Revision as of 10:31, 15 May 2017 (view source) Scipediacontent (talk | contribs) (Created page with "==Abstract== In this paper, five national meteorological stations in Anhui province are taken as typical examples to explore the effects of local urbanization on their therma...")	Latest revision as of 10:34, 15 May 2017 (view source) Scipediacontent (talk | contribs) m (Scipediacontent moved page Draft Content 580093000 to Shi et al 2015b)
(No difference)

---

Latest revision as of 10:34, 15 May 2017

Abstract

In this paper, five national meteorological stations in Anhui province are taken as typical examples to explore the effects of local urbanization on their thermal environment by using Landsat data from 1990 to 2010. Satellite-based land use/land cover (LULC), land surface temperature (LST), normalized difference vegetation index (NDVI) are used to investigate the effects. The study shows that LULC around meteorological stations changed significantly due to urban expansion. Fast urbanization is the main factor that affects the spatial-temporal distribution of thermal environment around meteorological stations. Moreover, the normalized LST and NDVI exhibit strong inverse correlations around meteorological stations, so the variability of LST can be monitored through evaluating the variability of NDVI. In addition, station-relocation plays an important role in improving representativeness of thermal environment. Notably, the environment representativeness was improved, but when using the data from the station to study climate change, the relocation-induced inhomogeneous data should be considered and adjusted. Consequently, controlling the scale and layout of the urban buildings and constructions around meteorological stations is an effective method to ameliorate observational thermal environment and to improve regional representativeness of station observation. The present work provides observational evidences that high resolution Landsat images can be used to evaluate the thermal environment of meteorological stations.

Keywords

Urbanization ; Thermal environment ; Representativeness ; Land surface temperature ; Normalized difference vegetation index (NDVI)

1. Introduction

Driven by the recent thirty-year economic booming, China has undergone rapid development and urbanization. Many meteorological stations used to be in rural area now are in urban area and their observational environments have changed dramatically (Ren et al., 2010 ; Zhang et al., 2010 ; Ren and Ren, 2011 ; Shao et al., 2011 ; Yang et al., 2011 ; Yang et al., 2013  ;  Li et al., 2015 ). The changed thermal environment around meteorological stations significantly influences the observations (Yan et al., 2010 ; Li et al., 2012 ; Shao et al., 2011  ;  Ren and Ren, 2011 ), which will further disturb local weather and climate analysis, such as the evaluation of heat island effect, one of the main features of a modern city (Kalnay and Cai, 2003 ; Li et al., 2004 ; Zhou et al., 2004 ; Chen et al., 2007 ; Ren et al., 2007 ; Ren et al., 2008 ; Shi et al., 2011a ; Zhao et al., 2013  ;  Yang et al., 2013 ). For these reasons, objectively quantifying changes in thermal environment is crucial to evaluate the representativeness of meteorological stations.

Land surface temperature (LST) is an important index to represent the thermal environment around meteorological stations. Satellite remote sensing, as a recently developed technique, provides a unique opportunity to monitor and study macroscopical and dynamic-continuous LST in different spatial scales. Satellite-derived LST images have been widely used in land use classification, urban heat island research, thermal environment and hydrological investigation in an urban or even larger scales (Yang, 2000 ; Weng, 2001 ; Zhang et al., 2005 ; Zhang et al., 2011 ; Hung et al., 2006 ; Shi et al., 2011b  ;  Shi et al., 2013 ). However, only in recent years have satellite-derived LST data been used specifically for evaluating thermal environment and observational stations' representativeness in China. For example, by using 1-km resolution MODIS LST dataset, Wang et al. (2011) evaluated the representativeness of 142 weather stations and investigated the relationship between the representativeness of a station and its surrounding conditions. Ren et al. (2010) and Ren and Ren (2011) proposed to combine remote sensing images from Google Earth and MODIS LST data to evaluate the representativeness of observational stations and investigate the effect of urban heat islands. Using Landsat remote sensing data, Li et al. (2015) calculated the normalized LST and the heat effect contribution index (HECI) of different land use/land cover (LULC) type to classify the stations' observational environment into three types (urban, sub-urban and rural), and these two indexes could be used conveniently, effectively and quantitatively to choose a reference station when analyzing observational data in weather and climate research. However, thermal environment and representativeness around many Chinese meteorological stations are still not clear, especially the ones located in the southeastern China where has undergone rapid urbanization. Therefore, it is essential to use satellite-derived LST to monitor spatial-temporal variations of thermal environment and representativeness of these meteorological stations.

In this study, taking five typical meteorological stations as samples in Anhui province, a southeastern province in China, where the tremendous growth of urban sprawl, population, vehicles and economy have occurred since the 1990s (Shi et al., 2008 ; Li et al., 2012 ; Yang et al., 2011  ;  Yang et al., 2013 ), spatio-temporal variations of LULC and thermal environment (LST) around meteorological stations are systematically explored by using Landsat remote sensing data. Moreover, the effects of urbanization on the thermal environment of meteorological stations are quantitatively evaluated by using LULC change around these meteorological stations. Finally, the relationship between LST and normalized difference vegetation index (NDVI) are quantitatively investigated.

2. Data and method

2.1. Data

The Landsat-5 remote sensing data used in this study are obtained from the Open Spatial Data Sharing Project, which was launched by the Institute of Remote Sensing and Digital Earth of the Chinese Academy of Sciences (http://ids.ceode.ac.cn/en/ ). The TM (ETM) sensor which is on board of the Landsat-5 satellite has seven bands, and the sixth band (TM6, with the band wavelength 10.40–12.50 μm) is selected here for LST retrieval (Li et al., 2015 ). In addition, the fourth band (TM4, with the near infrared band wavelength 0.62–0.69 μm) and the third band (TM3, with the red band wavelength 0.76–0.96 μm) are also selected here for NDVI retrieval (Shi et al., 2011b ). In order to better capture the local vegetation information in a similar season, only data during the vegetation growth periods (i.e., May to September in China) in 1990, 2000 and 2010 are selected.

To investigate the temporal-spatial variability of the meteorological stations' surrounding environment, the LULC types in the buffer zone of the station are categorized into three types, vegetation (including farmland, forest and grass land), water (including lakes, rivers and pools), and construction (including buildings and roads), which are derived exactly through supervised classification together with visual interpretation (Li et al., 2015 ).

Five stations under rapid urbanization are selected as typical cases (Fig. 1 ), i.e., Suzhou (SZ), Hefei (HF), Chuzhou (CZ), Anqing (AQ) and Wuhu (WH); and five reference stations in rural areas are also selected corresponding to the five urban stations, i.e., Lingbi (LB), Feixi (FX), Quanjiao (QJ), Huaining (HN) and Nanling (NL). In addition, the underlying surfaces around the five reference stations experienced little changes during 1990–2010. The annual mean air temperatures for the period 1980–2010 were recorded at the above mentioned 10 meteorological stations in Anhui province.

Fig. 1. The locations of five urban meteorological stations (indicated by stars), i.e., Suzhou (SZ), Hefei (HF), Chuzhou (CZ), Anqing (AQ) and Wuhu (WH), and five reference stations (indicated by triangle), i.e., Lingbi (LB), Feixi (FX), Quanjiao (QJ), Huaining (HN) and Nanling (NL) in Anhui province, China.

2.2. Method

2.2.1. Critical radius for the buffer zone around station

The buffer zone here is defined as the area within 2 km from the station based on two following reasons: 1) In Anhui province, through field survey it has been found that most of the meteorological stations were initially built 2 km away from the edge of cities. Therefore, by studying the 2 km radius area of a station, the influence of city expansion on the stations' observational environment can be captured. 2) According to recent studies (Ren et al., 2010 ; Yang et al., 2013  ;  Li et al., 2015 ), the influence of urbanization on a meteorological station in China can be reflected by the observational environment within its 2-km radius. Therefore, 2 km is chosen as the critical radius for the buffer zone.

2.2.2. Retrieval of LST

Brightness temperature is usually retrieved from thermal infrared band data. First, digital number values (Q_DN ) of TM6 images are converted to thermal radiation intensity (R ) ( Weng, 2003 ), following,

${\displaystyle R={\frac {Q_{\mbox{DN}}}{Q_{\mbox{MAX}}}}\left(R_{\mbox{MAX}}-\right.}$${\displaystyle \left.R_{\mbox{MIN}}\right)+R_{\mbox{MIN}}}$

( 1)

Here, Q_DN is the digital number of a pixel in the TM6 image. R_MAX and R_MIN are the maximum and minimum radiation intensity received by TM6, which are set as 1.896 and 0.1534 mW cm⁻²  sr⁻¹  mw⁻¹  μm⁻¹ , respectively. The maximum and minimum digital number Q_MAX and Q_MIN are 255 and 0, respectively. As such, Equation (1) can be written as:

${\displaystyle R=0.1534+0.006833725Q_{\mbox{DN}}}$

( 2)

Brightness temperature T_B (units: K) for a black body is then calculated following:

${\displaystyle T_{\mbox{B}}={\frac {K_{2}}{ln\left(K_{1}/R+1\right)}}}$

( 3)

Here, K₁  = 60.776 and K₂  = 1260.56. However, the Earth is not a black body and has different underlying surface conditions at difference places. LST, which is different from T_B , is then calculated using ( Artis and Carnahan, 1982  ;  Weng, 2001 )

${\displaystyle LST={\frac {T_{\mbox{B}}}{1+\left(\lambda T_{\mbox{B}}/\rho \right)ln\quad \epsilon }}}$

( 4)

Here, λ (≈11.5 μm) is the effective wavelength of TM6. ρ  = hc/σ  = 1.438 × 10⁻²  mK, where σ is the Boltzmann constant (1.38 × 10⁻²³  J K⁻¹ ), h is the Planck constant (6.626 × 10⁻³⁴  J s) and c is the light velocity (2.998 × 10⁸  m s⁻¹ ). ε is emissivity whose value is 0.9925, 0.95 and 0.92 for water, vegetation and construction surfaces, respectively ( Nichol, 1994 ).

2.2.3. Normalized LST

To reveal the effects of changing underlying surface conditions on the stations' surrounding thermal environment during the recent 20 years, it is important to exclude the influence of background weather conditions. To do so, LST is normalized as I_i following Li et al. (2015) :

${\displaystyle I_{i}={\frac {T_{i}-T_{min}}{T_{max}-T_{min}}}}$

( 5)

Where T_i is the LST of pixel i . T_max and T_min are the maximum and minimum LST in the considering area. I_i is the normalized LST index for pixel i , which ranges 0–1. With the data of all the pixels in the 2 km buffer zones of the 5 stations, I is categorized into 5 zones (i.e., high, sub-high, medium, sub-low and low temperature zones) using the Natural Breaks method in ArcGIS ( Table 1 ).

Table 1. Five zones of the thermal environment categorized by normalized LST.

Normalized LST	Thermal environment zone
[0, 0.276)	Low temperature zone
[0.277, 0.411)	Sub-low temperature zone
[0.412, 0.522)	Medium temperature zone
[0.523, 0.643)	Sub-high temperature zone
[0.644, 1]	High temperature zone

In addition, linear regression was applied to calculate the relationship between normalized LST and NDVI around the five stations in 1990, 2000, and 2010. Due to 60-m resolution of TM6, about 3000 pixels can be obtained in the 2 km buffer zones. After qualities controlled, about 2600 pixels in the 2 km buffer zones per panel are selected as samples for linear regression at each station. All correlation coefficients were tested using a t -distribution. A p value associated with this test was applied.

2.2.4. Heat effect contribution index (HECI)

In the surrounding area of an observational station, various land use types contribute differently to the thermal environment. To quantitatively evaluate the heat contribution from each type of land use, an index (HECI ) is introduced following Li et al. (2015) :

${\displaystyle HECI_{i}={\frac {\sum _{j=1}^{N_{i}}T_{ij}}{{\overline {T}}N}}\times 100\%}$

( 6)

Where ${\textstyle {\overline {T}}}$ is the average LST of the whole area. N is the total pixel number within the area. T_ij is the LST of type i LULC in its j th pixel. N_i is the pixel number of the area covered by type i LULC. HECI_i represents the heat contribution LULC from type i , and it varies from 0 to 100%. Larger HECI_i means larger contribution of type i LULC to the thermal environment. The sum of HECI_i overall types LULC equals 100%.

3. Results

3.1. Impacts of urbanization on LULC around meteorological stations

Fig. 2 shows the temporal variation of LULC in the 2 km buffer zone of the five urban stations. Generally, it can be seen that the LULC changed significantly primarily due to urban expansion. Taking CZ as an example, in 1990, the construction area covered approximately 30% of the buffer zone (Fig. 2 a), which then increased to approximately 45% in 2000 (Fig. 2 b), while to approximately 70% in 2010 (Fig. 2 c). Similarly, it is clear that other four stations have been in urban areas due to the tremendous urban construction sprawl since the 1990. In fact, they were in rural areas before the 1980s. Ten years later, the construction areas of the buffer zone increased continuously, accompanying with vegetation and water reductions around these four stations (Fig. 2 b). Until 2010, SZ and AQ had entered into city passively, except HF and WH (Fig. 2 c). It is evident that the measurements can be strongly affected by surrounding buildings. Therefore, in order to avoid disturbances from urban developments, HF and WH moved to rural areas in 2004 and 2006, respectively. After their relocations, vegetation areas increased significantly, while construction decreased drastically.

Fig. 2. LULC around the 2 km buffer zones of the five urban stations.

3.2. Impacts of LULC on thermal environment around observational stations

Fig. 3 shows the five-level thermal environment based on the normalized LST in the 2 km buffer zones of the five stations. For SZ/AQ/CZ, the high and sub-high temperature zones with the construction LULC type kept increasing significantly, while the area of low, sub-low and medium temperature zones, as well as the vegetation and water LULC type kept decreasing from 1990 to 2010 (Figs. 2 and 3 ). As a result, the HECI of construction for SZ/AQ/CZ also kept increasing, e.g., 65.47%/54.20%/48.50% in 1990, 80.96%/71.37%/50.66% in 2000, and 89.81%/80.17%/61.32% in 2010, respectively (Table 2 ), while that of vegetation and water also kept decreasing. Similarly, increase/decrease trends for thermal environment of construction/vegetation and water are found for HF and WH during 1990–2000 (Fig. 3 , Table 2 ). However, after their relocation, the high and sub-high temperature zone and the construction LULC type in the 2 km buffer zone were drastically reduced, and the low, sub-low and medium temperature zone, as well as the vegetation and water LULC type were increased. The HECI of construction for HF and WH were 69.72% and 79.04% in 2000, while decreased to 21.90% and 29.97% in 2010, respectively (Table 2 ). The above results clearly indicate that the observational environment of a meteorological station can be greatly influenced by LULC under local fast urbanization.

Fig. 3. Same as Fig. 2 but for thermal environment based on normalized LST. LT, SLT, MT, SHT and HT represent low, sub-low, medium, sub-high and high temperature zone, respectively.

Table 2. The HECI of different LULC type around stations (unit: %).

HECI	SZ			HF			CZ			AQ			WH
	1990	2000	2010	1990	2000	2010	1990	2000	2010	1990	2000	2010	1990	2000	2010
Construction	65.5	81.0	89.8	50.6	69.7	21.9	48.5	50.7	61.3	54.2	71.4	80.2	63.3	79.0	30.0
Vegetation	29.0	12.9	5.5	40.5	22.6	73.7	46.1	42.8	33.2	32.3	14.4	13.1	26.4	13.2	64.5
Water body	5.6	6.1	4.7	8.9	7.7	4.4	5.4	6.5	5.5	13.5	14.2	6.7	10.3	7.8	5.5

Overall, in low and sub-low temperature zones, vegetation and water are the main contributors, while in sub-high and high temperature zones, construction becomes dominant. In the medium temperature zone, vegetation and construction are mixed. Consequently, controlling the scale and layout of the urban buildings around the stations is an effective method to ameliorate observation thermal environment and to improve region representativeness of station observation.

3.3. Relationship between normalized LST and NDVI around meteorological stations

NDVI, as an important indicator of LULC, can be derived by visible/infrared (VIR) sensors aboard most satellites with the opportunity of high sampling frequency. Therefore, it is very important to investigate the relationship between LST and NDVI to study the thermal environment around meteorological stations, which can be further used to effectively improve and populate the technique that using VIR measurements to monitor the thermal environment around meteorological stations. Previous studies indicated that LST and NDVI were significantly negative correlation in urban areas (Shi et al., 2011b ; Shi et al., 2013  ;  Zhang et al., 2011 ). To assess quantitative relationship between NDVI and LST around meteorological stations, the linear regressions of NDVI on the normalized LST in the 2 km buffer zone of each station are shown for the year 1990, 2000, and 2010, respectively (Fig. 4 ). In general, the normalized LST and NDVI exhibit great negative correlations, with the normalized LST increasing while NDVI decreasing around the stations (Fig. 4 ). The correlation coefficients are statistically significant at the 99% confidence level. These results indicate that NDVI can reflect the changes of thermal environment conditions in respond to LULC changes around meteorological stations affected by urbanization. Therefore, the variability of thermal environment could also be monitored through evaluating the variability of NDVI.

Fig. 4. The linear regressions of NDVI and normalized LST in 1990, 2000 and 2010 in the 2 km buffer zones for the five urban stations. The linear regression coefficient a and correlation coefficient r are shown in the panels. All the correlations are significant at the 99% confidence levels. About 2600 pixels per year in the 2 km buffer zones are selected for sample number at each station.

4. Discussion

Different remote-sensing products have different advantages and disadvantages. For example, although some historical images are missing, images provided by Google Earth usually have high resolution (approximately 15 m) and can be used to evaluate the representativeness of observational stations in recent years (Ren et al., 2010  ;  Ren and Ren, 2011 ). Large quantities of MODIS LST products have been gathered for more than 14 years since the launch of MODIS, and they provide a unique opportunity to study the long-term representativeness of observational stations (Wang et al., 2011  ;  Ren and Ren, 2011 ). However, the resolution of MODIS LST products is 1 km, which is too coarse for investigating the surrounding conditions within a stations 1 km buffer zone. Moreover, because of the low spatial resolution of these products, it is difficult to precisely construct the relationship between LULC and LST. In contrast, remote-sensing images provided by the Landsat project have both high spatial resolution (approximately 30 or 60 m) and long-term coverage (from 1979 to present). Landsat images have been widely used for land use classification and thermal environmental assessment around stations on a 1-km or smaller scale (Shao et al., 2011 ; Yang et al., 2013  ;  Li et al., 2015 ). In the present study, Landsat images were reliable for constructing a relationship between LULC (or NDVI) and LST. Consequently, we found advantages in the combined LULC and LST method for station-type classification.

To evaluate urbanization-related warming, Fig. 5 shows the time series of the differences between the observed temperatures at five urban stations and at reference stations between 1980 and 2010. These differences can be treated as the intensities of UHI for urban stations (Liu et al., 2007  ;  Ren et al., 2008 ). The figure clearly shows that UHI values for the five urban stations increased with time over the past 20 years. In particular, the UHI values for the SZ, CZ, and AQ stations continued to increase significantly after 2000; however, this was not observed for the HF and WH stations because of their relocation. This implies that urbanization-related warming cannot be ignored (Ren and Zhou, 2014  ;  Ren et al., 2014 ). Wang et al. (2010) also indicated that the urban heat island effect and the decadal variation in ocean thermohaline circulation were responsible for the continuing warming in China, although a global warming hiatus occurred after 2000. In addition, Stewart and Oke (2012) pointed out that the effects of terrain and local wind patterns would influence the representativeness of the stations and the local UHI. Therefore, the exact mechanisms underlying warming in China are complex and can vary from place to place. The analysis and separation of these mechanisms are beyond the scope of this paper but will be addressed in future research.

Fig. 5. Time series of the differences (UHI) between the observed LST of the five urban stations and that of the reference stations during 1980–2010.

Abrupt changes (i.e., significant drops) in UHI values occurred at the HF (Fig. 5 b) and WH stations (Fig. 5 e) in 2004 and 2006, respectively, as indicated by the arrows. These detectible inhomogeneities were mainly associated with the station relocations in 2004 and 2006, which were also consistent with the LULC and thermal environment changes (Table 1 , Fig. 2  ;  Fig. 3 ). After relocation, the environmental representativeness improved, but the inhomogeneities in the data must be taken into account. There are methods available for adjusting some of these inhomogeneities (Yan et al., 2010 ; Li and Yan, 2010 ; Cao and Yan, 2012 ; Yang et al., 2013  ;  Yang and Li, 2014 ).

We selected LST rather than in situ air temperatures for the present study because of the fact that LST can be monitored using macroscopic, large-scale, dynamic, continuous satellite remote-sensing images. In contrast, in situ air temperatures, as observed by meteorological stations, are taken from a single space point and therefore have weak spatial representativeness, particularly over heterogeneous urban surfaces. In fact, air temperature and LST are closely related, and air temperature can be estimated from satellite-based LST using the temperature/vegetation index (TVX) method and a statistical regression approach (Prihodko and Goward, 1997 ; Florio et al., 2004 ; Stathopoulou et al., 2006  ;  Shen and Leptoukh, 2011 ). Therefore, the use of LST rather than in situ air temperatures was more advantageous in the present study.

5. Summary

We used Landsat-based LULC, LST, NDVI values to study the effects of local urbanization on the thermal environments of five national meteorological stations in the Anhui province from 1990 to 2010. The results obtained in this study are summarized as follows:

First, LULC around the observational stations changed significantly because of rapid local urban expansion between 1990 and 2010. Construction areas increased continuously, accompanied by reductions in vegetation and water bodies. To avoid disturbances from these urban developments, the HF and WH stations were moved to new locations in 2004 and 2006, respectively. Therefore, in the areas surrounding these two stations, vegetation increased and construction decreased after the relocation, accompanied by improved environmental representativeness and inhomogeneity in the data.

Second, the observational environment of a meteorological station can be greatly influenced by LULC changes. In particular, vegetation and water are the main contributors in low and sub-low temperature zones, while construction is dominant in sub-high and high temperature zones. In the medium temperature zone, the influences of vegetation and construction are mixed.

Finally, the normalized LST and NDVI values significantly exhibit negative correlations, with LST increasing and NDVI decreasing around the stations. Therefore, it is reasonable to use NDVI as another indicator of thermal environmental conditions around meteorological stations. By evaluating the variability of NDVI values, the variability of the thermal environment can also be monitored. The scale and layout of construction around a station can be changed by its relocation, and this is an effective way to improve the observational thermal environment and to improve regional representativeness of station observations. In addition, when choosing a new site for a station, this study provides a method for examining the thermal environment around the intended location with high-resolution Landsat images.

Acknowledgements

This study was supported by the National Natural Science Foundation of China (41205126 and 41475085 ), Anhui Provincial Natural Science Foundation (1408085MKL60 and 1508085MD64 ) and Meteorological Research Fund of Anhui Meteorological Bureau (KM201520 ). We also appreciated the constructive comments and suggestions by the editors and two anonymous reviewers.

References

1. Artis and Carnahan, 1982 D.A. Artis, W.H. Carnahan; Survey of emissivity variability in thermography of urban areas; Remote Sens. Environ., 12 (4) (1982), pp. 313–329
2. Cao and Yan, 2012 L.-J. Cao, Z.-W. Yan; Progress in research on homogenization of climate data; Adv. Clim. Change Res., 3 (2) (2012), pp. 59–67
3. Chen et al., 2007 Z.-H. Chen, H.-J. Wang, G.-Y. Ren; Asymmetrical change of urban heat island intensity in Wuhan, China; Adv. Clim. Change Res., 3 (5) (2007), pp. 282–286 (in Chinese)
4. Florio et al., 2004 E.N. Florio, S.R. Lele, C.Y. Chang, et al.; Integrating AVHRR satellite data and NOAA ground observations to predict surface air temperature: a statistical approach; Int. J. Remote Sens., 25 (15) (2004), pp. 2797–2994
5. Hung et al., 2006 T. Hung, D. Uchihama, S. Ochi; Assessment with satellite data of the urban heat island effects in Asian mega cities; Int. J. Appl. Earth Obs., 8 (1) (2006), pp. 34–38
6. Kalnay and Cai, 2003 E. Kalnay, M. Cai; Impact of urbanization and land use change on climate; Nature, 423 (6939) (2003), pp. 528–531
7. Li et al., 2004 Q. Li, H. Zhang, X. Liu, et al.; Urban heat island effect on annual mean temperature during the last 50 years in China; Theor. Appl. Climatol., 79 (3–4) (2004), pp. 165–174
8. Li et al., 2015 Y.-B. Li, T. Shi, Y.-J. Yang, et al.; Satellite-based investigation and evaluation of the observational environment of meteorological stations in Anhui province, China; Pure Appl. Geophys., 172 (6) (2015), pp. 1717–1733 http://dx.doi.org/10.1007/s00024-014-1011-8
9. Li and Yan, 2010 Z. Li, Z. Yan; Application of multiple analysis of series for homogenization (MASH) to Beijing daily temperature series 1960–2006; Adv. Atmos. Sci., 27 (4) (2010), pp. 777–787 http://dx.doi.org/10.1007/s00376-009-9052-0
10. Li et al., 2012 Z.-H. Li, J. Yang, C.-E. Shi, et al.; Urbanization effects on fog in China: field research and modeling; Pure Appl. Geophys., 169 (5–6) (2012), pp. 927–939
11. Liu et al., 2007 W. Liu, C. Ji, J. Zhong, et al.; Temporal characteristics of the Beijing urban heat island; Theor. Appl. Clim., 87 (1–4) (2007), pp. 213–221 http://dx.doi.org/10.1007/s00704-005-0192-6
12. Nichol, 1994 J.E. Nichol; A GIS-based approach to microclimate monitoring in Singapores high-rise housing estates; Photogramm. Eng. Rem. S., 60 (10) (1994), pp. 1225–1232
13. Prihodko and Goward, 1997 L. Prihodko, S.N. Goward; Estimation of air temperature from remotely sensed surface observations; Remote Sens. Environ., 60 (3) (1997), pp. 335–346
14. Ren and Zhou, 2014 G.-Y. Ren, Y.-Q. Zhou; Urbanization effect on trends of extreme temperature indices of national stations over mainland China, 1961–2008; J. Clim., 27 (6) (2014), pp. 2340–2360
15. Ren et al., 2007 G.-Y. Ren, Z.-Y. Chu, Z.-H. Chen, et al.; Implications of temporal change in urban heat island intensity observed at Beijing and Wuhan stations; Geophys. Res. Lett., 34 (2007) http://dx.doi.org/10.1029/2006GL027927
16. Ren et al., 2008 G.-Y. Ren, Y.-Q. Zhou, Z.-Y. Chu, et al.; Urbanization effects on observed surface air temperature trends in North China; J. Clim., 21 (6) (2008), pp. 1333–1348
17. Ren et al., 2010 G.-Y. Ren, A.-Y. Zhang, Z.-Y. Chu, et al.; Principles and procedure for selecting reference surface air temperature stations in China; Meteor. Sci. Technol., 38 (1) (2010), pp. 78–85 (in Chinese)
18. Ren et al., 2014 G.-Y. Ren, Y.-Y. Ren, Q.-X. Li, et al.; An overview on global land surface air temperature change; Adv. Earth Sci., 29 (8) (2014), pp. 934–946 (in Chinese)
19. Ren and Ren, 2011 Y.-Y. Ren, G.-Y. Ren; A remote-sensing method of selecting reference stations for evaluating urbanization effect on surface air temperature trends; J. Clim., 24 (13) (2011), pp. 3179–3189
20. Shao et al., 2011 Q.-Q. Shao, C.-Y. Sun, J.-Y. Liu, et al.; Impact of urban expansion on meteorological observation data and overestimation to regional air temperature in China; J. Geogr. Sci., 21 (6) (2011), pp. 994–1006
21. Shen and Leptoukh, 2011 S.-H. Shen, G.-G. Leptoukh; Estimation of surface air temperature over central and eastern Eurasia from MODIS land surface temperature; Environ. Res. Lett., 6 (4) (2011) http://dx.doi.org/10.1088/1748–9326/6/4/045206
22. Shi et al., 2008 C. Shi, M. Roth, H. Zhang, et al.; Impacts of urbanization on long-term variation of fog in Anhui province, China; Atmos. Environ., 42 (11) (2008), pp. 8484–8492
23. Shi et al., 2011a T. Shi, Y.-J. Yang, Y.-L. Jiang, et al.; Impact of the variation of urban heat island intensity on temperature series in Anhui province; Clim. Environ. Res., 16 (6) (2011), pp. 779–788 http://dx.doi.org/10.3878/j.issn.1006-9585.06.13 (in Chinese)
24. Shi et al., 2011b T. Shi, Y.-J. Yang, A.-M. Zhang, et al.; Study of thermal environment of Hefei city based on TM and GIS; Remote Sens. Technol. Appl., 26 (2) (2011), pp. 156–162 (in Chinese)
25. Shi et al., 2013 T. Shi, Y.-J. Yang, J. Ma, et al.; Study on spatial-temporal characteristics of urban heat island in Anhui province based on MODIS: typical cases study; J. Appl. Meteror. Sci., 24 (4) (2013), pp. 484–494 (in Chinese)
26. Stathopoulou et al., 2006 M. Stathopoulou, C. Cartalis, N. Chrysoulakis; Using midday surface temperature to estimate cooling degree-days from NOAA-AVHRR thermal infrared data: an application for Athens, Greece; Sol. Energy, 80 (4) (2006), pp. 414–422
27. Stewart and Oke, 2012 I.D. Stewart, T.R. Oke; Local climate zones for urban temperature studies; Bull. Amer. Meteorol. Soc., 93 (12) (2012), pp. 1879–1900
28. Wang et al., 2010 S.-W. Wang, X.-Y. Wen, Y. Luo, et al.; Does the global warming pause in the last decade: 1999–2008?; Adv. Clim. Change Res., 1 (1) (2010), pp. 49–54
29. Wang et al., 2011 Y.-Y. Wang, G.-C. Li, Y. Zhang; Regional representativeness analysis of national reference climatological stations based on MODIS/LST product; J. Appl. Meteror. Sci., 22 (2) (2011), pp. 214–220 (in Chinese)
30. Weng, 2001 Q. Weng; A remote sensing-GIS evaluation of urban expansion and its impact on surface temperature in the Zhujiang Delta, China; Int. J. Remote Sens., 22 (10) (2001), pp. 1999–2014
31. Weng, 2003 Q. Weng; Fractal analysis of satellite-detected urban heat island effect; Photogramm. Eng. Rem. S., 69 (5) (2003), pp. 555–565
32. Yan et al., 2010 Z. Yan, Z. Li, Q. Li, et al.; Effects of site-change and urbanisation in the Beijing temperature series 1977–2006; Int. J. Climatol., 30 (8) (2010), pp. 1226–1234
33. Yang, 2000 L. Yang; Integration of a numerical model and remotely sensed data to study urban/rural land surface climate processes; Comput. Geosci., 26 (4) (2000), pp. 451–468
34. Yang and Li, 2014 S. Yang, Q.-X. Li; Improvement in homogeneity analysis method and update of China precipitation data; Prog. Inquisitiones De. Mutat. Clim., 10 (4) (2014), pp. 276–281 (in Chinese)
35. Yang et al., 2011 Y.-J. Yang, T. Shi, W.-A. Tang, et al.; Study of observational environment of meteorological station based on remote sensing: cases in six stations of Anhui province; Remote Sens. Technol. Appl., 26 (6) (2011), pp. 791–797 (in Chinese)
36. Yang et al., 2013 Y.-J. Yang, B.-W. Wu, C.-E. Shi, et al.; Impacts of urbanization and station-relocation on surface air temperature series in Anhui province, China; Pure Appl. Geophys., 170 (11) (2013), pp. 1969–1983
37. Zhang et al., 2010 A.-Y. Zhang, G.-Y. Ren, J.-X. Zhou, et al.; On the urbanization effect on surface air temperature trends over China; Acta Meteor. Sin., 68 (6) (2010), pp. 957–966 (in Chinese)
38. Zhang et al., 2011 H.-Q. Zhang, Y.-J. Yang, S.-P. Xun, et al.; Seasonal and spatial variability of vegetation and land surface temperature in Anhui province; J. Appl. Meteror. Sci., 22 (2) (2011), pp. 232–240 (in Chinese)
39. Zhang et al., 2005 J.-H. Zhang, Y.-Y. Hou, G.-C. Li, et al.; The diurnal and seasonal characteristics of urban heat island variation in Beijing city and surrounding areas and impact factors based on remote sensing satellite data; Sci. China Earth Sci., 48 (2005), pp. S220–S229
40. Zhao et al., 2013 Z.-C. Zhao, Y. Luo, J.-B. Huang; Are there impacts of urban heat island on future climate change?; Adv. Clim. Change Res., 4 (2) (2013), pp. 133–136
41. Zhou et al., 2004 L.M. Zhou, E.D. Robert, Y.H. Tian, et al.; Evidence for a significant urbanization effect on climate in China; Proc. Natl. Acad. Sci. U. S. A., 101 (26) (2004), pp. 9540–9544