Observed surface warming induced by urbanization in east China

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 116,, doi: /2010jd015452, 2011 Observed surface warming induced by urbanization in east China Xuchao Yang, 1,2 Yiling Hou, 3 and Baode Chen 1 Received 7 December 2010; revised 7 April 2011; accepted 6 May 2011; published 28 July [1] Monthly mean surface air temperature data from 463 meteorological stations, including those from the ordinary and national basic reference surface stations in east China and from the National Centers for Environmental Prediction and National Center for Atmospheric Research (NCEP/NCAR) Reanalysis, are used to investigate the effect of rapid urbanization on temperature change. These stations are dynamically classified into six categories, namely, metropolis, large city, medium sized city, small city, suburban, and rural, using satellite measured nighttime light imagery and population census data. Both observation minus reanalysis (OMR) and urban minus rural (UMR) methods are utilized to detect surface air temperature change induced by urbanization. With objective and dynamic station classification, the observed and reanalyzed temperature changes over rural areas show good agreement, indicating that the reanalysis can effectively capture regional rural temperature trends. The trends of urban heat island (UHI) effects, determined using OMR and UMR approaches, are generally consistent and indicate that rapid urbanization has a significant influence on surface warming over east China. Overall, UHI effects contribute 24.2% to regional average warming trends. The strongest effect of urbanization on annual mean surface air temperature trends occurs over the metropolis and large city stations, with corresponding contributions of about 44% and 35% to total warming, respectively. The UHI trends are C and 0.26 C decade 1. The most substantial UHI effect occurred after the early 2000s, implying a significant effect of rapid urbanization on surface air temperature change during this period. Citation: Yang, X., Y. Hou, and B. Chen (2011), Observed surface warming induced by urbanization in east China, J. Geophys. Res., 116,, doi: /2010jd Introduction [2] Detection and attribution of regional and global climate change, particularly climate warming stemming from natural and anthropogenic activities, are the central issues in current climate change research [Ren et al., 2008]. In general, the detection of surface warming caused by enhanced greenhouse gases (GHG) emission in urban areas is largely based on observational surface air temperature records. These records are usually considered inaccurate as they do not remove the so called urban heat island (UHI) effect, which gives rise to long term warming along with that contributed by GHG emissions. The UHI effect is mostly regarded as one of the major errors or sources of uncertainty in current surface warming studies [Gong and Wang, 2002; Heisler and Brazel, 2010]. Conversely, the effects of land use and land cover changes are usually disregarded despite their being first order climate forcing agents; these factors have also played an 1 Shanghai Typhoon Institute of China Meteorological Administration, Shanghai, China. 2 Institute of Meteorological Sciences, Zhejiang Meteorological Bureau, Hangzhou, China. 3 Shanghai Climate Center, Shanghai, China. Copyright 2011 by the American Geophysical Union /11/2010JD important role in alterations in regional and global climates [Pielke, 2005]. In terms of land use change, urbanization is one of the extreme processes [Shepherd, 2005]. As pointed out by Seto and Shepherd [2009], climate change and urbanization are two of the most pressing global environmental issues of the 21st century and are becoming increasingly interconnected. [3] Within the international climate community, there currently exist divergent views on urbanization and its degree of influence on regional and global mean temperature trends. The fourth assessment report of the Intergovernmental Panel on Climate Change (IPCC) states that urban heat island effects are real but local and have a negligible influence on global warming trends [IPCC, 2007]. Nevertheless, some research results indicate that this effect may play a more significant role in temperature trends estimated at multiple geographic scales; such results should be accorded more consideration in the mitigation of climate change [Pielke, 2005; Stone, 2009]. [4] Investigations on the effect of urbanization on observed surface air temperature trends are typically conducted by comparative analysis of stations in urban and rural areas. In general, either population data [e.g., Easterling et al., 1997; Hua et al., 2008; Ren et al., 2008] or satellite measured nighttime light data [e.g., Gallo et al., 1999; Hansen et al., 1of12

2 Table 1. Results of Recent Studies on the Urban Heat Island (UHI) Effect in China at Regional and National Scales Study Area Method Time Period UHI Warming ( C decade 1 ) References Southeast China OMR a Zhou et al. [2004] East of 110 E over China OMR Zhang et al. [2005] China OMR Yang et al. [2009] Mainland China UMR b <0.012 Li et al. [2004] China UMR Hua et al. [2008] North China UMR Ren et al. [2008] China Comparison with SST c Jones et al. [2008] Northwest China UMR Fang et al. [2007] Southwest China UMR for large and medium sized Tang et al. [2008] cities for small cities for national stations Yangtze River Delta UMR Du et al. [2007] Northeast China UMR Li et al. [2010] a OMR, observation minus reanalysis. b UMR, urban minus rural. c SST, sea surface temperature. 2001; Peterson, 2003; Du et al., 2007] are used to classify meteorological stations into urban and rural types. Recently, Stone [2007] categorized stations based on both population and nighttime light data in an analysis of urban and rural temperature trends over the United States. [5] China has experienced rapid economic development and a dramatic growth in its urban population and city areas over the past 30 years. Jones et al. [1990] performed a comparative analysis of mean temperature change obtained from urban and rural stations in east China and found that urbanization has had little effect on mean surface temperature change. Wang et al. [1990] and Portman [1993] indicated that the UHI exerts a significant influence on temperature trends. On the basis of homogenized temperature data from 390 national stations, Li et al. [2004] revealed that the UHI is not a significant contributor to the regional warming in mainland China. The recent study on northeast China, in which homogenized temperature data from 187 stations were used, also supports this conclusion [Li et al., 2010]. In northwest China, Fang et al. [2007] indicated that the average UHI effect from 1961 to 2000 was only 0.02 C because of the relatively low urbanization level in this region. On the other hand, some studies have shown that the effect of the UHI on regional surface air temperature trends is quite considerable. On the basis of the urban minus rural (UMR) data of 191 station pairs across China, Hua et al. [2008] investigated the effect of urbanization on temperature and showed that the UHI effect was 0.05 C decade 1 for large cities and 0.03 C decade 1 for mediumsized cities and small town stations from 1961 to Using 282 stations from two different networks (national and ordinary weather stations) across northern China, Ren et al. [2008] comprehensively analyzed the urbanization effects on observed surface air temperature trends in this region. The authors showed that the regional average annual mean temperature series, calculated using the data from 95 national stations, is significantly influenced by urban warming and that the urban warming trend, determined by comparing urban data with those from the rural network (63 sites), was 0.11 C decade 1 over the period 1961 to Using the data from 322 national stations and ordinary weather stations, Tang et al. [2008] found that the contributions of urban warming to overall temperature change are relatively large in southwest China and that the urban warming rate from 1961 to 2004 was C decade 1 for large and medium sized cities. Using nighttime light imagery from the Defense Meteorological Satellite Program (DMSP) Operational Linescan System (OLS) as a basis, Du et al. [2007] classified 99 stations in the Yangtze River Delta into megacity and nonmegacity regions, and found that the UHI effect was C decade 1 from 1961 to The UHI that is closely associated with the accelerated development of the Yangtze River Delta megacity region since the 1990s may be regarded as a critical climate signal. Table 1 summarizes some of the results of the UMR approach to the UHI effect in China at national and regional scales. [6] An important issue presented by the surface air temperature series, particularly in China, is that few weather stations are located in completely or perfectly rural locations; the possible underestimation of the UHI effect might have been a major source of error in the determination of regional average temperature trends [Ren et al., 2008]. Furthermore, most studies of urban related warming in China, derived from the difference in temperature trends between a single city and its adjacent rural stations, usually represent the characteristics of the stations only at a very small scale and more regional scale investigations are needed. [7] Kalnay and Cai [2003] introduced a new method for estimating the effect of urbanization and other land uses on surface temperature change in the United States. This method involves subtracting the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) Reanalysis (NNR) data [Kalnay et al., 1996] from the observed temperature (observation minus reanalysis, or OMR). The essence of the OMR method lies in the fact that the NNR does not assimilate surface observations of temperature, moisture, and wind over land, and the surface temperatures are estimated from atmospheric values so that the NNR is insensitive to land surface properties and the changes in these properties [Kalnay and Cai, 2003]. Subsequently, Lim et al. [2005] expanded the study area into the Northern Hemisphere to estimate the sensitivity of temperature changes to land types and found that desert and urban areas show the strongest OMR trends. Nuñez et al. [2008] applied the OMR method to surface stations in Argentina to estimate the effect of land use changes and found that the OMR trends show a warming contribution to mean temperature of 0.07 C decade 1 and a decrease in 2of12

3 Figure 1. stations. Terrain of the study area and locations of the diurnal temperature range of 0.08 C decade 1. Recently, Fall et al. [2010] investigated the sensitivity of temperature trends to land use change over the conterminous United States using the OMR trend from station observations and the North American Regional Reanalysis, which has been developed as an improvement upon the earlier NCEP/NCAR and National Centers for Environmental Prediction/Department of Energy (NCEP/DOE) data in terms of both resolution (32 km grid increments) and accuracy [Mesinger et al., 2006]. The results indicated that land use and land cover types are strong drivers of surface air temperature change. The regions converted into urban areas show a positive (warm) OMR trend, whereas regions converted into croplands display a cooling trend (presumably because of irrigation and increased evaporation). Moreover, the results for the areas of OMR warming and cooling over the United States agree well with those obtained by Hansen et al. [2001] when they defined urban/ rural stations based on nighttime lights [Fall et al., 2010]. [8] Observation minus reanalysis has also been recently used to estimate the effect of land surface forcing on temperature change in China. By comparing the surface observation temperature with the NCEP/DOE Reanalysis data, Zhou et al. [2004] found that the rate of temperature warming brought about by urbanization in southeast China was 0.05 C decade 1 from 1979 to Using similar data and methods, Zhang et al. [2005] determined the 1960 to 1999 change rate caused by urbanization and land use change in China. Their results are as follows: 0.12, 0.2, and 0.03 C decade 1 for mean, minimum, and maximum temperatures, respectively. In investigating the sensitivity of temperature change to land use/cover types in China, Yang et al. [2009] subtracted the European Centre for Medium Range Weather Forecasts 40 year Reanalysis temperature from the Climatic Research Unit observational surface air temperature. Their results revealed that warming in urban land areas reached 0.14 C decade 1 between 1960 and In a recent study by Hu et al. [2010], the OMR method was used to assess the effect of land surface forcing on extreme temperature in eastern China from 1979 to They demonstrated that the land surface change effect may explain one third of the observed temperature increase for annual warm nights and nearly half of the observed decrease for annual cold nights. Overall, the UHI effect on temperature change, detected using the OMR method, is more significant than that determined using the UMR approach (Table 1). However, no detailed comparisons have been performed to clarify this issue further. [9] The inconsistencies in the aforementioned studies may have resulted from several factors, such as the density of station network analyzed, the criteria for defining urban and rural stations, analysis methods, time periods, and the regional span studied. Most important, except for one or two Figure 2. Number of pixels of nighttime lights with values larger than 6 for 1992, 2000, and 2007 in east China. 3of12

4 Table 2. Thresholds of Nighttime Lights for Urban Areas at Different Time Slices Over Each Province Anhui Fujian Jiangsu Jiangxi Shandong Shanghai Zhejiang studies that utilized satellite nighttime lights to classify stations [e.g., Du et al., 2007] and data from ordinary weather stations to analyze the UHI effect [e.g., Ren et al., 2008; Tang et al., 2008], most of the previous investigations are based on the records of national reference climate stations and basic weather stations, the majority of which are located near cities or towns. The accuracy of the regional average temperature series obtained from the data of the rural station group may have been compromised by the urbanization effects. Moreover, China s rapid urbanization in the past three decades led to a quick transition of stations from rural into urban within a very short period. Note, however, that almost all of the previous studies did not consider this factor in their UMR analysis. The type of station remained fixed throughout an entire analysis period once it was identified as rural or urban. Thus, disregarding the effect of the conversion of stations from rural to urban on temperature records may give rise to a considerable underestimation of the UHI effect. [10] East China is densely populated, especially around the Yangtze River Delta, where dramatic economic development and growth have occurred since China s reform and opening up in the late 1970s. This area has been experiencing rapid urbanization over the past three decades. With urban land use area continuously growing and population increasing, regional UHI problems are becoming increasingly serious. Systematically and quantitatively examining the urbanization effect on surface warming is therefore an urgent requirement. In the current study, we attempt to quantitatively determine the potential magnitude of possible urban related effects on regional scale temperature trends during a rapid urbanization period. To this end, more indepth observations are made. An objective approach to dynamically categorizing urban and rural stations is developed and employed based on the DMSP/OLS nighttime light data of , population census data, and geographical information system (GIS) technology. Finally, OMR and UMR methods are applied to 463 categorized stations to examine the UHI effect on temperature change in east China during [11] The rest of the paper is organized as follows: section 2 provides geographical information on the study area and a detailed explanation of the data and analysis approaches, section 3 presents the results obtained, and section 4 provides a brief discussion of the results and presents the concluding remarks. 2. Description of the Study Area, Data, and Analysis Methods 2.1. Study Area [12] In general, east China is a geographical and loosely defined cultural region that covers the eastern coastal area of China. In the current research, east China is selected as the one administratively defined by the Chinese government, including the provinces of Anhui, Fujian, Jiangsu, Jiangxi, Shandong, and Zhejiang, as well as the municipality of Shanghai (Figure 1). This region comprises part of the North China Plain in the north, the Yangtze River Delta Plain at the center, and the mountainous area in the south. In particular, the urban agglomeration in the Yangtze River Delta has the highest city density and urbanization level in China. It is composed of Shanghai and 14 prosperous cities in Jiangsu and Zhejiang provinces where four city clusters, namely, Nanjing Zhenjiang Yangzhou, Suzhou Wuxi Changzhou, Shanghai, and Hangzhou Bay, form a belt with a zigzag shape Data [13] Monthly mean surface air temperature data from 463 weather and climate stations were used in the study. The Figure 3. Changes in the number of station groups derived from nighttime light and population census data from 1992 to of12

5 Figure 4. Defense Meteorological Satellite Program Operational Linescan System (DMSP/OLS) nighttime light imagery and distribution of stations for six categories in 1992, 1997, 2002, and 2007 in east China. data cover the period and include observations at elevations below 500 m from national reference climate stations, national basic weather stations, and ordinary weather stations (Figure 1). Obvious inhomogeneity caused by site relocation was adjusted for 21 stations. [14] In addition, the NNR 2 m air temperature data for the same period were used. Following the same techniques and procedures put forward by Kalnay and Cai [2003] in processing data, we linearly interpolated the NNR temperature data into the data from the observational station sites. In the analysis, the annual cycles were removed from both station observations and the NNR data, and only the anomalies were considered, which reduces the effect of systematic errors in the reanalysis. [15] Nighttime light imagery was obtained from the DMSP/OLS. This sensor is sensitive to the faint nighttime lights produced by cities, towns, fire, and lightning, making it unique among environmental remote sensing satellites. The Version 4 stable nighttime light products ( ) with 1 km spatial resolution, downloaded from the National Geophysical Data Center [ download.html], were used to classify the stations. The imagery shows lights from cities, towns, and other sites with persistent lighting, including gas flares. The DMSP/OLS image has digital number values ranging from 0 to 63. Areas with zero cloud free observations are represented by the value Classification of Stations [16] Classifying stations into different types is a key issue in UHI research. In the present study, DMSP/OLS nighttime light data from 1992 to 2007 were employed to dynamically categorize urban and rural stations. The large nighttime light value is most likely representative of urban areas. Figure 2 shows the number of pixels with nighttime light values larger than 6 for 1992, 2000, and 2007 over east China. From 1992 to 2000, the number of pixels with values larger than 6 slightly increased from about 100,000 to 109,000. However, the number of pixels with large values grew to more than 300,000 in 2007, indicating that the urbanization process intensified after [17] He et al. [2006] developed a technique for quickly and efficiently deriving urban land information from the non radiance calibrated DMSP/OLS nighttime light imagery and statistical data of the administrative, unit based urban land area in China. In the present study, we essentially followed the approach of He et al. [2006] to determine the urban thresholds under two basic premises: [18] (1) The existing statistical data of the administrative, unit based urban land area for Shanghai and the other six provinces can largely reflect the quantitative characteristics of the urban land in each province as a whole; hence, the derived total urban land area from the DMSP/OLS imagery should be close or equal to the urban land area from the statistical data. [19] (2) Within the 1 km resolution DMSP/OLS imagery, the derived urban land in east China has been continuously growing year by year, so an urban patch detected in the earlier DMSP/OLS imagery should remain in later imagery. [20] Within a particular province, the number of grids (i.e., 1 km 1 km) is accumulated according to the nighttime light value at the grid from the maximum in a descending order; once the area (total number of grids) is approximately equal to the statistical urban land area, the nighttime light value at the last grid accumulated is defined as the threshold for an urban area, i.e., a grid with nighttime light value no less than the threshold. Table 2 provides the nighttime light thresholds for urban land at different time slices for each province. The thresholds of urban land are consistent with the local economic development levels. Shanghai has the highest level of urbanization. The Yangtze River Delta region of Jiangsu and Zhejiang and the coastal area in Fujian also have higher urbanization levels. Nighttime lights over Shandong are numerous because of its huge population. Inland regions, such as Anhui and Jiangxi, have relatively lower urbanization levels. [21] To determine the type of station, the means of nighttime light value over a circular area with the center at the station were calculated for radii of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, and 20 km. As observed, the means of nighttime light values no longer change much after the radius increases to 7 km. Therefore, the mean of nighttime light values over a 5of12

6 Figure 5. DMSP/OLS nighttime light imagery of 2007 and annual mean temperature trends from (a) station observations (OBS), (b) the NCEP/NCAR Reanalysis (NNR), (c) the observation minus reanalysis (OMR) in C decade 1 at sites located below 500 m in east China; C, C, and C decade 1 are the mean trends of all OBS, NNR, and OMR, respectively. circular area with a 7 km radius around the station was used to identify the type of station. If the mean for a station exceeds the threshold of its province, the station is categorized as an urban station; otherwise, it is classified as rural. [22] Furthermore, urban stations classified by the above mentioned approach were grouped into four types using nonagricultural population data from China City Statistical Yearbook, including (1) a small city with a nonagricultural population of million, (2) a medium sized city with a nonagricultural population of million, (3) a large city with a nonagricultural population of million, and (4) a metropolis with a nonagricultural population of over 1.0 million. Suburban stations were classified based on their proximity to a large city or metropolis within a radius of 30 km and a continuous nighttime light patch. [23] Figure 3 shows the station number of each group from 1992 to Rural stations accounted for more than 60% of the total number of stations in During the rapid urbanization in east China over the past two decades, many rural stations were converted into small cities within a very short period. The number of small cities exceeded the number of rural stations in In 2007, the number of rural stations accounted for only about 30%. [24] Figure 4 shows the classification of stations and the DMSP/OLS nighttime light imagery in 1992, 1997, 2002, and 2007 over east China. The most significant urbanization occurred in the Yangtze River Delta, followed by Jiangsu, Shandong, and the coastal area Calculation of Time Series of the Surface Air Temperature Anomalies for Different Station Categories [25] In the current research, we primarily considered the effect of gradually growing urbanization on regional average temperature series and developed a new method to calculate the time series of temperature anomalies for each station category. The surface air temperature anomalies with respect to the mean annual cycle (based on a 27 year climatology) for station observations and the NNR were calculated. The temperature anomalies were then averaged over six station categories to create six time series according to the dynamic classification for Because the DMSP/OLS nighttime light data were recorded only beginning in 1992, the Figure 6. Correlation coefficients between the surface air temperature for station observations and the NNR; is the average. 6of12

7 Figure 7. Observational, NNR, and OMR time series of temperature anomalies for each of the station groups in east China during ; denoted are temperatures from station observations (solid lines with squares) and NNR (solid lines with dots), OMR (bars), and OMR linear trends (gray lines). categorization applied to stations before 1992 was that used for All trends were calculated using simple linear regression, and the degree of significance was assessed using related P values Urban Minus Rural (UMR) Analysis [26] Referring to Karl et al. [1988] and considering the large population in east China, we created a pair of urban and rural stations by selecting the rural station within a radius between 50 and 100 km for medium sized and small cities and between 100 and 150 km for metropolises, large cities, and suburban areas. For a given urban station, there may be several station pairs, and in that case their average is calculated. Because some rural stations were transformed into urban stations, the number of station pairs may have changed in a particular year from 1992 to The time series of the surface air temperature anomalies and the UMR were calculated in a manner similar to that described in section Results 3.1. Temperature Trends [27] Figures 5a and 5b show the 27 year surface air temperature linear trend (in C decade 1 ) of the station 7of12

8 Table 3. Temperature Trends From Station Observations and NNR, and the Differences Between the Observations and NNR (i.e., OMR) for Different Station Groups in East China, , With Statistical Significance at 0.05 (Single Asterisk) and 0.01 (Double Asterisk) a Annual Spring (MAM) Summer (JJA) Autumn (SON) Winter (DJF) Rural OBS 0.502** 0.556** 0.255* 0.446** 0.757** NNR 0.421** 0.435** 0.252** 0.386* 0.610* OMR Suburban OBS 0.651** 0.745** 0.464** 0.614** 0.780** NNR 0.513** 0.562** 0.350** 0.516** 0.623** OMR Small city OBS 0.614** 0.646** 0.356** 0.594** 0.854** NNR 0.447** 0.467** 0.295** 0.413* 0.612* OMR Medium city OBS 0.674** 0.722** 0.424** 0.644** 0.900** NNR 0.460** 0.471** 0.306** 0.436* 0.627* OMR Large city OBS 0.742** 0.836** 0.527** 0.721** 0.883** NNR 0.482** 0.576** 0.328** 0.448** 0.576* OMR Metropolis OBS 0.904** 1.010** 0.689** 0.841** 1.077** NNR 0.506** 0.525** 0.377** 0.513** 0.611* OMR a The unit of measure is C decade 1. substantially less than that in the station observations, indicating possible effects from atmospheric circulation and GHG concentration. [29] Figure 6 displays the correlation between the surface temperature from the station observations and from the NNR at the stations located at elevations below 500 m. The average correlation of implies that the NNR may identify surface temperature variations caused by atmospheric circulation such as storms, advection of warm/cold air, and variations in the frequency or track of major storms. Given that the NNR is not affected by changes in surface properties, it can be used in the OMR method to estimate the urbanization effects on temperature change [Kalnay and Cai, 2003] Effect of Urbanization on Surface Air Temperature Changes Over East China [30] Figure 5c shows the difference between the observational and NNR trends (i.e., OMR). The distribution of the OMR trends are quite similar to that of the observational temperature change in Figure 5a, indicating a significant effect of urbanization on surface warming in east China in the past 27 years. There are 66 stations showing cooling (negative) OMR trends, and most of them are located in areas with low nighttime light values. Other land use changes such as croplands with irrigation may have contributed to these negative OMR trends. [31] Figure 7 presents the time series of annual mean temperature anomalies from the station observations, the NNR, and their differences (OMR) averaged over six station groups (as defined in section 2.4) for the period Both the station and NNR annual mean temperatures show increasing trends. The NNR agrees well with the station observations in records and the NNR for 463 sites located at elevations below 500 m in east China, as well as the DMSP/OLS nighttime light imagery in The station observations (Figure 5a) indicate that, on average, the mean temperature in east China reflects an increase by a rate of C decade 1 over the past 27 years, with the warming pattern closely related to the intensity of the nighttime lights. Stations in the urban agglomeration of the Yangtze River Delta, where the nighttime light values are highest, show the most significant warming trends. Temperatures in medium sized city stations with relatively high nighttime lights also display rapid increases. Conversely, sites with weak warming trends are mainly rural or small city stations, which are mostly located in the provinces of Anhui, Jiangxi, and Fujian, where the economy is less developed and the nighttime lights are low. A broad range of spatial disparity exists in the temperature linear trends over east China, with a minimum of and a maximum of C decade 1. [28] By contrast, the NNR estimation (Figure 5b) shows that, as a whole, the temperature in east China reflects an increase at a mean warming rate of C decade 1 from 1981 to 2007; spatial difference ranges from C to C decade 1. Note that the NNR temperature in the Yangtze River Delta also shows the strongest warming trend but cannot reach the amplitude from the station observations. The spatial disparity in the NNR temperature trends is Table 4. Urban Minus Rural (UMR) Results for Station Observations, With Statistical Significance at 0.05 (Single Asterisk) and 0.01 (Double Asterisk) a Annual Spring (MAM) Summer (JJA) Autumn (SON) Winter (DJF) Suburban Urban 0.651** 0.745** 0.464** 0.614** 0.780** Rural 0.551** 0.652** 0.355** 0.520** 0.759** UMR Small city Urban 0.611** 0.646** 0.363** 0.588** 0.840** Rural 0.534** 0.571** 0.299* 0.495** 0.773** UMR Medium city Urban 0.670** 0.714** 0.432** 0.633** 0.878** Rural 0.535** 0.562** 0.319* 0.507** 0.767** UMR Large city Urban 0.802** 0.897** 0.605** 0.772** 0.884** Rural 0.595** 0.693** 0.372** 0.549** 0.759** UMR Metropolis Urban 0.870** 0.925** 0.637** 0.823** 1.011** Rural 0.585** 0.653** 0.358* 0.550** 0.781** UMR a The unit of measure is C decade 1. 8of12

9 Figure 8. Mean OMR ( C decade 1 ) at different time slices: (a) , (b) , and (c) The DMSP/OLS nighttime light imagery of 2007 is also shown in Figure 8c. identifying the interannual variability and long term warming trends for all the station groups. Nevertheless, the station observations exhibit a stronger warming trend than does the NNR. As a result, the OMR shows a positive trend, with the strongest being in the metropolis group, followed by the large city, medium sized city, small city, and suburban stations. The rural stations show the weakest trend. The most substantial increase in OMR value occurred after the early 2000s, implying a significant effect of rapid urbanization on surface air temperature change during this period. [32] Table 3 presents the temperature trends from the station observations and NNR, along with their differences (OMR). Table 4 shows the temperature trend differences between urban and rural sites (UMR); the differences were derived from the station observations. In Tables 3 and 4, the values for spring, summer, fall, and winter are shown, along with the annual means for each station group. [33] From 1981 to 2007, the largest increase in stationobserved annual mean surface temperature occurred at the metropolis stations with an annual linear trend of C decade 1. From the sites of large to small cities, the linear trends show monotonic descent from C to C decade 1. The rural sites display the weakest warming trend with a rate of C decade 1. The NNR annual mean temperature changes, which reflect those mainly associated with changes in circulation and greenhouse warming, generally exhibit a weaker warming trend than do station observations. In addition, the NNR trends among various station groups are nearly uniform, with a low range of C to C decade 1 (Table 3). [34] The annual OMR trends, which may be attributed to the intense UHI effect in the past 30 years, indicate strong warming in metropolises and large cities with averages of C and C decade 1, contributing % and %, respectively, to total warming. The OMR trends of medium sized cities, small cities, and suburban areas display moderate warming with C, C, and C decade 1, contributing %, %, and %, respectively, to total warming. The OMR trend for rural stations shows almost imperceptible warming with C decade 1. [35] The seasonal mean temperature trends from station observations and NNR of various station groups (Table 3) all appear to be most significant in winter, followed by spring and fall; relatively weak warming is observed in summer. In the four seasons, the strongest warming trends in the station observations all occur at the metropolis stations, while the weakest warming trends occur at rural stations; the NNR seasonal temperature does not show the same pattern. In addition, the seasonal OMR trends also depict strong warming in metropolises and large cities and weak warming in rural stations. Generally, the OMR trends in the winter half year are stronger than those in the summer half year. [36] From the difference in observational temperature trends between urban and rural stations (i.e., UMR, Table 4), the annual mean UMR trends are C decade 1 for metropolises, C decade 1 for large cities, C decade 1 for medium sized cities, and C decade 1 for suburban sites. These reflect a contribution of %, %, %, and %, respectively, to total warming. The lowest annual UMR trend can be seen over small cities at C decade 1, which accounts for % of total warming. Similarly, the seasonal mean of the UMR trends from the station observations are highest in metropolises and lowest in small cities. Compared with the OMR trend in Table 3, the UMR results are lower possibly because the temperature trends of rural stations close to urban sites are stronger than those reflected by the NNR, especially in metropolises, large cities, and suburban areas. The OMR results in Table 3 indicate that a mean warming of C decade 1 occurs over rural stations in east China. If this value is added to the UMR, the results will be quite consistent with those of the OMR. Therefore, both methods 9of12

10 each station to determine the contribution rate of urbanization on observational surface air temperature change in east China during (Figure 9). In the urban agglomeration in the Yangtze River Delta, urbanization has contributed more than 40% to climate warming since The rapid development of coastal cities in Zhejiang and Fujian has also contributed to recent warming. The contribution rates are also relatively large in Shandong Province because of its huge population. Note that the contributions of the UHI effect in many stations over Jiangxi Province are also quite large. Nonetheless, the reasons for such occurrences require further investigation. Figure 9. Percentage of contribution from urbanization to the surface air temperature change during and the difference in DMSP/OLS nighttime lights between 2007 and indicate that intense urbanization imposes significant effects on surface air temperature change in east China. [37] Of the four seasons, summer reflects the largest contribution of urban warming to total warming from the UMR, whereas winter shows the smallest contribution. The small contribution in winter indicates that the UHI is not a major contributor to the rapid wintertime warming in east China, which agrees with the result obtained by Ren et al. [2008] for north China. [38] Following the procedure of Kalnay and Cai [2003], we computed the mean OMR in each decade for every station (Figure 8). During , the mean OMR value for almost all the stations was the largest in comparison with those of the 1980s and 1990s. The mean OMR value shows a spatial pattern highly similar to the linear trends of the observational surface air temperature in Figure 5a and the OMR trends in Figure 5c. The OMR and the linear observational temperature trend both show significant surface warming induced by urbanization in east China, especially in the Yangtze River Delta and coastal areas, with the most intensive UHI effect occurring in the 2000s. [39] We computed the ratio between the OMR (Figure 5c) and the observational temperature trends (Figure 5a) for 4. Conclusions and Discussion [40] On the basis of the partly homogeneity adjusted monthly mean temperature data of 463 stations (including national stations and ordinary weather stations) located at elevations below 500 m, we conclude that the annual mean temperature over east China increased at a rate of C decade 1 from 1981 to With an increase rate of C decade 1 derived from the NNR during the same period, the contribution of urbanization and other land uses to overall regional warming is determined to be 24.22%. The results from the OMR method are generally quite consistent with those from the UMR. [41] In the current study, an objective and fast method was developed to dynamically classify urban and rural stations based on DMSP/OLS nighttime light data and GIS technology. Among the nonrural station groups, metropolis stations exhibit the strongest warming trends in annual and seasonal mean temperatures, as well as the most significant UHI effect. Annual mean urbanization warming reaches C decade 1 as detected by the OMR and C decade 1 as determined by the UMR, accounting for % of total warming as measured by the OMR and % as determined by the UMR. The annual mean UHI warming rates estimated for the other city station groups are also significant, with C, C, and C decade 1 for large cities, medium sized cities, and small cities as determined by the OMR and C, C, and C decade 1, respectively, as measured by the UMR. [42] Compared with studies on the UHI warming over other regions in China (Table 1), our investigation adopted much denser data sites. As a result, a more evident UHI effect on temperature trends was obtained. The clarity of results obtained may be attributed to the availability of more rural observational sites, reasonable calculation of background warming rate based on dynamic station classification approach, the choice of a time period with the most rapid urbanization, and consideration of the growing UHI effect under constant urban development. Moreover, there is very good agreement between the observed and NNR temperature anomaly over rural stations (Figure 7f), showing that good regional background temperature change was captured for both observations and the NNR over rural areas. Such results can also be attributed to the objective and dynamic classification of stations. [43] As meteorological stations in China were mostly set up near cities or towns, and finding rural stations completely free of UHI effects is difficult, the rural stations selected in the present study are the only currently available stations that are relatively less influenced by the UHI effects. Because of the limited coverage period of the DMSP/OLS 10 of 12

11 nighttime light data, the growing UHI effect before 1992 was excluded. In effect, the UHI warming trends and their contributions to the overall warming over east China provided in this paper can still be regarded as conservative. [44] Urban heat islands differ from city to city because of the varied features and background climatic characteristics of each site. Note that although the UHI is a reality, what matters is not the temperature bias introduced by the town but whether the bias changes over time, which can occur if the surroundings of the sites change slowly [Strangeways, 2009]. Therefore, detailed information on the sites and their surroundings is crucial when conducting climate research. Satellite observations of nighttime light emissions can provide comparatively objective information on urban development but cannot provide a broad range of details on the physical nature of the sites and their surroundings. The current study draws attention to an important issue in the evaluation and classification of weather station sites and in the investigation of the regional UHI effect. Extensive remotesensing observation of sites and their surroundings will be needed in future work. [45] The investigation of the effects of urbanization and other land use changes on local and regional scale climate change is an urgent requirement in some regions, such as China, where rapid urbanization has occurred. However, the role of land use such as urbanization in climate warming is a key climate related factor that has not been widely covered by the media. For instance, the mean surface air temperature trend of seven metropolis stations in Shanghai was C decade 1 from 1981 to 2007, whereas the NNR trend was C decade 1 in the same period, indicating an amplification of the background warming rate of 0.4 C decade 1 caused by the UHI effect. Therefore, for metropolises and large cities in east China, the significant contribution of urbanization to temperature change may be comparable to that of GHG concentration, suggesting that land surface processes can play a vital role in shaping future climate change [Feddema et al., 2005]. If such trends continue, certain metropolitan areas may experience a rate of warming well beyond the range projected by the global climate change scenarios of IPCC [Stone, 2007]. The increasing divergence between urban and rural surface temperature trends highlights the limitations of the response policy to climate change; these policies focus only on GHG reduction [Stone, 2009]. Policymakers need to address the impact of land use such as urbanization and deforestation on climate change in addition to that of GHG emissions. Serious measures for broadening the range of management strategies beyond GHG reductions and a land based mitigation framework should be included in the scheme for mitigating climate change [Betts, 2007; Pielke et al., 2002; Stone, 2009]. The results presented in the current work suggest that a more complete metrics for the representation of anthropogenic contributions to climate change should be developed. The effects of land surface conditions and other processes should be considered as well in climate change mitigation strategies in east China. [46] Acknowledgments. The authors are very grateful to the three anonymous reviewers for their helpful comments and constructive suggestions, which led to a significant improvement of the original manuscript. This work was supported by the Climate Change Special Foundation of the China Meteorological Administration (grant CCSF 09 10), National Natural Science Foundation of China (grants and ), and Special Fund for Meteorological Science and Technology of Zhejiang Province, China (grant KF ). References Betts, R. (2007), Implications of land ecosystem atmosphere interactions for strategies for climate change adaptation and mitigation, Tellus, Ser. B, 59(3), , doi: /j x. Du, Y., Z. Xie, Y. Zeng, Y. Shi, and J. Wu (2007), Impact of urban expansion on regional temperature change in the Yangtze River Delta, J. Geogr. Sci., 17(4), , doi: /s Easterling, D. R., et al. (1997), Maximum and minimum temperature trends for the globe, Science, 277, , doi: /science Fall, S., D. Niyogi, A. Gluhovsky, R. A. Pielke Sr., E. Kalnay, and G. Rochon (2010), Impacts of land use land cover on temperature trends over the continental United States: Assessment using the North American Regional Reanalysis, Int. J. Climatol., 30(13), , doi: /joc Fang, F., H. Bai, H. Zhao, and S. Yang (2007), The urbanization effect in northwestern China and its contributiontotemperaturewarming [in Chinese], Plateau Meteorol., 26(3), Feddema, J. J., K. W. Oleson, G. B. Bonan, L. O. Mearns, L. E. Buja, G. A. Meehl, and W. M. Washington (2005), The importance of land cover change in simulating future climates, Science, 310, , doi: / science Gallo, K. P., T. W. Owen, D. R. Easterling, and P. F. Jamason (1999), Temperature trends of the US historical climatology network based on satellite designated land use/land cover, J. Clim., 12(5), , doi: / (1999)012<1344:ttotus>2.0.co;2. Gong, D. Y., and S. W. Wang (2002), Uncertainties in the global warming studies [in Chinese], Earth Sci. Front., 9(2), Hansen, J., R. Ruedy, M. Sato, M. Imhoff, W. Lawrence, D. Easterling, T. Peterson, and T. Karl (2001), A closer look at United States and global surface temperature change, J. Geophys. Res., 106(D20), 23,947 23,963, doi: /2001jd He,C.Y.,P.J.Shi,J.G.Li,J.Chen,Y.Z.Pan,J.Li,L.Zhou,and T. Ichinose (2006), Restoring urbanization process in China in the 1990s by using non radiance calibrated DMSP/OLS nighttime light imagery and statistical data, Chin. Sci. Bull., 51(13), , doi: / s Heisler, G. M., and A. J. Brazel (2010), The urban physical environment: Temperature and urban heat islands, in Urban Ecosystem Ecology, edited by J. Aitkenhead Peterson and A. Volder, pp , Am. Soc. of Agron., Crop Sci. Soc. of Am., Soil Sci. Soc. of Am., Madison, Wis. Hu, Y., W. Dong, and Y. He (2010), Impact of land surface forcings on mean and extreme temperature in eastern China, J. Geophys. Res., 115, D19117, doi: /2009jd Hua, L. J., Z. G. Ma, and W. D. Guo (2008), The impact of urbanization on air temperature across China, Theor. Appl. Climatol., 93(3 4), , doi: /s Intergovernmental Panel on Climate Change (IPCC) (2007), Climate Change 2007: The Physical Science Basis: Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge Univ. Press, Cambridge, U. K. Jones, P. D., P. Y. Groisman, M. Coughlan, N. Plummer, W. C. Wang, and T. R. Karl (1990), Assessment of urbanization effects in time series of surface air temperature over land, Nature, 347, , doi: / a0. Jones, P. D., D. H. Lister, and Q. Li (2008), Urbanization effects in largescale temperature records, with an emphasis on China, J. Geophys. Res., 113, D16122, doi: /2008jd Kalnay, E., and M. Cai (2003), Impact of urbanization and land use change on climate, Nature, 423, , doi: /nature Kalnay, E., et al. (1996), The NCEP/NCAR 40 year reanalysis project, Bull. Am. Meteorol. Soc., 77, , doi: / (1996) 077<0437:TNYRP>2.0.CO;2. Karl, T. R., H. F. Diaz, and G. Kukla (1988), Urbanization: Its detection and effect in the United States climate record, J. Clim., 1(11), , doi: / (1988)001<1099:uidaei>2.0.co;2. Li, Q., H. Zhang, X. Liu, and J. Huang (2004), Urban heat island effect on annual mean temperature during the last 50 years in China, Theor. Appl. Climatol., 79(3 4), , doi: /s Li, Q., W. Li, P. Si, X. Gao, W. Dong, P. Jones, J. Huang, and L. Cao (2010), Assessment of surface air warming in northeast China, with emphasis on the impacts of urbanization, Theor. Appl. Climatol., 99(3 4), , doi: /s Lim, Y. K., M. Cai, E. Kalnay, and L. Zhou (2005), Observational evidence of sensitivity of surface climate changes to land types and urbanization, Geophys. Res. Lett., 32, L22712, doi: /2005gl of 12

12 Mesinger, F., et al. (2006), North American regional reanalysis, Bull. Am. Meteorol. Soc., 87(3), , doi: /bams Nuñez, M. N., H. H. Ciapessoni, A. Rolla, E. Kalnay, and M. Cai (2008), Impact of land use and precipitation changes on surface temperature trends in Argentina, J. Geophys. Res., 113, D06111, doi: /2007jd Peterson, T. C. (2003), Assessment of urban versus rural in situ surface temperatures in the contiguous United States: No difference found, J. Clim., 16(18), , doi: / (2003)016<2941: AOUVRI>2.0.CO;2. Pielke, R. A., Sr. (2005), Land use and climate change, Science, 310, , doi: /science Pielke, R. A., Sr., G. Marland, R. A. Betts, T. N. Chase, J. L. Eastman, J. O. Niles, D. D. S. Niyogi, and S. W. Running (2002), The influence of landuse change and landscape dynamics on the climate system: Relevance to climate change policy beyond the radiative effect of greenhouse gases, Philos. Trans. R. Soc. London, Ser. A, 360, , doi: / rsta Portman, D. A. (1993), Identifying and correcting urban bias in regional time series: Surface temperature in China s northern plains, J. Clim., 6, , doi: / (1993)006<2298:iacubi>2.0. CO;2. Ren, G., Y. Zhou, Z. Chu, J. Zhou, A. Zhang, J. Guo, and X. Liu (2008), Urbanization effects on observed surface air temperature trends in North China, J. Clim., 21(6), , doi: /2007jcli Seto, K. C., and J. M. Shepherd (2009), Global urban land use trends and climate impacts, Curr. Opin. Environ. Sustainability, 1, 89 95, doi: /j. cosust Shepherd, J. M. (2005), A review of current investigations of urbaninduced rainfall and recommendations for the future, Earth Interact., 9, 1 27, doi: /ei Stone, B., Jr. (2007), Urban and rural temperature trends in proximity to large US cities: , Int. J. Climatol., 27, , doi: /joc Stone, B., Jr. (2009), Land use as climate change mitigation, Environ. Sci. Technol., 43(24), , doi: /es902150g. Strangeways, I. (2009), Using Google Earth to evaluate GCOS weather station sites, Weather, 64(1), 4 8, doi: /wea.334. Tang, G., G. Ren, and J. Zhou (2008), Change of urban heat island intensity and its effect on surface mean air temperature records in southwest China [in Chinese], J. Appl. Meteorol. Sci., 17(6), Wang, W. C., Z. Zeng, and T. R. Karl (1990), Urban heat islands in China, Geophys. Res. Lett., 17(13), , doi: /gl017i013p Yang, X. C., Y. L. Zhang, L. S. Liu, W. Zhang, M. J. Ding, and Z. F. Wang (2009), Sensitivity of surface air temperature change to land use/cover types in China, Sci. China, Ser. D, 52(8), , doi: / s Zhang, J., W. Dong, L. Wu, J. Wei, P. Chen, and D. K. Lee (2005), Impact of land use changes on surface warming in China, Adv. Atmos. Sci., 22(3), , doi: /bf Zhou, L., R. E. Dickinson, Y. Tian, J. Fang, Q. Li, R. K. Kaufmann, C. J. Tucker, and R. B. Myneni (2004), Evidence for a significant urbanization effect on climate in China, Proc. Natl. Acad. Sci. U. S. A., 101(26), , doi: /pnas B. Chen and X. Yang, Shanghai Typhoon Institute of China Meteorological Administration, 166 Puxi Rd., Shanghai , China. (baode@mail.typhoon.gov.cn) Y. Hou, Shanghai Climate Center, Shanghai , China. 12 of 12