**4. Application of geochemical image in geology**

The formed geochemical atlas can provide the prospecting target area just like conventional geochemical method and may also conduct multielement geochemical analysis. The geochemical image can accomplish the structural interpretation, e.g., linear structure and ring structure in geology just like what is fulfilled by the optical remote sensing. This paper does not restate these traditional methods but will mainly introduce the following three kinds of application in geology in the northern Daxinganling metallogenic belt.

#### **4.1 Assisting in geological mapping**

The geochemical atlas of 39 geochemical elements was generated in the northern Daxinganling metallogenic belt, including major elements and trace elements. The full use of all the elements will better assist geological mapping. Especially, in the Daxinganling Mountains, the outcrop is scarce because of the forest cover and that the field work of geological mapping encounters a great deal of difficulties. Therefore, boundaries of the geological bodies are indistinct, and the final boundaries are somehow judged by subjective experience. To employ unsupervised classification method may provide the reference for determining the boundaries of rock in the working area. As shown in **Figure 11**, the Chaihe area in the northern Daxinganling metallogenic belt was taken as an example; this working area belongs to stream sediment survey of the 1:200,000 Wuchagou sheet. The 39 geochemical element images are classified by K-Mean classification, and the geological interpretation map is created as the following one.

It can be seen that the geochemical mapping (**Figure 12**) may relatively clearly distinguish γ<sup>5</sup> 2(2) alkali feldspar granite from monzogranite. However, the boundary is different from that in the geological map (**Figure 11**). In the north and south, it was verified; but in the east of the map sheet, the rock which was delineated by geochemical images (**Figure 12**) was not presented in the geological map (**Figure 11**). Other Wuchagou basalt can also be easy to identify; two signs were manifested in the north, same as the geological map. Because Baoshi formation and Fujiawazi formation are volcanic, it is sometimes difficult to classify them. As a result, the interpenetration phenomenon is frequent. In the field work, it is hard to distinguish the volcanic rocks. For example, both Fujiawazi formation and Baoshi formation contain tuff; sometimes, the difference between intermediate lava and acidic lava is weak in the field. In this case, the divided geological map is worse than the geochemical classification.

#### **4.2 Prospecting target selection**

There is plenty of research on the methods of the prospecting target selecting using data-driven and knowledge-driven modes. In the past, selecting prospecting area was primarily based on the anomaly of the major ore-forming elements. The area with high anomaly value of a single element or integrated anomalies was selected

**63**

**Figure 12.**

*same as* **Figure 12***).*

**Figure 11.**

*Geological sketch of Chaihe area (1:200,000).*

*The Geochemical Data Imaging and Application in Geoscience: Taking the Northern…*

as prospecting target. Although the large area of geochemical working had been carried out, fewer researchers utilize all elements for prospecting target selecting. Combining the characteristics of the geochemical atlas in the northern Daxinganling metallogenic belt, the geochemical spectrum method may be adopted to exhibit the similarity with the known deposits on target locating. The most frequently used method is spectral angle mapper method (SAM) [33, 34]. SAM method utilize N-dimensions angle to match image elements and reference

*Map of each unit of K-Mean classification of 39 geochemical elements in Chaihe area (legend codes are the* 

*DOI: http://dx.doi.org/10.5772/intechopen.84725*

*The Geochemical Data Imaging and Application in Geoscience: Taking the Northern… DOI: http://dx.doi.org/10.5772/intechopen.84725*

#### **Figure 11.**

*Applied Geochemistry with Case Studies on Geological Formations, Exploration Techniques…*

**4. Application of geochemical image in geology**

Daxinganling metallogenic belt.

**4.1 Assisting in geological mapping**

tation map is created as the following one.

**4.2 Prospecting target selection**

distinguish γ<sup>5</sup>

firstly gives category, whereas the unsupervised one is determined by the statistics characteristics of image data itself. The classification method used for remote images are suitable for the geochemical atlas. Usually employed methods include multilevel slice classifier, decision tree classifier, minimum distance classifier, maximum likelihood classifier, and the like (e.g., method of fuzzy theory, expert system method, etc.). SAM method mentioned later is one of the supervised classification

The formed geochemical atlas can provide the prospecting target area just like conventional geochemical method and may also conduct multielement geochemical analysis. The geochemical image can accomplish the structural interpretation, e.g., linear structure and ring structure in geology just like what is fulfilled by the optical remote sensing. This paper does not restate these traditional methods but will mainly introduce the following three kinds of application in geology in the northern

The geochemical atlas of 39 geochemical elements was generated in the northern

Daxinganling metallogenic belt, including major elements and trace elements. The full use of all the elements will better assist geological mapping. Especially, in the Daxinganling Mountains, the outcrop is scarce because of the forest cover and that the field work of geological mapping encounters a great deal of difficulties. Therefore, boundaries of the geological bodies are indistinct, and the final boundaries are somehow judged by subjective experience. To employ unsupervised classification method may provide the reference for determining the boundaries of rock in the working area. As shown in **Figure 11**, the Chaihe area in the northern Daxinganling metallogenic belt was taken as an example; this working area belongs to stream sediment survey of the 1:200,000 Wuchagou sheet. The 39 geochemical element images are classified by K-Mean classification, and the geological interpre-

It can be seen that the geochemical mapping (**Figure 12**) may relatively clearly

is different from that in the geological map (**Figure 11**). In the north and south, it was verified; but in the east of the map sheet, the rock which was delineated by geochemical images (**Figure 12**) was not presented in the geological map (**Figure 11**). Other Wuchagou basalt can also be easy to identify; two signs were manifested in the north, same as the geological map. Because Baoshi formation and Fujiawazi formation are volcanic, it is sometimes difficult to classify them. As a result, the interpenetration phenomenon is frequent. In the field work, it is hard to distinguish the volcanic rocks. For example, both Fujiawazi formation and Baoshi formation contain tuff; sometimes, the difference between intermediate lava and acidic lava is weak in the field. In this case, the divided geological map is worse than the geochemical classification.

There is plenty of research on the methods of the prospecting target selecting using data-driven and knowledge-driven modes. In the past, selecting prospecting area was primarily based on the anomaly of the major ore-forming elements. The area with high anomaly value of a single element or integrated anomalies was selected

2(2) alkali feldspar granite from monzogranite. However, the boundary

**62**

methods.

*Geological sketch of Chaihe area (1:200,000).*

#### **Figure 12.**

*Map of each unit of K-Mean classification of 39 geochemical elements in Chaihe area (legend codes are the same as* **Figure 12***).*

as prospecting target. Although the large area of geochemical working had been carried out, fewer researchers utilize all elements for prospecting target selecting.

Combining the characteristics of the geochemical atlas in the northern Daxinganling metallogenic belt, the geochemical spectrum method may be adopted to exhibit the similarity with the known deposits on target locating. The most frequently used method is spectral angle mapper method (SAM) [33, 34]. SAM method utilize N-dimensions angle to match image elements and reference

spectra. The geochemical spectra were regarded as vector, whose number of dimensions is the same as the number of waveband. Then using the angle algorithm for calculating the angles' inter-element geochemical spectra, the similarity of two geochemical spectra could be determined. The geochemical spectra of

#### **Figure 13.**

*Comparison map between the prospecting target and actual deposits in the Manzhouli region. The prospecting target was obtained by applying spectra angle method to some porphyry copper-molybdenum deposit; a is spectral angle map; b is the prospecting target formed by threshold segmentation of spectral angle map; and c is the corresponding location map between prospecting target and actual deposit. A is Wunugetushan deposit, B is Babayi deposit, and C is Badaguan deposit.*

**65**

attained by them.

*The Geochemical Data Imaging and Application in Geoscience: Taking the Northern…*

in N-dimensions space. The smaller angle indicates that it fits better

results have been verified by deposits of the same type (**Figure 13**).

stratigraphic unit classification in geological mapping.

**5. Discuss and future prospects**

locations on the known deposits are regarded as end-member spectra, and then SAM is used to compare end-member spectra with the angles of each pixel vector

with the geochemical spectra of the discovered deposits. This method fully utilized the information of geochemical spectra and makes every elements involved in the classification. Additionally it emphasizes the shape characteristics of the geochemical content and greatly reduces the information such as the main ore-

In this study, the geochemical spectral of the Wunugetushan copper deposit was taken as reference spectra, the SAM method was adopted, and the classification

Since the late 1950s, Webb and his colleagues presented to collect fine granular sedimentary from drainages which stands for the average content of the catchment basins [35]. The subsequent regional geochemistry survey mainly based on their theory and method, namely, the sample collected, may stand for the contribution of all matters in the surrounding area of this sampling position. This is the same as to the so-called mixed spectra in remote sensing. Because large areal distributional mixed pixels evidently affect the calculation and classification of the remote sensing image, many researchers put forward the method of decomposing mixed pixels. Nowadays, methods of decomposition of mixed pixels are mainly classified into two classes, one is the linear spectral decomposition, which is based on the linear additivity of brightness of pixels, and the other is the fuzzy decomposition method. In the process of geological mapping, the stratigraphic unit needs to be divided, and it includes various kinds of rocks. The Manitu formation on the Xiaodonggou section in the northern Daxinganling metallogenic belt served as an example. The standard strata, which were distributed between upper Baiyingaolao formation and lower Manketouebo formation, are 690.6 m thick. From bottom to top, the section involves green andesite (101.6 m), light gray andesitic-rhyolitic breccia tuff (219 m), dark gray, yellow gray andesite interlayered with debris tuff (190.3 m), and purple gray-dark andesite (179.7 m). In the fieldwork, it is difficult to observe all the rock types mentioned due to a few outcrops. As a result, the stratigraphic division can only be based on the limited artificial outcrops. Under this condition, the method of decomposing mixed pixels was used. Through decomposing the mixed pixels, the shares of various kinds of rocks can be achieved; thus it can assist

In the past, regional geochemistry has made significant achievements in geology and mineral exploration. However, all of these relied on vector data, and the number of geochemical elements is limited, which narrowed the application of geochemical data. This paper only aims to supplement and modify the shortcomings of previous methods, rather than to overthrow or criticize the achievements

The rasterized geochemical image possesses many advantages. The geochemical image is vivid for the visual interpretation. Additionally, data can be compatible for statistical analysis. That vectorized geochemical data accomplished can be achieved by the rasterized data in most cases. Furthermore, the imaged geochemical data could be processed with hyperspectral tools, which cannot be used in vector data.

*DOI: http://dx.doi.org/10.5772/intechopen.84725*

**4.3 The classification on mixed rocks**

forming elements.

#### *The Geochemical Data Imaging and Application in Geoscience: Taking the Northern… DOI: http://dx.doi.org/10.5772/intechopen.84725*

locations on the known deposits are regarded as end-member spectra, and then SAM is used to compare end-member spectra with the angles of each pixel vector in N-dimensions space. The smaller angle indicates that it fits better with the geochemical spectra of the discovered deposits. This method fully utilized the information of geochemical spectra and makes every elements involved in the classification. Additionally it emphasizes the shape characteristics of the geochemical content and greatly reduces the information such as the main oreforming elements.

In this study, the geochemical spectral of the Wunugetushan copper deposit was taken as reference spectra, the SAM method was adopted, and the classification results have been verified by deposits of the same type (**Figure 13**).

## **4.3 The classification on mixed rocks**

*Applied Geochemistry with Case Studies on Geological Formations, Exploration Techniques…*

spectra. The geochemical spectra were regarded as vector, whose number of dimensions is the same as the number of waveband. Then using the angle algorithm for calculating the angles' inter-element geochemical spectra, the similarity of two geochemical spectra could be determined. The geochemical spectra of

*Comparison map between the prospecting target and actual deposits in the Manzhouli region. The prospecting target was obtained by applying spectra angle method to some porphyry copper-molybdenum deposit; a is spectral angle map; b is the prospecting target formed by threshold segmentation of spectral angle map; and c is the corresponding location map between prospecting target and actual deposit. A is Wunugetushan deposit,* 

**64**

**Figure 13.**

*B is Babayi deposit, and C is Badaguan deposit.*

Since the late 1950s, Webb and his colleagues presented to collect fine granular sedimentary from drainages which stands for the average content of the catchment basins [35]. The subsequent regional geochemistry survey mainly based on their theory and method, namely, the sample collected, may stand for the contribution of all matters in the surrounding area of this sampling position. This is the same as to the so-called mixed spectra in remote sensing. Because large areal distributional mixed pixels evidently affect the calculation and classification of the remote sensing image, many researchers put forward the method of decomposing mixed pixels. Nowadays, methods of decomposition of mixed pixels are mainly classified into two classes, one is the linear spectral decomposition, which is based on the linear additivity of brightness of pixels, and the other is the fuzzy decomposition method.

In the process of geological mapping, the stratigraphic unit needs to be divided, and it includes various kinds of rocks. The Manitu formation on the Xiaodonggou section in the northern Daxinganling metallogenic belt served as an example. The standard strata, which were distributed between upper Baiyingaolao formation and lower Manketouebo formation, are 690.6 m thick. From bottom to top, the section involves green andesite (101.6 m), light gray andesitic-rhyolitic breccia tuff (219 m), dark gray, yellow gray andesite interlayered with debris tuff (190.3 m), and purple gray-dark andesite (179.7 m). In the fieldwork, it is difficult to observe all the rock types mentioned due to a few outcrops. As a result, the stratigraphic division can only be based on the limited artificial outcrops. Under this condition, the method of decomposing mixed pixels was used. Through decomposing the mixed pixels, the shares of various kinds of rocks can be achieved; thus it can assist stratigraphic unit classification in geological mapping.
