GISdevelopment.net ---> AARS ---> ACRS 1989 ---> Poster Session 2

Land Use Mapping by Remote Sensing

Zhou Bomin

Since 1960’s being not satisfied with his with his perceptual knowledge on the physical elements of the Earth, man has progressed to learn the relationship between man and his environment, to know and control the dynamic laws governing the environmental changes and the changes of natural resources. China is vest in territory and rich inland resources, but the land shared by each person is rather less because of its large population and therefore the land resources are very precious. The further research and mapping of land resources series of the whole country are fundamental work of land management of China.

The paper will discuss and summarize some procedures and methods on land use map compilation by using remote sensing information.

I. Map-liked Processing of Remote Sensing Information

I. Mathematical Basis
A map should be a conformal projection, this is the say that all points within the map sheet are equally scaled, while an uncorrected remote sensing image is a multi-centered projection , all points with the picture are differently scaled. However it must be pointed out that an unrefined multi-characteristic demonstrated by the conformal projection, i.e. it has an approximate map projection. When wanted map materials lack and the same named object points can not be obtained for photograph rectification, the unrefined image can be expediently used as a map. A refined picture is generally controlled by the same-named object points can not be obtained for photograph rectification, the unrefined image can be expediently used as a map. A refined picture is generally controlled by the same-named object points and at least four such control points are selected for the purpose. The pixel plane needs to be displaced in order that the accuracy requirement set by the same kind of maps is to be meted. The projection difference caused by altitude difference of an object can be computed according to the equation dh=h/H.r.n., As a rule, the maximum projection deference of a pixel should be. SSmaller than 0.5mm. Ordinarily, projection difference can be neglected in land use man compilation. After its geometric correction by pixel plane displacement, a remote sensing image can have the same mathematical basis and geometric accuracy as a map.

2. Symbol and Annotation System
Transmitting information by coordinate combination of symbols and annotions, a map is readable without any difficulty. Although remote sensing imagery is abundant in panoramic perpendicular images, it lacks relevant symbols and annotions and therefore it is necessary to introduce some special symbol and annotion system into the remote sensing imagery. For example, river images matched with their annotion, mountains and their peaks matched with their names and altitude annotation as well as unpicturable administrative boundaries or other linear symbols drawn complementarily etc. can make the remote sensing imagery to be integrated with both map characteristics and contents simultaneously.

II. Geographic Base Map Compilation
Any thematic map needs geographic base map as its carrier of thematic contents , so does a land use map. When transmitting one or several special thematic contents on to a base map, the elements of the base map are taken as the control and support of qualitative, quantitative and positional determination of the thematic contents, At the same time, there exist mutual conditional and mutual –controlled relationships between the base map elements and the thermatic contents. The distribution law and the occurrence and the development of the thematic contents are integrated into a specially organized geographic environment and hence the scientific level and the practical value of the thematic contents are improved.

The base map elements of a land use map are as following:
  1. water system. It included rivers, canals, irrigation ditches, lakes (ponds), reservoirs, oceans and marsh lands, wells and springs etc., They make up the fundamental structure of land use maps and are an important element of base maps. In addition, the water system is a basic type of land use. so index and capacity selection of the system should be made in greater detail and not only is its capacity demonstrated, but its structure and characteristics as well.
  2. Landform. Landform is a basic condition affecting the thematic contents of land use. The altitude and location of a terrain, its gradient and sloping direction and its material making up etc. directly influence and determine land use types and land use features. According to the practical land use map compilation requirements, the total or partial relief contour lines and elevation points should be selected. Land use maps can be compiled without showing their contour lines to avoid unnecessary confusion with the presentation of land use the metric contents.
  3. Residential area. Residential area is both the base map element and the thematic content of land use maps. The residential area selection can be made according either to the real map blotch size or to the consideration of both the administrative level and the block size involved. Also it can only select such index as administrative level and population of residential areas should be chosen depends on map compilation objective and map application purpose. In addition, some smaller residential areas which have special meanings and close relationship with land use thematic contents should be chosen.
  4. Transportation network. Here transportation network mainly refers to railways and highways. But when mapping lands are vast in area and sparse in population or maps are to be compiled on large scale, some lower level roads must be taken into account, such as tractor-plugging roads, cart roads, earth roads and pastureland paths etc.,
  5. boundary. What a boundary , provincial boundary, country boundary, commune boundary, drainage boundary, natural boundary and the boundary of economy are several kinds of boundary areas to be named here.
Once a map scale is determined, the same-scaled relief maps at their possible latest publications should be consulted to carry out the content element selection, delineation and drawing in an order of water system first, followed by topography, residential area, transportation network and boundary. At the same time, the newest rectified and processed remote sensing data should be used to make amending measurement and renewal of the content elements.

III. Extraction of Remote Sensing Information on thematic Contents
Remote sensing imagery is rich in information which is not always application to a special thematic content. In addition, the information sometime may not be in much accordance with a specific them. In order that usable information is enhanced, CCT or remote sensing imagery should be processed purposively before interpretation. Image processing methods can be summed up in the following:

1. image processing by electronic and optical machinery
False-color combination: With MSs-4,5,7 bands or TM-4,5,. 7 bands, the combination of red, green and blue is most in common, because such pseudo-color combination has a greater application scope and is more suitable for the classification of TM-5,6,7, bands suits the differentiation. The combination can achieve good results in the differentiation. of woodlands and farmlands or their sublevel classification. The combination can achieve good results in the differentiation.

True Color combination: The combination of three TM wave bands of visible lights can produce bright true-color combination imagery with great chromatic aberration. Such combination suits the interpretation of such water physical and chemical

Characteristics as water turbidity, water depth, water passivity and water algae distribution etc.,

Pseudo-color slicing: A pseudo-color slicing picture can be produced on a single-banded black and white copy. In general, the pseudo-color slicing imagery obtained in this way is rather applicable to the high-level or sub high-level classification of land use with unitary background.

Mask correlation: Firstly by reproducing imagery on single banded black and white copies using a certain number of negatives with different exposures and then combination of them can acquired pseudo-color mask correlation pictures. The mask correlation processing method plays a positive role in edge enhancement, chromatism stregthening and color richness of imagery produced and the multi-combination of the mask correlation images can obtain one or several distinctive land use types which have close or similar characteristics and can not be distinguished easily by regular combinations.

2. Image processing by computers
Radiant composition: Making use of various combination of man values, exponents and logarithms of CCT radiant flux can perform many kinds of picture processing. Generally, the ratio processing of 5/7 of MSS or 3/6 or TM is more applicable to the interpretation of vegetation and land use classification.

Contour composition: According to the nonlinear features of surface reflection of an object, the contour of the object is represented by linear imaging masks and if the differential convolution processing is used, the contour enhancement can be achieved. the geometric contour enhancement of objects suits the interpretation in land use classification of some objects with vague contours or boundaries, such as interpretation of residential area and marsh lands etc.,

Square blotch composition: Classification or grading is made based on he average value of radiant flux of one square blotch unit and the land use maps with multi-grey levels or multi-colored square blotches can be produced in the form of codes.

In addition, there exist multi-information combination in image processing by computers, including combinations of different satellite photographs, of satellite photos with aerial photographs and of remote sensing images with maps etc. Such kinds of combinations can improve classification accuracy and reveal trends of dynamic changes involved.

The thematic abstraction of remote sensing information and the computer-aided land use mapping method leave much to be studied.

IV. Principles and Methods of Remote Sensing Imagery Interpretation
Remote sensing imagery interpretation in land use classification is ordinarily limited by its imaging ration and resolution. Now taking imagery of LANDSAT I, II, III, an example. On imagery scaled 1:1,000,000, 1mm2 is made up of 220.58 pixels representing 1,500mu real: land, but on images scaled 1:500,000, 1mm2 consists of in smaller areas with complicated land use types such as in southern China’s hilly land areas or in extensively cultivated plains, even within an area of 1mm2 of the picture with a scale of 1:250,000, the all pixel points do not belong to the same kind of pixel points representing the same object. Each pixel is almost a composite picture element. Wing to its environmental difference like water content, vegetation canopy density and crop growth condition etc., an object can have a varied radiant flux which results in a variation of picturing grey level of the object. So in land use classification interpretation, depending only on remote sensing information is not enough.

As a component part of geographic environment, the land use type is influenced by various natural factors such as relief altitude social and economic factors such as local economic state, production structure, transportation condition and distance between cities and country sides. As a result, in the classification interpretation, the multi-information centered on remote sensing message should be applied to make geographic correlation analysis comprehensively and to make deduction and demonstration mutual-conditionally and mutual-causally. Only in this way can the interpretation accuracy be improved.

V. Equal-sized Land Use Map compilation
The common technological process of map compilation is to edit large scaled maps first and then to get contract compilation step by step. When there exists a greater difference in scale between the maps to be compiled and the maps to be taken as basic data of the map compilation, it is necessary to compile maps with transitional scales. In this case, the map making period will be longer, depending on data characteristics of map compilation and on the ways of data acquisition. Because of limitation of the carried information and the picture-forming scale of aerial photographs, the indoor map making using aerial data still follows the above-mentioned process. The shuttle remote sensing imagery, however, is rich in carried information, for example, one MSS or CCT image carries 230 billion bytes. in addition, such data acquisition is rather easy. so the equal-sized map compilation can b performed directly by using remote sensing information.

For LANDSAT MSS imagery, the optimal scale scope for the equal sized map compilation is 1:500,000 to 1:1,000,000;
for MSS-CCT, the optimal scale is from 1:500,000;
For TM imagery, the optimal scale is about 1:200,00;
For TM,-CCT, the optimal scale is greater then 1:100,000;

As a result of its high resolution—20m and 10m, the spot imagery can be used as the basic materials for equal-sized compilation of land use maps on large, medium and small scales.

VI. Land Use Map Compilation Procedure
The map compilation by remote sensing imagery should reverse the common procedure of carrying out field investigation first and indoor map compilation afterward, that is: the indoor interpretation should go before field filling-up investigation and the indoor map making after the field investigation. Such a procedure renovation has many advantages.

1. Preparations
  1. remote sensing data acquisition, geometric rectification and processing of remote sensing imagery, abstraction of usable information and obtaining imagery with abstracted land use contents and uniform scale, and at the same time, introduction of symbol and annotation system into the imagery and achieving map-liked imagery.
  2. Matching the remote sensing imagery with same-scaled relief maps, delineation, drawing and selection of base map elements, data rectification and measurement and renewal, geographic base map compilation.
  3. Collection of relevant thematic maps, relief maps, treatises, and statistics etc.,
  4. Getting familiar with geographic characteristics of mapping lands and having good knowledge of land use distribution trend and distribution law.
  5. preliminary setting up of remote sensing imagery interpretation indices of mapping lands and setting up of experimental areas to have some tests if necessary.
  6. Preliminary determination of land use classification systems.
2. Indoor preliminary interpretation
According to the preliminarily determined classification systems and interpretation indices, a 0.05-0. 1mm thick coarse polyester sheet is covered on the processed image to make classification interpretation. The delineation and drawing is carried out by making use of natural masks and object images. The interpretation process is: to deal with simple objects first, difficult ones afterword and with smaller objects first, the larger ones afterword (putting small map blotches into a large map blotch ). In which we have some doubts should be noted down, in great detail before we complete land use preliminary interpretation maps.

3. Field filling-up investigation
According to the problems recorded during the preliminary interpretation, the field investigation plans are set up and the optimal investigation routes selected and then the corresponding on-the-spot investigation is done solve the problems found in the preliminary interpretation. in the preliminary interpretation.

4. Indoor map compilation
  1. countering the problems found in the preliminary and the followed field investigation, the classification system are further rectified and determined.
  2. Based on the on-the-post investigation, the analysis and the demonstration of other data, all problems involved in the classification are finally solved and the final version of land use thematic contents is determined.
  3. The thematic contents are transferred to and drawn on the geographic base map and then after some map matching and map appearance, the manuscript map is accomplished.
VII. Concluding remarks
  1. through the practical work it is known that based on the image formed by remote sensing thematic information abstraction, the map making classification using expericed visual interpretation is simple, easy and unexpensive to operate. The classification method is practical and the classification accuracy is high. Besides, the map making classification can make full use of the strong points of high interpretation level and comprehensive analysis of Chinese specialists .
  2. the application of multi-information centered on remote sensing and the comprehensive geographic correlation analysis can reduce errors in the interpretation and enhance the classification accuracy.
  3. making remote sensing imagery like a map before geographic based map compilation having thematic information abstraction before interpretation and performing indoor interpretation before field filling-up investigation can promote work efficiency and enhance map compilation speed.
  4. the equal-sized compilation can greatly shorten the mapping period.