THERMAL REMOTE SENSING: CONCEPTS, ISSUES AND APPLICATIONS Anupma PRAKASH ITC, Geoogica Survey Division prakash@itc.n Working Group WG VII / 3 KEY WORDS: Therma infrared, radiant temperature, emissivity, back body, vocanoes, fires. ABSTRACT In the ast few decades remote sensing has reached from an experimenta to an operationa eve. The increase in the number of earth observation sateites, the advancement in toos and processing techniques, and the use of data for new appications has been phenomena. However, the major part of the efforts were directed in the past towards the use of optica data and now aso to the use of microwave data. The avaiabe iterature underines the fact that the use of data acquired in the therma infrared region has been reativey imited within the scientific and appication community. The imited use of therma data is inked to severa facts such as the imitation of the sensor capabiities, the nature of data itsef, and the reuctance of many to expore the potentias of therma remote sensing. This paper deas with the concepts and issues of therma remote sensing and presents a variety of appications where therma data finds its way. The benefits and imitations of therma data are discussed and the potentia of therma remote sensing, speciay in ight of future high resoution sateites, is highighted. The paper concudes with the author's views on the importance of these aspects speciay in the standard remote sensing educationa programmes. INTRODUCTION Therma remote sensing is the branch of remote sensing that deas with the acquisition, processing and interpretation of data acquired primariy in the therma infrared (TIR) region of the eectromagnetic (EM) spectrum. In therma remote sensing we measure the radiations 'emitted' from the surface of the target, as opposed to optica remote sensing where we measure the radiations 'refected' by the target under consideration. Usefu reviews on therma remote sensing are given by Kahe (980), Sabins (996) and Gupta (99) It is a we known fact that a natura targets refect as we as emit radiations. In the TIR region of the EM spectrum, the radiations emitted by the earth due to its therma state are far more intense than the soar refected radiations and therefore, sensors operating in this waveength region primariy detect therma radiative properties of the ground materia. However, as aso discussed ater in this artice, very high temperature bodies aso emit substantia radiations at shorter waveengths. As therma remote sensing deas with the measurement of emitted radiations, for high temperature phenomenon, the ream of therma remote sensing broadens to encompass not ony the TIR but aso the short wave infrared (SWIR), near infrared (NIR) and in extreme cases even the visibe region of the EM spectrum. Therma remote sensing, in principe, is different from remote sensing in the optica and microwave region. In practice, therma data prove to be compementary to other remote sensing data. Thus, though sti not fuy expored, therma remote sensing reserves potentias for a variety of appications. The next sections discuss the main concepts and issues of therma remote sensing and continue to present a brief overview of the appication of therma data. The artice concudes with the advantages and imitations of therma remote sensing and the need for incuding this topic in remote sensing educationa programmes. 2 CONCEPTS In therma remote sensing, radiations emitted by ground objects are measured for temperature estimation. These measurements give the radiant temperature of a body which depends on two factors - kinetic temperature and emissivity. Figure presents the various factors affecting the radiant temperature and these are further discussed in Internationa Archives of Photogrammetry and Remote Sensing. Vo. XXXIII, Part B. Amsterdam 2000. 239
subsection 2.2. Subsection 2.3 presents the concept and measurement of radiant temperature which is the basis ef estimating radiant temperature from therma remote sensing data. However, before presenting these concepts, subsection 2. briefy discusses the waveength ranges which are of interest for therma remote sensing. RADIANT TEMPERATURE Kinetic temperature Emissivity Heat budget Therma properties Soar heating Longwave upweing radiation Downweing radiation Anomaous heat sources Soar eevation Coud cover Atmospheric conditions Topographic atitude Therma conductivity Specific heat Heat capacity Therma diffusivity Therma inertia Figure. Factors controing radiant temperature 2. Waveength / Spectra Range The infrared portion of the eectromagnetic spectrum is usuay considered to be from 0.7 to,000 µm. Within this infrared portion, there are various nomencatures and itte consensus among various groups to define the subboundaries. In terrestria remote sensing the region of 3 to 35 µm is popuary caed therma-infrared. As in a other remote sensing missions, data acquisitions are made ony in regions of east spectra absorption known as the atmospheric windows. Within the therma infrared an exceent atmospheric window ies between 8- µm waveength. Poorer windows ie in 3-5 µm and 7-25 µm. Interpretation of the data in 3-5 µm is compicated due to overap with soar refection in day imagery and 7-25 µm region is sti not we investigated. Thus 8- µm region has been of greatest interest for therma remote sensing. 2.2 Spectra Emissivity and Kinetic Temperature Therma remote sensing expoits the fact that everything above absoute zero (0 K or -273.5 C or 59 F) emits radiation in the infrared range of the eectromagnetic spectrum. How much energy is radiated, and at which waveengths, depends on the emissivity of the surface and on its kinetic temperature. Emissivity is the emitting abiity of a rea materia compared to that of a back body (see 2.2.), and is a spectra property that varies with composition of materia and geometric configuration of the surface. Emissivity denoted by epsion (ε) is a ratio and varies between 0 and. For most natura materias, it ranges between 0.7 and 0.95. Kinetic temperature is the surface temperature of a body/ground and is a measure of the amount of heat energy contained in it (see 2.2.2). It is measured in different units, such as in Kevin (K); degrees Centigrade ( C); degrees Fahrenheit ( F). 2.2. Back body is a theoretica object that absorbs and then emits a incident energy at a waveengths. This means that the emissivity of such an object is by definition. Needess to say, such an object is ony imaginary and no natura substance is an idea back body. 2.2.2 Factors Affecting the Kinetic Temperature can be categorised in two broad groups - heat energy budget and therma properties of the materias (figure ). Heat energy budget incudes factors such as soar heating, ongwave upweing and downweing radiations, heat transfer at the earth-atmosphere interface and active therma sources such as fires, vocanoes etc. Therma properties of materia incude factors such as therma conductivity, specific heat, 20 Internationa Archives of Photogrammetry and Remote Sensing. Vo. XXXIII, Part B. Amsterdam 2000.
density, heat capacity, therma diffusivity and therma inertia of the materia. An exceent expanation of these factors is given by Kahe (980). 2.3 Radiant Temperature The radiant temperature (T R ) is the actua temperature obtained in a remote sensing measurement and, as mentioned earier, depends on actua or kinetic temperature (T K ) of the body and the its emissivity (ε). The tota radiations emitted by non back body (natura surfaces) is given by W = e s s T K = T R () where σ is the Stefan-Botzmann's constant. This defines the reation between the radiant temperature and kinetic temperature of a body as T R = e T K (2) From the above equation, and the knowedge that a natura materias are non-back bodies with emissivity ess than one, it is cear that the radiant temperature (temperature estimated by remote sensing data) is aways ess than the actua surface temperature of the body by a factor e. The tota amount of radiations emitted by a body can aso be estimated using Panck' equation which gives: W 2p h c = 5 2 e h c k T e (3) 3 - where W is the spectra emittance, h is the Panck's constant ( 6.62 0 Js), c is the speed of ight ( 3 0 8 ms ), is the waveengtn metres, k is Botzmann's constant (.38 0 23 JK - ), T is the temperature in K and e is the spectra emissivity. This formua aso impies that with the rise in temperature of the ground objects, there is an increase in the intensity of the emitted radiations, with the peak shifting towards shorter waveengths. Inverting Panck's equation we get C 2 T = e n C 5 + W () where h c C =. k 6-2 C = 2p h c = 3.72 0 Wm and 2 = 0.0 mk For converting digita vaues from remote sensing data to spectra radiance and to radiant temperatures the reader is referred to the paper by Markham and Barker (986) as we as Prakash and Gupta (998). 3 ISSUES Due to the fundamenta difference between remote sensing in the therma infrared region and the other regions of the EM spectrum, there are some issues pecuiar and pertinent for therma remote sensing. Some of these reate to the mode of acquisition, caibration, radiometric and geometric correction, and are discussed in the foowing sections: 3. Data acquisition: Modes and patforms There are three different aspects which must be considered whie taking about the mode of therma data acquisition. These are Internationa Archives of Photogrammetry and Remote Sensing. Vo. XXXIII, Part B. Amsterdam 2000. 2
3.. Active versus passive mode: Most of the therma sensors acquire data passivey, i.e. they measure the radiations emitted naturay by the target/ground. Data can aso be acquired in the TIR activey depoying aser beams (LIDAR). However, these techniques are not we researched and are ony in the infancy. 3..2 Broad band versus mutispectra mode: For the broad band therma sensing, in genera the 8 to µm atmospheric window is utiised. However, some spaceborne therma sensors such as Landsat Thematic Mapper Band 6 operate in the waveength range of 0. to 2.6 µm to avoid the ozone absorption peak which is ocated at 9.6 µm. The mutispectra therma channes, such as those in the ASTER patform, are targeted speciay for geoogica appications. 3..3 Daytime versus night-time acquisition: Therma data can be acquired during the day and during the night. For some appications it is usefu to have data from both the times. However, for many appications night-time or more specificay pre-dawn images are preferred as during this time the effect of differentia soar heating is the minima. The patforms for such data acquisitions range from sateites, aircrafts to ground based scanners. 3.2 Spatia resoution and geometric correction Most therma sensors have onboard recording and caibration systems. Two back bodies (BB) commony known as BB and BB2 are setup which contro the radiometric caibration of the acquired data. As the sensors measure emitted radiations, there is aso a heating effect and constant cooing of the sensors is required. This poses a physica imit to the measuring capabiity of the sensors and therefore the spatia resoution of the acquired data. The coarse spatia resoution, speciay of sateite borne broad band therma data poses some additiona probems in geometricay registering it to other data, speciay when the atter have much higher spatia resoution. Identification of corresponding reiabe contro points on data sets with such wide differences in spatia resoution is not ony difficut but when tried may resut in unacceptabe transformation resuts. Aternate approaches of co-registration must be thought of. This may be done by first registering the therma image to another image with intermediate spatia resoution and in the next step to the target high resoution image. For detais on this two step transformation the readers are referred to the paper by Prakash et a. 999. APPLICATIONS Therma property of a materia is representative of upper severa centimetres of the surface. As in therma remote sensing we measure the emitted radiations, it proves to be compementary to other remote sensing data and even unique in heping to identify surface materias and features such as rock types, soi moisture, geotherma anomaies etc. The abiity to record variations in infrared radiation has advantage in extending our observation of many types of phenomena in which minor temperature variations may be significant in understanding our environment. Therma remote sensing reserves immense potentia for various appications. The foowing is a ist of some of the areas in which therma data is put to use Identification of geoogica units and structures Soi moisture studies Hydroogy Coasta zones Vocanoogy Forest fires Coa fires Seismoogy Environmenta modeing Meteoroogy Medica sciences Vetenary sciences Inteigence / miitary appications Heat oss from buidings Others 22 Internationa Archives of Photogrammetry and Remote Sensing. Vo. XXXIII, Part B. Amsterdam 2000.
Detais on how therma data is processed and used in these various appications is beyond the scope of this artice. Numerous references can be found in iterature for the use of therma data for most of these appications. A modest coection of these reference sources is aso avaiabe at http://www.itc.n/~prakash/research/therma_ref.htm 5 CONCLUSIONS From the issues addressed in section 3 of this artice, some imitations of therma remote sensing are cear. On the other hand, from the variety of possibe appications highighted in section, the advantages and potentias of therma remote sensing are aso obvious. One more fact that is now cear is that new sateite and airborne patforms with new and improved therma sensors aso promise to bring more interest and chaenge in this reativey ess expored fied. Therefore, now there is a definite need to promote the understanding and the use of therma data by the scientific and appication community. These deveopments shoud encompass among other things fundamenta research in the principes of therma remote sensing aboratory measurements of spectra response of natura materias in the therma infrared region deveopment of more sophisticated sensor technoogy finay to appication oriented research where the aready expored appication fieds can be refined and new appication areas can be tapped. One of the most promising way to ensure that such a goa is achieved in by introducing the topic of therma remote sensing in greater depth in the remote sensing educationa programmes. The 'spin off ' effect of such a venture woud be that more and more researchers with fresh ideas woud expore the therma data and its possibiities. REFERENCES Gupta R.P., 99. Remote Sensing Geoogy (Berin-Heideberg:Springer-Verag). Kahe A.B., 980. Surface therma properties. In Remote Sensing in Geoogy, edited by B.S. Siega, and A.R. Giespie (New York; John Wiey), pp.257-273 Markham, B.L., Barker J.L., 986. Landsat MSS and TM post caibration dynamic ranges, exoatmospheric refectances and at-sateite temperatures. EOSAT Landsat Technica Notes, August 986, Earth Observation Sateite Co. (Lanham, Maryand), pp. 3-8. Prakash A., Gens R., Vekerdy Z., 999. Monitoring coa fires using muti-tempora night-time therma images in a coafied in North-west China. Internationa Journa of Remote Sensing, 20(), pp. 2883-2888.. Prakash A., Gupta, R.P., 998. Land-use mapping and change detection in a coa mining area - a case study of the Jharia Coafied, India. Internationa Journa of Remote Sensing, 9(3), pp. 39-0. Prakash A., References on therma remote sensing. http://www.itc.n/~prakash/research/therma_ref.htm Sabins F.F. Jr, 996. Remote Sensing: Principes and Interpretation, 3rd edn. (New York: W.H. Freeman). Internationa Archives of Photogrammetry and Remote Sensing. Vo. XXXIII, Part B. Amsterdam 2000. 23