Temperature Change Characteristics in Gansu Province of China

Abstract

The applicability of reanalysis data has been widely addressed in climate and hydrology studies over the past two decades. In this study, we analyzed spatiotemporal variations in ERA-Interim temperature data from four climate zones within Gansu Province from 1979 to 2017 by using linear regression model and Mann-Kendall mutation test. Results showed that: (1) The highest temperature was found in the subtropical monsoon climate zone, and the lowest in the plateau mountain climate zone. Temperatures in high-elevation regions were lower than those in low-elevation regions; (2) The annual mean temperature increased across Gansu Province from 1979 to 2017. The lowest warming rates of annual mean, annual maximum, and annual minimum temperatures were found in the subtropical monsoon climate zone, and these were 0.334, 0.300, and 0.336 °C/10a, respectively. The highest warming rates of annual mean and annual minimum temperature were found in the temperate monsoon climate zone, and these were 0.420 and 0.464 °C/10a, respectively. The highest warming rate of annual maximum temperature was found in the temperate continental climate zone (0.471 °C/10a); (3) The Mann-Kendall analysis showed that the mutation times of annual mean temperature of the subtropical monsoon, temperate monsoon, and temperate continental climate zones in Gansu Province were all in 1997. The mutation times of annual maximum temperature were found in the subtropical monsoon climate zone (1997) and temperate monsoon climate zone (1997). The mutation times of annual minimum temperature were found in the temperate continental climate zone (1997) and plateau mountain climate zone (1994). ERA-Interim reanalysis data are reliable for capturing mutation time of temperature, especially in the high-elevation areas with rare meteorological station. This study can provide a reference when analyzing climate change at different climatic zones using reanalysis data.

1. Introduction

Climate change is one of the major issues facing the international community [1,2,3,4,5]. Climate warming has led to considerable changes in the global climate and environment, resulting in the more frequent occurrence of severe weather events, such as floods and droughts, which damage the ecological environment and have a marked impact on production and sustainable economic development [6,7]. Therefore, strengthening research on regional climate change plays an important guiding role in production optimization and natural disaster prevention.

Reanalysis data have been used and analyzed widely in recent years because of their advantages of higher resolution and longer time series [8,9,10,11,12,13]. ERA-Interim, the third-generation reanalysis data provided by the European Centre for Medium-Range Weather Forecasts (ECMWF), have been widely used in previous studies [12,14]. Many studies have verified that ERA-Interim temperature can successfully capture the observational data and can be applied to climate research. For example, Zhao et al. [15] evaluated the errors of ERA-Interim temperature in the Qilian Mountains of China and showed that it could successfully capture temperature trends in that region. Gao et al. [13] used ERA-Interim temperature data in the Tian Shan Mountains of China and constructed a high-resolution dataset, which they say will benefit climate change research in that area. Qin et al. [16] used the ERA-Interim reanalysis dataset to assess changes in the ground surface freezing and thawing condition on the Qinghai-Tibet Plateau (QTP); their results showed that the calibrated reanalysis ground surface temperature (GST) provided a higher resolution dataset for monitoring the thermal state of permafrost on the QTP, and the calibrated reanalysis GST data could also be used to reflect changes in the active layer thickness.

Gansu Province is located at the boundary of the temperate continental, temperate monsoon, plateau mountain, and subtropical monsoon climate zones within China. The climate and terrain conditions are complex, which results in a limited number of meteorological stations within the province. Drought and flood disasters often occur in Gansu Province, causing serious impacts on the ecological environment, so it is vital to research climate change in this region. Wen et al. [17] used the daily observation dataset at 29 meteorological stations over Gansu during 1951–2015 to analyze the spatiotemporal variability of temperature. The results revealed that the change trends in annual and seasonal precipitation over Gansu were not significant. Liang et al. [18] used observations and found that the change range of extreme temperatures in central southern Gansu were larger than the northwest, especially in the Hexi Corridor. Ma et al. [19] used the daily temperature data from 80 observational stations during 1951–2010 to analyze the seasonal, annual, and inter-decadal changes of temperature in Gansu Province. The results showed that the temperature during 1951−2010 in Gansu Province increased significantly, with an increasing rate of 0.175 °C/10a.

Despite these previous studies describing the climate in the Gansu province, there have been relatively few previous studies related to temperature change within each climate zone in Gansu Province, probably because the Gansu Province consists mostly of plateau and mountainous regions, meaning that meteorological stations in each climate zone are rare; hence, meteorological stations alone are insufficient to reflect climate change in this region. In this study, we used a linear regression model and Mann-Kendall trend analysis to analyze the variability characteristics in ERA-Interim temperature data within the four different climate zones and different time scales in Gansu Province during the period 1979–2017.

2. Materials and Methods

2.1. Study Area

Gansu Province, located in Northwest China, covers an area of 453,700 km². Due to its continental setting, the climate in most areas is relatively dry, with large temperature differences between the morning and evening. The annual precipitation is 36.6–734.9 mm, which is mainly concentrated from June to August. Spatially, precipitation decreases from southeast to northwest in Gansu Province. The climate zones of China include tropical monsoon, subtropical monsoon, temperate monsoon, plateau mountainous, temperate continental, and tropical rain forest. In Gansu Province, four main climate zones exist: temperate continental, temperate monsoon, plateau mountainous, and subtropical monsoon [20].

2.2. ERA-Interim Reanalysis Temperature

ERA-Interim temperature data were downloaded from the ECMWF’s official website (link: https://apps.ecmwf.int/datasets/data/interim-full-daily/levtype=sfc/, accessed on 1 May 2022 ). In this study, we used the 6-hourly forecast temperature (2m) data initialized at 00:00 UTC, with the forecast times of 06:00, 12:00, 18:00, and 24:00 UTC. The spatial resolution was 0.25° × 0.25°, the geographical range was 32.5°–43° N, 92.5°–109° E, and the period was January 1979 to December 2017. ERA-Interim data were processed into daily average temperature (T_mean), daily maximum temperature (T_max), and daily minimum temperature (T_min). The maximum and minimum values in 06:00, 12:00, 18:00, and 24:00 UTC in a day were determined as daily maximum temperature and daily minimum temperature, respectively [13,21]. Figure 1 shows the distribution of ERA-Interim grids and four climate zones as well as elevation information of Gansu Province.

2.3. Methods

2.3.1. Inverse Distance Weight Method

The inverse distance weighted (IDW) method was used to analyze the spatial variations of temperature based on the ERA-Interim reanalysis data in Gansu Province [22]. The IDW method is a useful and common spatial interpolation method that is included in almost any ArcGIS software [22]. In this IDW technique, two formulas are formed as follows:

$$R_p={{\displaystyle \sum }}_{i=1}^n{w}_i{R}_i$$

(1)

where R_p is the interpolated temperature at a resolution of 0.04 km, R_i represents the known temperature from ERA-Interim grid point, and n is the total number of ERA-Interim grid point. The weighting of each ERA-Interim grid point, w_i, is computed as follows:

$$w_i=d_i{}^{-a}/\left({{\displaystyle \sum }}_{i=1}^n{d}_i{}^{-a}\right)$$

(2)

where d_i represents the distance from each ERA-Interim grid point to the unknown area, and a represents the control parameter and power.

2.3.2. Linear Regression Model

The linear regression model was used to analyze the increasing trend, the equation is defined as:

y = ax + b

(3)

where the a and b are the slope and intercept, respectively. The y and x are the dependent variable and the explanatory variable, respectively. y represents the temperatures (T_mean, T_max, and T_min) on annual timescales, x is the time series from 1979 to 2017.

2.3.3. Mann-Kendall Mutation Test

The Mann-Kendall test method was applied to examine the abrupt change-point time of the annual temperature evolution from 1979 to 2017. We set up an order list for time series x₁, x₂, x₃, …, x_n with n samples. A rank sequence of r_i was constructed where r_i was the sample cumulative number of x_i > x_j (1 ≤ j ≤ i). The definition of S_k is as follows:

$$S_k={{\displaystyle \sum }}_{i=1}^k{r}_i\hspace{1em}\hspace{1em}k=2,3,\mathsf{\Lambda },n$$

(4)

$$r_i=\{\begin{array}{c}1, x_i>x_j\\ 0, else\end{array}\hspace{1em}\hspace{1em}j=1,2,\mathsf{\Lambda },i$$

(5)

The mean value and variance of S_k are defined as follows:

$$E\left(S_k\right)=\frac{k\left(k\mp 1\right)}{4}\hspace{1em}Var\left(S_k\right)=\frac{k\left(k-1\right)\left(2k+5\right)}{72}$$

(6)

Under the assumption of the stochastic independent time series, the statistic is defined as follows:

$${\mathit{UF}}_k=\frac{S_k-E\left(S_k\right)}{\sqrt{Var\left(S_k\right)}}\hspace{1em}\hspace{1em}k=1,2,\mathsf{\Lambda },n$$

(7)

In the formula, UF₁ = 0. UF_k is a standard normal distribution. If the level of significance was α, then |UF_i| > U_α/2, indicating that there was a significant trend change in the time series, x_i. This above-mentioned way was repeatedly applied to the anti-sequence, and the former calculated values multiply by −1 to obtain UB_k.

UB_i = −UF_i  i = n, n − 1, Λ, 1

(8)

In the formula, UB₁ = 0, we graphed UF_k and UB_k, where values of UF_k and UB_k > 0 indicated that the temperature was increasing, and values < 0 indicated that the temperature was decreasing. Values of UF_k and UB_k passing the critical line indicated that the upward or downward trend was significant (±1.96). If the values exceeded the critical line (±1.96), it represented a significant increasing or decreasing trend. If the UF_k and UB_k curves intersect and the intersection was between critical lines (±1.96), the year corresponding to the intersection was accepted as the year when temperature shifted [23,24,25,26].

3. Results

3.1. Daily Temperature at Each Climate Zone

Table 1. Annual ERA-Interim temperature data for different climate zones.

Climate Types	T_mean (°C)	T_max (°C)	T_min (°C)
Subtropical monsoon climate	10.193	13.968	7.160
Temperate monsoon climate	6.747	11.904	2.581
Temperate continental climate	6.686	12.328	1.460
Plateau mountain climate	−0.312	5.577	−5.215

shows the T_mean, T_max, and T_min for the different climate zones within Gansu Province. In general, the T_mean, T_max, and T_min was highest in the subtropical monsoon climate zone. The lowest temperatures were found in the plateau mountain climate zone. Figure 2 shows the spatial distribution of T_mean, T_max, and T_min in Gansu Province based on the IDW method; temperatures in the high-elevation regions were clearly lower than those in the low-elevation regions. The west of Gansu Province was the eastern Qinghai Tibet Plateau (TP), with an average altitude is above 4000 m. The temperature decreases with the increase of altitude, so the temperature in plateau mountain climate zone was lowest. The latitude of subtropical monsoon climate zone in Gansu Province was the lowest compared with other three climatic zones, so the temperature in this region was highest. T_max was higher in the northern part, while T_min was higher in the south-east part, suggesting that there was a higher amplitude between T_min and T_max in the north part of the Gansu Province.

3.2. Temperature Increasing Trend

Table 2. Rates of increase of annual ERA-Interim temperature data for the four different climate zones within Gansu Province during the period 1979–2017 (°C/10a).

Climate Types	T_mean	T_max	T_min
Subtropical monsoon climate	0.334	0.300	0.336
Temperate monsoon climate	0.420	0.405	0.464
Temperate continental climate	0.407	0.471	0.424
Plateau mountain climate	0.387	0.464	0.398

and Figure 3 show the increasing trends of T_mean, T_max, and T_min for the four climate zones within Gansu Province during the period 1979–2017. Temperature showed a clear upward trend for all four climate zones from 1979 to 2017 in general. All the warming trends passed the significance test of r = 0.01. In order to remove the edge effects, we analyze the change of temperature for different periods (

Table 3. Temperature increasing rates at different periods for the four different climate zones within Gansu Province at each period (°C/10a).

Periods	Subtropical Monsoon Climate			Temperate Monsoon Climate			Temperate Continental Climate			Plateau Mountain Climate
	T_mean	T_max	T_min	T_mean	T_max	T_min	T_mean	T_max	T_min	T_mean	T_max	T_min
1979–2008	0.371	0.378	0.306	0.462	0.485	0.481	0.500	0.579	0.502	0.474	0.548	0.491
1980–2009	0.395	0.382	0.335	0.493	0.513	0.513	0.499	0.582	0.503	0.491	0.565	0.511
1981–2010	0.393	0.392	0.324	0.503	0.551	0.510	0.494	0.574	0.510	0.483	0.578	0.491
1982–2011	0.371	0.330	0.323	0.480	0.499	0.506	0.436	0.501	0.467	0.450	0.540	0.464
1983–2012	0.347	0.279	0.332	0.451	0.435	0.505	0.406	0.466	0.440	0.389	0.459	0.420
1984–2013	0.366	0.307	0.352	0.437	0.428	0.483	0.403	0.484	0.429	0.342	0.435	0.354
1985–2014	0.340	0.253	0.364	0.405	0.371	0.466	0.317	0.396	0.342	0.290	0.383	0.296
1986–2015	0.354	0.272	0.381	0.398	0.362	0.462	0.314	0.386	0.342	0.287	0.380	0.296
1987–2016	0.367	0.310	0.383	0.397	0.379	0.445	0.316	0.380	0.342	0.291	0.389	0.290
1988–2017	0.397	0.362	0.401	0.439	0.436	0.476	0.368	0.423	0.403	0.328	0.411	0.341
Average	0.370	0.327	0.350	0.447	0.446	0.485	0.405	0.477	0.428	0.383	0.469	0.395
SD	0.019	0.048	0.030	0.037	0.061	0.022	0.072	0.077	0.065	0.081	0.077	0.085

). The average standard deviations (SD) at each climatic zone were all lower than 0.1, suggesting that these values were closer to the average. The average temperature increasing rates for all periods were similar with the corresponding warming rates calculated directly from 1979 to 2017 in temperate continental climate zone and plateau mountain climate zone. However, there exists higher differences between the average warming rates for all periods and the corresponding increasing rates calculated directly from 1979 to 2017 in subtropical monsoon climate and temperate monsoon climate. Temperature performs higher increase in temperate climate and less in subtropical. In temperate climates, the continental climate shows a higher increase for T_max, while in monsoon climate, T_min increases more. In monsoon climate, the subtropical monsoon climate and temperate monsoon climate all show a lower increase for T_max. The plateau mountain also shows a higher increase for the T_max.

3.3. Mann-Kendall Mutation Analysis of Temperature

Temperature mutation is a factor to be considered in climate prediction and simulation. At present, the Mann-Kendall (MK) method is considered the most accurate and objective method to analyze temperature mutation. Figure 4 and

Table 4. Mutation time of annual mean temperature for the four different climate zones within Gansu Province during the period 1979–2017 (°C/10a).

Climate Types	T_mean	T_max	T_min
Subtropical monsoon climate	1997	1997	No
Temperate monsoon climate	1997	1997	No
Temperate continental climate	1997	No	1997
Plateau mountain climate	No	No	1994

show the MK mutation tests of T_mean, T_max, and T_min for the four climate zones within Gansu Province based on the ERA-Interim reanalysis data. According to the intersection point of two curves of UF_k and UB_k between the critical lines (±1.96), we can learn that the abrupt change of temperature in Gansu Province occurred in the 1990s in general. In the subtropical monsoon climate zone and temperate monsoon climate zone, the T_mean and T_max all showed a shift rising in 1997. In the temperate continental climate zone, the T_mean and T_min indicated an abrupt increase in 1997. In the plateau mountain climate zone, the T_min showed an abrupt rise in 1994. From the

Table 5. Temperature mutation time at different periods for the four different climate zones within Gansu Province at each period.

Periods	Subtropical Monsoon Climate			Temperate Monsoon Climate			Temperate Continental Climate			Plateau Mountain Climate
	T_mean	T_max	T_min	T_mean	T_max	T_min	T_mean	T_max	T_min	T_mean	T_max	T_min
1979–2008	1997	1997	1997	1994	1995	1997	1993	1993	1996	No	No	1994
1980–2009	1997	1997	1997	1995	1994	1997	1993	1993	1996	No	No	1994
1981–2010	1997	1994	1997	1995	1994	1997	1993	1993	1996	No	No	1994
1982–2011	1994	1994	1997	1994	1994	1997	1993	1993	1994	1993	No	1993
1983–2012	1994	1997	1997	1994	1993	1997	1987	1989	1993	1993	1994	1994
1984–2013	1994	1993	1997	1994	1994	1997	1993	1993	1993	1993	1996	1993
1985–2014	1997	1997	No	1997	1993	1997	1997	No	1996	1997	1997	1996
1986–2015	1997	1997	2003	1997	1997	2001	1997	1997	1996	1997	1997	1996
1987–2016	1997	1997	No	2000	1997	No	1997	No	1998	1997	2002	1998
1988–2017	1997	1997	No	1997	1996	No	1997	No	1998	1997	1997	1997

, we can learn that there is no mutation in some periods, which may explain why there was no sudden change in temperature from 1979 to 2017 in general.

4. Discussion

Under the background of the global warming, Gansu Province has become warmer during the period of 1979–2017. The significant abrupt increases of annual mean temperature in subtropical monsoon climate zone, temperate monsoon climate zone, and temperate continental climate zone were found in 1997. For the annual mean temperature in plateau mountain climate zone, there was no obvious temperature mutation. Since 1990, the large-scale positive sea temperature anomaly and the weakening of trade winds in the tropical Pacific and the frequency of El Nino Events in the central Pacific were higher [27,28]. The rise of sea surface temperature (SST) led to the weakening of the East Asian summer monsoon and the less northward extension of the southwest summer monsoon, which accelerated the warming trend in the northwest and slowed down the warming trend in the Qinghai-Tibet region in summer [27,28]. The warming rates for T_max in monsoon climate zone were lower, and the temperature mutation happened in 1997. In the year of El Nino, the summer monsoon in China was weak, and the northern region of China was often prone to drought and high temperature in summer [28,29]. For the T_min in temperate continental climate zone, the mutation time happened in 1997, which was also caused by the EI Nino. In winter, the Asian continent is controlled by the strong Siberian cold and high pressure, while the ocean is much warmer, and the air pressure is relatively low. The air flows from the high-pressure part to the low-pressure part and flows from the mainland to the ocean. Therefore, the northwest air flow from high latitude prevails in most of China. There are often cold tides and cold air weather transits, and the climate is cold and dry [28,29]. The higher temperature increasing trend of T_max in plateau mountain climate may be due to the intensified melting of ice and snow caused by extreme high temperature in mountainous areas (Qilian Mountains), which reduces the ground albedo and promotes faster temperature rise [30]. For the T_max in plateau mountain climate zone, there was no obvious temperature mutation. The T_min in plateau mountain climate zone mutated in 1994, which was mainly caused by El Nino in 1994.

Wen et al. [3] found that annual mean surface air temperature over mainland China generally rose by 0.230 °C/10a. The warming rate in northwest China was 0.290 °C/10a, lower than our results from Gansu Province, which may be because the forest coverage in Gansu Province was less than the national average, resulting in less absorption of greenhouse gases and faster temperature rise [31]. Additionally, we found that the temperature trends based on the ERA-Interim reanalysis data were larger than those based on the observations in Gansu Province. For example, Wang et al. [32] found that temperatures in Hexi oasis in Gansu Province increased at a rate of 0.320 °C/10a, and Li et al. [33] found that the annual extreme minimum air temperature in same region had a significant linear increasing trend (0.337 °C/10a), which were all lower than that in corresponding area (temperate continental climate zone) in Gansu Province in this study. The spare observations in the western Gansu Province are possibly responsible for the difference.

Previous studies have verified that mutation times in Gansu Province and surrounding area occurred approximately in 1997. For example, Yang et al. [34] found that a temperature mutation in Gansu Province occurred in 1998. Teng et al. [35] investigated changes in climate in the Wushaoling region of Gansu Province during the period 1951–2016 and found an abrupt change in air temperature in 1997. The Wushaoling region, located in the temperate continental climate zone, showed a similar temperature mutation time to that observed herein. Dou et al. [36] found that temperature mutations in Gansu Province and the Hedong region occurred in 1993, while in the Hexi region, a temperature mutation occurred in 1994. Zhao et al. [37] found that a sharp change of temperature occurred in the Hedong region in 1997 by using observational data. Wang et al. [38] used observations to analyze the mutation time of extreme temperature during 1960–2009 in Gansu Province; results found that the mutation time of extreme temperature in Gansu Province existed in 1997. Li et al. [33] used the observations to analyze the mutation time in Hexi Corridor in Gansu Province; results found that the mutation time of extreme maximum air temperature occurred in 1996 and extreme minimum air temperature occurred in 1993. Jiao et al. [39] analyzed the spatiotemporal change of extreme temperature in the Hedong region in Gansu Province by using observational data; results found that the mutation point of most indexes is in the middle and late 1990s and the early 21st century. Huang et al. [40] found that the extreme temperature index suddenly changed in 1997 in the Hedong region of Gansu Province by using observational data. These previous results are consistent or similar with our study, which indicates that the reanalysis data can well reflect the shift change time of temperature and are reliable for capturing temperature mutation time.

Factors that affect temperature change are very complex. Some previous researches have indicated that temperature is also influenced by terrain. In the future, we need to study different terrain areas to explore the impact of terrain on the temperature in Gansu Province. Additionally, the changes of cloud cover may be also the main factor that affects the changes of temperature [30].

5. Conclusions

In order to investigate temperature variations across different climate zones within the Gansu Province, we used ERA-Interim temperature data from the period 1979–2017; these data were obtained from ECMWF and analyzed using a linear regression model and Mann-Kendall mutation test. We conclude that:

(1) The highest ERA-Interim temperature was found in the subtropical monsoon climate zone and the lowest was found in the plateau mountain climate zone. ERA-Interim temperatures in high-elevation regions were lower than those in low-elevation regions.

(2) The annual ERA-Interim temperature in Gansu Province increased from 1979 to 2017. The lowest rates of increase of annual mean, maximum, and minimum ERA-Interim temperatures were found in the subtropical monsoon climate zone; these were 0.334, 0.300, and 0.336 °C/10a, respectively. The highest rates of increase of annual mean and minimum ERA-Interim temperatures were found in the temperate monsoon climate zone, being 0.42 and 0.464 °C/10a, respectively. The highest rate of increase of annual maximum ERA-Interim temperature was found in the temperate continental climate zone (0.471 °C/10a). The temperature increasing trends in Gansu Province were larger than the national average (0.29 °C/10a), which may be caused by the lower forest coverage in Gansu Province.

(3) Mann-Kendall mutation analysis showed that the mutation times of annual mean temperature of the subtropical monsoon, temperate monsoon, and temperate continental climate zones in Gansu Province were all in 1997. The mutation times of annual maximum temperature were found in the subtropical monsoon climate zone (1997) and temperate monsoon climate zone (1997). The mutation times of annual minimum temperature were found in the temperate continental climate zone (1997) and plateau mountain climate zone (1994). The mutation times in Gansu Province based on the ERA-Interim reanalysis data are consistent with previous results based on the observations, indicating that the reanalysis data can well reflect the shift change time of temperature and are reliable for capture temperature mutation time especially in the high-elevation areas with rare meteorological station.