Home Articles The extraction method of surface temperature in agricultural area using Satellite Remote...

The extraction method of surface temperature in agricultural area using Satellite Remote Sensing and GIS

Myung-Hee Jo, Hyoung-Sub Kim
Urban Information Cadastral Engineering of Kyungil University
[email protected]
[email protected]

Hyang-Hee Um
Meteorological Research Institute
[email protected]

Joon-Bum Kim
Korea Forest Research Institute
[email protected]

Today the gathering information, which is synthetic and systemic, is possible from social activities to natural environment based on divers spatial information technology such as GIS and remote sensing. It, therefore, is processing not only the forecast of meteorological environment in the whole scale of the earth, but also the study about the local area, being possible to get the input datum which need that relative meteorological models have been developed with remote sensing for the decades.

The purpose of this study is that the development of the area by the urbanization of the agricultural area interprets more reasonably influencing the area of meteorological environment characteristic, fractionating both the diverse civilization and the environmental information of the nature around the basin of Ansung-watershed area based on population distribution information in addition to the observation of AWS (Automatic Weather Station) around the basin, and also analyzes systemically the pattern of the landcover change, surface temperature as well as NDVI distribution in the agricultural area, applying the technique of GIS spatial analysis and the multi-temporal of Landsat TM satellite image for Ansung-watershed area of Korea.

Fig. 1 Study area

The data and method of study
This study applied the data such as multi temporal of Landsat TM image (May 20th 1987, May 20th 1993, May 7th 2000), 1:25,000 digital map, topographic map, AWS observation information of in site(1996∼1999), population statistics information to analyze more systemically the meteorological environment characteristic in the basin of Ansung-watershed (585.188㎢).

  1. Landsat TM image processing and the information extraction of temperature environment
  2. To analyze the temperature distribution characteristic in the area of the study object which is spatial-temporal by applying ERDAS Imagine 8.5, it extracted the surface temperature using the value of the radiance got from the value of DN(Digital Number) in Landsat TM band 6. At the same time in classifying landcover, selected the training area using the 1:25,000 digital map as well as topographical map, and it was classified into the 6 classes such as forest, paddy field, field, stream, urban, barren with Maximum Likelihood method of the supervised classification applying Landsat TM false color image (band 4/3/2) after the survey of ground truth. In addition produced NDVI to analyze the vegetation vitality chart around the basin.

  3. Preparation of GIS digital map and analyzing topography environment information
  4. To analyze the more detail environmental information about basin of Ansung-watershed and surrounding area, this study constructed GIS DB such as administration boundary, drainage network, contour, and road network with both 1:25,000 digital map and topography map using ARC/INFO 8.1, ArcView 3.2. And also produced river density, stream frequency, catchments concentration chart, and bifurcation getting the stream order in the basin by Strahler (1957) stream order calculation formula by using the stream network of these built GIS DB.

    The aspect and slope map based on the digital elevation model using both contours at intervals of 10m and elevation points were also calculated. This study prepared choroplethe map with population statistics information, which was constructed with DB from 1995 to 2000.

  5. Construction the meteorological information DB using GIS
  6. Information was converted to vector data in contour to use AreView 3.2, using the point data of both longitude and latitude in the network of AWS meteorological observation which has 2 points in the basin of Ansung-watershed area and 5 points around the basin, building as well as classifying the data by day and month which obtained the information such as the highest average temperature, the lowest average temperature, the average temperature, the highest temperature, the lowest temperature, the average wind velocity, the maximum wind velocity, the maximum wind direction, precipitation generating the number of day.

Spatial meteorological analysis of Ansung watershed area

  1. Surface temperature distribution characteristic of landcover class

  2. Fig. 10 Multi-temporal landcover classification map

    Fig. 11 Multi-temporal surface temperature distribution map
    As a result of landcover classification in the area, forest was distributed to the most of there and it showed following the order such as paddy field, field, urban, barren, and stream.

    In result of the , it is showed that both forest and paddy field were decreased generally, while both urban and field were increased especially in centering around Ipjang-Myeon where was much increased to field. This was to be analyzed with the result indicated by the increase of the cultivation grape from 1990 when was active. Both Ansung-eup and Cheonan-city area were showed in raising the most active in tie studied area, developing the streets of the city along most roads, classifying many urban area centered by the location of Myeon or Eup. On the other hand the classified barren was to be identified with wasting land mostly to develop either of the golf or crop field. In 2000 the forest, paddy field, and field were decreased and otherwise the stream as well as the barren was increased.

    While the average surface temperature in the basin of Ansung-watershed area extracted from multi-temporal Landsat TM band 6 presented both 21.68℃ in 1987 and 23.16℃ in 1993, the result which was classified again to 7 classes of it is the same that . As a result of comparing the temperature between ’87 and ’93 in these of them, the forest and stream showed the lowest temperature distribution, as well as the pattern of the surface temperature distribution ordering field > paddy field> forest > stream, the location of Myeon and Eup showed the highest temperature distribution. In addition, this study analyzed the change pattern of surface temperature in the basin of Ansung stream by the change of landcover such as the , obtaining the value of surface temperature by multi-temporal, selecting the training area to each class with the standard landcover classification of (Figure 10).

    Fig. 12 The change of surface temperature on the change of landcover

    It could be identified the temperature increase from the minimum 0.9℃ to the maximum 2.15℃ according to decrease of forest, paddy field, field and stream area. While the temperature of urban and barren were increased from 0.03 to 1.83℃ because the sizes of area were increased. To look over the temperature change by landcover classes concretely about them, the surface temperature was increased to 1.55℃ while the forest was decreased as ’93 compared with ’87. As the paddy field was decreased to 19.1㎢, the temperature was increased to 2.15℃, and, as the urban area was increased 14.86㎢, the surface temperature was increased to 1.83℃. On the other hand it showed that the surface temperature was decreased to -0.45℃, though the field was increased to 22.39㎢.

    It could be analyzed the change pattern of the temperature environment of the agricultural area by the urbanization based on the both the increasing and decreasing phenomenon in the surface temperature according to landcover change in the basin of Ansung stream.

  3. Surface temperature spatial properties of based on topography and NDVI
  4. As a result analyzing the area, aspect, slope, temperature, and NDVI according to elevation in the basin of Ansung-stream, the highest value of the area is represented between 0 and 50m level, and it could be identified that the more elevation, the lower proportion of area is.

    Collectively aspect inclines toward the south, 351∼400m level and 501∼570 level is under 10° and the rest of sections is over 10°which has the steep gradient. The average temperature shows descending trends according to raising the elevation, and NDVI shows ascending trends according to raising the value. In 1987 28℃∼31℃ which is the highest surface temperature class distributes that the average elevation is 35.22m topographically and the slope distributes 1.31°.

    It cloud be found that as the temperature shows the high value, slope, elevation, and average value of NDVI are descended. This is analyzed as the result because the level of the high temperature is generally formed in the flat area where is centered except some of vegetation area in city. It could be identified that the NDVI, elevation, slope has the consistent correlation according to the temperature.

    Fig. 13 Distribution map of NDVI

    The higher NDVI value, the higher elevation and slope value is. It is showed that the proportional relation, as the result in the correlation analysis, the lower NDVI value, the higher temperature is.

  5. The correlation of NDVI and surface temperature.
  6. It could be identified that analyzing the result of the produced value of the vegetation index by using the Landsat TM band 3, band4. The golf area and industrial complex when was increased in 1993 in comparison with 1987 changed the vegetation index of some area. With the basis of this result this study implemented both the correlation and regression analysis to examine the correlation between surface temperature and NDVI. The result is in and showed that there was clearly negative (-) correlation between surface temperature and NDVI. Moreover, in result of the regression analysis with the basis of the confidence which has about average 69% for periods, the experimental formula was extracted which can be interpret the relation between surface temperature and NDVI.

    Fig. 14. Regression analysis between NDVI and surface temperature

    The spatial analysis of watershed property factor

    The drainage density, drainage frequency, and catchment concentration rate and bifurcation were calculated on the basis of the catchment area, perimeter an stream length in the basin of Ansung stream using GIS TOOL. As a result, catchment area 585.18 stream length 569.60 the perimeter 137.5, drainage density 0.97, the total numbers of streams 854.00, drainage frequency 1.45, catchment concentration 7.31 were produce. As a consequence of surface temperature to each area, Gumkwang-Myeon, which is high drainage density to basin is 20.5℃, Gongdo-Myeon, which is low, is 23.6℃

  7. The correction between the population distribution and meteorological characteristics
  8. As an influential factor, which affects landcover change and surface temperature, as a result of investigating population distribution around basin 1995 to 2000, such as in Ansung-city, Asan-city, Cheonan-city the population has been increasing continually. These regions indicate that the temperature is rising according to urbanization, forming housing, roads, commercial and industrial zones. On the other hand, in proportion to the decrease of population, the temperature has been low. The temperature is changed by population distribution for each area.

    Fig. 15 The temperature change of each area on population distribution

  9. Spatial distribution characteristics of high temperature area
  10. As a result of extracting topographical information each administrative regions, in high temperature area around Ansung-watershed area, the regions such as Ansung-city, Cheonahn-city, which are characteristic of urbanization, are showed the high surface temperature. And also compare to distribution the average elevation, 79.08m, that of this area is low in 35.61m, for the average slope 6.26˚, the average of this area is 2.26˚. The higher the temperature, the lower NDVI value is, the larger population, the higher temperature is.

    As a result of analyzing the characteristic of surface temperature distribution to topographical feature nearby river, the surface temperature is high in Ansung-city and Daedeok-Myeon, average latitude height 20∼40m, slope less than 2˚, area mainly faced on South-Western aspects high surface temperature zone more than 9℃ was identified.

    The analysis of landcover environment showing high surface temperature zone indicates urban area and industrial complex of suburbs.

    Fig. 16 High temperature distribution map surrounding Ansung-watershed area

This study aims to analyze the meteorological environmental around Ansung watershed area, extracting human and physical environment factors in the rural area, using Landsat TM image and GIS. The conclusions are summarized as followings.

  1. Average surface temperature extracted Landsat TM band 6 to the period, 21.68℃ in 1987, 23.16℃ in 1993, Ansung-eup and Sungwhan-Eup in the basins are showing the highest temperature distribution zone, generally barren, field > paddy field > forest > stream in the order named.
  2. Not only Due to the area decrease in forest, paddy field and area regions, the temperature is higher, but also due to the area increase in urban, barren area, the temperature rises from 0.03℃ to the minimum to 1.83℃ to the maximum. On the basis of the rising of surface temperature by landcover change like this, the heat environment change pattern in rural area because of urbanization is identified.
  3. As a result of analyzing temperature distribution characteristics between the topographic factors and NDVI, the average temperature is according to elevation rising, while NDVI is high. The region where is high in temperature is normally low in slope, elevation and NDVI. This is mainly because the high temperature zone forms in plain low zone except the partial forest zones.
  4. Between The surface temperature and NDVI distinctively negative correlation exists, as result of regression analysis in two periods the relation of NDVI and surface temperature on the basis of average confidence 69%.
  5. After calculating the characteristic factors to each basin, the average surface temperature nearby Ansung watershed extracted from Landsat TM band 6. Geumkwang-myeon, which has high drainage density in the basin, has 20.5℃, Gongdo-myeon, which has low drainage density, 23.6℃.
  6. According as the population increases forming residential and industrial zone by urbanization phenomenon, the temperature is rising, while in proportion to decrease of population, the temperature is low.
  7. As a consequence of analyzing spatial distribution characteristic of high temperature area nearby Ansung watershed such as Ansung-city, Cheonan-city, which has large population and the characteristic of urbanization, showed high surface temperature distribution zone, and to the landcover, urban area and industrial zones in the suburbs have high temperature.
  8. Through this study, it be could clarified the surface temperature characteristic of spatial distribution for urbanization in the rural area, using satellite image and GIS. Also, in the near future it will be expected the basic data and foundation construction for formation technique development concerned with more detail meteorological information in agriculture area.


  1. Y.S.Kim, A.S.Suh, M.H.Jo. 1998. Introduction of remote sensing. Dong-A technique press co.
  2. K.J.Lee, 2002. Extraction method of temperature distribution properties in urban area using remotely sensed data and GIS, Thesis of master degree of Kyungil University.
  3. M.H.Jo, J.H. Shin, K.J.Lee, A.S.Seo, 2001, Spatial characteristic analysis of temperature environment in rural area using RS and GIS, Korean association of geographic information studies conference.pp.228-236
  4. K.J.Lee, M.H.Jo, H.S. Kim, D. H. Shin, Extraction method of temperature distribution properties in urban area using remotely sensed data and GIS. Proceedings of the 2002 Joint Spring Meeting.pp.145-151
  5. Myung-Hee Jo, Sung-Nam Oh, Ae-Sook Suh, Ji-Hyun Shin, Kwang-Jae Lee, Surface Temperature Analysis of Urban Area using RS and GIS, Proceedings of The 21st Asian Conference on Remote Sensing, pp.234-239.
  6. Myung-Hee Jo, Kwang-Jae Lee, Yun-Won Jo, Spatial Analysis of surface Temperature Distribution in Metropolitan Area using Landsat TM Images and GIS, Pressented at AURISA 2001-The 29th Annual Conference of Aurisaa, 2001.11.
  7. Markham, B. L and J. L. Becker, 1986, Landsat MSS and TM post-calibration dynamic ranges, exoatmospheric reflectances and at-satellite temperatures, EOSAT Landsat Tech Notes, 1:3-7.
  8. GUPTA. R. K, PRASAD. S, SESHA SAI M. V. R. and T. S. VISWANADHAM, 1997, The estimation of surface temperature over an agricultural area in the state of Haryana and Panjab, India, and its relationship with the Normalized Difference Vegetation Index (NDVI) using NOAA-AVHRR data, International Journal of Remote Sensing, Vol.18, No.18, pp.3729-3741.
  9. Hafner and Jan, 1999, Urban heat island modeling in conjunction with satellite-derived surface/soil parameters, Journal of applied meteorology, Vol.38, No.4, pp.448.
  10. Prakash. A. and R. P. Gupta, 1999, Surface fires in JHARIA coalfield, India-their distribution and estimation of area and temperature from TM data, International Journal of Remote Sensing, Vol.20, No.10, pp.1935-1946.