A Technical Support Project For The Land Cover Classification Using Modis-take Beijing-Tianjin-Hebei Area For Example

Based on the MODIS 1B images transponded by the DVBS (Digital Video Broadcast by Satellite) system, relying on the ENVI / IDL software, integrate MODIS 1B image pre-processing technique flow, propose a new idea for clear sky image synthesis. Pre-processing includes three main components of destriping, geometric correction and atmospheric correction. In destriping, improve equalization curve compensation, obtain good results.There are two kinds of stripes in MODIS 1B images transponded by DVBS(Digital Video Broadcast by Satellite)system: horizontal and vertical stripes, and horizontal stripes have three types: single stripes, wide stripes and multi-line stripes. The destriping methods mainly include the frequency domain method, the spatial domain method and the statistic adjustment method. MODIS sensor has 36 spectral bands, using only one destriping method can not reach desired effect. Based on compensation and elimination of equalization curve method, this article proposes a new set of algorithm on rapidly detecting and removing stripes. Given a definite stripe detection judgment parameter,namely: the correction factor Ci of 10 groups of sensors,if the value Ci of A-frame and B-frame are separated by 0, moreover ABS (Ci)> 0.008, then there exists wide stripes; if the maximum of correction factor Ci ' is greater than 0.05, then there exists single stripes. In the wide destriping model, the current column correction factor of each sensor can be extracted from the data that the current column as the first column to continuous 100 columns of image. The experimental results show that: the striping detection model can effectively identify the single stripes and the wide stripes, multi-line stripes need visual identification .The removal effect of single and wide stripes is better than NASA website's data, although multi-line stripes are not completely removed, image quality is significantly improved after the treatment.Based on ENVI software, this article summarizes a set of pre-processing process for MODIS 1B image, including three main components of destriping, geometric correction and atmospheric correction. In this paper, Use Georeference MODIS module in ENVI to wipe off bow-tie and do geometric correction; Use FLAASH module to do atmospheric correction for VNIR radiation intensity bands that after geometric correction, use thermal infrared atmospheric correction module (Thermal Atm Correction) to do atmospheric correction for thermal infrared band emissivity that after geometric correction. Image quality is markedly improved after the pre-processing , advantage for retrieving the MODIS products. Image pre-processing procedures should be based on retrieval a?lgorithm of different products , in general, the data after geometric correction can be used to retrieve atmospheric products, image after atmospheric correction can be used to develop the land and ocean products. When retrieve the land and ocean products, if the indicator or algorithm has the characteristic that can reduce the atmosphere impact, then can directly use image data after geometric correction.Compared the images composed by b1/b7 ration method with MOD09GA released by NASA,the conclusion can be leaded to that this method is good at the land part clear image composition; by contrast, the max saturation method was sutible for the water part. Combined these two methods, the special method was proposed,which regarded the land piexl of the minimum value of b1/b7 in a certain days as the clear piexl;considered the water piexl of the maximum value of saturation as the clear piexl of the water. According to the index image, producted during the compound progress of the land surface reflectance, named as HC_time, the compound clear images of Apparent reflectance, radiance, emitivity or land suface emitivity could be producted. As the time span change, the 5 days clear image,10 days clear image and month clear image could be produced. This algrithm of compound clear image improved the data basis for the research of land-use spatio-temporal evolution rule around Beijing ,Tianjin and Hebei province.