Linear trends in mean and extreme temperature in Xiongan New Area, China

QIAN Cheng and CAO Li-Juan

aCAS Key Laboratory of Regional Climate-Environment for Temperate East Asia, Institute of Atmospheric Physics, Chinese Academy of Sciences,Beijing, China; bUniversity of Chinese Academy of Science, Beijing, China; cNational Meteorological Information Center, China Meteorological Administration, Beijing, China

ABSTRACT On 1 April 2017 China established Xiongan New Area in Hebei Province, which was described as‘a(chǎn) strategy crucial for a millennium to come’. A point of interest for the public is to be aware of the historical climate change in this new area; however, results from previous global-scale or largerregional-scale averages provide relatively limited information because of the distinct regional differences in climate change. This study analyzes the changes in mean and extreme temperature in this area, based on homogenized daily temperature data for the period 1960–2016. The results show a significant warming in the indices of annual, summer, and winter mean temperature (Tmean),maximum temperature (Tmax), and minimum temperature (Tmin). The linear rate of annual Tmean is 0.34 °C/decade. Temperatures on the hottest day, the warmest night, the coldest day, and the coldest night, every year, all show increasing trends, with the trends in the two nighttime-related indices being significant. An increasing occurrence of warm days, warm nights, hot days, and tropical nights,but a decreasing occurrence of cold days, cold nights, icing days, and frost days, are found, all of which are significant, except for the occurrences of hot days and icing days. A significant extension of the length of the thermal growing season is also found. The magnitudes of change in most of the temperature indices in Xiongan New Area are larger than those of the adjacent Jing-Jin-Ji and North China regional mean. These results could provide valuable information for policymakers, city planners, engineers, and migrants to this new area.

KEYWORDS Xiongan New Area; climate change; linear trend; mean temperature; extreme temperature

1. Introduction

On 1 April 2017 China established Xiongan New Area in Hebei Province, as part of measures to advance the coordinated development of the Beijing–Tianjin–Hebei region(otherwise commonly referred to as the Jing-Jin-Ji region).This ‘new area of national significance’ followed previous similar projects such as Shenzhen Special Economic Zone and Shanghai Pudong New Area, and was described as ‘a(chǎn) strategy crucial for a millennium to come’. The move will help phase out functions from Beijing that are not related to the capital, and explore a new model of optimized development in densely populated areas (http://www.chinadaily.com.cn/china/2017-04/01/content_28774796.htm). The new area, about 100 km southwest of downtown Beijing, includes the three counties of Xiongxian,Anxin, and Rongcheng (Figure 1(a)). One important point of interest for the public in establishing this new area and making it a success is an awareness of the historical climate changes that have taken place there.

Figure 1. (a) Map of Xiongan New Area, including Xiongxian station (Code: 54636, blue marker), Anxin station (Code: 54605,red marker) and Rongcheng station (Code: 54503, purple marker).The inset map in the upper-right corner shows the location of Xiongan New Area in China. (b) Map of the North China region and its national stations (dots), with the red dots representing the national stations within the Jing-Jin-Ji region.

Change in mean temperature is a globally important topic in climate science (IPCC 2013). Likewise, change in extreme temperature has also gained wide interest both abroad (e.g. Alexander et al. 2006; Donat et al. 2013) and in China (e.g. Yan et al. 2002; Zhai and Pan 2003; Qian and Lin 2004; Qian et al. 2011b; Zhou and Ren 2011; Xu et al. 2013; Sun et al. 2014; Yin et al. 2015; Zhou et al.2016). However, as climate change has distinct regional characteristics–especially at urban sites, where local urbanization effects play a noticeable role (e.g. Zhou and Ren 2011; Qian 2016b)–results from global or larger-region averages provide relatively limited information for the public and other stakeholders with an interest in this new area (especially the project’s planners), where the effect of urbanization is still relatively small at this stage.In addition, with the interference in recent years of the so-called ‘global warming hiatus’ as some researchers claimed, it is also necessary to investigate whether there is ‘warming hiatus’ phenomenon in this new area, i.e. in order to provide the project’s planners with the correct scientific information.

Accordingly, this study focuses on the linear trends in mean and extreme temperature in this new area since modern meteorological observations began, based on a newly homogenized high-quality temperature dataset compiled from 2419 meteorological stations (Cao et al. 2016), including those in the three counties of Xiongan New Area which lets this study become possible. Continuous daily temperature series for Xiongan New Area starting from 1960 are also produced. The results are compared with those in the adjacent Jing-Jin-Ji region and North China region. The study could provide valuable scientific information for policymakers, planners, engineers,and migrants to the area.

2. Data and methods

2.1. Surface air temperature data

A new homogenized temperature data-set called the China Homogenized Historical Temperature Data-set version 1.0 (Cao et al. 2016), including daily temperature series from 2419 meteorological stations distributed throughout mainland China for the period 1951–2016, is used in this study. The original daily mean, maximum, and minimum temperature data had already been quality-controlled and the outliers removed (Ren et al. 2012). The metadata,which are archived in text files by the China Meteorological Administration (CMA) for each station, include detailed information on changes in observational instrumentation,times, locations, and the evolution of the surrounding environment. The homogenized data-set applied the RHtests V3.0 software (Wang and Feng 2010) to detect inhomogeneities in individual station series with detailed metadata information and to adjust the discontinuities caused by non-climate changes such as changes to observing sites,instrumentation, and observation environments (Cao et al. 2016). The daily mean temperature (Tmean), maximum temperature (Tmax), and minimum temperature (Tmin) in the three counties of Xiongan New Area are available in this data-set, but for different periods. The station with the longest time series, beginning in 1960, is Anxin, while the other two stations (Rongcheng and Xiongxian) have observations available from 1968 and 1974, respectively(Table 1).

2.2. Method

2.2.1. Interpolation

Since the data at Anxin and Rongcheng have missing values, we interpolate them as follows: (1) For Tmeanat Anxin,the missing value on 1 January 1960 is interpolated using the mean value of its Tmaxand Tmin, which are available.The missing values during 11–23 August 1963 are interpolated using those available from Xushui station (38.59°N,115.39°E, 13.1 m), which is very close to Anxin station and has similar elevation. The correlation coefficient between these two stations reaches 0.93. Therefore, it is reasonable to use Xushui station for interpolation. The missing valuesfrom 1 September 1968 to 8 May 1969 are interpolated using those available from Rongcheng station. (2) For Tmeanat Rongcheng, the missing value on 28 September 2012 is interpolated using the corresponding mean value of its Tmaxand Tmin, which are available. (3) For Tmaxand Tminat Anxin, the missing values from 11 August to 27 September 1963 are interpolated using those available from the adjacent Xushui station. The missing values from 1 September 1968 to 8 May 1969 are interpolated using those available from Rongcheng station. After interpolation, the Tmean,Tmax, and Tmindata at Anxin and Rongcheng are continuous for the period 1960–2016 and 1968–2016, respectively.Along with the continuous daily temperature records at Xiongxian, a continuous daily time series of the average for Xiongan New Area for the period 1960–2016 can be produced by calculating the arithmetic mean values available from the three stations in the new area. Since the three stations are very close and have similar elevations (Table 1), this calculation is reasonable.

Table 1. Details of the three meteorological stations in Xiongan New Area.

2.2.2. Calculation of temperature indices

The mean temperature indices analyzed include annual Tmean, Tmax, and Tmin(Table 2). The corresponding indices of the two extreme seasons, i.e. summer (June–July–August,JJA) and winter (December–January–February, DJF), are also analyzed (Table 2). The 16 extreme temperature indices analyzed are listed in Table 2. These are adopted from the 27 core indices of the Expert Team on Climate Change Detection and Indices (ETCCDI) (Zhang et al. 2011; also see http://etccdi.pacificclimate.org/list_27_indices.shtml), except for the index of hot days (HD35), which is revised from the index of summer days according to the local threshold of 35 °C frequently used in China (e.g. Zhai and Pan 2003; Sun et al.2014; Qian 2016b) to fit regional characteristics.

Version 1.1 of the RClimDex software package (Zhang and Yang 2004) is used to ensure consistency in the calculation of the indices with other regions. The percentiles required for some of the extreme temperature indices are calculated from the base period of 1961–90 using a bootstrap method to avoid possible inhomogeneities (Zhang et al. 2005). The base period of 1961–90, as recommended by ETCCDI, is used to facilitate comparison of the results with those at other global sites, because using different base periods would result in different mean annual cycles and thus different anomalies, making the results difficult to compare with others (Qian et al. 2011a).

North China in this study is defined as the region covering (33°–43°N, 108°–120°E), as in Ren et al. (2008) and Zhou and Ren (2011). Since national-level Reference Climatic and Basic Meteorological Stations (hereafter, ‘national stations’) have been frequently used by researchers in China, especially those outside the CMA, for analyzing the long-term changes in mean and extreme climate, the Jing-Jin-Ji and North China region results, to which Xiongan’s results are compared, are based on data from this type of station. There are 124 such stations analyzed in the North China region (Figure 1(b)), not including Baoding station, because of unresolved recent relocation problems. Records with standard deviations five times greater than the average are considered as outliers and marked as missing on those days. The average time series for the Jing-Jin-Ji region and North China region are calculated as grid area–weighted averages on the basis of longitude by latitude grids of 2° × 2° (Figure 1(b)), after calculating the mean and extreme temperature indices at each station.

2.2.3. Trend calculation

The linear trends in the mean and extreme temperature indices at Anxin station, as well as those of the three-station-averaged temperature in Xiongan New Area, for the period 1960–2016, are analyzed and compared to those of the regional average in the Jing-Jin-Ji region and North China region. The linear trends in the mean temperature indices are estimated using the ordinary least squares(OLS) method, and the 95% or 99% confidence interval is estimated by allowing positive first-order auto-regressive dependence in the data (Hartmann et al. 2013; Qian 2016b). Since extreme temperature indices may not necessarily follow a Gaussian distribution (e.g. Qian 2016b),their trend slopes and significance testing are estimated using the nonparametric Sen’s slope estimator and Mann–Kendall test, taking into account the first-order autocorrelation calculated through an iterative process proposed by Zhang et al. (2000) and refined by Wang and Swail (2001).This method is referred to as WS2001 in this paper.

3. Results

3.1. Mean temperature

The mean values of annual, summer, and winter Tmeanfor Xiongan New Area (at Anxin station) during the base period of 1961–1990 are 11.2 °C (11.2 °C), 24.8 °C (24.7 °C), and?4.5 °C (?4.3 °C), respectively. As can be seen from Table 2,the annual (Figure 2(a)), summer (figure not shown), and winter (figure not shown) Tmeanfor Xiongan New Area all show warming trends during the analyzed period, at rates of 0.34 °C/decade, 0.21 °C/decade, and 0.50 °C/decade,respectively. These trends are all statistically significant at the 0.01 level. In other words, the annual Tmeanincreases by about 1.9–2 °C from 1960 to 2016. The results at Anxin station are similar to those for Xiongan New Area, except for a slightly smaller trend slope magnitudes for the annual and winter situations. It should be mentioned that, even after 1998, the linear trend in annual Tmeanstill increases at a rate of 0.28 °C/decade, i.e. without a warming hiatus(Figure 2(a)).

在如今的化妝品市場上，一百多元的單品精華定價(jià)并不算高，還稱不上是高端產(chǎn)品。但是你要知道，大寶最為消費(fèi)者所熟知的明星單品“大寶SOD蜜”100 mL容量在天貓超市的售價(jià)僅為9.9元。與大寶一貫的低價(jià)相比，這款定價(jià)110元的精華，已經(jīng)算是高端。

For Xiongan New Area, the annual (Figure 2(b) and (c)),summer (figure not shown), and winter (figure not shown)Tmaxand Tminall show warming trends and are statistically significant at the 0.01 level, except for the summer Tmax,which is at the 0.05 level. Besides, all the Tminindices warm at a faster (more than double) rate than their Tmaxindex counterparts, resulting in a reduction in the diurnal temperature range (DTR) at a rate between 0.2 and 0.5 °C/decade. These trends in the annual (Figure 2(d)), summer(figure not shown), and winter (figure not shown) DTR are all statistically significant at the 0.01 level. It should be noted that the minimum value in 1964 for the annual DTR (Figure 2(d)) is an outlier, which would make the OLS-estimated linear trend an underestimated one because this method is sensitive to outliers. This outlier is a regional feature, since it also exists in the DTR index of the North China regional mean calculated without the stations in Xiongan New Area, mainly due to an extremely cold Tmaxin that year (figures not shown). This outlier may also be partly due to the relocation of Anxin station to the countryside on 1 October 1964, which reduced its elevation by 7 m. It is adjusted in the Tmeanand Tmin, but not in the Tmax,as it is not significant in the RHtest result. Nevertheless,the WS2001 estimator used in this study overcomes this problem. The results at Anxin station are similar, but with slightly larger trends in Tmaxindices, smaller trends in Tminindices and DTR indices, and lower statistical significance for winter DTR. In addition, the winter Tmean, Tmax, and Tminin this region all show warming trends at a much faster rate than their summertime counterparts, resulting in weakening annual cycles.

3.2. Extreme temperature

As can also be seen from Table 2, for Xiongan New Area,four extreme value indices (TXx, TNx, TXn, and TNn) all show warming trends (Figure 3(a)–(d)), at rates of 0.13,0.32, 0.35, 0.99 °C/decade, respectively, although only TNx and TNn are statistically significant (at the 0.01 level), indicating that temperatures on the hottest day (Figure 3(a)),the warmest night (Figure 3(b)), the coldest day (Figure 3(c)), and the coldest night (Figure 3(d)), every year, all show increasing trends. The mean values of TXx, TNx, TXn,and TNn during the base period of 1961–90 are 37.6, 25.0,?5.6, and ?20.1 °C, respectively.

Figure 2. Annual time series (solid lines) of (a) Tmean, (b) Tmax, (c) Tmin, and (d) DTR, and their corresponding linear trends (dashed lines), for the period 1960–2016, for Xiongan New Area.

Figure 3. Time series (solid lines) of four extreme value indices: (a) TXx, (b) TNx, (c) TXn, and (d) TNn; and of four relative threshold indices: (e) TX90p, (f) TN90p, (g) TX10p, and (h) TN10p; and their corresponding linear trends (dashed lines), for the period 1960–2016,for Xiongan New Area.

The linear trends in the four relative threshold indices(TX90p, TN90p, TX10p, and TN10p) are all statistically significant at the 0.01 level, with increasing annual occurrence of warm days (TX90p, Figure 3(e)) and warm nights (TN90p,Figure 3(f)), at rates of 1.38% and 5.39%/decade, respectively, but decreasing annual occurrence of cold days(TX10p, Figure 3(g)) and cold nights (TN10p, Figure 3(h)),at rates of ?1.08% and ?2.14%/decade, respectively. The mean values of TX90p, TN90p, TX10p, and TN10p during the base period of 1961–90 are 10.3%, 10.4%, 10.5%, and 10.4%, respectively.

Figure 4. Time series (solid lines) of four absolute threshold indices: (a) HD35, (b) TR, (c) ID, and (d) FD; and of three other indices: (e) GSL,(f) WSDI, and (g) CSDI; and their corresponding linear trends (dashed lines), for the period 1960–2016, for Xiongan New Area.

The linear trend results for the four absolute threshold indices (HD35, TR, ID, and FD) are similar to those of the relative threshold indices, with increasing annual occurrence of hot days (HD35, Figure 4(a)) and tropical nights(TR, Figure 4(b)), at rates of 0.40 and 5.39 days/ decade,respectively, but decreasing annual occurrence of icing days (ID, Figure 4(c)) and frost days (FD, Figure 4(d)), at rates of ?1.67 and ?4.40 days/decade, respectively. The increasing trends in TR and the decreasing trends in FD are both statistically significant at the 0.01 level. The mean values of HD35, TR, ID, and FD during the base period of 1961–90 are 8.5, 38.9, 22.9, and 136.5 days, respectively.

The growing season length (GSL, Figure 4(e)) extends at a rate of 2.88 days/decade, and is statistically significant at the 0.01 level. The mean value of GSL during the base period of 1961–90 is 240.6 days. The warm spell duration index (WSDI, Figure 4(f)) and cold spell duration index(CSDI, Figure 4(g)) show no obvious linear trend. The maximum value of WSDI is 28 days in 1987 and the minimum in many years, especially in the 1960s, is zero (Figure 4(f)).The maximum value of CSDI is 13 days in 1981 and the minimum in many years, especially in the 2000s, is zero(Figure 4(g)).

For Anxin station, the corresponding results are similar,albeit with slightly different trend slopes. The linear trends for its TNx, TNn, TX90p, TN90p, TX10p, TN10p, TR, FD, and GSL are all statistically significant, at least at the 0.05 level(Table 2).

3.3. Comparison with the Jing-Jin-Ji and North China regions

Compared with the Jing-Jin-Ji and North China regions(Table 2), the basic characteristics of the linear trends in the temperature indices in Xiongan New Area are similar, with the same sign of change in almost all the indices except HD35, whose trend is trivial. However, the magnitudes of the linear trends in all the mean temperature indices and most (14 out of 18) of the extreme temperature indices of Xiongan New Area are larger than those of the adjacent regional mean, especially for the high temperature indices.The exceptions are ID, WSDI, and CSDI, for which North China has larger and statistically significant linear trends.

4. Discussion

The basic characteristics of the changes in the mean and extreme temperature indices reported in this study are similar to those in other recent studies, such as Ren et al. (2008) and Zhou and Ren (2011) for North China,and Zhou and Ren (2011), Xu et al. (2013), and Zhou et al.(2016) for China as a whole. However, the magnitudes of these changes are different because of the different analysis periods, methods, and regions. Here, the comparisons between Xiongan New Area and adjacent regions are performed using the same data source, analysis period,and method, and the results show an overall faster rate of change in Xiongan New Area compared to the adjacent regional mean. This is useful information for the planning of this new area.

5. Summary

Linear trends in mean and extreme temperature indices for the newly established Xiongan New Area are analyzed,based on homogenized daily temperature data for the period 1960–2016. The main conclusions can be summarized as follows:

(1) All the mean temperature indices show warming trends, statistically significant at least at the 0.05 level. A decreasing trend in the DTR and a weakening trend in the annual cycle are found.

(2) Temperatures on the hottest day, warmest night,coldest day, and coldest night, every year, all show increasing trends, with the two nighttime-related indices being statistically significant.

(3) An increasing occurrence of warm days, warm nights, hot days, and tropical nights, but a decreasing occurrence of cold days, cold nights,icing days, and frost days, is found, all of which are statistically significant, except for the occurrences of hot days and icing days.

(4) A statistically significant extension of the length of the growing season is found.

In short, local climatic warming in terms of mean temperature is continuing in Xiongan New Area without a ‘hiatus’, especially after 2013; and so the effect of this, along with related impacts, should be taken into account during its construction. In addition, the magnitudes of the linear trends in most of the temperature indices for Xiongan New Area are larger than those of the adjacent Jing-Jin-Ji and North China regional mean, especially for the high temperature indices, indicating a potentially greater danger under future global warming and development-induced local urbanization effects. Thus, green development planning seems important for this new area.

Funding

This study was sponsored by the National Key R&D Program of China (grant number 2016YFA0600404), Key Technology of Integration of Meteorological and Application Projects (grant number CMAGJ2015Z16), the Youth Innovation Promotion Association of CAS (grant number 2016075), and the Jiangsu Collaborative Innovation Center for Climate Change.

ORCID