3.Application and Validation
The proposed algorithm was applied to AVHRR images over Inner Mongolia region. The principal land cover types within this region are semi-and grasslands, sand-land and shrub vegetation. 10 scenes of NOAA-14/AVHRR over afternoon pass from 1st to 10th August 1996 were prepared. Firstly, by using ground control points and nearest neighbor resembling algorithm, these images were registered to a plate care projection images with 1024*1024 pixels of about 1km resolution. While qc,qs ,Df of each pixel were calculated by referring to eh satellite orbit parameters. Secondly, a composite image was synthesized by picking up the pixels with the maximum NDVI value among 10 scenes AVHRR images. Lastly, the proposed atmospheric correction algorithm with the LUTs was applied to the composite image, where the subarctic summer atmospheric model with visibility of 60 km was assumed. The elevation data of each pixel was referred to GTOP30 (Global Land One-Km Elevation data).
Figure 1. Dependence of A_i B_i and S_i on q_c, q_s, Df , z and V based on the atmospheric profil models as shown an - Tropical - Midlatitude summer - Subarctic summer for AVHRR Channels 1 and 2 of NOAA- 14
Figure 2 shows the histograms of Channels 1 and 2 as well as NDVI before and after the atmospheric corrections for the composite image. Figure 3 shows the results of ground measuring spectral reflectance's compared to AVHRR data. The ground measuring was undertaken at 18 sites of homogenous vegetation in the region of semi-arid grasslands from 30 July to 10August in 1996. The ground spectral reflectance of each site was calculated as the mean of measured values at 7 points around the site. AVHRR observing values of each site was calculated as the mean of 3*3 pixels centered at the site in the composite image. Comparing with uncorrected AVHRR data, the corrected results show the following effects: (1) the overall decrease in Channel 1 reflectance, (2) the overall increase in Channel 2 reflectance, (3) the overall increase in NDVI of about 0.1-0.25, (4) closer to the ground measuring results in Channels 1 and 2 as well as NDVI. The differences between Channels 1 and 2 are due to the correction for Rayleigh scattering, which is more pronounced at the lower at the shorter wavelength, and the attenuation caused by water vapor.

Figure 2. Histogram of Channels 1 and 2 as well as NDVI before and after the atmospheric correction by using LUTs for the AVHRR composite image in Aug.1-10 1996.



Figure 3. Comparison between atmospheric correction results of AVHRR data and ground measuring results for 18 ground sites in Inner Mongolia grasslands region, Aug. 1-10 1996.
4.Conclusion
An atmospheric correction methodology have been developed for AVHRR data. The algorithm works with a catalogue of atmospheric correction function, stored in look-up tables, and was evaluated with AVHRR data in Inner Mongolia. It is adaptive to the mass AVHRR data, and is implemented in a AVHRR processing system for composting the cloud-free mosaic AVHRR images for the entire Asia. It is necessary to furtherly verify this algorithm for other land cover types and different areas.

Acknowledge
This research was supported by Grant-in-aid for scientific research of No.0824204, 1996-98, Ministry of Education of Japanese Government.

References

• Eidenshink, J.C. and Faundeen, J.L. (1994), the 1 km AVHRR global land data set: first stage in implementation. INT. J. REMOTE SESNING, Vol. 15, No.15, pp.3443-3462.
• Hoben, B.N., (1986), Characteristics of maximum-value composite images from temporal AVHRR data. INT.J. Remote sensing, Vol.7, No.11, pp.1417-1434.
• Moran, M.S. Jackson R.D., Shelter P.N. and Teillet P.M. (1992), Evaluation of simplified procedures for retrieval of land surface reflectance factors from satellite Sensor output, remote sens. Environ. Vol.41, pp.169-184.
• Tanre, D., Hoblen, B,N., and Kaufan, Y. F., Volz, F.E., and Garing, J.S. (1972), Optical Properties of the Atmosphere, 3rd ed., Report AFCRL-720497, Environmental Research Papers, No. 411, Air Force Cambridge Research Laboratory. Hanscom Field, Bedford, MA, 113 pp.
• Markham,B.L., Halthore, R.N., and Goetz, S.J. (1992), Surface reflectance retrieval from satellite and aircraft sensor: results of senor and algorithm comparison doing FIFE, Journal of Geographysical Research, Vol. 97, No. D17, pp. 785-795.
• Vermote, E.F., Tanre, D., Deuze J.L., Herman M., and Morcrette J.J. (1997), Second simulation of the satellite Signal in the solar spectrum, 6S: an overview, IEEE Trans. Geosci. Remote Sensing, Vol. 35, No. 3, pp.675-686.