(Aerogeophysical Remote Sensing Center, Ministry of Geology and Mineral Resources, Beijing 100083)
The phenomenon of "banding" (below without quotation marks) in the aerial geophysical exploration results map widely exists in various parameter results maps and conversion processing maps, especially in the aerial gamma-ray spectrum measurement results map. The strip in the map not only affects the beauty of the map, but also distorts the geological structure characteristics, which brings inconvenience to the interpretation work. Therefore, the research on stripping methods and technologies has been widely concerned by peers at home and abroad. The published methods mainly include ripple filtering method (Wang Maoji et al., 199 1), image restoration technology (Zhang Yujun et al., 1990), observation and correction of daily variation of atmospheric radon (Shui Enhai et al., 1987) and correlation analysis method (A.A.
The causes of banding are quite complicated, which can roughly include two aspects, that is, the measurement conditions are inconsistent (meteorology, temperature, humidity, time, atmospheric radiation background, flight altitude, etc.). ) and improper data processing. Although the causes of the bands are various, they share a common feature, namely, the long wave anomaly along the survey line and the rectangular wave anomaly perpendicular to the survey line (step-like). Because any geophysical anomaly has the characteristics of continuous gradual change and there is no step jump, a special filter can be designed to remove band interference.
First, the method principle.
Stripe, in essence, is a strip-shaped field value increasing and decreasing profile with a certain width that appears in the aerogeophysical map and extends along the survey line direction. It is characterized by a long wavelength along the survey line direction and a stepped feature perpendicular to the survey line direction. In other words, stripes are the exception of culture. Because the regional geophysical field has the characteristics of continuous gradual change, there is no step jump. Therefore, in theory, a special filter can be designed to remove this band, while retaining the inherent characteristics of the regional geophysical field to the greatest extent.
(A) directional filtering method
Because fringes are long-wavelength anomalies distributed along the direction of the survey line, fringe anomalies can be separated theoretically by directional filtering. By comparing the main characteristics of anomalies in spatial domain and frequency domain, we can draw the following conclusions: spatial domain has a greater resolution for horizontal superposition anomalies, while frequency domain has a greater resolution for vertical superposition anomalies. Therefore, the directional filtering method of fringes should be carried out in spatial domain.
It is assumed that the measurement data P(x, y) is composed of real data f(x, y) and strip data t(y), that is
Selected Papers on Aerogeophysical Remote Sensing Research
Where x is the survey line direction and y is the direction perpendicular to the survey line. For the one-dimensional filter operator H(x) acting on P(x, y), there are
Selected Papers on Aerogeophysical Remote Sensing Research
If the one-dimensional filter operator H(y) acts on H(x)P(x, y), there are
Selected Papers on Aerogeophysical Remote Sensing Research
Subtract formula (3) from formula (2) to obtain
S(x，y)=H(x)f(x，y)-H(y)H(x)f(x，y)-H(y)t(y)
Then there is
Selected Papers on Aerogeophysical Remote Sensing Research
Through human-computer interaction or statistical analysis, we can find a suitable H(y) to make S(x, y)→0, so there are
Selected Papers on Aerogeophysical Remote Sensing Research
Substitute equation (5) into equation (1) and move each term, and the calculation formula of data correction is as follows.
Selected Papers on Aerogeophysical Remote Sensing Research
(B) differential commercial law
As mentioned above, the strips are bar codes of the same shape distributed along the direction of the survey line. Therefore, banding anomalies can be extracted and eliminated through enhancement, computer recognition and human-computer interaction.
1. Stripping exception type
Seen from the plane, the bar is similar to a bar code. However, strip data and useful data are mixed together, and there is still no suitable filter to completely separate them. Therefore, the method of identifying and separating stripes is put forward by studying the manifestations of stripes on the cross section of vertical survey line.
In cross section, the strip is stepped. If the horizontal difference quotient is calculated along the profile, single pulse anomaly, double pulse anomaly, positive and negative pulse anomaly and multi-pulse anomaly can be obtained, as shown in figure 1.
Figure 1 ideal pattern diagram of banding anomaly
1- The original field curve has no band; 2- original field curve with strip; 3- horizontal difference quotient curve without band; 4— Horizontal Difference Quotient Curve with Strip
2. Enhancement of banding anomalies
In practical work, in addition to the horizontal difference quotient of stripes, various errors will also cause the above pulse anomalies. Therefore, it is necessary to enhance the pulse anomalies caused by banding and suppression errors.
Firstly, it is necessary to filter the original data along the survey line direction to reduce the interference of random errors. Secondly, the cross section should be set in the static field to reduce the interference of useful anomalies. Secondly, the horizontal difference quotient is transformed by exponent or power to enhance the banding anomaly (because the amplitude of the horizontal difference quotient pulse of banding anomaly is usually greater than the pulse anomaly caused by accidental error). Fig. 2 shows the contour curves of horizontal difference quotient before and after power conversion. The profile curve (1) before reconstruction has obvious weak jump anomaly; The transformed profile curve (2) has only band anomalies, and the weak jump anomalies basically do not exist.
3. Identification of banded anomalies
After the above enhancement and suppression, banded anomalies usually show strong and steep peak anomalies, while other genetic anomalies are weak peak anomalies or wide and slow anomalies, as shown in Figure 2. As can be seen from the figure, the visual method can well distinguish banded anomalies from non-banded anomalies. Similarly, an appropriate strain filter can be designed to identify banding anomalies.
Fig. 2 contrast diagram before and after abnormal strengthening of the strip.
1-before enhancement; 2- After enhancement
4. Removal of abnormal bands
Let the horizontal difference quotient on both sides of the abnormal bar be △P 1 and △P2, then the points in the abnormal bar can be linearly interpolated with the values of these two points. After interpolating all the banded anomalies, the regional field after banded removal is obtained by integrating along the profile. To facilitate understanding, here are some examples.
There is a cross-sectional data series {-60, -50, -40, 50, 60, -30, 0, 10, 20}, and its first-order forward difference quotient △P is {*, 10, 10, 90. 10, 10, 10, 10, {10} (* is no value), then the cross-sectional data sequence f after strip removal is {-60,-. The above process can be shown in fig. 3.
The above example shows that this method can not flatten the strip by traditional filtering method. Therefore, this method can keep the distribution characteristics of real data well. Fig. 4 is a real cross-sectional view before and after correction, which further shows the effectiveness of this method.
Fig. 3 Schematic diagram of striping process by horizontal difference method
A- original cross section p; B- original horizontal difference quotient △ p; C- corrected horizontal difference quotient △ f; D- corrected cross section
Fig. 4 Comparison of cross-sectional curves in Kanggurtag area
1-original section curve; 2- corrected contour curve; 3— horizontal difference quotient curve before correction; 4- Corrected horizontal difference quotient curve
5. Stripe data in the original data is eliminated.
Where m sections P 1(l= 1, 2, ..., m), and the corrected data is F 1, the corrected value of point j in the ith row in this area can be obtained as follows.
Selected Papers on Aerogeophysical Remote Sensing Research
Therefore, the correction value of point j of line I is
Selected Papers on Aerogeophysical Remote Sensing Research
Where pij and fij are the values of I-line J before and after correction, respectively.
Second, the application example
The application software based on PC is developed for the above methods, and it is used to process the airborne gamma-ray spectrum measurement data in Qaidam Basin in Qinghai and Kanggurtag area in Xinjiang, as well as the vertical first derivative data of airborne magnetism in Chaoshui-Yabulai Basin in Gansu and Yili area in Xinjiang, and achieved good results. The following describes the application in Kanggurtag area.
(1) General situation of airborne geophysical prospecting and geology in Kanggurtag area
1990, the Aerogeophysical Remote Sensing Center completed a comprehensive survey of1:25,000 aerogeophysical prospecting (magnetic, electrical and gamma spectrum) with an area of 6300km2 in Kangurtage, Xinjiang, and the survey line was north-south, obtaining high-quality raw data. Nevertheless, there are still obvious bands in the north-south direction on the plane isoline map of airborne gamma spectrum (Figure 5). The main geological structure framework in this area (Figure 6) is as follows: the Kanggurtag deep fault zone is the main fault zone in the north, and the secondary faults on both sides basically strike northwest, northeast and near east-west; Yamansu fault zone is the main fault zone in the south, and the secondary faults on both sides of it are NW-trending and NE-trending. The geological bodies with high content of radioactive elements in this area are mainly intermediate-acid intrusive rocks, and their long axis direction is consistent with the distribution direction of fault structures in this area. In addition, it can be seen from the aeromagnetic result map that there are no main geological structures in the north-south direction in this area. Therefore, the spectral bands in airborne gamma-ray spectrogram are of non-geological origin and should be eliminated.
(2) Stripping treatment
Due to the limitation of space, this paper only introduces the process of eliminating banding interference in potassium content data of aviation gamma-ray spectrum.
Fig. 5 Isogram of original data of potassium content in airborne gamma-ray spectrum in Kangurtage area.
Fig. 6 Schematic diagram of geological structure in Kanggurtag area
1- intermediate acid intrusive rocks; 2- Qiugemintash-Huangshan ductile shear zone; 3- Archie Mountain-Amansu Island Arc Belt; 4- Tuha Depression; 5- fracture
Fringe processing by 1. directional filtering method
According to the length (or width) of the local geological body in the north-south direction in this area, it can reach 10km, and banded anomalies generally run through the north and south. Repeated tests show that the filtering radius along the survey line is 15km, and the effect is better.
When filtering along the direction perpendicular to the survey line, the banding effect under different filtering radii r is tested. When r is 1km, most of the strips (usually strips with small width) can be removed, and the regional field is basically undistorted. When the value of r increases gradually, more and more bands are removed, and the distortion phenomenon of regional field is gradually revealed. When R reaches 9km, there are no stripes on the map, but the regional field is obviously distorted.
2. Horizontal differential stripping process
Firstly, the filtering radius is 0.7km along the survey line, which suppresses the interference that noise and other factors may cause to the abnormal identification zone. Then make the filtered data into images and display them on the computer screen; Draw a horizontal line on the vertical bar with no obvious abnormality to get the profile data, use special software to calculate the wechat service of the horizontal line of the profile data and display it on the computer screen (Figure 4), then mark the position of the bar by visual inspection, and the computer will automatically correct it; Finally, the results are displayed, checked and corrected with images. The above process has been repeated for dozens of times and satisfactory results have been obtained.
Fig. 7 Isogram of potassium content in Kangurtage area after stripping by airborne gamma-ray spectrometry.
3. Comprehensive application of horizontal differential service method and directional filtering method in stripping process.
First, the horizontal difference method is repeated for more than ten times until the wide and strong band anomalies no longer appear in the image. Then the above results are filtered along the survey line with the filtering radius of 15km, and the regional anomaly is obtained. Filtering along the direction perpendicular to the survey line with a filtering radius of 0.8km can well eliminate banding anomalies. Finally, the original data is corrected with the above results to obtain corrected data, as shown in Figure 7.
By comparing Figure 5, Figure 6 and Figure 7, it can be found that the corrected data well retains the basic characteristics of the original data, and the banding interference no longer exists, thus providing an original map with more reliable quality for interpreters.
Three. Concluding remarks
In order to better explain the airborne geophysical data, all kinds of interference including banding should be eliminated before mapping. The striping method proposed in this paper has been applied in four work areas, and the results show that: ① The original basic features are well preserved while the striping interference is eliminated. ② Horizontal difference commercial method and directional filtering method have their own advantages and disadvantages. The former has high fidelity but tedious, while the latter is convenient but low fidelity. When the two methods are used together, the former is used to remove the strips with larger width and amplitude, and the latter is used to remove the strips with smaller width and amplitude, and the effect is better.
Compared with the existing methods at home and abroad, this method has the following advantages: ① this method can be used to process not only airborne gamma-ray spectrum measurement data, but also aeromagnetic data; (2) Because this method corrects the background field, it will not cause abnormal distortion, which is necessary to maintain the basic characteristics of the original data; (3) The developed software is integrated by man-machine dialogue, and the intermediate results are displayed in color images, which makes the marking process simple and clear, and experts can control the marking process at any time.
refer to
1. Bear. Transformation and filtering technology of magnetic (gravity) anomalies. Beijing: Metallurgical Industry Press, 1990.
2. Interpretation method and application of airborne geophysical prospecting. Beijing: Geological Publishing House, 1992.
3. The Green One, translated by Wang Hong, etc. Correction of airborne gamma radiation data by using the relationship between channels. Aerial remote sensing, 1989, (4): 92 ~ 98.
A new technique for eliminating stripe interference of airborne geophysical data
Fan zhengguo
(Aerogeophysical Remote Sensing Center, Beijing 100083)
abstract
Stripe interference in aerogeophysical maps is a difficult problem that has not been satisfactorily solved. In this paper, two new methods to eliminate stripe interference of airborne geophysical data are proposed. According to the gradual change characteristics of regional geophysical field, the long wavelength characteristics of fringe interference along the survey line, and the sudden change and short wavelength characteristics of fringe interference perpendicular to the survey line, this technology uses fringe edge enhancement technology and computer (vision) recognition technology to extract fringe interference data, and uses special filters to eliminate fringe interference. Practice shows that this technology can not only eliminate the fringe interference, but also preserve the basic characteristics of the original data.