Retrieval Model of Infrared
Surface Emissity Based On NOAA Satellite Data Fang Yonghua and Xun
Yulong AbstractAnhui Institute of Optics & Fine Mechanics Hefei, Anhui, 230031, China Changchun Institute of Optics & Fine Mechanics Changchun, Jilin, 130022, China E-mail: yhfang@aiofm.ac.cn In this paper, we accomplish research on retrieval model of surface emissivity. We establish appropriate calculation model and method based NOAA satellite data to acquire actual thermal radiation characteristic of any area in real time. Introduction The infrared emissity of surface is an important parameter that describng the interaction of earth’s surface with atmosphere in research on thermal radiation characteristic of earth’s surface. Nowadays, they’re two kinds of elementary research approachs: the first method is based on groundwork observation, establishing model with direct measure and long-term statistic. This method is direct and reliable but troublesome. The second method is to adopt satellite data. Remote sensing image data in thermal infrared band is primary research area (such as channel 4 and channel 5 of NOAA satellite). The application of NOAA’s image data is the most universal on account of its high repeat rate and simple atmosphere correct arithmetic. We can investigate quickly the thermal radiation characteristic of any area in real time according to the retrieval model in despite of some technology difficulty. The emissity of seawater is approximately 1 in thermal infrared waveband, which change hardly with the change of wavelength, sea condition and seawater component. Compared with seawater, the absorption coefficient of terrestrial surface (such as rock, vegetation etc.) is variable along with the surface characteristic and wavelength. Research on infrared waveband emissity (temperature) of land is advanced task in infrared remote sensing. Currently, there is no universal and effective method. We should establish appropriate calculation model and method to acquire actual thermal radiation characteristic of land. Influence of sun reflectance in thermal infrared waveband Around l=10mm, the irradiance of land surface from sun is about (the atmosphere transmissivity supposed 1): E=5×10-1w/m2·mm The radiance is: B=rE/p =1.6×10-1w/m2·sr·mm Where surface reflectance r is supposed to 1(the strongest). In normal temperature (300K), thermal infrared radiant intensity of surface on land is: B(300K)=1.0×101w/m2·sr·mm By all appearances, thermal radiation occupies a leading place. Solar reflectance can be ignored. Emissivity e Generally, the infrared emissivity of most water is close to 1; the infrared emissivity of vegetation is about 1 too. For most rock and soil, there are some characteristic absorption in 8-12mm waveband (mostly due to SO2 oscillation). Statistically, their reflectance spectrum look as a bell cover (nearly Gauss distribution), which max value is located in 9.5-10mm, and which peak values are not entirely uniform. So, e depends on surface types, surface states and wavelength variety. For enough thick surface which transimissivity is zero, we have: r+e=1, Therefore, we should establish retrieval model to seek reflectance r or transimissivity e. Infrared transimissivity model of rock and soil Our research is based on two infrared waveband (channel 4 and 5) on NOAA satellite. The image gray degree D of two channels is directly related to radiance. We can calculate radiance L from D and instrument function of sensor and vice versa. Where a and k are the plus and the offset of sensor respectively. They refer to involved handbook. Surface’s infrared radiance model can be expressed as: L'(D4)=e 4 B(T,l 4)t4 +L (1) L'(D5)=e 5 B(T,l5)t5+L A5 (2) Where B is the radiance of blankbody at the same temperature which can be gotten from Plank formula. l4 and l5 are central wavelength or average wavelength wavelength of channel 4 and channel 5 respectively. e4 and e5 are surface emissivity of two channels respectively. t4 and t5 are atmosphere transimissivity which can be calculated from Lowtran 7 model. Besides, L1 is the atmosphere thermal radiation, which don’t contain any surface information and Should be deducted in retrieval process. Atmosphere influence deduction First, near the research area, we work over water surface or an area having lower emissivity (approximately black body), and calculate the temperature T4° and T5° of two channels from Plank formula by the method of seawater retrieval. There are some components of atmosphere radiation in T4°and T5°. Practically, there is a ratio between two channels atmosphere radiation. The model is such as: Ts°=T4°+(T4°-T5°)a+b Where a and b are content, Ts° is the actual temperature. We can calculate atmospheric influence from formula (1) and (2). relation of e between two channels Statistically, for rock and soil, there is a ratio between two average spectral reflectance (NOAA channel 4 and 5). r4/r5=a a»2 And e4=1-r4, e5=1-r5 , So we have: e5 =1-(1-e4 )/a (3) Accurate a can be calculated directly from reflectance spectrum. Emissivity model From (1), (2), (3), we have: e4=R(4/5) /(a b-R(4/5)) (4) Where R(4/5)=L(D4)/L(D5)=(L'(D4)-LA4 )/(L'(D5)-LA5) B(T,l4 )/ B(T,l5)=b Initial bcan be defined as 1. From e4 , e5 , (1) and (2), we have T41 and T51, adopting Ts1=(T41 +T51)/2. Modification of emissivity The emissivity and temperature calculated above are not finally ideal result. To get higher precision, we should reiterate to remove the influence of atmosphere and instrument until two channels temperature are equal. Conclusion In this paper, we accomplish research on retrieval model of surface emissivity. Its specialty is to acquire surface thermal radiation characteristic by establishing retrieval model based NOAA satellite data. Using this method we can investigate quickly the thermal radiation characteristic of any area in real time The validity of model will be presented for the future. References
Since 1986 he have been engaged in research work on Remote Sensing. Recently, his esearch interests are remote sensing retrieval methods, radiation calibration and atmosphere correction . Xun Yulong graduated in July,1966 as a graduate student from Changchun Institute of Optics & Fine Mechanics. He is a senior research fellow, tutor of student for Ph.D.. He has been engaged in optics for tens years. Recently, his research interests are laser induced fluorescent remote sensing, application of pattern recognition, neural network and wavelet techniques to recognition of remote sensing spectra & image, and laser lidar. |