J. Geogr. Sci. 2012, 22(2): 245-260 DOI: 10.1007/s11442-012-0924-3 © 2012
Science Press
Springer-Verlag
The combined influence of background climate and urbanization on the regional warming in Southeast China SI Peng1, REN Yu2, LIANG Dongpo2, LIN Bingwen3 1. Tianjin Meteorological Information Center, Tianjin Meteorological Bureau, Tianjin 300074, China; 2. Tianjin Climate Center, Tianjin Meteorological Bureau, Tianjin 300074, China; 3. Xiamen Meteorological Bureau, Xiamen 361100, Fujian, China
Abstract: Based on China homogenized land surface air temperature and the National Centers for Environmental Prediction/Department of Energy (NCEP/DOE) Atmospheric Model Intercomparison Project (AMIP)-Ⅱ Reanalysis data (R-2), the main contributors to surface air temperature increase in Southeast China were investigated by comparing trends of urban and rural temperature series, as well as observed and R-2 data, covering two periods of 1954–2005 and 1979–2005. Results from urban-rural comparison indicate that urban heat island (UHI) effects on regional annual and autumn minimum temperature increases account for 10.5% and 12.0% since 1954, but with smaller warming attribution of 6.2% and 10.6% since 1979. The results by comparing observations with R-2 surface temperature data suggest that land use change accounts for 32.9% and 28.8% in regional annual and autumn minimum temperature increases since 1979. Accordingly, the influence of land use change on regional temperature increase in Southeast China is much more noticeable during the last 30 years. However, it indicates that UHI effect, overwhelmed by the warming change of background climate, does not play a significant role in regional warming over Southeast China during the last 50 years. Keywords: Southeast China; urban heat island; land use; climate warming; contributor
1
Introduction
Warming of the climate system is unequivocal, as is now evident from observations of increasing in global average air and ocean temperatures, and it is likely that there has been significant anthropogenic warming in the past 50 years averaged over each continent (IPCC, 2001; IPCC, 2007). Anthropogenic warming mainly involves the emission of greenhouse gases or aerosols and land use change, in which urbanization and urban heat island (UHI) can be regarded as local forms of land use change. Gallo et al. (1996a; 1996b) found that
Received: 2011-02-16 Accepted: 2011-09-15 Foundation: Urban Meteorological Research Fund of CMA, No.UMRF201009 Author: Si Peng (1983–), Master and Engineer, specialized in climate change, climate data analysis and processing. E-mail:
[email protected]
www.geogsci.com
springerlink.com/content/1009-637X
246
Journal of Geographical Sciences
observations in daily temperature range based on urban surroundings were lower than those based on rural surroundings, and the transformations of land use in these two different environments might greatly affect temperature trend change. Moreover, some other studies also set out that anthropogenic land use change was an important contributor to regional climate change, especially on temperature and precipitation (Li et al., 2006; Lian et al., 2009). The observed climate change is a comprehensive result of natural change and anthropogenic activities. In the past 100 years, the average near surface air temperature over mainland China significantly increased in the context of global warming, especially in winter. It was reported that most of records in situ stations in China had been affected by rapid urbanization and enhanced UHI (Chu and Ren, 2005; Chao and Sun, 2009; Si et al., 2010a; Deng et al., 2010; Zhu et al., 2010; He et al., 2011). In recent decades, China has been regarded as one of the main areas of land use change in the world, and as a result of rapid development of urbanization, especially in the eastern and southern China, the land cover/use was changed greatly (Seto et al., 2000), which was also evident from trend change in the normalized difference vegetation index (Piao et al., 2003; Fang et al., 2003). Accordingly, it is necessary to quantify the urbanization effect in Southeast China, and what is more, with the results of impact of urbanization on surface air temperature increase in Northeast China (Si et al., 2010b), the main factor affecting climate warming of different regions in China and its contribution can be further understood. In this paper, the estimation of land use change would be based on comparison of reanalysis data and observations, which could be reevaluate the quantifications by numerical simulation in the reported studies before, on the other hand, it may remedy the single method to assess the impact of urbanization by this way.
2
Data and methods
In this study, the range of Southeast China was defined as the land area between 15°–35°N and 105°–135°E, including most areas of East China, Central China, South China, parts of Shaanxi-Gansu region and east of Southwest China. 2.1
Data
Two climate datasets were used in this research. One was China Homogenized Historical Temperature Datasets (CHHT) derived from China Meteorological Administration (CMA) (http://data.cma.gov.cn), from which monthly mean temperature, maximum temperature, and minimum temperature covering 1954–2005 of 225 situ meteorological stations were selected. The methods of data selection and missing data processing in the study by Si et al. (2010b) were adopted here. The other one was NCEP/DOE AMIP-Ⅱ Reanalysis data (R-2) derived from National Oceanic and Atmospheric Administration (NOAA), with the global grid points of 192×94 (http://www.cdc.noaa.gov). In R-2 data, surface daily mean temperature, maximum temperature, and minimum temperature covering 1979–2005 were selected. A remote sensing dataset of stable nighttime lights as well as the statistics of urban land area and population data were also employed here to determine urbanization change over Southeast China. The stable nighttime lights image used were Version 4 DMSP-OLS Stable
SI Peng et al.: The combined influence of background climate and urbanization on the regional warming
247
Nighttime Lights Products for 1999–2004 derived from National Geophysical Data Center (NGDC) (http://www.ngdc.noaa.gov). The spatial resolution of the digital image products covering Southeast China is 0.00833º×0.00833º, and data values range from 1 to 63. The statistics of urban land area data were from China Land and Resources Almanac covering 2000–2005. The annual settlements and mining land area data of the provinces between 15°–35°N and 105°–135°E were used to calculate the proportions of land use during 1999–2004 for each province. The statistics of population data were from the fifth national census in 2000. 2.2
R-2 reanalysis data assessment
R-2 data is an updated 6-hourly global analysis series, which fixes the known processing errors in the NCEP-NCAR reanalysis (R-1), and surface temperatures are estimated from the atmospheric values, thus are not sensitive to changes in land surface (Kanamitsu et al., 2002). IPCC AR4 (2007) pointed out that the reanalysis products at present were reliable to some extent for assessment on climate trend change on short-term scale from 1979 to present. Gong et al. (2006) found that weekend effect on diurnal temperature range in China with R-2 data could well reproduce major features of that with observed, and the high reproducibility also further supported their findings over East China. Additionally, the quality of R-2 surface air temperature data was evaluated by Zhou et al. (2004) by analyzing correlations of maximum and minimum temperature between R-2 and observations, and found that the best correlation performance obtained in the eastern part of China, on the other hand, inspection of the sensitivity of R-2 data to the urbanization effect showed that it was not significant for urbanization effect on R-2 data compared with observations, furthermore, in their research, R-2 surface air temperature data was also used to estimate the impact of urbanization on climate in Southeast China, and the result of that was a first step in the development of a quantitative basis for assessing the consequences from temperature of land use change associated with Chinese urbanization. In order to ensure the reliability of conclusions in the study here, the rationality and representative of R-2 Reanalysis data in Southeast China was analyzed and estimated. 2.2.1
Correlation analysis
As shown in Figure 1, the average correlations between R-2 and observations in the whole region are significant, and coefficients for annual minimum, maximum, and mean temperature are all over 0.8. Observation stations with coefficients of annual mean temperature of 0.8 or more are in the percentage of more than 76%, 48% of which are over 0.9, and the proportion of those with correlation coefficients of 0.8 or more for annual minimum and maximum temperature are 68% and 71%, respectively. For the seasons, correlations of autumn and winter air temperature between R-2 and observed are much better, the statistically significant (95%-level) regional average coefficients are over 0.8, and coefficients of spring and summer air temperature also come to 0.7 or more, except for summer minimum and maximum temperature, with coefficients of less than 0.7, but they are still significant (Table 1). Similarly, the same as annual mean temperature series, correlations of seasonal mean temperature between R-2 and observed are the best, and observation stations with coefficients of 0.8 or more are 81%and 90% in autumn and winter,
248
Journal of Geographical Sciences
Figure 1 Correlations of annual air temperature between R-2 and observations covering 1979–2005 in Southeast China
Table 1 Regional average correlations of annual and seasonal air temperatures between R-2 and observations from 1979 to 2005; Statistically significant coefficients at the 95% level are noted by asterisk Annual
Spring
Summer
Autumn
Winter
Minimum
0.813*
0.786*
0.652*
0.850*
0.878*
Maximum
0.810*
0.764*
0.672*
0.827*
0.861*
Mean
0.840*
0.776*
0.727*
0.853*
0.894*
respectively, 58% of which are over 0.9 in winter. Although less relevance in summer, the statistically significant (95%-level) correlation coefficients of 0.8 or more still account for about 50%. For seasonal minimum temperatures, observation stations with coefficients of 0.8 or more are 85%and 91% in autumn and winter, separately, 51% of which are over 0.9 in winter, correspondingly, observation stations with coefficients over 0.9 for winter maximum temperature account for 52%. And the proportion of observation stations with statistically significant (95%-level) correlation coefficients of 0.8 or more are all about 70% for spring minimum, maximum, and mean temperature. 2.2.2
Error analysis
Two statistics of Standard Error (SE) and Mean Absolute Error (MAE) (Ma et al., 2008) were used to evaluate the consistency and errors of R-2 reanalysis data from ground-based measurements in Southeast China for the period of 1979–2005 (Figure 2). SE is the statistics inferring reliability, and smaller SE represents better consistency and representative of observations for R-2 reanalysis data, so the more representativeness, the better reliability. 1/2
⎡ 1 n 2⎤ SE = ⎢ φi′ − φ ′) ⎥ ( ∑ ⎣n − 1 i =1 ⎦
(1)
SI Peng et al.: The combined influence of background climate and urbanization on the regional warming
249
Figure 2 Frequency distribution of SE (a) and MAE (b) for annual air temperature differences between R-2 and observed (R-2-Obs) covering 1979−2005 in Southeast China
φi′ = φi − φi
(2)
1 n ∑φi′ n i =1
(3)
φ′ =
where φi denotes difference sequence between R-2 reanalysis data and observations. MAE is usually used to express average errors between model-predictions and observations. It sums up the absolute values of the errors to obtain the ‘total error’, and it is the most natural and an unambiguous measure of average error magnitude (Willmott and Matsuura, 2005). MAE =
1 n ∑φi′ n i =1
(4)
Frequency distribution of SE (Figure 2a) shows that observation stations with values concentrating in [0.1℃, 0.3℃) for annual minimum, and mean temperature are about 75%, and there are 76% of observation stations with values concentrating in [0.3℃, 0.6℃) for annual maximum temperature. Correspondingly, observation stations with values of MAE (Figure 2b) concentrating in [0.2℃, 0.4℃) for annual minimum, maximum, and mean temperature are 56%, 89%, and 44%, respectively. Seasonally (curves omitted), the values of SE in autumn and winter are relatively small, and the proportions of stations with values focusing on [0.3℃, 0.5℃) for autumn minimum, maximum, and mean temperature are 73%, 40%, and 65%, separately, and those for winter are 55%, 40%, and 46%. Observation stations with values of SE focusing on [0.3℃, 0.5℃) in spring are about 44%, which are 51% and 59% for summer minimum and mean temperature. Frequency distribution characteristics of MAE, are similar to those of SE, but with smaller values, which are mostly concentrating in [0.3℃, 0.5℃) for seasonal R-2-Obs, and there are 58% and 50% of observation stations for summer minimum and mean temperature with values in [0.1℃, 0.3℃). In conclusion, there is much better consistency existing between R-2 reanalysis data and ground-based measurements in Southeast China, no matter from results of correlation analysis or error analysis, and also characteristics of ground-based temperature trend change can be represented well by R-2 reanalysis data. Furthermore, combining with objectivity of R-2 reanalysis data and its applicability in eastern China demonstrated in section 2.2, R-2 data used to analyze urbanization effect on climate change in Southeast China is relatively reasonable and reliable.
250
Journal of Geographical Sciences
2.3 Dynamic classification of urban-rural stations based on DMSP/OLS nighttime lights image and census data
In previous studies, demographic data was only used for the division of urban-rural stations. But Balling Jr and Idso (1989) indicated that significant urbanization effects existed in the city even if the number of population was less than 1000. Therefore, it is not comprehensive to divide different types of stations only using demographic data. Recently, many researchers have used remote sensing nighttime lights image for classification of urban-rural stations (Owen et al., 1998; Hansen et al., 2001; Peterson, 2003). Taking into account realities of China, however, such as imbalance of development of national economic level, lifestyle, awareness of energy conservation and other factors make this approach to be not objective in parts of China. In this research, two criteria were used for classification of different types of stations in Southeast China, one was by combining demographic data with surrounding environment where the station is located (Si et al., 2010b). The other was based on information of statistical urban land area, using DMSP/OLS nighttime lights remote sensing data to extract spatial information of urban land use, which referred to He et al. (2005; 2006). We assumed that the ratio of nighttime lights area to land area in each province was equal to that of settlements and mining land area to the whole province land area, and found the closest level to the latter one by calculating the area ratios of each stable nighttime lights intensity value and above to determine the thresholds of nighttime lights for urban areas in Southeast China, as the criteria classifying urban-rural stations. Shown as the results we calculate, the thresholds of nighttime lights in developed regions are significantly greater than those in underdeveloped ones, and rapidly developing areas, Zhejiang and Jiangsu for example, have obviously increasing thresholds, but overall, most of the provinces in Southeast China have nearly stable thresholds of nighttime lights for urban areas. Due to the stations used here are all from national reference climate stations and national basic weather stations, the mean of 31×31 pixels over an area of about 700 km2 around the pixel where the station is located is used to identify the station type. If the mean for a station exceeds the threshold of its province, it is committed to an “obvious UHI station”, otherwise, it is a “non-obvious UHI station”, and the classification results are obtained (Figure 3). Figure 3 gives the number of different types of “UHI stations” in Southeast China classified by two methods above, respectively. Results obtained from dynamic classification based
Figure 3 Numbers of different types of “UHI stations” derived from DMSP/OLS and census data covering 1999–2005 in Southeast China
SI Peng et al.: The combined influence of background climate and urbanization on the regional warming
251
on DMSP/OLS nighttime lights image show that the proportion of obvious UHI and non-obvious UHI stations divided is roughly equal from 1999 to 2004. But the number of non-obvious UHI stations is decreasing obviously classified based on census data in combination of surrounding environment information in 2005, even 21 more than those in 2001, so the distribution of meteorological stations in Southeast China in 2001 and 2005 are depicted in Figure 4.
Figure 4 In situ meteorological station distribution in Southeast China in 2001 (a) and 2005 (b) Triangle symbol presents obvious UHI stations, and rotundity presents non-obvious UHI stations
As indicated in Figure 4, the coincidence rate of classification results in 2001 and 2005 reaches 80% or more, and the same for the other years during 1999–2004. For the inconsistent stations, they are almost classified from “non-obvious stations” into “obvious UHI” ones, locating at “urban” with developed economy, like Fujian, Zhejiang, Guangdong and so on. From the objective perspective of urban development, urbanization should be sustained, therefore, it is relatively reasonable for the type changes of those inconsistent stations. In addition, there are a small part of stations classified as “non-obvious UHI” from “obvious UHI”, but these stations are all located at areas where economic development is relatively slow with surrounding environment being “suburban” or “rural”. Considering the method of using threshold of nighttime lights as classification criterion, on one hand it is related to the consistency of satellite remote sensing images, data quality, and the way of digital images production. While the other related with the statistical error in urban land area may cause a small part of instability existing in station type classification. However, the types divided using census data with information of surrounding environment in 2005 is more reasonable and appropriate for the research here. 2.4
Processing of R-2 surface temperature data
Inverse Distance to a Power: Supposing a series of scatters is distributed on the plane, with the location is (xi, yi) and the property values are (=1, 2, …, n). p(x, y) is any of the grid knots, whose property value is obtained using those scatters all around by the method of inverse distance to a power. Selecting evenly four scatter values from four directions around p, the interpolation P(z) is expressed below, n Zi P( z ) =
∑ d ( x, y ) u i =1
[
]
i
n
1
∑ d ( x, y ) u i =1
[
i
]
(5)
252
Journal of Geographical Sciences
where Zi is the i scatter from four directions around p, di(x, y) is the distance between scatter (xi, yi) and p(x, y), shown as follows,
di ( x, y ) = ( x − xi )2 + ( y − yi ) 2
(6)
and the power parameter u=2 in this research. To generate R-2 regional background anomalies, R-2 surface daily mean temperature, maximum temperature, and minimum temperature matching regional location of Southeast China (15°N–35°N, 105°E–135°E) were needed to interpolate into the location in situ meteorological station correspondingly, using the method of inverse distance to a power, and then interpolation data were obtained. According to the method of building regional average temperature anomalies used in the study by Si et al. (2010b), R-2 regional background anomalies were generated. 2.5
The methods of evaluating urbanization effect
Two means would intend to be used here to assess the impact of urbanization, one was urban heat island (UHI) estimated by comparing linear trends between regional average temperature anomaly series with and without “obvious UHI stations”, and the difference in regional average temperature anomalies trends between observed and R-2 data was expressed as land use effect. Long-term linear regression trends for each regional average temperature anomaly were computed by using a linear regression model at the 95% confidence level.
3 3.1
Contribution of UHI effect to temperature increase since 1954 Analysis of regional background temperature change
Curves of regional background temperature anomalies in Southeast China are depicted in Figure 5. Shown in this picture, the warming trend in Southeast China during the last 50 years is significant, especially in minimum temperature series (0.172℃ per decade), but smaller than that in Northeast China (shown in Figure 6), which reaches to 0.405 ℃ per decade on the same time scale (Si et al. 2010b), particularly after the 1980s, the warming change is more clearly, but in Southeast China, the course of regional warming has mainly occurred after the early 1990s, and the warmest years all exist in 1998, except for annual maximum temperature, with the other warmest year of 2004. For the seasonal change (curves omitted), the characteristics of Southeast China agrees with those of Northeast China, the most significantly warming change exists in winter, with
Figure 5 Regional annual temperature anomalies in Southeast China covering 1954–2005 (excluding “obvious UHI stations”). The smoothed line is based on the lowess smoother (William, 1979)
SI Peng et al.: The combined influence of background climate and urbanization on the regional warming
Figure 6
253
Regional annual temperature anomalies in Northeast China covering 1954–2005 (Si et al., 2010b)
the magnitudes of minimum, mean, and maximum temperature covering 1954–2005 are 0.310℃ per decade, 0.251℃ per decade, and 0.172℃ per decade in Southeast China, respectively, followed by those in spring and autumn, while in summer it is an order of magnitude smaller, which even appears some cooling trends. In the inter-decadal oscillation, the warming processes happening for different temperature elements in each season are different, they mostly focus on the late 1980s, early 1990s and late 1990s, but the warmest year for different temperature elements is generally in 1998. 3.2
Regional annual temperature trends
Figure 7 displays spatial distribution of annual temperature trends in Southeast China (the left ones), and corresponding ones for Northeast China (the right ones). Shown in Figure 7 (left), influenced by east monsoon, the climate of parts of Shaanxi-Gansu at southeast area and the east coast is warm, and the annual warming change has been relatively larger, especially in annual minimum temperature. The warming change for areas inland is an order of magnitude smaller, where even appears some small continuous cooling trends in maximum temperature at western area. Moreover, as many high warm centers are displayed in Figure 7 (left), particularly for minimum temperature, it is easy to find that they are the areas with prominent urbanization development expressed by distribution of obvious UHI stations shown in Figure 4b. It indicates that temperature trend change in Southeast China would be impacted by urban land use forms to some extent. And analysis of regional background climate in Southeast China in section 3.1 points that it is larger for minimum temperature increase, but smaller for maximum temperature, which agrees with the characteristic of the whole regions, it also illustrates that Southeast China has a universal warming change. Seasonally (curves omitted), they still show the characteristics expressed in annual temperature change above, but for the increasing amplitudes, they are relatively larger for the strength of high warming centers in winter temperature trend distribution, and weaker in summer. Similarly, the above regional temperature change features in time and space also fully exist in Northeast China, but for the spatial distribution (Figure 7 right), the warming magnitudes in Northeast China are much larger than those in Southeast China. And the geographical differences of warming change were also expressed in Li et al. (2010) about estimation of uncertainty in Chinese temperature changes in nearly 100 years. Thus, the preliminary analysis shows that there are some similarities for climate warming in Southeast and Northeast China, and also expresses that it is universal for the warming phenomenon in eastern China. But further research is needed for Southeast China, whether
254
Journal of Geographical Sciences
Figure 7 Trend distribution of regional annual temperature in Southeast (left) and Northeast China (right) (after Si et al., 2010b) covering 1954–2005 (℃ per decade)
the factors influence on surface air temperature increase as those in the Northeast, mainly by regional background climate change, not urban land use change (Si et al., 2010b). 3.3
UHI effect on regional warming during the last 50 years
The average values of regional annual and seasonal temperature for both obvious UHI stations and non-obvious UHI stations covering 1954–2005 and 1979–2005 are calculated, respectively. Results show that the average values of obvious UHI stations are higher than those of non-obvious UHI ones, which express a typical characteristic of UHI, but the warming change due to UHI during the last 30 years is not significantly larger than that in the period from 1954 to 2005.
SI Peng et al.: The combined influence of background climate and urbanization on the regional warming
255
Average UHI effect on temperature series from 1954 to 2005 is calculated (Table 2). Results indicate that annual mean temperature increase in Southeast China due to UHI over the past 50 years is 0.013℃ per decade, which is 9.3% of the regional background warming, and the warming change in annual minimum temperature is the largest, with the increasing magnitude of 0.018℃ per decade, the contribution is 10.5%, yet for annual maximum temperature there is some decreasing change, with the magnitude of 0.001℃ per decade. Table 2 Regional annual temperature trends with and without “obvious UHI stations” in Southeast China from 1954 to 2005 Annual temperature
With UHI (℃ per decade)
Without UHI (℃ per decade)
Differences (℃ per decade)
UHI contribution (%)
Minimum temperature
0.190
0.172
0.018
10.5
Mean temperature
0.153
0.140
0.013
9.3
Maximum temperature
0.100
0.101
–0.001
–
For regional seasonal temperature change caused by UHI effect (Table 3), minimum temperature increase is still most significant, followed by mean temperature, and there is some decreasing change in maximum temperature, which performs noticeably well in spring, autumn, and winter. And temperature increase in autumn is more significant than that in the other seasons, the warming trends in minimum temperature and mean temperature are 0.020℃ per decade and 0.016℃ per decade, respectively, which are 12.0% and 11.3% as compared with corresponding regional background warming change. It may be due to special climate characteristic of Southeast China in autumn, of less cloud cover, stable atmospheric stratification condition, and smaller wind speed, which is helpful for UHI effect increasing. Minimum temperature increases in spring and winter are 0.018℃ per decade and 0.016℃ per decade, respectively, which are 12.5% and 5.2% of the regional background warming. Owing to UHI, maximum temperature in summer expresses warming change, with the magnitude of 0.006℃ per decade, which may be related to the increase of anthropogenic Table 3 Regional seasonal temperature trends with and without “obvious UHI stations” in Southeast China from 1954 to 2005 (℃ per decade)
Minimum temperature
Mean temperature
Maximum temperature
Spring
Summer
Autumn
With UHI
0.162
0.078
0.186
Winter 0.326
Without UHI
0.144
0.065
0.166
0.310 0.016
Differences
0.018
0.013
0.020
UHI contribution (%)
12.5
20
12.0
5.2
With UHI
0.154
–0.007
0.157
0.261
Without UHI
0.143
–0.020
0.141
0.251
Differences
0.011
0.013
0.016
0.010
UHI contribution (%)
7.7
–
11.3
4.0
With UHI
0.136
–0.077
0.115
0.168
Without UHI
0.138
–0.083
0.124
0.172
Differences
–0.002
0.006
–0.009
–0.004
UHI contribution (%)
–
–
–
–
256
Journal of Geographical Sciences
heat in southern cities, such as air-condition opening, accordingly, how to lessen UHI effect and heatstroke prevention in summer are the main problems for current urban planning. The analysis above suggests that UHI has been one of the urban climate characteristics in Southeast China over the past 50 years, and UHI effect has an active expression in summer. However, although the warming changes in regional annual and seasonal minimum temperature due to UHI are most significant, analyzed in sections 3.1 and 3.2, the contribution of which to regional background warming is just 20%, much smaller. It would be said that UHI is not a significant contributor relative to natural or other human-induced effect.
4
Contribution of urbanization to temperature increase since 1979
4.1
UHI effect on regional warming during the last 30 years
Table 4 shows the UHI effect since 1979, the warming changes are much larger than those of the whole 50 years, and temperature increase in autumn is still the largest. Annual and autumn minimum temperature increases due to UHI are 0.025℃ per decade and 0.037℃ per decade, respectively, but with smaller contributions to regional background warming of 6.2% and 10.6% than the whole 50 years. Table 4 Regional annual and seasonal temperature trends with and without “obvious UHI stations” in Southeast China from 1979 to 2005 (℃ per decade)
Minimum temperature
Mean temperature
Maximum temperature
4.2
Annual
Spring
Summer
Autumn
Winter
With UHI
0.429
0.554
0.303
0.386
0.624
Without UHI
0.404
0.531
0.281
0.349
0.607
Differences
0.025
0.023
0.022
0.037
0.017
UHI contribution (%)
6.2
4.3
7.8
10.6
2.8
With UHI
0.422
0.621
0.287
0.407
0.610
Without UHI
0.406
0.605
0.279
0.381
0.599
Differences
0.016
0.016
0.008
0.026
0.011
UHI contribution (%)
3.9
2.6
2.9
6.8
1.8 0.640
With UHI
0.471
0.773
0.314
0.480
Without UHI
0.475
0.771
0.330
0.480
0.648
Differences
–0.004
0.002
–0.016
0.000
–0.008
UHI contribution (%)
–
0.3
–
–
–
Average effect of land use on regional temperature increase
Table 5 lists the estimation of land use effect in Southeast China since 1979 by means of the other urbanization effect evaluation. It shows that regional temperature increase changes due to land use are larger than those caused by UHI of the same period, and contributions are also relatively significant. Warming trend in annual minimum temperature is still the largest, with the magnitude of 0.141℃ per decade, followed by annual mean temperature, with the warming change of 0.075℃ per decade, which are 32.9% and 17.8% of the regional background warming, respectively, yet there is still a little decline in annual maximum temperature change. For the seasons, the warming trend in mean temperature in winter owing to land use is 0.054℃ per decade, coinciding with Zhou et al. (2004), who used the same evaluating
SI Peng et al.: The combined influence of background climate and urbanization on the regional warming
257
method in Southeast China. The contributions of minimum temperature and mean temperature in autumn to regional background warming are 28.8% and 8.4%, respectively, which are more significant than UHI effect of the same period. In addition, the contributions to temperature increase in summer caused by land use change are the biggest, in which minimum temperature is even 90.4% of the regional background warming, and warming trend in maximum temperature is 0.011℃ per decade, with a contribution of 3.5%. Accordingly, whether for UHI during the last 50 years or for land use effect during the last 30 years, it is incontestable for the fact that there is an increasing change in regional maximum temperature in summer, and it is relatively prominent for temperature increase in autumn due to UHI of different periods, that means, to some extent, warming changes in Southeast China owing to urbanization effect mostly exist in warm seasons. Table 5
Regional annual and seasonal temperature trends for observations and R-2 (℃ per decade)
Minimum temperature
Mean temperature
Maximum temperature
5
Annual
Spring
Summer
Autumn
Observations
0.429
0.554
0.303
0.386
Winter 0.624
R-2
0.288
0.429
0.029
0.275
0.538 0.086
Differences
0.141
0.125
0.274
0.111
Contribution (%)
32.9
22.6
90.4
28.8
13.8
Observations
0.422
0.621
0.287
0.407
0.610
R-2
0.347
0.531
0.164
0.373
0.556
Differences
0.075
0.090
0.123
0.034
0.054
Contribution (%)
17.8
14.5
42.9
8.4
8.9
Observations
0.471
0.773
0.314
0.480
0.640
R-2
0.504
0.790
0.303
0.608
0.688
Differences
–0.033
–0.017
0.011
–0.128
–0.048
Contribution (%)
–
–
3.5
–
–
Discussion
Comparison of trends between urban and rural temperature series, as well as observations and R-2 data, suggested that land use change played an important role in regional warming in Southeast China over the past 30 years, yet in which UHI effect was not a significant contributor. However, in this paper, it was discussed only from urbanization effect, the true sense of land use contains urbanization, agriculture, animal husbandry, forest harvesting, desertification and so on caused by anthropogenic activities (Zhang, 2005). So the influence of land use assessed here, to some extent, may include other forms of land cover change effect on regional temperature increase except for urbanization, of course UHI is included. To detect the main contributor to temperature increase in Southeast China further, here regional background temperature series (Non-obvious UHI stations) from 1954 to 2005 were analyzed once more. A sequential algorithm was used for testing climate regime shifts (taking annual mean temperature series for example) (Rodionov, 2004), shown as Figure 8. This method can produce consistent sequence of shifts caused by regime shift, and it is less sensitive to the presence of trends in a time series that may be easy to falsely identify as a shift point of the center of this time series.
258
Journal of Geographical Sciences
Figure 8
Shifts in the mean for annual temperature anomalies covering 1954–2005
It is indicated that an abrupt increase by about 0.8℃ around 1997 occurs in annual mean temperature anomalies in Southeast China, and also in autumn series, but around 1998. The abrupt increases are also detected in the annual and autumn minimum and maximum temperatures. Regional background temperature anomalies in Southeast China during the last 50 years were emphatically analyzed in section 3.1, it also found that the warming years for different temperature elements on annual or seasonal scale almost existed around 1998. Li et al. (2006) pointed out that constant temperature increase events had already happened since 1998, which were similar to those in the Northern Hemisphere, and 1998 was the warmest year over the past 50 years. And it is found in the research by Li et al. (2009) that the abrupt change round 1988 mainly accounted for warming in Northeast China, but not UHI. By this token, as like Northeast China (Si et al., 2010b), temperature increase change in Southeast China is just the reflection of Chinese climate change, and also is the mapping of global climate warming in the area of China. As analyzed in section 3.2, there are similarities for climate warming in Southeast and Northeast China, and it is universal in eastern areas of China for the warming phenomenon.
6
Conclusions
Based on China homogenized land surface air temperature data and R-2 reanalysis data, the contributors that inducing regional temperature increase in Southeast China are estimated, and conclusions are as follows. (1) There is a significant warming trend in Southeast China during the last 50 years, especially in east coast area, and minimum temperature series express much larger, but the warming range is much smaller than that in the whole Northeast China due to geographical differences. (2) Regional background climate analysis suggests that Chinese climate change mainly accounts for temperature increase in Southeast China over the past 50 years. (3) In addition, land use change has played a key role in regional warming over Southeast China during the last 30 years, by comparing different methods evaluating urbanization effect from two periods of 1954–2005 and 1979–2005.
References Balling Jr R C, Idso S B, 1989. Historical temperature trends in the United States and the effect of urban population growth. Journal of Geophysical Research, 94(D3): 3359–3363. Chao Lumen, Sun Jianxin, 2009. Temporal change in annual air temperature and heat island effect in a coastal city
SI Peng et al.: The combined influence of background climate and urbanization on the regional warming
259
and an inland city at mid-latitude in China during 1956–1998. Chinese Journal of Applied Ecology, 20(12): 2839–2846. (in Chinese) Chu Ziying, Ren Guoyu, 2005. Change in urban heat island magnitude and its effect on mean air temperature record in Beijing region. Acta Meteorologica Sinica, 63(4): 534–540. (in Chinese) Deng Yujiao, Kuang Yaoqiu, Huang Feng, 2010. Research of urban heat island in Guangzhou based on landsat/TM data. Meteorological Monthly, 36(1): 26–30. (in Chinese) Fang Jingyun, Piao Shilong, He Jinsheng et al., 2003. Activities in vegetation coming to enhance in China for nearly 20 years. Science in China (Series C), 33(6): 554–565. (in Chinese) Gallo K P, Easterling D R, Peterson T C, 1996a. The influence of land use/land cover on climatological values of the diurnal temperature range. Journal of Climate, 9(11): 2941–2944. Gallo K P, Tarpley J D, 1996b. The comparison of vegetation index and surface temperature composites for urban heat-island analysis. International Journal of Remote Sensing, 17(15): 3071–3076. Gong D Y, Guo D, Ho C, 2006. Weekend effect in diurnal temperature range in China: Opposite signals between winter and summer. Journal of Geophysical Research, 111: D18113, doi: 10.1029/2006JD007068. Hansen J, Ruedy R, Sato M et al., 2001. A closer look at United States and global surface temperature change. Journal of Geophysical Research, 106(D20): 23947–23963. He Chunyang, Li Jinggang, Chen Jin et al., 2005. The urbanization model and process in Bohai Sea surrounding area in the 1990s by using DMSP/OLS data. Acta Geographica Sinica, 60(3): 409–417. (in Chinese) He C Y, Shi P J, Li J G et al., 2006. Restoring urbanization process in China in the 1990s by using non-radiancecalibrated DMSP/OLS nighttime light imagery and statistical data. Chinese Science Bulletin, 51(13): 1614–1620. He Ping, Li Hongbo, Huang Hui, 2011. The analysis of climate decadal change characteristics and urban climate about Chuxiong city of Yunnan Plateau during 1960–2009. Progress in Geography, 30(1): 65–72. (in Chinese) IPCC, 2001. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. UK: University of Cambridge. IPCC, 2007. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. UK: University of Cambridge. Kanamitsu M, Ebisuzaki W, Woollen J et al., 2002. NCEP-DOE AMIP-Ⅱ REANALYSIS (R-2). Bulletin of the American Meteorological Society, 83(11): 1631–1643. Li Qiaoping, Ding Yihui, Dong Wenjie, 2006. A numerical simulation on impact of historical land-use changes on regional climate in China since 1700. Acta Meteorologica Sinica, 64(3): 257–270. (in Chinese) Li Qingxiang, Dong Wenjie, Li Wei et al., 2010. The estimation of uncertainty in Chinese temperature changes nearly a hundred years. Chinese Science Bulletin, 55(16): 1544–1554. (in Chinese) Li Qingxiang, Li Wei, Ju Xiaohui, 2006. Analysis on the persistent extreme warm period over China since 1998. Science & Technology Review, 24(4): 37–40. (in Chinese) Li Q X, Li W, Si P et al., 2009. Assessment of surface air warming in Northeast China, with emphasis on the impacts of urbanization. Theoretical and Applied Climatology, 99(3/4): 469–478. Lian Lishu, Shu Jiong, Li Chaoyi, 2009. The impacts of grassland degradation on regional climate over the origin area of three rivers in Qinghai-Tibet Plateau, China. Acta Meteorologica Sinica, 67(4): 580–590. (in Chinese) Ma L J, Zhang T J, Li Q X et al., 2008. Evaluation of ERA-40, NCEP-1, and NCEP-2 Reanalysis air temperatures with ground-based measurements in China. Journal of Geophysical Research, 113: D15115, doi: 10.1029/2007JD009549. Owen T W, Gallo K P, Elvidge C D et al., 1998. Using DMSP-OLS light frequency data to categorize urban environments associated with U.S. climate observing stations. International Journal of Remote Sensing, 19(17): 3451–3456. Peterson T C, 2003. Assessment of urban versus rural in situ surface temperatures in the contiguous United States: No difference found. Journal of Climate, 16(18): 2941–2959.
260
Journal of Geographical Sciences
Piao S L, Fang J Y, Zhou L M et al., 2003. Interannual variations of monthly and seasonal normalized difference vegetation index (NDVI) in China from 1982 to 1999. Journal of Geophysical Research, 108(D14): 4401, doi: 10.1029/2002JD002848. Rodionov S N, 2004. A sequential algorithm for testing climate regime shifts. Geophysical Research Letters, 31: L09204, doi: 10.1029/2004GL019448. Seto K C, Kaufmann R K, Woodcock C E, 2000. Landsat reveals China’s farmland reserves, but they’re vanishing fast. Nature, 406(6792): 121. Si Peng, Li Qingxiang, Li Wei et al., 2010a. Urbanization-induced surface air temperature change in Shenzhen. Transactions of Atmospheric Sciences, 33(1): 110–116. (in Chinese) Si Peng, Li Qingxiang, Li Wei, 2010b. Impact of urbanization on surface air temperature warming in Northeast China. Meteorological Monthly, 36(2): 13–21. (in Chinese) William S C, 1979. Robust locally weighted regression and smoothing scatterplots. Journal of American Statistical Association, 74(368): 829–836. Willmott C J, Matsuura K, 2005. Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance. Climate Research, 30(1): 79–82. Zhang Jingyong, 2005. Impact of land cover/use on the regional climate change [D]. Beijing: Graduate University of Chinese Academy of Sciences, 2–22. (in Chinese) Zhou L M, Dickinson R E, Tian Y H et al., 2004. Evidence for a significant urbanization effect on climate in China. Proceedings of the National Academy Sciences, 101(26): 9540–9544. Zhu Yan, Zhu Lianfang, Xu Yongming et al., 2010. Study on the urban heat island of Suzhou city based on Landsat remote sensing data. Plateau Meteorology, 29(1): 244–250. (in Chinese)