The marmot preferred temperature between 5°C and 20°C in the daylight. The extremely low temperature below -8.38°C would limit their activities. Extremely low and high temperature of land surface seems negatively affect the distribution of host animals. The temperature of land surface may influ- ence the plague persistence in ‘marmot-flea’ communities in complex ways. We suspected extreme temperature may negatively affect marmot and flea ecology, and blockage of vector fleas . Host marmots were observed and pre- dicted in the sparely vegetated areas, but in our modeling, NDVI variation had no significant limitation in host ani- mal distribution (Figure 3). This result suggested that plague prefer alpine desert/semi-desert grasslands con- firmed previously, such as alpine meadow, alpine grass- land and alpine shrub [5,39], while habitats with extravagant vegetation seemed to be unsuitable for enzo- otic plague. This could be due to that luxuriant vegetation obstructs them to protect from predators, and the high subterranean biomass make them hard to burrow under- ground .
Available hydrological impact assessment methods com- monly used in literature include the catchment paired exper- iments, statistics analysis, and measurements with hydro- logical models , but seldom have these been involved in streamflow regime studies [12, 13]. Moreover, the experimen- tal and statistical methods treat the study basin as a black box and rarely examined the complexities of precipitation changes, underlying surface conditions, and the interactions between climate change and the respective hydrological processes . Hydrological models, however, offer a viable framework for conceptualizing and investigating the relation- ships between climate, underlying surface, and hydrological processes in various categories of time and space [14, 15] and the approach of one-factor-at-a-time (OFAT) has been widely applied [16, 17]. For instance, Karlsson et al.  modeled the combined effects of land use and climate changes on the hydrology for a catchment located in Denmark. Zhao et al.  evaluated the climate variability and land use influences on green and blue water resource in Weihe River basin, NW China. These methods considered the hydrological processes and their interactions with the environment. It should be noted that hydrological models are more effective as they relate the model parameters directly to the physically observable land surface characteristics . However, the assumption within OFAT method is that in the course of evaluating the influence of a given factor on hydrological processes, the other factors’ effects are not considered. In fact, the other factors not considered are changing concurrently over the entire period of observation which can contribute as an apparent bias in the separation results. Separation of the individual influence of the two factors on hydrological processes thus warrants the need for inputting the applicable status of two factors from the baseline period to the entire period. Yang et al.  recommended that a combined statistical and modeling method could be used to resolve this bias and further identified the climate and land cover change impacts on the hydrological processes in the Heihe River. In that study, the results were more reasonable and offered greater accuracy than the traditional OFAT and other statistical methods.