Sensors & Transducers, Vol. 180, Issue 10, October 2014, pp. 67-71 SSSeeennnsssooorrrsss &&& TTTrrraaannnsssddduuuccceeerrrsss © 2014 by IFSA Publishing, S. L. http://www.sensorsportal.com

Correlation Analysis of Urban Land Surface Temperature and Fluxes Based on Remote Sensing Technology

1, 2 Wen-Xia QIU 1 University of Technology, Chengdu, 2 Institute of Engineering Surveying, College of Architectural Technology, , China E-mail: [email protected]

Received: 24 June 2014 /Accepted: 30 September 2014 /Published: 31 October 2014

Abstract: Land surface temperature and fluxes (soil heat flux, sensible heat flux and latent heat flux) of the city in Sichuan Province on September 20, 2007, were retrieved using Landsat ETM+ images. Then based on sampling points in study area, surface temperature and each flux were compared with scatter diagram, and simulated the function to explore their correlation. The research results are the following: 1) Land surface temperature had positively correlation with both soil heat flux and sensible heat flux, but negatively correlation with latent heat flux. 2) Surface temperature was absolutely affected by sensible heat flux. Copyright © 2014 IFSA Publishing, S. L.

Keywords: Land surface temperature, Soil heat flux, Sensible heat flux, Latent heat flux, Correlation analysis.

1. Introduction fluxes has not yet been studied specially. Based on this, taking the Nanchong city in Sichuan Province as In recent years, with the development of an example, the correlation between surface urbanization, the scale of city has expanded temperature and heat fluxes, which was retrieved continuously, the population of which has increased respectively using remote sensing technology, was sharply, and large amount of natural surface, for explored to provide basis for research work of Urban example, woodland, farmland and grassland, etc, Heat Island. have changed into building, road, etc., which uses cement as main materials, in addition a lot of heat, which is from cars, computers, air conditionings, 2. Research Region and Data Source refrigerators, mobile phones, and IPADs, etc., has discharged into air. Theses have caused obvious Heat The Nanchong City lies in the northeast of Island Effects, which is seriously affecting city Sichuan Basin, and locates in the middle reaches of people’s quality of life. the Jialing River (see Fig. 1). The topography is At present, there are two methods of urban heat mainly hilly, and the altitude is from 26 to island research using remote sensing technology. The 480 meters. The research region is the urban area of first is retrieving land surface temperature, the the Nanchong city, which locates in the west of the second is retrieving land surface fluxes. But the Jialing River and the east of the Xi River, including relationship between surface temperature and heat the Shunqing , the and the http://www.sensorsportal.co m/HTML/DIGEST/P_2440.htm 67 Sensors & Transducers, Vol. 180, Issue 10, October 2014, pp. 67-71

Gaoping District. The and the Lmax  Lmin Jialing District lies in the west of the Jialing River, L    (Q  Q )  L , (1)  Q Q  min min  and the Gaoping District lies in the east of the Jialing max min River. The Jialing River flows through the east side of the urban from north to south. The urban area is where λ is the band value, Lλ is the spectral radiance -2 -1 -1 surrounded by the highway. The highway from by the sensor (W·m ·sr ·μm ), Qλ is the digital Chengdu to Nanchong runs across the south number of analyzed pixel, Qmax is the maximum of urban area. L recorded (255), Qmin is the minimum recorded, max and L are the maximum and minimum spectral min radiance, detected for Qmin and Qmax.

3.2. Radiation Calibration

1) Retrieving brightness temperature (BT). Based on spectral radiation value of pixels on sensor, BT can be calculated directly by Planck's radiation function or an approximation Formula (2) [2-3].

T=K2/ln(1+K1/Lλ), (2)

where T is the BT of pixels and its unit is K, K1 and K2 are the pre-launch calibration constants, as for -2 -1 -1 ETM+ band62, K1 is 666.093W·m ·ster ·μm , and K2 is 1282.708 K. 2) Retrieving land surface temperature (LST). LST can be calculated according to Formula (3) [4].

Trad , Ts  (3) 1 ( T / )ln rad

where Ts is the LST and its unit is K; Trad is the BT; λ is the center wavelength (11.4 μm);   h c / , where h is the Planck constant ( 6.6261034 J  s ), c is the velocity of light ( 2.998108 m / s ),  is the Boltzmann constant Fig. 1. General situation of research region. ( 1.381023 J / K ); ε is the surface emissivity,

when NDVI<0.05, ε=0.973, when NDVI>0.7,

This research used Landsat-7 ETM+ data ε=0.99, when 0.05≤NDVI≤0.7,   0.004Pv 0.986, obtained on September 20, 2007, DEM and where P is the vegetation proportion in pixel, atmosphere temperature. The projection of the ETM+ v , images was UTM 48N, the ellipsoid and reduced Pv  ( NDVI  NDVI s ) /( NDVI v  NDVI s ) plane of which was WGS84, the spatial resolution of NDVI  (   ) /(   ) , 3 and 4 are the which was 30 m. The quality of images was good. 4 3 4 3 surface reflectance acquired in the visible (red) and DEM was downloaded from Global Mapper, and its near-infrared band, respectively [5-6]. (0.7) projection and resolution was same as ETM+ images. NDVI v

and NDVI s (0.05) stand for the NDVI value of vegetation and bare land. 3. Retrieving Land Surface Temperature

3.1. Radiation Calibration 4. Retrieving Land Surface Fluxes

Radiation calibration is a process of This research is mainly based on the surface converting the digital value of remote sensing data energy balance Equation (4) [7]. into spectral radiance value of sensor. The model R  LE  H  G  PH , (4) is Formula (1) [1]. n

68 Sensors & Transducers, Vol. 180, Issue 10, October 2014, pp. 67-71 where R is the net radiation, LE is the latent heat formula. In this paper, the soil heat flux is calculated n based on the Formula (6) [12] proposed by flux, H is the sensible heat flux, G is the soil heat Bastiaanssen in 2000. 2 flux, and the unit is w / m . PH is the energy for plant photosynthesis and biomass increased, and can G  R T (0.0038  0.0074 2 )  n s , (6) be neglected in actual calculation because its value is (1 0.98NDVI 4 ) / very small.

where Rn is the surface net radiation, TS is the surface temperature,  is the surface albedo. 3) Estimating sensible heat flux (H). The sensible heat flux is the exchange energy between land surface and atmosphere, and is determined by the surface temperature, are temperature and air resistance. The model is Formula (7) [12].

 C  dT H  air p , (7) r ah

  349.635 (T  0.0065Z) /T 5.26 /T , (8) air  

3 where air is the air density ( kg  m ), and is

calculated by Formula (8) [13-14], Cp is the air specific heat at constant pressure (1004J kg 1  K 1 ),

dT is the temperature difference at the height of Z1 Fig. 2. Land surface temperature distribution and Z2, Z1 (0.01 m) is the bare land roughness length, on September 20, 2007. Z (2m) is the reference height of meteorological 2 data, dT  a(Ts  0.0065 Z )  b , a and b are 1) Estimating surface net radiation (Rn). constants, and their values are obtained by selecting The surface net radiation (Rn) is obtained the “cold point” and “hot point” from remote sensing according to the Formula (5) [8]. image; rah is the corrected aerodynamic resistance, R  Q(1 )   T 4   T 4 , (5) unstable, and can be calculated through multiple n   s s recursive calculation. 4) Estimating Latent heat flux (LE). where Q is the total solar radiation, Q  Gd sw cos , 2 The latent heat flux is the heat energy of G is the solar constant (1367w/ m ), d is the sun- evaporation or condensation water, can be calculated earth distance,  is the zenith angle,   90   , by the surface heat balance Equation (9).  is the solar elevation,  is the surface albedo, LE  Rn  H  G (9)   0.356Rs (1)  0.130Rs (3)  0.373Rs (4)  0.085Rs (5)

 0.072R (7)  0.0018 [9], R (i) is the band s s surface reflectance,  is the Stefan-Boltzmann 5. Correlation Analysis of Land Surface 8 2 4 constant, and its value is 5.67 10 W  m  K , Temperature and Fluxes 4 T is the atmospheric long wave radiation,  is the atmospheric emissivity, To analyze the correlation between surface temperature and heat fluxes, 163 sampling points 1.08( ln )0.265 [10]  is the atmospheric     sw sw were uniformly selected in study area. Then, the 5 4 values of land surface temperature, soil heat flux, transmittance,  sw  0.75  210  Z [11],  sTs sensible heat flux, and latent heat flux of sampling is the surface long wave radiation, is the surface  s points were obtained through spatial analysis between sampling points data and surface emissivity, T is the near surface air temperature, temperature, soil heat flux, sensible heat flux, latent Ts is the land surface temperature, Z is the altitude at heat flux, respectively. Finally, surface temperature the measurement point and its value is from DEM. and soil heat flux, sensible heat flux, latent heat flux 2) Estimating soil heat flux (G). of each sampling point were compared with scatter The soil heat flux is the heat stored in the soil diagram respectively, and least square method was layer, and often estimated using the empirical used to simulate the function (Fig. 4~6).

69 Sensors & Transducers, Vol. 180, Issue 10, October 2014, pp. 67-71 )

K 305 ( 304 y = 0.0052x2 - 0.7455x + 325.88 303 302 301 300 299 298 Land surface temperature 75.00 80.00 85.00 90.00 95.00 100.00 Soil heat flux (G)(W/s2)

Fig. 4. Correlation diagram between land surface temperature and soil heat flux.

305 304 303 302 301 y = 0.1694x + 298.74 a) Soil heat flux (G) 300 299 298 Land surface temperature(K) 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 Sensible heat flux (H)(W/s2)

Fig. 5. Correlation diagram between land surface temperature and sensible heat flux.

305 304

303 y = -8×10-5x2 + 0.0684x + 286.85 302 301 300 299 298 Land surface temperature(K) 520.00 540.00 560.00 580.00 600.00 620.00 640.00 660.00 Latent heat flux (LE) (W/s2)

Fig. 6. Correlation diagram between land surface b) Sensible heat flux (H) temperature and latent heat flux.

The analysis results showed that, 1) Land surface temperature (y) and soil heat flux (x) was positively related, the fitting relationship: y=0.0052x2-0.7455x+325.88, and the correlation coefficient was 0.66. All data distributed in the right part of the function symmetry axis, so land surface temperature was increased smoothly with increasing soil heat flux. 2) Land surface temperature (y) had absolutely linear positive correlation with sensible heat flux (x), the fit function was: y = 0.1694x + 298.74, and the correlation coefficient was 1. So land surface temperature was increased linearly with increasing sensible heat flux. 3) Land surface temperature (y) was negatively correlated with latent heat flux (x), the fit -5 2 function was y=-8×10 x +0.0684x+286.85, and the c) Latent heat flux (LE) correlation coefficient was 0.55. All data distributed in the right part of the function symmetry axis, so Fig. 3. Land surface heat fluxes distribution land surface temperature was decreased smoothly on September 20, 2007. with increasing latent heat flux.

70 Sensors & Transducers, Vol. 180, Issue 10, October 2014, pp. 67-71

6. Conclusion and Discussion Transactions on Geoscience and Remote Sensing, Vol. 41, No. 11, November, 2003. [3]. J. F. Mustard, M. A. Camey, A. Sen, The use of The Nanchong city in Sichuan Province was satellite data to quantify thermal effluent impacts, taken as an example, the correlation between land Estuarine Coastal and Shelf Science, 49, 1999, surface temperature and heat fluxes were researched. pp. 509-524. The main conclusions are the following: [4]. Artis D. A., Carnahan W. H., Survey of emissivity 1) Land surface temperature had positively variability in thermography of urban areas, Remote correlation with both soil heat flux and sensible heat Sensing of Environment, Vol. 12, No. 4, 1982, flux, but negatively correlation with latent heat flux. p. 313-329. 2) Land Surface temperature was absolutely [5]. Goward S. N., Markham B., Dye D. G., Dulaney W., effected by sensible heat flux. Yang J., Normalized difference vegetation index measurements from the advanced very high 3) Influence mechanism of heat island effect is resolution radiometer, Remote Sensing of that the impervious surface area has higher soil heat Environment, Vol. 35, 1991, pp. 257–277. flux, lower latent heat flux, because it cuts off the [6]. Santos P., Negri A. J., A comparison of the soil-atmosphere heat exchange, its evapotranspiration normalized difference vegetation index and rainfall is small, energy conservation ability is weak, speed for the Amazon and Northeastern Brazil, Journal of of absorbing or releasing energy is high, and higher Applied Meteorology and Climatology, Vol. 36, sensible heat flux, because its roughness is lower, 1997, pp. 958-965. and its albedo is higher, most of solar radiation is [7]. Zhao Yingshi, Analysis theory and method of remote reflected to the atmosphere, more heat in life is sensing application, Science Press, Beijing, 2003. [8]. Liebe H. J., Hufford G. A., Cotton M. C., discharged in addition, and conversely, the natural Propagation Modeling of Moist air and suspended surface or vegetation cover area has lower soil heat water/ice particles at frequencies below 1000 GHz, in flux, higher latent heat flux, and lower sensible heat Proceedings of the AGARD 52nd Special Meeting of flux. So the land surface temperature of impervious the Electromagnetic Wave Propagation Pand, Vol. 3. surface area is higher, which of natural surface or 1993, pp. 1-10. vegetation cover area is lower. [9]. Liang Shunlin, Narrowband to broadband conversions of land surface albedo I algorithms, Remote Sensing of Environment, Vol. 76, 2000, pp. 213-238. Acknowledgements [10]. Bastiaanssen W. G. M, Menenti M., Feddes R. A., et al., A remote sensing surface energy balance This work was supported in part was supported algorithm for land (SEBAL) formulation, Journal of by the open foundation of the Laboratory of Geo- Hydrology, 1998, pp. 212-213. special Information Technology Ministry of Land [11]. Tasumi M., Allen R. G., Application of the SEBAL and Resources, Chengdu University of Technology methodology for estimating consumptive use of China (No. KLGSIT2013-10), and by the research water and stream flow depletion in the Bear River project of Sichuan College of Architecture Basin of L daho through remote sensing, Final Report Submitted to the Raytheon Systems Company, Technology (2011) and by Key Project of Education Earth Observation System Data and Information Department of Sichuan Province under Grant System Project, 2000. 12ZA255 and by The key science and technology [12]. Bastiaanseen W. G. M., SEBAL-based sensible and project of the Deyang City (No. 2013ZZ074-05). latent heat fluxes in the irrigated Gediz Basin, Journal of Hydrology, Turkey, Vol. 229, 2000, pp. 87-100. References [13]. Burman R. D., Jensen M. E., Allen R. G., Thermodynamic factors in evapotranspiration, in Proceedings of the Irrigation Systems for the 21st [1]. Xu Hanqiu, Image-based Normalization Technique Century Conference, 1987, pp. 140-148. Used for Landsat TM/ETM+ Imagery, Geomatics [14]. Smith M., Allen R. G., Monteith J. L., et al., Report and Information Science of Wuhan University, Vol. on the expert consultation on procedures for revision 32, No. 1, 2007, pp. 62-67. of FAO guidelines for prediction of crop water [2]. Gyanesh Chander and Brian Markham, Revised requirements, Land and Water Development Landsat-5 TM Radiometric Calibration Procedures Division, United Nations Food and Agriculture and Post-calibration Dynamic Ranges, IEEE Service, Rome, Italy, 1992.

______

2014 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com)

71