(1) Image processing method based on ETM data
According to the analysis of spectral characteristics of ground object images in the Rujigou coal fire area, the image band ratio ETM7/4 abnormal pair identification and extraction The information of burnt rock and open fire is favorable, and the multi-band waveform characteristics of ground objects ETM5>ETM7>ETM1>ETM3>ETM2>ETM4 are good indicators for identifying underground coal fires. The specific image processing method designed to extract remote sensing anomalies of coal fire disasters is as follows.
(1) Use coalfield boundary data to generate a coalfield distribution mask image;
(2) Under masking conditions, use spectral angle technology to extract coal fire area features The distribution information of the coal field is generated to generate a coal field feature distribution image, as shown in Figure 3?4?1;
Figure 3-4-1 The image of the feature distribution information extracted by the spectral angle technology in the coal fire area
p>
(3) Calculate the band ratio ETM7/4 of the coalfield distribution image, and use the method of adding several times the standard deviation of the ratio image to extract the ratio anomaly as remote sensing information for identifying burnt rocks;
(4) Use the template ETM5>ETM7>ETM1>ETM3>ETM2>ETM4 to extract the waveform anomaly information in the ETM image as information for identifying spectral anomalies of underground coal fire features;
(5) Compare the above two The similar remote sensing information is superimposed and integrated to generate an underground coal fire disaster remote sensing information map.
Figures 3?4?2 are the remote sensing anomaly information of the Rujigou coal fire extracted according to the above method. These include abnormal ETM7/4 ratio of images in coal fire areas, segmentation threshold 1.60, and 3×3 mean filtering; abnormal multi-band waveforms in coal fire areas, waveform matching threshold 0.04, and 3×3 mean filtering.
(2) Image processing method based on ASTER data
1. Data processing
ASTER used for coal fire detection research in Rujigou and Wuda coalfields The data are radiometrically and geometrically corrected ASTER?1B data. Among them, the Rujigou coal fire area data image is clear, rich in information, and free of banding and noise interference; the Wuda coal fire area image has some cloud cover, covering an area of ??less than 40, and the cloud-free area has high definition and no banding and noise interference. Noise interference can be used in coal fire detection research.
The projection mode of the original ASTER?1B data is UTM, Zone 48 North, and the datum is North America 1927. It needs to be projected and converted to GK Zone 18 (Pulkova 1942); then according to the coal fire area Image data is cropped based on geographical coordinate range. Since ASTER data have different spatial resolutions, in order to facilitate multi-band data combination and processing, the SWIR and TIR data are resampled and geometrically registered to make their image resolution consistent with the VNIR resolution, and generate a uniform spatial resolution ASTER multi-band image of the coal fire area.
2. Extraction of burnt rock remote sensing information
Figure 3-4-2 Rujigou coal fire disaster remote sensing information extraction image
The burnt rock information is Studying important indicators of coalfield fire areas is also an important basis for determining the migration direction of underground coal fires and the scope of fire changes. Taking the short-wave infrared spectral curves of the burned sandstone in the dead fire area at Point 9 and the burned rock in the burning area at Point 7 in the Rujigou coal fire area as target parameters, the spectral angle technology was selected based on the principle of waveform similarity to analyze each part of the coal fire area image. The pixels are automatically scanned one by one, and the image pixel information similar to the short-wave infrared waveform of the burnt rock is extracted at a given threshold as the burnt rock information distribution image. In the Rujigou coal fire area, the spectral angle threshold of burned sandstone was selected as 0.01, and the spectral angle threshold of burned coarse gravel sandstone in the combustion area was selected as 0.005. Figure 3?4?3 is the distribution map of burnt rock remote sensing information extracted using short-wave infrared data.
(3) Analysis of the extraction effect of coal fire disaster information
Read the center coordinates of the spectrum abnormal area on the remote sensing anomaly image of the Rujigou coal fire area, and search for it based on the GPS display in the field Image abnormal areas and investigate the ground properties of the abnormal areas.
Table 3?4?1 is a list of some ground inspections for remote sensing anomalies. In the Rujigou area, ground inspections were carried out for ETM7/4 anomalies, waveform anomalies and ASTER spectral anomalies. The main results show that the ETM7/4 abnormal area mainly reflects the burnt rocks and coal seam information of coal-bearing strata, and the waveform abnormal area mainly reflects the rock and coal fire information of coal-bearing strata.
The ASTER spectral anomaly area can partially reflect the underground coal seam combustion information, such as points 7 and 8; and partially reflect the coal seam information, such as point 11. Compared with ETM, the ASTER spectral anomaly area is mainly concentrated in the fire zone. It can be seen from the extracted remote sensing information images that the information distribution is concentrated and has certain regularity. It is mainly distributed near both sides of the Rujigou syncline axis in the northeast-southwest direction. The burnt rock overlying the coal at the eastern end of the syncline They are widely distributed, mostly on ridges and slopes. The distribution direction of burnt rock is consistent with the combustion history of underground coal fires in coal fire areas, and is distributed in extinguished fire areas; that is, it is mainly distributed in Baijigou Mining Area, Weidong Coal Mine, Dafeng Mine, Rujigou Mine and Gulaben Mine Surrounding area is larger. The evolution process of coal seam spontaneous combustion in this area can be seen from the distribution pattern of burned rocks. The spontaneous combustion of coal seams in this area develops from north to south along the two wings of the syncline. The rock information of the burning area is currently mainly distributed in Rujigou, Dafeng Mine, Weidong Coal Mine and Rujigou Taiyanggou area, which is consistent with the current fire situation and field inspection in the mining area. It can be seen that using multispectral information, some information related to coal fire disasters can be extracted. The images of the Rujigou coal fire area have the following conditions.
Figure 3-4-3 Extracted image of burnt rock remote sensing information in Rujigou coal fire area
Table 3-4-1 Typical remote sensing anomaly information field in Rujigou coal fire area Checklist
Continued list
(1) ETM7/4 anomalies are mostly distributed in small areas, with relatively high thermal anomaly values, which are consistent with the distribution range of coal seams. This may be related to the strong oxidation of surface rock soil due to the baking and burning of coal fires, which enriches the iron ions in the rock and soil and makes the texture dense and solid. It is a favorable basis for distinguishing burnt rock areas and active fire areas.
(2) The anomalies extracted using ETM5>ETM7>ETM1>ETM3>ETM2>ETM4 spectral features are mostly in point distribution, which may be due to coal seam burning changing the mineral composition and stratigraphic structure of the overlying strata in the burning area. , structure, weathering resistance and other characteristics. In places with relatively low surface thermal anomalies, it can be used as a basis for identifying dead fire areas.
(3) ASTER short-wave infrared data is more beneficial for identifying burnt rock information in burning areas and dead fire areas.