The application of remote sensing technique to identifying and classifying the quaternary sediments in south fringe of NAN XIANG basin

The application of remote sensing technique to identifying and classifying the quaternary sediments in south fringe of NAN XIANG basin

Zhao Buyi
Remote Sensing Application Center, Zhongshan university, 510275 China

Liu Yaotin
Wuhan Technital University of Survey and Mapping Wuhan, 430070 China

Abstract
In this paper, we have discussed the application of remote sensing "quantitative approach" to investigate of regional geology and national land in the quaternary sediment covered area, in order to improve the precision and efficiency. Based on the exploration of the typical profile of the Quaternary strata in the southern fringe of Nan-Xiang Basin, and combined with the regional geological investigation at the southern fringe of Nan-Xiang Basin, and combined with the regional geological investigation at the scale of 1:200000, we took the main aim at the method how to extract the information of the Quaternary layer from the remote sensing data. The J-M distance method is introduced as the criterion of selecting the optimum bands and band combinations. The methods of identifying and classifying the quaternary strata by using spectrum date assisted by TDM is discussed. On the basis of the transformation of geometrical projection for MSS images, we have tried to use computer to process the Quaternary geological map in the sub area (Guwan) at the scale of 1:50000.

The general situation of study area
With the development of economical construction, the investigation tasks of regional geology and national land in the quaternary sediment covered area are increasing rapidly. The investigations of the type of formation causes for Quaternary sediment, they're attributing, of Quaternary strata, and the mapping of Quaternary geology are the main point of the study.

These basic data are playing important roles in national planning and constructing. Generally, the landscape in this area is not rolling and was reconstructed by men. (e.g. cultivated)

It is difficult to enhance the precision and efficiency by using the conventional method and visual interpretation for the tasks above, especially for the precision requirements in the investigation of regional geology at medium scales 1:50000 to 1:200000, and the method is efficient for enhancing the precision and efficiency,

XiangFan is located at the Northern part of HuBei Province (112°00' ~ 113°00' 32°00' ~ 32040'N), and the area is about 6960.53 km², 80% of the study area (southern finge of Man-Xiang, Basin) is covered by the quaternary sediment. The strata of quaternary developed completely over there with bigger thickness and various type of contributing factors. There is close relationship between space distributing of Quaternary strata and the landforms there. (see

Table 1: The Table of Quaternary Strata in Nan-Xiang Basin and Visual Interpretation Symbol of Remote Sensing image.
Table 1 is missing

The study was done by combining typical section analysis of quaternary strata with remote sensing quantitative approach, that is, using synthetically method of Quaternary geology for typical section analysis establishing the order of strata, classifying stratum units and making details study in litho logy character of each unit and their depositing environment. The emphasis is put on using remote sensing techniques (especially, quantitative approach) for extracting information about Quaternary strata so that the space distribution of strata units is defined and the map of quaternary geology is finished.

Enhancement of Remote Sensing information of quaternary strata and extraction of the feature information
The main signs of interpretation on remote sensing image for quaternary strata are the difference of grey, texture landforms and the river systems as well as the frequency features. The information of Quaternary strata in the original image which was influenced by various factors during image formation is weak and boundaries between the types are indistinct. In order to extract the information, the image must be enhanced. (a) Contrast enhancement, used for extending contrasts between classes of Quanternary strata. (b) Operational enhancement such as addition, multiplication, that is values of the same pixel in bands were added or multiplied, quantitated and produced a new image, The result o this method is magnifying the distance of pixel values. For example, values of MSS 5 multiplying the values of MSS 4, it is obvious for distinguishing Q₂ from Q₂¹ Q₂ fom Q₃, Q₃ from Q₄. (c) Convolution filter, for enhancing the character of edges and texture of classed. After the operation of 3x3 convolution core, the texture of Q₃¹ And Q₃¹ are easier to distinguish, Q31 presents point shape and Q32 shows linear shape and the boundaries of classes are more obvious. (d) Ratio enhancement, available method for identifying Quanternary strata, but the key point is how to select optimum ration bands their combination, which are feature parameters of identification and classification, so J-M distance was introduced as quantitation standards for degree of separation for classes, and selections of features and optimum band combinations. For example, the degree of separation between two types of remote sensing data can be represented by J-M distance between two types.

The average J-M distance of different band ratio combinations can be calculated. The bigger J-M distance, the easier separation of classes in Table 2, we listed average J-M distance for different bands combinations.

The definition of J-M distance is as follows:

Where:
J_ij: J-M distance between class I and class j in afeature parameter;.
m_i: mean value of class j;
S_i: Covariance matrix of class j.
S_j: Covariance matrix of class j.

The definition of average J-M distance is as follows:

Where:
Java: average distance;
P(w_i): prior probability of class I;
P(w_j): Prior probability of class j.

Table 2 Average J-M distance of different band combination

Band combinations	average J-M distance
MSS7, MSS5, MSS4	0.4084
MSS7/MSS4/MSS4/MSS5, MSS5/MSS6	0.6548
MSS6/MSS7,MSS7/MSS5,MSS5/MSS4	0.2503
MSS7, MSS5	0.7236
MSS6/MSS5,MSS7/MSS4,MSS6/NSS5	0.2765
MSS7+MSS5/MSS7,MSS7+MSS5/MSS5,MSS7+MSS5/MSS	4 0.2627
K1 K2 K3	0.4537

(NOTE: G stands for topographic data image, k₁, k₂, k₃ separatedly for three major component images.)

It is known from the Table 2, that the optimal band combination is MSS7, MSS5 and G. The second is MSS7/MSS4, MSS4/MSS5 and MSS5/MSS4. For exaple Q₄, Q₃², Q₃¹, Q₃¹, Q₂² and Q₂¹ repressent different colors in the composed image of MSS7/MSS4, MSS4/MSS5, MSS5/MSS4.

Classified process of quaternary strata.
The combinations of optimal band for distinguishing Quanternary strata were decided by selecting character parameters of the original image. In order to know the distribution attribute of the Quanternary strata, we try to make data processing according to different classes. Since the resolving powers of the space and spectrum of the sensor are limited, it is difficulty to get expecting results of classification only using spectral data. In order to improve the classifying precision, we put forward a method using spectrum data assisted by DTM data, In view of the fact that the quaternary strata in the study area present, the feature of ladder like cut, this is, at different times the strata appear at different heights, and the strata at the same time emerge at almost the same heights, it is helpful to use this method for improving the precision. The size of grids for forming DTM is 1x1 cm². The data of altitude were collected on the topographer map of 1:50000, and input into the digital image-processing device. In the device the data were amplified and resample, and the DTM image was formed, which was matched with remote sensing data for classified processing.

In order to get better classification result, we compared the advantages and disadvantages of different schemes of classifications, and we chosed the improved mixed scheme of supervised and unsupervised method. The main steps of this method are: (a) making unsupervised classifications in order to have a general idea about the whole area though this rough classification for later selecting of training units. (b) Deciding the training units. (c) Making supervised classification in the whole study area.

The method, which we took, is not only simple (showing one of the advantages of unsupervised classification), but also of high precision (presenting one of the advantages of supervised classification). And it can give better results too, which are proved to be satisfactory through field check.

Geometric transfromation of mass image and automatic mapping in the study area.
Since the image taken by satellite is centric projection and geological mapping requires its corresponding fundamental base (gauss-Kruger projection), it is necessary to make geometric transformation for the image, so that the geometric distortion of the image can be deleted. The geometric transformation of the image is actually done through (a) digital analoging of geometricd distortion f the original image which was found by comparing with ground control points; (b) defining relationship of space coordinates between original image and standard space; (c) and transferring all pixels of the original image to the standard space. The whole process was done by computer. It included two steps: (a) transforming of geometric positions of pixels and determine conjugate position in the original image of every pixed in the standard space; (b) determing grey values of conjugate points and transferring these values to the standard space.

In order to determine the grey values of grey values of conjugate points, resampling needs to be done, which is combined with the selection of control points. Three mathematic approaches through which the shift of projections is accomplished in the study area are mentioned here of cubic convolution can give better results.

After geometric transformation, the image can match with the topographic map, with the digital device we got the base file. The boundaries of various types on the classified image were used to form the graphic file. Through connecting interface with drawer, and adding letterings and colors, we got a computer drawn color map of the quaternary geology at the scale of 1:50000 in the sabarea (juwan) of the study area.

Conclusions

The method which we took-connecting typical section investigation on the spot with remote sensing echniques (including visual interpretation and quantitative analysis), is an effective way in which the precision and efficiency can be improved.
In roder to get better results of classification, J-M distance between various types was introduced as the basis for selecting optimum bands and band combinations, and analysing the degree of differences between various types, with which satisfying results can be achieved. In the classification of the quaternary strats, the method of using spectrum data assisted by DTM along with the combination of supervised and unsupervised approaches is a perfect way for improving the results of classification.
We have tried to use computer to draw the map of the quaternary geology based on geometric transformation of MSS image, and extraction for remote sensing data of the quaternary strata. The precision of this map can mainly meet the fundamental demand of the classification.

References

Zhao Buyi, Remote sensing technique and geological application, 1986
Quiel. F, Trends and development in the classification of multispectral data ISPRS. No. 9 1987.