Prolonged dry spells in recent decades over north-central China and their association with a northward shift in planetary waves

Transcription

1 INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. 3: (1) Published online 1 April 1 in Wiley Online Library (wileyonlinelibrary.com) DOI: 1.1/joc.4337 Prolonged dry spells in recent decades over north-central China and their association with a northward shift in planetary waves Jie Zhang, a,b * Laurent Li, a,c Zhiwei Wu a and Xingmin Li d a Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters/Key Laboratory of Meteorological Disaster of Ministry of Education, Nanjing University of Information Science & Technology, Nanjing, China b Key Laboratory of Arid Climatic Changing and Reducing Disaster of Gansu Province and CMA, Lanzhou, China c Laboratoire de Météorologie Dynamique, IPSL/CNRS, Université Pierre et Marie Curie, Paris, France d Institute of ShanXi Meteorology, Xi an, China ABSTRACT: Summer rainfall in north-central China, accounted for 7% of the annual rainfall, is sensitive to anomalies in the East Asian summer monsoon (EASM) and westerlies due to lying in the EASM marginal belt. Summer rainfall and the durations and frequency of dry spells were investigated using daily rainfall data from 33 weather stations, the European Centre for Medium-range Weather Forecasting (ECMWF) re-analysis data and Climate Research Unit (CRU) surface temperature data. The results demonstrate that summer rainfall exhibits an early wetting later drying pattern with a weak increasing in June and a dramatic decreasing trend in July and August (JA). Decreased rainfall in JA is associated with prolonged dry spells; the frequency and durations of dry spells exceeding 7 days increase, especially for dry spells exceeding 1 days. This increase in dry spell duration is positively correlated with a northward shift in planetary waves expressed by the 7-gpm isoline at hpa and the westerly jet at hpa, which lead north-central China to close to the left of jet entrance and a positive vorticity convergence region on the high level, it is helpful to downward movements and less rainfall. The northward shift in planetary waves is affected by boundary forcing. Dramatic increases in temperature over the Tibetan Plateau (TP) result in a large temperature gradient and thermal contrast between the TP and mid-high latitude, and likely strengthening and widening the South Asian High (SAH) pattern and pressure gradient. These conditions contribute to the northward shift in planetary waves and prolonging dry spells. KEY WORDS prolonged dry spells; planetary wave; temperature gradient; northward shift Received 7 December 14; Revised 1 March 1; Accepted 13 March 1 1. Introduction Northern China is the main agricultural, industrial and cultural centre in China. The region is situated within the marginal belt of the East Asian summer monsoon (EASM) or to the north of the EASM; therefore, the climate in this region is sensitive to anomalies in the EASM and to westerly. In the past 3 years, northern China has undergone severe dry spells during the rainy season (Cheng and Zhou, 13). Dry spells are characterized by continuous days without substantial rainfall (>1mmday 1 ) (Zolina et al., 13). Long-duration or persistent dry spells lead to droughts, and several droughts have occurred in northern China over the past 3 years. Moreover, it is well know that eastern China exhibits a southern flooding and northern drought pattern, which refers to the increased rainfall over the middle to lower reaches of the Yangtze River basin and the decreased rainfall over Northern China * Correspondence to: J. Zhang, College of Atmospheric Science, Key Laboratory of Meteorological Disaster, Ministry of Education (KLME), Nanjing University of Information Science & Technology (NUIST), Ningliu Road 19, Nanjing 144, China. gs-zhangjie@13.com (Zhao et al., 1). Especially during the recent period of substantial global warming, droughts have become more intense and more frequent, such as the droughts in and 1 in northern Shanxi province and Henan province ( 11_China_drought). Droughts have major effects on agriculture, the ecological environment and the social economy. Therefore, the effective prediction of extreme droughts can minimize losses of goods and lives. Dry spells can be used as a reference index for potential droughts (Groisman and Knight, ). Moreover, it is important to understand the physical mechanisms involved in dry spell dynamics. However, less attention has been given to the long-term variability in the duration of dry spells. Dry spells are natural events that are associated with the duration of wet periods, which are the results of rainfall. That is first decided by moisture transports mainly transported by EASM, therefore, many previous studies addressed the rainfall change in northern China and its relations with EASM. For example, the southern flooding and northern drought pattern is associated with the weakening of the EASM circulation (Hu et al., 3), which is forced by either a phase transition of the Pacific Decadal 1 Royal Meteorological Society

2 43 J. ZHANG et al. Oscillation (PDO) or recent warming of the tropical Pacific and Indian Oceans (Li et al., 1a, 1b), furthermore, approximately 7% of the drying trend with multi-decadal variability is related to the PDO-negative phase (Cheng and Zhou, 13). EASM circulation is affected by the land ocean temperature gradients and thermal contrast (Li et al., 1a, 1b), anomaly land thermal also leads to EASM circulation change, especially thermal anomalies in spring and summer over the Tibetan Plateau (TP), reductions in the sensible heat flux over the TP have weakened the EASM circulation and postponed its seasonal changes (Duan et al., 1). As a factor affecting the TP temperatures, TP snow cover (TPSC) exerts an effect on El Niño-Southern Oscillation (ENSO) tele-connections with East Asia, whereas the ENSO only exhibits a significant positive correlation with the EASM during reduced TPSC summers (Wu et al., 1a). In addition, Yu et al. (4) attributed the decreased rainfall pattern over northern China to the summertime cooling in the upper troposphere over extra tropical East Asia, which was assumed to be associated with stratosphere troposphere interactions. Decreased rainfall and frequent droughts in northern China are associated not only with the EASM circulation but also with wave activities associated with westerlies because northern China just located in the northern marginal belt of the EASM. Moreover, frontal precipitation primarily contributes to wet conditions in summer due to the convergence of cold masses by westerlies and warm moist masses by EASM. The effects of the EASM and the westerlies exhibit non-uniform in northern China, north-east China is largely influenced by the EASM, whereas north-west China is primarily influenced by the westerlies. In recent decades, rainfall over north-west China has exhibited an increasing trend (Chen and Huang, 1), which contradicts the rainfall trend over north-east China. Decreased rainfall and droughts over north-central China are commonly influenced by Rossby waves generated by the westerlies and EASM moisture transport. Through westerly circulation, the North Atlantic Oscillation (NAO) related to the North Atlantic sea surface temperature (SST) distribution modifies the effects of the ENSO on the EASM, by enhancing a distinct Rossby wave train within the prevailing westerlies over northern Eurasia (Wu et al., 1b). The positive anomalies in the northwestern North Atlantic SST pattern also excites a distinct Rossby wave train, which leads to an anticyclonic anomaly near Lake Baikal; this phenomenon is related to drought/flood over north-central China (Hu and Feng, ; Zhang et al., 13). With respect to high-latitude forcing, Arctic ice loss is favourable for cyclonic anomalies in the lower troposphere over the Arctic, which results in a weaker polar cell and a stronger Ferrel cell. The final result is a strengthening of the anticyclonic anomaly near Lake Baikal, which is related to meridional shifts or a northward elongation and anomaly in planetary wave ridges near Lake Baikal (Zhang et al., 14). All the previous findings suggest that anomalous Rossby waves in the mid-latitude are related to rainfall and drought in north-central China. Reduced rainfall and drought in north-central China is primarily associated with the decreased frequency of precipitation events (Zhai et al., ), which corresponds to an increase in dry spells. In the present work, we investigate the possible mechanisms underlying drought and decreased rainfall based on the characteristics of dry spells, including the duration and frequency of dry spells during the rainy season in north-central China, because prolonged dry spells increase the frequency of droughts. Furthermore, the potential relationship between dry spells and atmospheric circulation anomalies is investigated to explore relevant mechanisms of drought events on decadal scales. In this study, the data and methods are described in Section. The temporal and spatial variability of dry spells and their relationships with planetary waves are discussed in Section 3. The possible mechanisms of planetary waves and prolonged dry spells are explored, and the surface temperature differences between the most recent three decades and the preceding three decades are analysed in Section 3. The model sensitivity simulations are also discussed in Section 3. Finally, conclusions are presented in Section 4.. Data and methods.1. Study area Northern China corresponds to the northern extent of China, encompassing the eastern, central and western portions of the country. Summer precipitation over northern China accounts for more than 7% of the annual precipitation. Summer rainfall over north-east China is primarily affected by the EASM. Moreover, summer rainfall over north-central China is affected by both the EASM and the westerlies. Western China is generally regarded as northwestern China, which corresponds to the plateau climate and arid climate. Summer rainfall in this region is mainly influenced by the westerlies. In this study, we focus on the region east of 1 E, the 33 weather stations are presented in Figure 1 and the study region is also showed in the Figure 1 (rectangular region, north-central China is defined by 37 4 N and E). North-central China encompasses the middle reaches of the Yellow River, the second largest river, and the climatic variability in this region is large. We selectively analyse rainfall events and dry spells over north-central China... Datasets and methods Three datasets are used in this study. Daily precipitation data in summer (JJA) from 3 weather stations in northern China are used, which correspond to 3a of data from 191 to 13, 4 stations from 3 stations are rejected due to discontinuous or moving stations, and finally 33 stations are selected. The discontinuous data include the rainfall gaps in the data records during JJA due to uncertainty reasons, and such gaps cannot be ensured whether it is rainy day or not, and there are more than two continuous days with gap in the data records, or total days exceed days in JJA. Other gaps in the data records are revised according to 1 Royal Meteorological Society Int. J. Climatol. 3: (1)

3 NORTHWARD PLANETARY WAVES CONTRIBUTE TO PROLONGED DRY SPELLS OVER NORTH-CENTRAL CHINA E 1 E 11 E 11 E 1 E 1 E sensitivity region province boundary river station Figure 1. Location of north-central China (rectangle; 37 4 N, E) and 33 stations distribution in Northern China. daily temperature and relative humidity within continuous lead-lag dry days, because temperature and humidity are recorded at that very day. If temperature difference at that very day with gap rainfall record with averaged temperature in continuous days is less than 1 and relative humidity difference is less than %, the gap record day must be dry day, otherwise it is uncertainty day. The data are used for calculating summer rainfall events and dry spells, which are two parameters that reflect dry/wet conditions. Dry spell is characterized by continuous days without substantial rainfall (>1mmday 1 )injja,this threshold excludes very light precipitation and partially accounts for the limited accuracy of rain gauges (Klein and Konnen, 3; Zolina et al., 13), and it is calculated for all stations. The study mainly aimed at the duration of dry spells longer than 7 days, so as to investigate its relations with planetary wave movements, because long-time dry spells correspond to stable and large-scale circulation of planetary wave. The European Centre for Medium-range Weather Forecasting (ECMWF) re-analysis Interim (ERA-Interim) data are used to analyse atmospheric circulation anomalies, such as potential heights and Rossby waves at hpa, the westerly jet at hpa (Bromwich et al., 7). For comparison with the atmospheric circulation before 19, ERA4 data are used to represent the mean Rossby wave position before 19 (i.e ). Although the ERA4 resolution is lower than that of the ERA-Interim dataset, the ERA4 dataset can be regarded as a reference state, by comparing the 7-gpm latitude positions of ERA4 with ERA-Interim in JJA during the overlapping periods (from 1979 to ) around our study region (7 1 E, 3 N), the systematic bias of latitude positions is.3,.17 and.31 N corresponding to June, July and August, respectively, it is regarded as reference for the difference of 7-gpm isoline between ERA4 and ERA-Interim. The surface air temperature data are from the Climate Research Unit (CRU) Ts dataset, which encompasses the period from 191 to 11 (Harris et al., 14). The data error is calculated based on the CRU 3.1 dataset at a resolution of.. ( nerc.ac.uk_atom_dataent). Surface air temperatures are used to analyse the temperature gradient over northern China and the adjacent regions. The following statistical metrics and methods are applied in this study: correlation coefficients, trend analysis and diagnostic analysis. Reliability tests are also performed by using Mente Carlo test, and it is proved that Mente Carlo method is better than Pearson s correlation (Stephen, 199). The article also shows correlation coefficient between trend line (linear and nonlinear) and original data distribution, and some correlation coefficients are showed in the Figures 3, 4, and 7, and 9% confidence level is., 9% confidence level is CAM model and experiments The atmospheric module of the community earth system model, i.e. the Community Atmospheric Model version.1 (CAM.1) (Neale et al., 11), was developed at the National Center for Atmospheric Research. The dynamical scheme includes four frameworks, including a finite volume method, a spectral element method, an Euler method and a half-lagrange method. This study applies the finite volume dynamic framework with a horizontal resolution of 1.9. with 3 vertical layers in the σ-p vertical coordinate. CAM.1 is coupled to the CLM4 land process model. In this study, several physical processes, including radiation processes, cloud effects, convection, boundary layer effects and other physical processes, are represented in the model according to the default options. Only the land surface data are changed in the sensitivity experiments. A detailed description of the model is available at This study attempts to change the surface characteristics over the TP to simulate increases in air temperature over the TP. The surface character changes include receding glaciers, permafrost degradation, desertification, decreased soil moisture and vegetation degradation. One of the sensitivity experiments includes a decrease in vegetation cover by 3% ( 4 N, 7 1 E); the decrease corresponds to a fit for the vegetation trend throughout the first of 1st century (Zhang et al., 14). The initial background data are the mean values from the recent three decades (i.e. 19 1), including atmospheric data and surface data. The control simulation and sensitivity simulation are performed for 1a; the results in JA for the last 1a are analysed in this study..4. CMIP historical simulation In order to further test statistical results and CAM results, we analyse the historical simulations (19 ) from global warming A1B stabilization experiments (with an atmospheric CO concentration of history ppm) conducted 1 Royal Meteorological Society Int. J. Climatol. 3: (1)

4 43 J. ZHANG et al. Mean duration of dry 7+ days 1 (c) Duration trend of dry 7+ days E 1 E 11 E 11 E 1 E 1 E.1 1 E 1 E 11 E 11 E 1 E 1 E Frequency of dry 7+ days (d) Number trend of dry 7+ days E 1 E 11 E 11 E 1 E 1 E.3 1 E 1 E 11 E 11 E 1 E 1 E Figure. Spatial distributions of the mean duration of dry spells exceeding 7 days (a, unit: days), the number of dry spells exceeding 7 days, and the trend in the duration (c) and number (d) of dry spells exceeding 7 days. The black dashed lines in (c) and (d) represent the 9% confidence level by Mente Carlo test. The blue rectangle represents north-central China, and the thin pink lines represent provincial boundaries. by state-of-the-art ocean atmosphere general circulation models that participated in the Fifth Assessment Report (AR) coordinated by the Intergovernmental Panel on Climate Change (IPCC). The models include the Goddard Institute for Space Studies Model E-H (GISS-EH) and L Institut Pierre-Simon Laplace Coupled Model, version (IPSL CMA). The reasons to select CMIP models are that increasing temperature over TP is the result from nature variability and greenhouse gasses and its decadal variable is synchronous with anomaly SST and global warming (Klein et al., 4). AR historical simulation can realize their synchronous change due to considering SST and greenhouse gasses contents. 3. Results and discussion 3.1. Dry spell distribution over northern China The duration of dry spells is opposite to the trend for rainfall variability, therefore, the duration of dry spells could reflect dry/wet condition and drought/flood. Figure shows the spatial distribution in the mean duration of dry spells exceeding 7 days; the maximum of the mean duration is longer than 1 days, occurring over the northern portion of northwestern China. Moreover, the minimum of the mean duration is 9.3 days, occurring over the southern portion of northwestern China. Duration in north-central China is 1.1 days. There are three peak centres in the mean duration of dry spells, in northwestern (39 4 N, 1 1 E), northeastern (4 4 N, 1 1 E) and eastern (34 3 N, E) China, those durations are 1.1, 13. and 14.1 days, respectively. Moreover, two minima centres occur in the east of the TP and North China, i.e. near Beijing (37 4 N, E). These minima centres occur near mountains; therefore, the relatively low duration of dry spells is related to orographic precipitation. Figure shows the number of dry spells with the duration exceeding 7 days in summer. Most regions in northern China have exceeding three times dry spells; one maximum centred in the north-central China, it is. times; the other maximum and minimum values agree with the duration of dry spells presented in Figure. The maximum values are 4., 3.3 and 3.9 times, respectively. By multiplying by the duration of the dry spells, we find that the dry spells exceeding 7 days lasted for 7, 4 and days for three maximum centres, respectively, and days in north-central China, encompassing 3, 4, 3 and 74% of the summertime (JJA). These long durations could lead to summer droughts. Figures (c) shows the duration trend rate of dry spells. The duration trend rate of dry spells exhibits pronounced zonal variability. For example, dry spells occur more frequently in north of 37 N and east of 1 E; however, these events occur less often south of 37 N. The trend rate is. days per event per year; however, except for the east portion, increasing trend rate is not significant at the 9% confidence level. In Figure (d), the trend in the number of dry spells increases in most regions of northern 1 Royal Meteorological Society Int. J. Climatol. 3: (1)

5 NORTHWARD PLANETARY WAVES CONTRIBUTE TO PROLONGED DRY SPELLS OVER NORTH-CENTRAL CHINA JJA P Frequency of JJA rainfall events P (mm) Numbers. 1 r=.4 4. r= Figure 3. Temporal trend in summer (JJA) rainfall and the mean number of rainfall events in north-central China, dotted line is mean value and thick line is three order trend line, r is correlation coefficient of rainfall and the numbers of rainfall events with its three order trend, respectively, 9% confidence is. and 9% confidence is.3 for one-sided regression (it is similar hereinafter). 3 JJA rainfall Frequency of JJA rainfall events 3 r=.. 4. JA Jun mean trend P (mm) 1 Numbers r=.4 1. r= r= Figure 4. Temporal trends in June and JA rainfall and the mean number of rainfall events in June and JA in north-central China. The top lines are for JA; the bottom lines are for June, dotted line is mean value and thick line is trend line. r is correlation coefficient of rainfall and the numbers of rainfall events with its three order trend, respectively. China; there are a few small regions in which the increase rate is not observed. The pronounced increasing trend is significant at the 9% confidence level, in the region with an increasing occurrence of dry spells (3 4 N, E). The increasing dry spells events and weak increasing duration of dry spells suggest that drought events are becoming more likely in northern China. The study region [blue rectangle, north-central China (37 4 N, E)] is one of the regions exhibiting an obvious increase in the number of dry spells and partly weak increase in duration. Similarly, another region exhibiting an increase in both the duration and number of dry spells is defined by the boundaries of north-east China (3 4 N, E). Rainfall and dry spells in the north-east China are primarily affected by the EASM (Li et al., 1a, 1b); however, rainfall and dry spells in the north-central China need further exploration. 3.. Temporal change in summer rainfall over north-central China Figure shows a pronounced increasing trend in both the duration and number of dry spells over north-central China; these trends are associated with rainfall. Figure 3 shows the time series of total JJA rainfall (Figure 3) and the number of JJA rainfall events (Figure 3) over north-central China. The total JJA rainfall and the number of rainfall events exhibit the following variability: a decrease trend after 199, which precedes a mean value 1 Royal Meteorological Society Int. J. Climatol. 3: (1)

6 434 J. ZHANG et al. Percent (%) Duration of dry 1+ episode (days) Duration of dry 7+ episode (days) (c) Duration of dry 1+ episode (days) (d) Start date of dry 1+ episode Figure. Dry spells histogram (a, unit: %) and duration of dry spells exceeding 7 days for all stations in north-central China in JJA (b, unit: days). The duration of dry spells exceeding 1 days for all stations in north-central China in JJA (c, unit: days) and the corresponding start date (d, date) are also depicted. of rainfall, and infrequent rainfall events occurred after 199. The decrease rainfall and rainfall events contrast the increase trend in the duration and number of dry spells events. The time series of rainfall for each month, i.e. June, July and August, is also analysed (Figure 4). The trend for each month differs, especially in June, it exhibits a different rainfall trend compared with which in the other months. Rainfall and rainfall events in June exhibit a weak increasing trend, whereas rainfall and rainfall events in JA (July and August) have the same trend as the JJA rainfall trend. This pattern about summer rainfall exhibits an early wetting later drying pattern, which suggests that rainfall in JJA is affected by different mechanisms. A decrease in the number of rainfall events corresponds to prolonged dry spells, which possibly explains the early wetting later drying pattern. Therefore, we can explore decrease rainfall through prolonging dry spells and its mechanism Temporal change in dry spells over north-central China To explore the change regulation of dry spells, the duration of every dry spell from all stations over north-central China (i.e. the study region) is plotted. Figure shows a histogram of dry spell events from 1 day to 49 days in length; the occurrence frequency decreases as the duration increases. There are four peak frequency showing four time periods, those are 3 days, days, 1 1 days and 19 days. Moreover, 1% of the events are longer than 7 days, 3% of the events are longer than 1 days and events exceeding 1 days account for 1% of the events. These longer dry spell events should be given more attentions because they can more easily lead to droughts. Figure shows the duration of dry spells exceeding 7 days (Figure ), indicating that these longer duration events ranged from to more than days. From the scatter diagram, we can conclude that the durations of dry spell is highly asymmetrical as a function of time, i.e. most stations have recently reported longer dry spells than those during the early period; the average duration in the region is approximately 1 days. Moreover, the annual average duration of dry spell exhibits a pronounced increasing trend. Furthermore, events exceeding the mean value increase with time. Figure (c) shows a scatter diagram of events with a duration exceeding 1 days. This figure demonstrates that fewer dry spells exceeding days are found before However, after 1977, an increasing frequency of longer dry spell events occurs. This trend suggests that increasing stations experience prolonged dry spells or dry spells are more frequent at individual stations. Specifically, events exceeding days account for 4% of all those events with longer than 1 days durations (figure omitted); these events are primarily found after Does prolonged dry spell related to global warming? It is worthy of paying attentions, because global warming signal was just clear during the period. Figure (d) shows the start dates of long dry spells (exceeding 1 days); the start dates are primarily distributed in June (1 1) before 1977, however, most cases occurred in JA (11 4) after This shift shows that prolonged dry spells in JA is an important factor that contributes to decrease rainfall in JA, and it is favourable for an early wetting later drying pattern. This pattern is associated with an anomalous circulation of the EASM and Rossby waves caused by the prevailing 1 Royal Meteorological Society Int. J. Climatol. 3: (1)

7 NORTHWARD PLANETARY WAVES CONTRIBUTE TO PROLONGED DRY SPELLS OVER NORTH-CENTRAL CHINA 43 3 Latitude ( N) Latitude ( N) (c) 4 Latitude ( N) Z7 June Z7 July Dry s 9s After s 7s Z7 August Longitude ( E) (d) 4 Latitude ( N) (e) 4 Latitude ( N) (f) Latitude ( N) Jet June Jet July Jet August Longitude ( E) Figure. The 7-gpm isoline at hpa from June to August for the recent three decades, mean from 19 to 19, and anomaly years of dry spells (a c). The westerly jet position obtained from the U component of the wind at hpa from June to August for the recent three decades and anomaly years of dry spells (d f). (c)... 1 E 1 E 11 E 11 E 1 E 1 E. 1 E 1 E 11 E 11 E 1 E 1 E. 1 E 1 E 11 E 11 E 1 E 1 E. Figure 7. Correlation coefficients between dry spells exceeding 7 days in length and the position of the mean 7-gpm isoline (1 114 E) at hpa (left panel), and mean westerly jet at hpa (middle panel), and the EASM index from Shi in JA (right panel). Blue rectangle is the study region in north-central China. westerlies because long dry spells are also related to planetary wave anomalies Decadal change of planetary waves Fields of hpa are selected for analysing anomaly circulation because numerous radiosondes and satellite retrievals information are constrained in the fields, the atmosphere is free from surface effects and upper-level wave patterns appear in the fields, all these ensure the high precision information about planetary wave movement. The 7-gpm isoline at hpa passes through northcentral China in July and August, therefore, 7-gpm isoline is selected for reflecting planetary wave anomality that directly related to dry spells in JA over north-central China. Figure (c) shows the 7-gpm isolines in summer over the recent three decades from ERA-Interim and the mean isoline before 19 from ERA4 dataset. The figure also presents isolines from six anomalous years (i.e. 1991, 1999, 1,,, 1); the anomalous years with longer dry spells in JA are selected according to the duration of the dry spells longer than mean duration from those more than 1 days duration in JA. The isolines in the three decades are similar, exhibiting a slight northward 1 Royal Meteorological Society Int. J. Climatol. 3: (1)

8 43 J. ZHANG et al. (U) N U U = 3 China province Jet after from mean U Jet s from mean U Key region N 7 E E 9 E 1 E 11 E 1 E 13 E Figure. JA U velocity distributions (unit: m s 1 ) from ERA-Interim, JA Westerly jet position with larger than 3 m s 1 from mean U velocity (thin black line, unit: m s 1 ), the JA jet centre in 19s (blue thick line) and after s (red thick line), Yellow rectangle is the study region in north-central China. shift in June to the east of 1 E (Figure ). However, this pattern is different from the isolines in JA. Compared with the 19s and the mean isoline before 19, the 7-gpm isolines in JA (Figure and (c)) in the 199s and the s exhibit northward shifts between and 1 E. Moreover, at approximately 1 E or to the east of this location, the 7-gpm isolines are either similar or are shifted southward over the recent two decades. This result is in accordance with the findings presented in Figure and decrease rainfall in JA in Figure 4, i.e. the northward shift in the 7-gpm isoline is possibly associated with prolonged dry spells. The westerly jet is a component of the planetary system. To reveal further the northward shift in the atmospheric circulation, the westerly jet position is also analysed over the recent three decades (Figure (d) (f)). Compared with the 19s, the westerly jet during the recent two decades exhibits a northward shift to the west of 9 E; however, a southward shift is found between 9 and 13 Ein June (Figure (d)). Moreover, the westerly jet exhibits a northward shift during the recent two decades between and 14 E (except at approximately 1 E) in July (Figure (e)). Furthermore, the westerly jet exhibits a northward shift between 9 and 11 E and to the east of 13 E during the recent two decades; however, a southward shift is observed between 11 and 13 E in August (Figure (f)). These changes indicate that the westerly jet movement on the synoptic to planetary scale is within approximately longitudes; the westerly jet nearing the north-central China exhibits a northward shift in JA and a southward shift in June. 3.. Correlation of dry spells with planetary waves We first explore the early wetting later drying pattern by exploring direct circulations using the 7-gpm isoline position (latitude). Figure 7 shows the correlation coefficients of the regional average location (latitude) of 7-gpm isoline (1 114 E) over north-central China with the duration of dry spells in JA, and the reason to select the regional average location of 7-gpm isoline is that the isoline location just located between the ridge and rough of one planetary wave under 3 years on average condition, and the meridional changes of the position is almost synchronous, so, regional average positions are used to reflect wave movements and its relations with dry spells. A majority of northern China exhibits positive correlations, specifically, both the north-central and western regions exhibit high correlations that surpass the 9% significance level of Mente Carlo test. This finding suggests that prolonged dry spells correspond to a northward shifts in the 7-gpm isoline. The regional average of westerly jet in the range (1 11 E) is also performed, and the location of westerly jet is smaller than the location of 7-gpm isoline, because westerly jet from 11 to 114 E in August has different decadal change with the location from 1 to 11 E, the reason to select the regional average of westerly jet location is the same as selection of 7-gpm isoline location. Figure 7 shows the correlation coefficients of the averaged jet location with the duration of dry spells in JA. A majority of northern China exhibits positive correlation distributions, and both the north-central and 1 Royal Meteorological Society Int. J. Climatol. 3: (1)

9 NORTHWARD PLANETARY WAVES CONTRIBUTE TO PROLONGED DRY SPELLS OVER NORTH-CENTRAL CHINA N Wave Jet Trend Drought 4 N 44 N Latitude 39 N Jet latitude 4 N 3 N r =.4 3 N 3 N 3 N 4 N 44 N Wave latitude (c) Z (d) Z trend Drought 4 4 Z (gpm) Z (gpm) r =. 4 3 N 3 N 4 N 44 N Wave latitude Figure 9. Temporal trend in the location of the 7-gpm isoline at hpa and the westerly jet in JA and a scatter plot of their relationship and one-sided regression line. The temporal trend in the geopotential height at hpa in JA (c) and a scatter diagram between the 7-gpm isoline and the geopotential height are also depicted. r in and (d) is the correlation coefficient of one-sided regression, respectively. western regions exhibit high correlations that surpass the 9% significance level of Mente Carlo test. This finding suggests that prolonged dry spells also correspond to a northward anomaly of westerly jet position, and this relation is the same as dry spells with 7-gpm isoline shift. Both the analysis announces that dry spells are associated with north shift in planetary wave. Dry spells are also associated with moisture transport from Pacific and Indian ocean, which decided by EASM intensity and the position of the western Pacific subtropical high (WPSH), the study analyses correlation coefficients between the EASM index and the duration of dry spells in JA, so as to explore the effect of EASM intensity on dry spells in northern China. The EASM index from Shi and Zhu (199) as summarized and assessed by Wang et al. () is used in this study. A strong positive index reflects strong EASM and heavy rainfall. In a comparison of the recent a of the EASM, the EASM index shows that the EASM has been undergoing a weak period over the recent three decades. Furthermore, the correlation between the duration of dry spells and the EASM exhibits an east west distribution (Figure 7(c)), which is similar to that of the Mei Yu frontal pattern, i.e. a negative correlation belt is at 33 3 N; however, the correlation is positive in the north of this belt, which shows that a strong EASM corresponds to increased rainfall and decreased dry spells in the south of 3 N; this pattern is not favourable for rainfall in the north of 3 N and over north-central China, and a weakened EASM is not a unique contributor to prolonged dry spells, because the correlation coefficient between the EASM and dry spells does not pass the 9% significance level of Mente Carlo test in large range, except for some regional locations. A possible reason for reversed correlation between north-central China and south 3 Nis that the EASM could not extend to north-central China. The Figure 7 also indicates that planetary waves from the prevailing westerlies are the primary contributors to prolonged dry spells over north-central China. EASM contributes main moisture transport for rainfall in northern China, however, planetary waves contribute circulation background and vertical upward movement for rainfall, and good relationship between dry spells 1 Royal Meteorological Society Int. J. Climatol. 3: (1)

10 43 J. ZHANG et al. and planetary waves position shows that vertical upward movements change with northward shift in planetary waves. Westerly jet positions can reflect vertical movement and rainfall positions, the U vector larger than 3 ms 1 shows westerly jet range and the largest U vector position shows jet core position. Figure shows U vector distribution at hpa, and jet centre position from mean U in 19s and after s. It is found that jet in the west of 1 E is northward shift after s by comparing with jet in 19s; jet (covers the range within 3 m s 1 ) is wide, it is from 71 E to the east boundary and from 9 to 43 N, the jet core with U larger than 37 m s 1 is in the east of 11 E. North-central China just lies on the left of jet entrance, north shift of westerly jet results in north-central China closed to jet entrance, where there are cyclonic shear, positive vorticity and negative relative vorticity advection, which are favourable for convergence at high levels and divergence at low levels, and all which are further for downward movement and decreasing rainfall. Therefore, northward shift in westerly jet is one of important reasons leading to prolonged dry spells. The planetary system is the background of synopticscale systems, and a northward shift in the planetary wave necessarily leads to a northward shift in synoptic-scale systems in mid-latitude, such as shortwave troughs, which are directly related to rainfall. Therefore, a northward shift in planetary system is inferred favourable for a northward shift in shortwave trough, which also contributes to dry spells. Without doubt this inference still needs to be testified hereafter. 3.. A possible mechanism underlying the northward shift in planetary waves The decadal change in westerly jet is related to decadal change in northward shift in the location of 7-gpm isoline, and what the mechanism about the northward shift in planetary wave is worthy of discussion. Figure 9 shows that the JA westerly jet and 7-gpm isoline locations (Figure 9) exhibit a positive correlation (Figure 9). The westerly jet and the 7-gpm isoline locations in JA exhibit the same variations, which means that the westerly jet possibly follows the northward shift in the planetary wave. As noted above, global warming possibly leads to a lifting of the entire atmosphere, which assists in increasing the potential height (Hu, 1997). Figure 9(c) shows the potential height at hpa in JA, which is indicative of an approximate location for the 7-gpm isoline over north-central China. The potential height exhibits an increasing trend, i.e. it is low in the 19s and high in the 199s and the s. The correlation with the position of the 7-gpm isoline (Figure 9(d)) is weak; however, this relationship is not significant at the 9% significance level. Which indicates increasing height in low-middle troposphere is not the primary factor resulting in north shift in planetary wave, but other factors. Using orthogonal regression analysis, the total relation between the wave position and jet position and potential height is expressed (Equation (1)). The equation shows that the potential height item is ignored by orthogonal regression analysis, due to no significant level, which further shows good correlations between the jet positions and wave positions, in addition, increasing in the potential height was not the main contributor to wave position. L w =.7L j ( R =.1, F = 7.77, P =.9 ) (1) Where, L w is wave latitude position, L j is jet latitude position, R is multiple correlation coefficient, F and P are F statistic and its probability (P) value, respectively. As one of planetary systems, the westerly jet exhibits regional changes in the location throughout the midlatitudes (figure omitted) in a global warming scenario. In some regions, the westerly jet shifts northward, whereas in other regions, the jet steam shifts southward. Such a position change may lead to the spatial coexistence of floods and droughts through the mid-latitudes, which may be associated with the Indus River flood in Pakistan in 1 and other extreme events, such as the European heat wave in 3 and the heat wave in the United States in 11 (Petoukhov et al., 13). Li et al. (1a, 1b) found that the EASM rainfall responses to global warming are not related to changes in rainfall intensity but, rather, to changes in the rainfall location. EASM rainfall is also associated with a southward shift in the westerly jet over East Asia (considering only regions between 1 and 13 E). A meridional asymmetric warming with the surface warming over the mid- to high-latitude regions could explain southward shift in the westerly jet over East Asia. Moreover, Woollings et al. (1) found a northward shift in the North Atlantic eddy-driven pole-ward westerly jet over the North Atlantic. These findings not only describe regional characteristics of westerly jet shifts but also reveal the different mechanisms related to location and time-scale changes of the westerly jet. Considering that circulation changes are typically intimately coupled with low boundary forcing anomalies, it is possible that inter-decadal changes in low boundary forcing lead to inter-decadal northward shifts. To explore the possible mechanisms underlying the northward shift in the westerly jet over north-central China, we first analyse the boundary forcing using the surface temperature difference between the most recent three decades (191 11) and the preceding three decades (191 19). The surface temperature data used in this study are from the CRU Ts dataset (Figure 1). The temperature in recent three decades indicates the effects of global warming, and it indicates spatial coexistence in a global warming scenario. West of 1 E, there is a small increase or decrease in temperature in the mid- to high-latitudes (4 N, 1 E). Moreover, there is a concordant increase in temperature over the TP ( 4 N, 1 E). These spatial differences in increase temperatures result in an anomaly in the increase temperature gradient between TP and the north of TP, which opposes the north south temperature gradient anomaly to the east of 1 E. The zonally averaged temperature differences between and 1 E are plotted (Figure 1). 1 Royal Meteorological Society Int. J. Climatol. 3: (1)

11 NORTHWARD PLANETARY WAVES CONTRIBUTE TO PROLONGED DRY SPELLS OVER NORTH-CENTRAL CHINA 439 N N N N Latitude N E 7 E 94 E 11 E 1 E 11 E Longitude Latitude N dt ( C) Figure 1. The surface temperature difference between the most recent three decades (191 11) and the preceding three decades (191 19) and the absolute temperature difference between and 1 E (b, unit: C). The temperature differences exceed.3 C to the south of 4 N; however, the temperature differences are less than.1 C to the north of 4 N, which indicates an anomaly in the south north temperature gradient over TP due to the pronounced increase in TP temperature. A positive temperature gradient leads to an enhanced meridional thermal contrast and an increased meridional pressure gradient in the mid- to high latitudes. Moreover, the South Asian High (SAH) at hpa is northward wider to the south of the westerly jet (figure omitted), which provides physical evidence of a northward shift in the westerly jet at hpa. In addition, previous studies found a drastic northward shifts in inter-tropical convergence zone (ITCZ) with an anomalously strong north-minus-south SST gradient over the eastern equatorial Indian Ocean (9 11 E) (Weller et al., 14). Whether SST gradient in Indian ocean drives north shift ITCZ, and then it also contributes to north shift subtropical zone and north shift planetary waves in the mid-latitude, except for increasing TP temperature? Or, northward shift in westerly jets and SAH are the common results of ITCZ in Indian Ocean and increasing TP temperature? These questions also need further exploration Simulation of the northward shift and anomaly rainfall To simulate the planetary wave shift caused by the pronounced increase in temperature over the TP that ultimately resulted in a larger temperature gradient, this study attempts to change the surface characteristics over the TP to simulate increases in air temperature over the TP. The surface character changes include receding glaciers, permafrost degradation, desertification, decreased soil moisture and vegetation degradation. One of the sensitivity experiments includes a decrease in vegetation cover by 3% ( 4 N, 7 1 E) in the study area; the initial background data are the mean values from the most recent three decades (19 1), including relevant atmospheric data and surface data. The control simulation and the sensitivity simulations are performed for 1a; the results in JA for the final 1a are selected for analysis. Figure 11 shows the air temperature difference at m between the sensitivity and control simulations, and the 7-gpm isoline in the sensitivity and control simulations (Figure 11) from the coupled CAM model and CLM4. The results indicate that there are positive temperature centres over the TP, East of Lake Baikal and South China, whereas negative temperature are located over the middle and lower reaches of the Yangzi River, the mid- to high-latitudes ( N). The 7-gpm isoline of sensitivity simulation exhibits a northward shift. North shift in planetary waves associate with anomaly rainfall. The same as data section definition of dry spell, rainfall in model data is characterized by > 1mmday 1, Figure 11 shows mean JA rainfall calculated from control simulations (thin thick line), and anomaly JA rainfall (shadow), which is the difference between sensitivity and control simulations. It is found that north-central China exhibits negative anomaly rainfall, which is helpful for prolonging dry spells, and this result is agreed with our analysis as above. Comparing anomaly rainfall centres with mean rainfall centres, the anomaly rainfall centre is northward shift in the west China but south shift in the east China, and positive anomaly rainfall over TP occurred in the north of the rainfall centre over TP, negative anomaly rainfall covered wide range in the Asian mid-latitude (3 N, 7 11 E), showing obvious climate shift on the planetary scale; positive anomaly rainfall over east China (3 4 N, east of 11 E) occurred in the south of the rainfall centre (3 4 N). Anomaly rainfall shifts show the same change with 7-gpm isoline and westerly jet (figure omitted). Increasing temperature around Lake Baikal is significantly related to decreased intensity of northern EASM (Zhu et al., 1); however, the increasing temperature located in the east of Lake Baikal in CAM, it would affect the precipitation. In order to decrease uncertainty of CAM simulation, the study also analysis ensemble mean of two historical simulation results (19 ) from global warming A1B stabilization experiments conducted in the IPCC AR 1 Royal Meteorological Society Int. J. Climatol. 3: (1)

12 44 J. ZHANG et al. N N dt 1. Latitude N 7 E E 9 E 1 E 11 E 1 E 13 E Longitude N dp and P 3 N 4 4 Latitude 4 1 N 7 E E 9 E 1 E 11 E 1 E 13 E Longitude Figure 11. The air temperature difference at m (unit: C) and the 7 isoline of control (blue thick line) and sensitivity (pink thick line) simulation, mean precipitation (black line, unit: mm) and anomaly precipitation (shadow, unit: mm) between the sensitivity and control simulations from the CAM model and 7 isoline. Red rectangle is the study region in north-central China. Blue thin line is lake-land boundary, pink dash line is province boundary. (see Figure 1). The difference between the most recent three decades (191 ) and the preceding three decades (191 19) is analysed. There are positive temperature centres over the TP, south of Lake Baikal, whereas negative temperatures are located in the midto high-latitudes (4 N) and south of East Asia (south of 4 N and the east of 1 E). The 7-gpm isoline exhibits a northward shift (from 7 to 11 E). North-central China exhibits negative anomaly rainfall, which is helpful for prolonging dry spells, this result is well agreed with our analysis and CAM results as above, however, no obvious shifts in the position of anomaly rainfall. In all, increasing temperature over TP and temperature gradient between TP and mid-high latitude are helpful for northward shift in planetary waves, which further leads to decreasing rainfall and prolonging dry spells over north-central China. According to the above analysis, an increased temperature gradient between the TP and the mid- to high-latitudes is favourable for an increase in the thermal contrast and pressure gradient, which likely strengthens and widens the SAH and is also favourable for a northward shift in the westerly jet, furthermore, north-central China is closed to the left of jet entrance, negative relative vorticity advection leads convergence at the high level and downward movement, and finally prolonging dry spells. Therefore, increased temperature over the TP should exhibit nonlinear relationships with dry spells or rainfall events in the study region. However, the mechanism how increase temperature over the TP is related to a strengthened and widened SAH requires additional exploration. 4. Summary and conclusions Summer rainfall in north-central China exhibits an early wetting later drying pattern, which is indicated by a slight increase in rainfall in June and a subsequent decrease in JA. The JA rainfall trend is the same as the JJA rainfall trend. The pattern shows that rainfall in JJA is affected by different mechanism. Rainfall in JA is associated with dry spells. Two large regions exhibit an increasing trend in both the duration and number of dry spells; these regions include north-central China and north-east China. In our study region, the durations and numbers of dry spells exceeding 7 days exhibit a pronounced increase, especially 1 Royal Meteorological Society Int. J. Climatol. 3: (1)