Remote Sensing Interpretation and Mineralization Information Extraction

(a) Comprehensive interpretation of remote sensing

Remote sensing interpretation is the basic work of salt lake mineral resources investigation, which runs through the whole process and is roughly divided into three stages: preliminary interpretation, field verification and comprehensive arrangement. In the whole region,1∶ 500,000 TM 432 standard false color composite image is mainly used for visual interpretation; The main metallogenic favorable zones are 1 ∶ 10000 and 1 ∶ 250000, and the metallogenic prospect area (prospecting target area) is 1 ∶ 50000 ETM432 standard false color composite image, which is completed by man-machine interactive interpretation method. In order to give full play to the advantages of remote sensing technology, multi-scale, multi-temporal and multi-component remote sensing images are used for auxiliary interpretation. Specific interpretation methods include: direct interpretation, comparison, logical reasoning, image processing and so on. The explanation mainly includes lakes, regional metallogenic background and metallogenic information related to the formation of mineral resources in lakes and salt lakes.

(1) Lakes: including freshwater lakes, saltwater lakes, salt (water) lakes, semi-dry lakes, dry lakes and hydrochemical lakes.

(2) Regional metallogenic background: including strata, magmatic rocks, geological structures (active structures) and landforms. See Table 2-2 for main stratigraphic interpretation marks and Table 2-3 for active structural interpretation marks in this area.

(3) Metallogenic information: including brine, brine containing crystalline salt bed, dry salt bed, wet salt bed, salt bed, ore-bearing formation, salt-bearing stratum, etc. See Table 2-4 for the interpretation marks of the main salt-bearing strata in the area.

(2) Extraction of remote sensing metallogenic information

Image processing is an important means to extract remote sensing metallogenic information. In order to improve the resolution of TM and ETM ++ images and obtain more and better useful information, according to the spectral test results, the methods of local contrast enhancement, anti-frequency enhancement, ratio synthesis and principal component analysis are mainly used to enhance and extract metallogenic information.

1. local contrast enhancement

Using the statistical results of image data in a local range, the variables are changed to the parts that need to be enhanced to highlight the linear image information. This method is mainly used to explain the metallogenic background.

Table 2-2 Interpretation Signs of Main Strata in Tibet

sequential

sequential

sequential

Table 2-3 Interpretation Signs of Active Structures in Tibet (TM74 1)

2. Anti-frequency enhancement

The anti-frequency enhancement of lake water can enrich water information. This method can suppress the reflection information of the main component (water body), enhance the reflection of weak information, and reflect the metallogenic information to some extent (plates 39 and 40).

3. False color equal density segmentation

False color equal density segmentation replaces image density with color, which can improve people's ability to distinguish density differences. In this project, this method is applied to the comparative analysis and research of lake mineral content (see chapter 4 for details).

Table 2-4 Interpretation Marks of Salt-bearing Strata in Tibet (TM74 1)

4. Ratio processing

Ratio processing is one of the most commonly used methods to enhance and extract TM(ETM++) data information. According to the spectral test results of the main salt minerals in the area, the basic ratio method, namely TM2/TM7, is selected to distinguish the salt deposition components (see Chapter 4 for details).

5. Principal component analysis

Principal component analysis (PCA) is an image transformation processing algorithm, which can remove redundant information between bands and compress multi-band image information into several effective conversion bands than the original band, so as to enhance and analyze the characteristics of remote sensing images simply and effectively. Its characteristics can be summarized as follows: ① the sum of variance before and after transformation remains unchanged, but the original variance is redistributed to the new component image; (2) The first component gets most of the variance (generally above 80%), and the variance of other components decreases in turn, and the amount of information also decreases sharply; ③ The information quantity and information characteristics reflected by each component are different, and the correlation coefficient is 0 or close to 0; ④ The first component is equivalent to the weighted sum of the original band, and the weight is directly proportional to the variance of the band, reflecting the total reflection intensity of the ground object. Other combinations are equivalent to weighted difference images of different components. According to the research purpose and the requirements of information characteristics, appropriate components can be selected to extract thematic information (see Chapter 4 for details).

(3) Verification of interpretation marks

In order to prove the reliability and accuracy of the interpretation marks, the project team verified the interpretation marks by remote sensing, and selected 2 ~ 3 lakes of the same type in different areas for verification, and each sample was sampled in a representative part of the lake. That is, according to the characteristics of remote sensing images, combined with the hydrogeological conditions of lakes, we generally avoid the river entrance and choose to shoot in a relatively stable water body with a certain depth and moderate concentration.

1. Verification of lake type interpretation mark

Eight lakes have been verified, including 2 freshwater lakes, 2 saltwater lakes and 4 salt lakes. Except the sampling analysis results are inconsistent with the interpretation results, the other seven lakes are all correct, and the interpretation accuracy rate is 88%. See Table 2-5 for the verification results.

Table 2-5 Verification Table for Interpretation Marks of Lake Types

2. Verification of lake hydrochemical type interpretation marks

There are 7 lakes, including 2 carbonate types, 2 sodium sulfate subtypes and 3 magnesium sulfate subtypes. Except for Siza's mistake, everything else is correct, with an accuracy rate of 86%. See Table 2-6 and Table 2-7 for the analysis results of the verification lake.

Table 2-6 Verification Table of Analysis Results of Lake Hydrochemistry Types

Table 2-7 Verification Table of Lake Hydrochemistry Types