Seasonal Palmer drought severity index reconstruction using tree-ring widths from multiple sites over the central-western Da Hinggan Mountains, China since 1825 AD

Abstract

Many evidences indicate that the drought in Northern China has progressively intensified, threatening people’s lives and properties. Drought has become one of the most severe environmental problems in Northern China, and thus has serious impacts on the regional social and economic development. The present study investigated the seasonal Palmer drought severity index (PDSI) over the central-western Da Hinggan Mountains (CW-DHM), northeastern China. 283 cores of Mongolian pine trees from 6 locations were used to generate a regional ring-width chronology. The chronology in CW-DHM was significantly correlated with the May–July PDSI, with an explained variance of 49% (r = 0.700, 1951–2013, p < 0.0001). The regional May–July PDSI (PDSI5−7) from the CW-DHM was reconstructed from 1825 to 2013 AD. The ensemble empirical mode decomposition method (EEMD) and multi-taper method (MTM) spectral analyses revealed that the cycles in the reconstructed PDSI5−7 were close to those of the ENSO and solar activity. This suggests that both the ENSO and solar activity have strong influence on the PDSI5−7 variation in the CW-DHM region. In addition, EEMD also revealed that the Pacific decadal oscillation and the Atlantic multi-decadal oscillation influenced the drought variation in this region. The PDSI5−7 reconstruction showed a long-term declining (dry) trend during the period of the 1950s–2010s. This drying trend was also detected in the PDSI data of other parts of China after the 1950s. We believe that these phenomena may be related to a large extent with the weakening of the East Asian summer monsoon.

Introduction

Drought is one of the most destructive natural disasters, and it has an extremely serious impact on regional agriculture, water resources, and environmental quality. Regional drought, as one of the characteristics of global climate change, has become a significant content of global climate change research (Fu 2008). The Palmer drought severity index (PDSI) is an indicator of the extent of drought (Palmer 1965; Dai 2011). The acquisition of the PDSI is of great importance in understanding the degree of climate change in the past. However, it is difficult to study PDSI variation in periods before the instrumental era because the period of contemporary meteorological observations is far too short. For example, more than 90% of the meteorological observations in China were recorded only since the 1950s (Zhang 1996).

China is a monsoonal country, and the climate in China displays complex variation. The PDSI variation in eastern China is always correlated with the strength of the East Asian summer monsoon (EASM) (Ding et al. 2008; Li et al. 2011). Drought and flood (or the dry and wet events) caused by a strong or weak EASM often bring huge damage to people’s lives and properties. Many studies have indicated that the drought in Northern China has progressively intensified (Fu 2008; Fu et al. 2008). Drought has become one of the most severe environmental problems in northern China and thus has serious influences on the region’s social and economic development (Ma and Fu 2005).

Using tree rings to investigate past PDSI variation is an important aspect of tree ring research. Analysis of tree ring data, with the advantages of high resolution, wide distribution, and easy sampling, has become an important approach to study past PDSI changes (Cook et al. 2010). Variation in the PDSI has been reconstructed based on the integrated impact of temperature and precipitation on tree growth within China (Song and Liu 2011; Cai and Liu 2013; Sun et al. 2012). However, these studies have mainly focused on one particular site, and thus regionally integrated research is scarce.

Although Chinese dendrochronologists have obtained a series of results from northeastern China (Bao et al. 2015; Liu et al. 2009, 2011, 2016; Sun et al. 2016; Peng et al. 2013; Zhu et al. 2009; Shi et al. 2015; Chen et al. 2011), and the levels of precipitation, temperature, relative humidity, and SPEI (Standardized precipitation evapotranspiration index) have been reconstructed for the past 200 years through these studies, there has been little PDSI research. In particular, our knowledge of PDSI variation in this region remains inadequate. In this paper, we present a regional PDSI reconstruction for the past 189 years using tree-ring width data from six locations in the central-western Da Hinggan Mountains (CW-DHM).

Materials and methods

Tree ring data and the CW-DHM regional tree-ring width chronology development

A total of 283 cores of Mongolian pine (Pinus sylvestris var. mongolica) trees from 6 sites (Fig. 1) were used to generate a regional ring-width chronology. Among these 6 sites, 5 were previously studied; and the new site was Hongweiqiao (HWQ, 48°16.53′N, 120°02.90′E, altitude 300–390 m), located on the western slope of the National Honghuaerji Forest Park, Inner Mongolia in the CW-DHM. In the HWQ site, 51 cores were collected from 28 trees in August 2009.

Fig. 1
figure1

Map of the study region (in the central-western Da Hinggan Mountains). The six sites are HWQ (this paper), HLBE (Liu et al. 2009), NGNE (Bao et al. 2015), SSQ (Liu et al. 2016), and BRT01 and BRT02 (Sun et al. 2016). In the upper right corner, three sites (red dots) are used for comparison with the present study. MSR denotes the Selenge River, Mongolia (Davi et al. 2006); NC is the North-Central China (Yi et al. 2012), and KT is the Mt. Kongtong on the Loess Plateau, China (Song and Liu 2011)

The samples from HWQ were dried and mounted following standard dendrochronological methods (Stokes and Smiley 1996; Phipps 1985). Samples were polished until the annual rings are clearly visible. The other five sites were Hulunbeier (HLBE, Liu et al. 2009), Nuogannuoer (NGNE, Bao et al. 2015), Shenshuiquan (SSQ, Liu et al. 2016), Baritu 01 (BRT01, Sun et al. 2016), and Baritu 02 (BRT02, Sun et al. 2016). The descriptions of all these six sites are given in Table 1.

Table 1 Descriptions of the six tree-ring sites used in the study

Since the 6 sites are closed and same tree species, we put all 283 original ring-width series were together to generate a regional chronology for CW-DHM using the ARSTAN program (Cook 1985). During the process, the relatively conservative negative exponential or linear functions were selected to eliminate climate-independent growth trends in each series. All of the series were then combined into a master chronology by computing a robust bi-weight mean. Then, the standard (STD), residual (RES) and autoregressive (ARS) chronologies were obtained. The STD chronology was selected in the following analysis (Fig. 2) since it contains both low- and high-frequency signals. The tree-ring chronology of CW-DHM spans the years 1748–2013 for a total of 266 years. The rate of missing rings was 0.08%; the mean correlation among all of the series (r1) was 0.54, and the average mean sensitivity was 0.25. These values suggest that the growth of trees and their controlling environmental factors in the CW-DHM area have been very consistent. An EPS (expressed population signal) value greater than 0.85 is generally considered to be an acceptable threshold for a reliable chronology (Cook and Kairiukstis 1990; Wigley et al. 1984). As a result, the valid period of our chronology was 1825–2013. The signal strength of the STD chronology was evaluated over time using moving correlations (Rbar) and EPS (D’Arrigo et al. 2005). The statistical characteristics of the STD chronology are shown in Table 2.

Fig. 2
figure2

a STD chronology of the central-western of the CW-DHM. b Number of cores. c Running EPS. d Rbar statistics

Table 2 Statistical characteristics of the STD chronology in the central-western Da Hinggan Mountains

Climatic data

To evaluate the climatic response of the regional tree-ring width chronology, two climatic parameters, precipitation (P) and mean temperature (T), were obtained from the Climatic Research Unit time-series (CRU TS) 3.24 (University of East Anglia Climatic Research Unit 2008) in 0.5° × 0.5° gridded data sets (47.5°N–50°N, 117.5°E–122.5°E, 1951–2013 AD). These two parameters were used to conduct further research. The distributions of the monthly temperature and precipitation are shown in Fig. 3.

Fig. 3
figure3

The monthly mean temperature and mean precipitation from the CRU TS 3.24 in 0.5° × 0.5° gridded datasets (47.5°N–50°N, 117.5°E–122.5°E, 1951–2013 AD)

The index is a comprehensive reflection of the changes in precipitation and temperature. In this paper, the PDSI data (Dai 2011, https://climexp.knmi.nl/) covering 47.5°N–50°N, 117.5°E– 122.5°E (1951–2013 AD) was used to test the ring width-drought response.

Data used in comparisons

To investigate the regional-scale climate signal variability, spatial correlation analyses were performed using the KNMI Climate Explorer (http://www.knmi.nl) (Mitchell and Jones 2005). To verify the reliability and spatial representativeness of our reconstructed PDSI, we used the following data for comparison tests.

  1. 1.

    Tree-ring width-based Selenge River, Mongolia (MSR) streamflow reconstruction (Davi et al. 2006);

  2. 2.

    Tree-ring width and historical documents based the North-Central China (NC) summer precipitation reconstruction (Yi et al. 2012);

  3. 3.

    Tree-ring width-based PDSI reconstruction for Mt. Kongtong (KT) on the Loess Plateau, China (Song and Liu 2011).

Statistical methods

To investigate the climate responses of the CW-DHM tree-ring width STD chronology, Pearson’s correlation (r) and partial correlations were calculated. The split calibration–verification method was used to test the stability and reliability of the transfer functions (Cook et al. 1999; Fritts 1991). These procedures were performed by calibrating climate data from one subperiod (1951–1980 and 1981–2013) and verifying the reconstruction using the remaining data (1984–2013 and 1951–1983). The verification of this result was evaluated by using the correlation coefficients (r), sign tests (ST), reduction of error tests (RE), the coefficients of efficiency (CE), and the product means (t). Values of RE and CE greater than zero indicate rigorous model efficiency (Cook et al. 1999).

The ensemble empirical mode decomposition method (EEMD, Wu et al. 2009) was employed to decompose our reconstructed data set. The first step in this method provides a relatively consistent reference size distribution by adding white noise to the target data. In the second step, data with white noise are decomposed into IMFs (intrinsic mode functions). After several iterations of steps one and two, the cumulative effect of the added white noise is reduced to a negligible level, and ensemble means of corresponding IMFs of decompositions with the same time scales are obtained (Sang et al. 2012; Shi et al. 2011).

Finally, to represent the regional-scale climate signal variability, spatial correlation analyses were performed using the KNMI Climate Explorer (http://www.knmi.nl) (Mitchell and Jones 2005).

Results

CW-DHM regional tree-ring width climate response

As the goal of this study was to achieve a multi-site based regional scale PDSI reconstruction, before pooling all sites together for reconstructing, it was necessary to calculate the correlations of all six chronologies and local PDSI data, since the other five studies dealt with precipitation, SPEI, and humidity. The chronologies from six sites were significantly correlated with local PDSI (Table 3).

Table 3 Correlation matrix of all six chronologies and CW-DHM STD chronology with the local PDSI data

The correlation between the STD chronology from CW-DHM and the climate records (temperature and precipitation) of the CRU TS3.24 did not meet the minimum requirement of the explained variance in dendroclimatic reconstructions. For example, the highest positive and negative correlations were 0.608 for total precipitation from the prior August to the current July and − 0.439 for the mean temperature from the prior December to the current July, respectively. Therefore, we calculated the correlation between the STD chronology and the regional PDSI. All the chronology has significant correlation with monthly PDSI. However, after the month’s combination, the greatest correlation was with May–July mean PDSI, which was determined to be the main factor limiting tree growth in the CW-DHM region. It also has clear physiological significance of tree growth. The correlation coefficient between the STD chronology and May–July PDSI (PDSI5−7) reached 0.70 at the 99.9% significance confidence level (n = 63) (Table 4; Fig. 4).

Table 4 Partial correlations between the CW-DHM STD chronology, PDSI5−7, precipitation (P5−7) and temperature (T5−7) from CRU TS3.24 (1951–2013)
Fig. 4
figure4

Correlations between the CW-DHM STD chronology and instrumental climate datasets. a Precipitation and temperature from CRU TS3.24 (47.5°N–50°N, 117.5°E–122.5°E, 1951–2013). b Regional PDSI (1951–2013) (https://climexp.knmi.nl/start.cgi?id=someone@somewhere, Dai 2011, 47.5°N–50°N, 117.5°E–122.5°E, 1951–2013). PDSI5−7 means the May–July PDSI, which is the reconstruction target we have chosen

It is worth noting that in order to make a comparison between the results from diverse detrending methods, the signal-free regional curve standardization (SF-RCS, Esper et al. 2003) was also used. However, as seen in the results, although the chronology acquired by the SF-RCS method showed significant correlation (0.627, p < 0.0001, n = 63) for the May–July PDSI, this was lower than the correlation coefficient value of 0.70 obtained by the negative exponential method. This demonstrates that the SF-RCS method was not suitable for our study. The analysis will thus focus on the CW-DHM STD chronology gained from the negative exponential method.

The CW-DHM STD chronology reflects the regional PDSI5−7 variations, but we wish to know the specific roles of temperature and precipitation in the changes in regional PDSI.

The results of the partial correlation analysis (Table 4) showed that the correlation between PDSI5−7 and the May to July mean temperature was − 0.15 (n = 63, p < 0.251) when the precipitation was fixed, and the correlation between PDSI5−7 and the May to July mean precipitation was 0.65 (n = 63, p < 0.0001) (Table 4) when the mean temperature was fixed. This was consistent with the results illustrated in Fig. 4, i.e., that tree rings in this area are controlled by the growth season water availability.

CW-DHM regional PDSI5−7 reconstruction during 1825–2013 AD

A transfer function was designed to reconstruct the May–July PDSI variation during 1825 to 2013 AD (189 years) in the CW-DHM region:

$${\text{PDS}}{{\text{I}}_{{\text{5}}-{\text{7}}}}={\text{ 6}}.{\text{76}} \times {\text{ST}}{{\text{D}}_{\text{t}}}-{\text{6}}.{\text{85}}$$
(1)

(n = 63, 1951–2013 AD, r = 0.700, p < 0.0001, R2 = 49%, R2adj = 48.1%, sd = 1.30, D/W = 1.46); where STDt is the CW-DHM regional tree-ring standard chronology at year t. The correlation coefficient between the reconstructed PDSI5−7 and the observed PDSI was 0.70; the percentage of explained variance was 49%. After an adjustment for the loss of degrees of freedom, the explained variance was 48.1%. The standard deviation (SD) was 1.30. The Durbin–Watson statistic (D/W, Durbin and Watson 1950) was 1.46. With a sample size n of 63, a D/W value between 1.45 and 2.55 indicates no autocorrelation in the series.

The original observation series and the reconstructed series matched quite well, as did their first-order difference (Fig. 5a, b). The correlation between the two series after the first-order differencing was 0.68 (n = 62, p < 0.0001).

Fig. 5
figure5

Comparisons between the observed and reconstructed May–July PDSI over the CW-DHM (1951–2013). a Original series. b First-order difference series. c The PDSI5−7 reconstruction from 1825 to 2013. The gray line is the reconstructed series; the smoothed dark line is an 11–year moving average

During the calibration–verification period, the results of the split-sample method (Table 5) indicated that the correlation coefficients (r) and the product means (t) as well as the sign tests (ST) in all of the periods were significant at the 99% confidence level. The reduction of the error (RE) and the coefficient of efficiency (CE) values in all verification periods were positive, indicating that the transfer functions (1) had reasonable predictive power (Cook et al. 1999). All of these statistical parameters indicated that the regression model was stable and reliable and that it could be used to reconstruct the CW-DHM PDSI5−7 variations.

Table 5 Statistics of the split calibration–verification model for the PDSI5−7 reconstruction of the CW-DHM region

Wet and drought events recorded in the CW-DHM regional PDSI5−7 reconstruction

We reconstructed the PDSI5−7 history for the CW-DHM region during 1825–2013 AD (Fig. 5c) using Eq. (1). The mean value of PDSI5−7 during 1825–2013 AD was − 0.21, and the standard deviation (σ) was 1.20. We defined an extreme wet year as one where the reconstructed PDSI5−7 value was greater than the mean + 1σ (0.99), while an extreme drought year was similarly defined as a value lower than the mean − 1σ (–1.41). Extremely wet and dry years accounted for 14.3% (27 years) and 15.3% (29 years), respectively. The top 10 drought years and wet years are listed in Table 6.

Table 6 Ten extreme dry and wet years from the 189–year reconstruction

Spatial representativeness of the reconstructed PDSI5−7 over the CW-DHM region and tele-connections with other regions

To examine the spatial representativeness of our PDSI5−7 reconstruction, a spatial correlation analysis was conducted using KNMI climate explorer (https://climexp.knmi.nl/). Spatial correlation results showed that the observed and reconstructed series of PDSI5−7 in the CW-DHM had similar spatial distribution patterns (Fig. 6), indicating that our reconstruction was highly reliable.

Fig. 6
figure6

Spatial correlation fields of the PDSI5−7 in the CW-DHM (1951–2013 AD). a Observation. b Reconstruction

Aside from the above spatial analysis, further comparisons were made between our PDSI5−7 reconstruction and Mongolian Selenge River streamflow reconstruction (Davi et al. 2006, Fig. 7a, a1), the North-Central China summer (June to August) precipitation reconstructions (Yi et al. 2012, Fig. 7b, b1), and PDSI reconstruction in the Loess Plateau, China (Song and Liu 2011, Fig. 7c, c1).

Fig. 7
figure7

Comparisons between the reconstructed PDSI5−7 in the CW-DHM (black lines) and Mongolian Selenge River streamflow reconstruction (Davi et al. 2006, gray lines in a, a1), the North-Central China summer (June–August) precipitation reconstructions (Yi et al. 2012, gray lines in b, b1) and PDSI reconstruction in the Loess Plateau, China (Song and Liu 2011, gray lines in c, c1). Left panels are the original series; right panels are the 20-year low pass values of the series shown in the left panels. Significance level of correlation coefficients in the right panels were tested through the calculating effective number of degrees of freedom (Yan et al. 2004)

Periodicity analysis of the reconstructed PDSI5−7 over the CW-DHM region

The MTM showed that the PDSI5−7 reconstruction in the CW-DHM region had 18.62- and 3.44-year quasi-cycles over the past 189 years at the 99% confidence level (Fig. 8). Note that there also existed 20.88-, 6.87-, 4.59-, 3.58-, 3.51-, 3.30-, and 2.30-year cycles at the 95% confidence level (Fig. 8).

Fig. 8
figure8

Spectral analysis results of the reconstructed PDSI5−7 over the CW-DHM from 1825 to 2013 AD. The smoothed lines indicate the 95% and 90% confidence levels

Ensemble empirical mode decomposition maximum entropy spectral analysis of the reconstructed PDSI5−7 over the CW-DHM region

The ensemble empirical mode decomposition maximum entropy spectral analysis (EEMD-MESA, Wu et al. 2009) method was used to derive five main intrinsic functions (Fig. 9, IMF1–IMF5) and two trends (IMF6 and IMF7). The variance contribution rates were 21.7%, 21.8%, 22.1%, 16%, 12.3%, 4.7%, and 1.3%, respectively. From IMF1 to IMF5, the main cycles were 3.42-, 8.98-, 18.97-, 35.33-, and 64.1-years at the 95% confidence level, respectively. Since the length of the series is only 189 years, there was no period for IMF6 and IMF7.

Fig. 9
figure9

The results of ensemble empirical mode decomposition maximum entropy spectral analysis (EEMD-MESA, Wu et al. 2009) of the reconstructed PDSI5−7 over the CW-DHM from 1825 to 2013 AD. The confidence level is 95%

Discussion

The 189-year PDSI5−7 variation characteristics over the CW-DHM region, and their spatial and temporal links to central-eastern China

The area east of the Da Hinggan Mountains is a temperate monsoon region that is impacted by the EASM (Fu et al. 1998; Zhao et al. 2002). The EASM brings abundant water vapor, with a mean annual precipitation of more than 400 mm. Since the Da Hinggan Mountains form a rain shadow, blocking the EASM, the area west of the mountains is characterized by a temperate continental climate with annual mean precipitation below 400 mm, and precipitation mainly occurs from June to August (Wang 1997).

The intensity of the EASM has a strong influence on the dry-wet/drought-flood cycles in the Da Hinggan Mountains region.

Our reconstruction represented the CW-DHM region, part of northeastern China (Fig. 6), and it captured many drought and wet years. The partial extreme drought years, such as 1851, 1892–1893, 1906, 1951, and 1986–1987 AD, and some wet events (Table 6) are consistent with previous studies. Overall, the frequency of wet years (14.3%) was slightly less than that of drought years (15.3%) in the CW-DHM region during the past 189 years. However, the Ding-Wu Disaster (1876–1878), the most serious drought event of the nineteenth century in central-northern China (Zhang and Liang 2010; Kang et al. 2013; Hao et al. 2010), did not appear in our reconstruction. This illustrates that during this remarkable drought event, there was no synchronization between the humid CW-DHM region and dry region of central-northern China.

The climate difference between the Northeast and central-northern China is also reflected in the factors that affected the variation in the PDSI. A very interesting observation is that the partial correlation analysis (Table 4) revealed that the PDSI5−7 in the CW-DHM region was mainly controlled by large-scale precipitation variation, not temperature. This was quite different to the results for the Loess Plateau in central-northern China, where the PDSI is mainly controlled by temperature instead of precipitation (Song and Liu 2011). This is because there are different precipitation and temperature in the CW-DHM region and the Loess Plateau.

After 21-year smoothing filtering, the persistent intervals of wet periods were the 1840s–1850 s, the mid-1860s–early 1870s, the 1910s–early 1920s, the late 1930s–late 1960s, and the mid-1970s–early 1980s. The dry periods were the 1830s–1840 s, the late 1870s–1880 s, the 1890s–1900 s, the late 1920s–early 1930s, and the late 1980s–2010 s.

Another notable feature displayed in our PDSI5−7 reconstruction is that the PDSI5−7 had a long-term declining (dry) trend during the period of the 1950s–2010s (Fig. 5c). This drying trend was also detected after the 1950s in the relative humidity series for Yaoshan in central China (Liu et al. 2017a), the relative humidity series for Tianmu Mountain in eastern China (Liu et al. 2018), the Mongolian Selenge River streamflow data (Davi et al. 2006; Fig. 7a, a1), the North-Central China summer (June–August) precipitation (Yi et al. 2012; Fig. 7b, b1), and the PDSI reconstruction for the Loess Plateau, China (Song and Liu 2011; Fig. 7c, c1). As we noted above, the eastern part of China is strongly influenced by the EASM; a stronger monsoon brings larger amounts of water vapor and vice versa. The precipitation-related factors, such as PDSI and relative humidity, have a tight relationship with the EASM. As such, the decreasing trends in all of these curves are related to the amount of monsoon precipitation in China having gradually reduced since the 1950s. At the same time, the temperatures in central-northern China were rising (Yi et al. 2012). That is to say, the slowly decreasing trends of PDSI5−7 and relative humidity since the 1950s may have been triggered by the weakening of the Asian summer monsoon in North China. The simulations from models showed that the forcing of sea surface temperatures, primarily in the central and eastern Pacific, was able to induce most of the observed weakening of the EASM circulation (Li et al. 2010). Increased anthropogenic aerosol emission strongly masks warm-ocean-warmer-land, negating to the rainfall increases due to the greenhouse gases warming, and leading to a further weakening of the EASM circulation, through increasing atmospheric stability (Lau and Kim 2017). There was an apparent rise in the last few years (Fig. 7), which was mainly due to the more precipitation anomaly in 2012 and 2013 (Li et al. 2015; Ma et al. 2013). In these 2 years, high over the Okhotsk sea were strong. Because of the blocking of high over Okhotsk, the cold air accumulated in the northwest region in Inner Mongolia. In addition, the anticyclone from the northwest Pacific continuously transported water vapor to this place. Therefore, the cold and warm air flow met here and led to more rainfall.

Possible mechanisms affecting PDSI5−7 variation in the CW-DHM region

The quasi-cycle of approximately 2 years (Fig. 8) obviously corresponds to that of the Tropospheric Biennial Oscillation (TBO, Meehl and Arblaster 2002) that is common in climate change and has been noticed in many previous tree-ring studies.

The main cycles of 18.62- and 20.88-year in our reconstruction (Fig. 8) apparently are related to solar activity double cycles (~ 22-year, Han and Han 2002). This cycle is also confirmed by an 18.97-year cycle in the IMF3 (Fig. 9c), which had the greatest variance of 22.1% in the set of IMFs in the EEMD decomposition (Fig. 9). Along with the IMF5 (Fig. 9e; see the following discussion), this demonstrated that solar activity had influence on PDSI5−7 variation in the CW-DHM region.

Other relatively notable cycles are from 3.3- to 6.87-year (Fig. 8). These appear to be clear signals of the ENSO (Kaplan et al. 1998; Liu et al. 2017b). The IMF1 (Fig. 9a) and IMF2 (Fig. 9b) also displayed those cycles. The contributions of these two modes account for 43.5% (21.7% + 21.8%) of the total variance. This suggests that ENSO, as suggested in a previous study (Davi et al. 2006), may play an obvious role in the PDSI5−7 change in the CW-DHM. When El Niño occurs, the sea-surface temperature in the equatorial east Pacific is higher, resulting in weakening of the Walker circulation and Hadley circulation in the west Pacific. Because of the weakening of the Hadley circulation, the intensity and location of the western pacific subtropical high are stronger and more southerly, respectively, and the convergence flow on its northern side becomes notably strong, therefore, there is less rainfall in northern China (Su and Wang 2007). So, the CW-DHW turns to be drought.

The 35.33-year cycle in the IMF4 (Fig. 9d) accounts for 16% of the total variance; the cycle may be related to the 35-year quasi-cycle of the PDO (Mohtadi et al. 2016). Previous studies have shown that the impacts of the PDO on the study areas were significant (Bao et al. 2015; Davi et al. 2006). Dry periods in the CW-DHM corresponded to warm phases of the PDO, and vice versa. The PDO influences on precipitation change in northern China by modulating the strength of the Asian summer monsoon and the location of the subtropical high (Shen et al. 2006). A study using observed data also revealed that there was a significant negative correlation between the PDSI in northern China and the PDO index during the period 1900–2010 (Qian and Zhou 2014).

In addition, the 35.33-year cycle is very consistent with the famous Brukner quasi-cycle in climatic processes (Zhang 1976), which can be interpreted as a result of the nonlinear effect of solar activity and variability on atmospheric processes (Raspopov et al. 2000). This cycle has been detected in many studies, including the precipitation from May to August over China’s lower reaches of the Yangtze River during 1885–1972 AD (Zhang 1976), the precipitation from May to August in Inner Mongolia of Northern China (Liu et al. 2001), and the precipitation from July to August in the Beijing and Tianjin regions during 1891–1972 AD (Zhang 1976).

The 64.1-year cycle of IMF5 (Fig. 9e) accounts for 12.3% of the total variance and may be coherent with solar activity and the Atlantic multi-decadal oscillation (AMO, Mohtadi et al. 2016). The solar radiation and AMO may influence climate variation with a cycle of around 60 years (Kerr 2000; Sutton and Hodson 2005; Mazzarella 2007). This cycle was also found in tree-ring temperature series from the Tibetan Plateau (Shao and Fan 1999; Liang et al. 2008).

Recently, the relationship between the AMO and the Asian monsoon climate has been receiving greater attention. The AMO can be regarded as a pacemaker of the Northern Hemisphere surface temperature (Zhang and Delworth 2007) and global air temperature (Kravtsov and Spannagle 2008). It has been observed that the warm-phase AMO tends to enhance air temperatures in East Asia and northern India, resulting in more rainfall, particularly in the summer (Wang et al. 2009). Usually, AMO variation can cause the change of the North Atlantic circulation system which likely affects the west wind belt trough system. Therefore, the climate in northeastern China is affected by AMO variations due to its downstream location (Qian et al. 2014). By combining the observational data and numerical simulation models, it has been shown that more precipitation is induced in northern China during the warm phase of the AMO (Li and Bates 2007).

The above discussion implies that IMF4 and IMF5 together indicated that the both PDO and AMO have influences on drought variation in this region.

IMF6 and IMF7 (Fig. 9f, g) indicate the megatrends variation of the PDSI5−7, and there is no apparent periodicity. This is reflected in the long-term tendency of the PDSI5−7 over the CW-DHM. However, these two trends have variance contributions to the total variation of the PDSI5−7, although less than the other functions.

Conclusion

In this study, the seasonal Palmer drought severity index (PDSI) was reconstructed using Pinus sylvestris tree-ring width data from the past 189 years at six locations in the central-western Da Hinggan Mountains (CW-DHM), northeastern China. The reconstruction captured many severe drought and wet events. Both the ensemble empirical mode decomposition method (EEMD) and MTM cycle analysis indicated that ENSO and solar activity have influences on PDSI5−7 variations in the CW-DHM region. In addition, the results of EEMD also revealed that the PDO and AMO influence drought variations in the study area. Overall, the factors that influence the PDSI5−7 variations on the CW-DHM region appear to be very complex.

The PDSI5−7 had a long-term declining (dry) trend during the period of the 1950s–2010s. To a large extent, this drying trend was caused by the weakening of the Asian summer monsoon, which may be potentially linked to the global warming caused by human activity during this period of time.

The phenomenon of the recent weakening of the East Asia Summer Monsoon, which causes decreased precipitation and enhanced drought tendency in northern China, is an important issue. Limitations of space prevent a more complete discussion of this topic. However, this is an issue of some significance because the drought in Northern China is threatening people’s lives and properties. We believe that this issue deserves further study and discussion.

References

  1. Bao G, Liu Y, Liu N, Linderholm HW (2015) Drought variability in eastern Mongolian Plateau and its linkages to the large-scale climate forcing. Clim Dyn 44:717–733

    Article  Google Scholar 

  2. Cai QF, Liu Y (2013) Climatic response of Chinese pine and PDSI variability in the middle Taihang Mountains north China since 1873. Trees Struct Funct 27:419–427

    Article  Google Scholar 

  3. Chen ZJ, He XY, Cook ER, He HS, Chen W, Sun Y, Cui MX (2011) Detecting dryness and wetness signals from tree-rings in Shenyang northeast China. Palaeogeogr Palaeoclimatol Palaeoecol 302(3):301–310

    Article  Google Scholar 

  4. Cook ER (1985) A time series analysis approach to tree-ring standardization. Dissertation, University of Arizona, Tucson

  5. Cook ER, Kairiukstis LA (1990) Methods of dendrochronology. Kluwer Academic Publishers, Dordrecht

    Google Scholar 

  6. Cook ER, Meko DM, Stahle DW, Cleaveland MK (1999) Drought reconstructions for the continental United States. J Clim 12(4):1145–1162

    Article  Google Scholar 

  7. Cook ER, Anchukaitis KJ, Buckley BM, D’Arrigo RD, Jacoby GC, Wright WE (2010) Asian monsoon failure and megadrought during the last millennium. Science 328:486–489

    Article  Google Scholar 

  8. D’Arrigo R, Mashig E, Frank D, Wilson R, Jacoby G (2005) Temperature variability over the past millennium inferred from Northwestern Alaska tree rings. Clim Dyn 24(2–3):227–236

    Article  Google Scholar 

  9. Dai AG (2011) Characteristics and trends in various forms of the Palmer Drought Severity Index (PDSI) during 1900–2008. J Geophys Res. https://doi.org/10.1029/2010JD015541

    Article  Google Scholar 

  10. Davi NK, Jacoby GC, Curtis AE, Baatarbileg N (2006) Extension of drought records for Central Asia using tree rings: West-Central Mongolia. J Clim 19:288–299

    Article  Google Scholar 

  11. Ding Y, Wang Z, Sun Y (2008) Inter-decadal variation of the summer precipitation in East China and its association with decreasing Asian summer monsoon. Part I: Observed evidences. Int J Climatol 28:1139–1161

    Article  Google Scholar 

  12. Durbin J, Watson GS (1950) Testing for serial correlation in least squares regression I. Biometrika 37:409–428

    Google Scholar 

  13. Esper J, Cook ER, Krusic PJ, Peters K, Schweingruber FH (2003) Tests of the RCS method for preserving low-frequency variability in long tree-ring chronologies. Tree Ring Res 59(2):81–98

    Google Scholar 

  14. Fritts HC (1991) Reconstructing large-scale climatic patterns from tree-ring data: a diagnostic analysis. The University of Arizona Press, Tucson

    Google Scholar 

  15. Fu CB (2008) Global change and regional aridification Chinese. J Atmos Sci 32(4):752–760

    Google Scholar 

  16. Fu CB, Wei HL, Chen M, Su BK, Zhao M, Zhen WZ (1998) Evolution of summer monsoon rain belts over East China in a regional climate model. Chin J Atmos Sci 22:522–534

    Google Scholar 

  17. Fu CB, Jiang Z, Guan Z, He J, Xu Z (2008) Climate extremes and related disasters in China. Region Clim Stud 313–344

  18. Han YB, Han YG (2002) Wavelet analysis of sunspot relative numbers. Chin Sci Bull 47(7):609–612

    Article  Google Scholar 

  19. Hao ZX, Zheng JY, Wu GF, Zhang XZ, Ge QS (2010) 1876–1878 severe drought in North China: facts impacts and climatic background. Chin Sci Bull 55:3001–3007

    Article  Google Scholar 

  20. Kang SY, Yang B, Qin C, Wang JL, Shi F, Liu JJ (2013) Extreme drought events in the years 1877–1878 and 1928 in the southeast Qilian Mountains and the air–sea coupling system. Quat Int 283:85–92

    Article  Google Scholar 

  21. Kaplan A, Cane MA, Kushnir Y, Clement AC, Blumenthal MB, Rajagopalan B (1998) Analyses of global sea surface temperature 1856–1991. J Geophys Res 103:18567–18589

    Article  Google Scholar 

  22. Kerr RA (2000) A North Atlantic climate pacemaker for the centuries. Science 288:1984–1985

    Article  Google Scholar 

  23. Kravtsov SV, Spannagle C (2008) Multi-decadal climate variability in observed and modeled surface temperatures. J Clim 21:1104–1121

    Article  Google Scholar 

  24. Lau WKM, Kim KM (2017) Competing influences of greenhouse warming and aerosols on Asian summer monsoon circulation and rainfall. Asia Pac J Atmos Sci 53(2):181–194

    Article  Google Scholar 

  25. Li SL, Bates GT (2007) Influence of the Atlantic multidecadal oscillation on the winter climate of East China. Adv Atmos Sci 24:126–135

    Article  Google Scholar 

  26. Li H, Dai A, Zhou T et al (2010) Responses of East Asian summer monsoon to historical SST and atmospheric forcing during 1950–2000. Clim Dyn 34(4):501–514

    Article  Google Scholar 

  27. Li Q, Wei F, Li D (2011) Interdecadal variation of East Asian summer monsoon and drought/flood distribution over eastern China in the last 159 years. J Geog Sci 21(4):579–593

    Article  Google Scholar 

  28. Li HY, Si YB, Hua WQ, He B (2015) Analysis on the summer precipitation anomaly in Inner Mongolia in 2013. Meteorol J Inner Mongolia 2:3–6

    Google Scholar 

  29. Liang EY, Shao XM, Qin NS (2008) Tree-ring based summer temperature reconstruction for the source region of the Yangtze River on the Tibetan Plateau. Glob Planet Change 61:313–320

    Article  Google Scholar 

  30. Liu Y, Cai QF, Ma LM, An ZS (2001) Tree ring precipitation records from Baotou and the east Asia summer monsoon variations for the last 254 years. Earth Sci Front 8(1):91–97

    Google Scholar 

  31. Liu Y, Bao G, Song HM, Cai QF, Sun JY (2009) Precipitation reconstruction from Hailar pine (Pinus sylvestris var mongolica) tree rings in the Hailar region Inner Mongolia China back to 1865 AD. Palaeogeogr Palaeoclimatol Palaeoecol 282(1–4):81–87

    Article  Google Scholar 

  32. Liu Y, Wang CY, Hao WJ, Song HM, Cai QF, Tian H, Sun B, Linderholm HW (2011) Tree-ring-based annual precipitation reconstruction in Kalaqin Inner Mongolia for the last 238 years. Chin Sci Bull 56:2995–3002

    Article  Google Scholar 

  33. Liu N, Liu Y, Bao G, Bao M, Wang YC, Zhang LZ, Ge YX, Bao WRG, Tian H (2016) Drought reconstruction in eastern Hulun Buir steppe China and its linkages to the sea surface temperatures in the Pacific Ocean. J Asian Earth Sci 115:298–307

    Article  Google Scholar 

  34. Liu Y, Liu H, Song HM, Li Q, Burr GS, Wang L, Hu SL (2017a) A 174-year Asian summer monsoon related relative humidity record from tree-ring δ18O in the Yaoshan region eastern central China. Sci Total Environ 593–594:523–534

    Article  Google Scholar 

  35. Liu Y, Cobb KM, Song HM, Li Q, Li CY, Nakatuska T, An ZS, Zhou WJ, Cai QF, Li JB, Leavitt SW, Sun CF, Mei RC, Shen CC, Chan MH, Sun JY, Yan LB, Lei Y, Ma YY, Li XX, Chen DL, Linderholm HW (2017b) Enhanced central Pacific El Niño variability compared to the last eight centuries. Nat Commun. https://doi.org/10.1038/ncomms15386

    Article  Google Scholar 

  36. Liu Y, Ta WY, Li Q, Song HM, Sun CF, Cai QF, Liu H, Wang L, Hu SL, Sun JY, Zhang WB, Li WZ (2018) Tree-ring stable carbon isotope-based April—June relative humidity reconstruction since AD 1648 in Mt Tianmu China. Clim Dyn 50(5–6):1733–1745

    Google Scholar 

  37. Ma ZG, Fu CB (2005) Decadal variations of arid and semi-arid boundary in China. Chin J Geophys 48(3):519–525

    Google Scholar 

  38. Ma QX, Gao J, Bao FX (2013) Cause analysis on large-scale and more rainfalls during the flood season in the Inner Mongolia area in 2012. Meteorol J Inner Mongolia 6:3–6

    Google Scholar 

  39. Mazzarella A (2007) The 60-year solar modulation of global air temperature: the Earth’s rotation and atmospheric circulation connection. Theor Appl Climatol 88:193–199

    Article  Google Scholar 

  40. Meehl GA, Arblaster JM (2002) The Tropospheric biennial oscillation and Indian Monsoon rainfall. Geophys Res Lett 28:1731–1734

    Article  Google Scholar 

  41. Mitchell TD, Jones PD (2005) An improved method of constructing a database of monthly climate observations and associated high resolution grids. Int J Climatol 25:693–712

    Article  Google Scholar 

  42. Mohtadi M, Prange M, Steinke S (2016) Palaeoclimatic insights into forcing and response of monsoon rainfall. Nature 533(7602):191–199

    Article  Google Scholar 

  43. Palmer WC (1965) Meteorological drought. US Department of Commerce, Washington DC

    Google Scholar 

  44. Peng JJ, Sun Y, Chen M, He XY, Davi NK, Zhang XL, Li T, Zhu CY, Cai C, Chen ZJ (2013) Tree-ring based precipitation variability since ad 1828 in northwestern Liaoning China. Quat Int 283:63–71

    Article  Google Scholar 

  45. Phipps RL (1985) Collecting, preparing, cross-dating, and measuring tree increment cores. US Geological Survey, Water Resource Investigation Report 85–4148

  46. Qian C, Zhou TJ (2014) Multidecadal variability of north China aridity and its relationship to PDO during 1900–2010. J Clim 27:1210–1222

    Article  Google Scholar 

  47. Qian C, Yu JY, Chen G (2014) Decadal summer drought frequency in China: the increasing influence of the Atlantic Multi-decadal Oscillation. Environ Res Lett 9(9):124004

    Article  Google Scholar 

  48. Raspopov OM, Shumilov OI, Kasatkina EA, Turenun E, Lindholm M (2000) 35-year climatic Bruckner cycle—solar control of climate variability?. In: The solar cycle and terrestrial climate, Solar and space weather Euro conference. ESA Publications Division, Santa Cruz de Tenerife, pp 517–520

    Google Scholar 

  49. Sang YF, Wang ZG, Liu CM (2012) Period identification in hydrologic time series using empirical mode decomposition and maximum entropy spectral analysis. J Hydrol 424–425:154–164

    Article  Google Scholar 

  50. Shao XM, Fan JM (1999) Past climate on west Sichuan Plateau as reconstructed from ring-widths of dragon spruce. Quat Sci 1:81–89

    Google Scholar 

  51. Shen CM, Wang WC, Gong W, Hao ZX (2006) A Pacific Decadal Oscillation record since 1470 AD reconstructed from proxy data of summer rainfall over eastern China. Geophys Res Lett 33:L03702

    Google Scholar 

  52. Shi F, Yang B, von Gunten L, Qin C, Wang ZY (2011) Ensemble empirical mode decomposition for tree-ring climate reconstructions. Theor Appl Climatol 109:233–243

    Article  Google Scholar 

  53. Shi ZJ, Xu LH, Dong LS, Gao JX, Yang XH, Lu SH, Feng CY, Shang JX, Song AY, Guo H, Zhang X (2015) Growth–climate response and drought reconstruction from tree-ring of Mongolian pine in Hulunbuir Northeast China. J Plant Ecol 9:51–60

    Google Scholar 

  54. Song HM, Liu Y (2011) PDSI variations at Kongtong Mountain China inferred from a 283-year Pinus tabulaeformis ring-width chronology. J Geophys Res 116:D22111. https://doi.org/10.1029/2011JD016220

    Article  Google Scholar 

  55. Stokes MA, Smiley TL (1996) An introduction to tree-ring dating. University of Arizona Press, Tucson

  56. Su MF, Wang HJ (2007) Relationship and its instability of ENSO—Chinese variations in droughts and wet spells. Sci China Ser D 50(1):145–152

    Article  Google Scholar 

  57. Sun JY, Liu Y, Sun B, Wang RY (2012) Tree-ring based PDSI reconstruction since 1853 AD in the source of the Fenhe River Basin Shanxi Province China. Sci China Ser D Earth Sci 55(11):1847–1854

    Article  Google Scholar 

  58. Sun B, Liu Y, Lei Y (2016) Growing season relative humidity variations and possible impacts on Hulunbuir grassland. Sci Bull 61(9):1–9

    Article  Google Scholar 

  59. Sutton RT, Hodson DLR (2005) Atlantic Ocean forcing of North American and European summer climate. Science 309:115–118

    Article  Google Scholar 

  60. University of East Anglia Climatic Research Unit (CRU) (2008) CRU Time Series (TS) high resolution gridded datasets. Available at: http://badc.nerc.ac.uk/view/badc.nerc.ac.uk__ATOM__dataent_1256223773328276

  61. Wang C (1997) Important climatic changes in Inner Mongolia. China Meteorological Press, Beijing

    Google Scholar 

  62. Wang YM, Li SL, Luo DH (2009) Seasonal response of Asian monsoonal climate to the Atlantic multidecadal oscillation. J Geophys Res 114:D02112

    Google Scholar 

  63. Wigley TML, Briffa KR, Jones PD (1984) On the average value of correlation time series with applications in dendroclimatology and hydrometeorology. J Clim Appl Meteorol 23:201–213

    Article  Google Scholar 

  64. Wu Z, Huang N, Chen X (2009) The multi-dimensional ensemble empirical mode decomposition method. Adv Adapt Data Anal 1(3):339–372

    Article  Google Scholar 

  65. Yan HM, Zhong M, Zhu YZ (2004) Determination of the degree of freedom of digital filtered time series with an application to the correlation analysis between the length of day and the Southern Oscillation index. Chin Astron Astrophys 28(1):120–126

    Article  Google Scholar 

  66. Yi L, Yu H, Ge J, Lai Z, Xu X, Qin L, Peng S (2012) Reconstructions of annual summer precipitation and temperature in north-central China since 1470 AD based on drought/flood index and tree-ring records. Clim Change. https://doi.org/10.1007/s10584-011-0052-6

    Article  Google Scholar 

  67. Zhang JC (1976) Climate change and its causes. Science Press, Beijing

    Google Scholar 

  68. Zhang P (1996) Cold/warm variation in China since 500 BP. In: Zhang P (ed) Climate of China and sea-level variation and their trends and influences. Shandong Science and Technology Press, Ji’nan, pp 345–351

    Google Scholar 

  69. Zhang R, Delworth TL (2007) Impact of the Atlantic multidecadal oscillation on North Pacific climate variability. Geophys Res Lett 34:L23708

    Google Scholar 

  70. Zhang DE, Liang YY (2010) A long lasting and extensive drought event over China during1876–1878. Adv Clim Change Res 6:106–112

    Google Scholar 

  71. Zhao HL, Zhao XY, Zhang TH, Zhou RL (2002) Boundary line on agro-pasture zigzag zone in north China and its problems on eco-environment. Adv Earth Sci 17:739–747

    Google Scholar 

  72. Zhu HF, Fang XQ, Shao XM, Yin ZY (2009) Tree ring-based February–April temperature reconstruction for Changbai Mountains in northeast China and its implication for East Asian winter monsoon. Clim Past Discuss 5:661–666

    Article  Google Scholar 

Download references

Acknowledgements

The authors thank Drs. Guang Bao, Bo Sun, Na Liu, Ying Lei, Yanchao Wang and Mr. Baoyin Shen for their great help. This study was jointly supported by grants from the Key Research Program of Frontier Sciences of Chinese Academy of Sciences (QYZDJ-SSW-DQC021, NSFC 41630531, XDPB05, GJHZ1777) and the Key Project of Institute of Earth Environment, CAS (01) and the State Key Laboratory of Loess and Quaternary Geology (020).

Author information

Affiliations

Authors

Corresponding author

Correspondence to Yu Liu.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Liu, R., Liu, Y., Li, Q. et al. Seasonal Palmer drought severity index reconstruction using tree-ring widths from multiple sites over the central-western Da Hinggan Mountains, China since 1825 AD. Clim Dyn 53, 3661–3674 (2019). https://doi.org/10.1007/s00382-019-04733-0

Download citation

Keywords

  • Central-western Da Hinggan Mountains (CW-DHM)
  • Northeastern China
  • Tree-ring width
  • PDSI reconstruction
  • East Asian summer monsoon (EASM)