1 quotation

It is of great significance to find out the situation of sloping farmland resources and make scientific evaluation for promoting economic development, scientifically formulating land resources development and utilization planning and ecological returning farmland planning. Slope farmland is the production base of dry grain, cash crops and fruit trees, which plays an important role in agricultural production. Because most of the sloping farmland is reclaimed along the slope, the ridge is not perfect, the topsoil is washed away, and the plough layer is desertification; Moreover, the emphasis on light tillage, extensive tillage and little or no application of organic fertilizer lead to low soil organic matter content, lack of nitrogen, phosphorus and potassium nutrients, high soil acidity, shallow tillage layer and drought, resulting in low crop yield. However, the sloping farmland has large area, deep soil layer, superior environmental conditions, sufficient water and heat, and great potential for increasing production. In view of the existing problems, it is necessary to find out the exact quantity, distribution and ownership of sloping farmland, make a scientific evaluation, and then formulate corresponding countermeasures for scientific improvement.

On the other hand, since mankind entered the agricultural civilization, it has written a history of deforestation and land reclamation. The United Nations Global Ecological Environment Outlook 2000 pointed out that human demand for trees and cultivated land reduced the global forest by 35%, of which 30% became agricultural land, which was difficult to support the development of human civilization. China is no exception. Deforestation has lasted for thousands of years in China, and "food is the most important thing for the people" is an important basis for governments to formulate various policies. From the founding of New China to the mid-1990s, the food problem has been puzzling the development of China, and the pace of deforestation and farmland expansion has never stopped. Up to now, the country has reclaimed 9 1 10,000 mu of sloping farmland above 25 degrees. Various food growth measures, including deforestation and land reclamation, have enabled China to feed 22% of its population with 7% of the world's arable land. This is a great contribution to all mankind. However, the ecological cost is also heavy. Due to unreasonable farming methods and deforestation in the middle and upper reaches of the Yangtze River and the Yellow River, the amount of sediment imported into the Yangtze River and the Yellow River reaches 2 billion tons every year, of which 2/3 comes from sloping farmland. The reclamation of sloping farmland caused soil erosion and desertification, which eventually led to the continuous deterioration of the ecological environment. It is imperative to return farmland to forests. Since the last few years of the 20th century, with the fundamental solution of the food problem, people may pay more attention to ecological problems. From 65438 to 0999, the CPC Central Committee and the State Council took the overall situation into consideration, sized up the situation, seized the favorable opportunity, and made a major decision to "implement the project of returning farmland to forests, grasslands and ecology", which will completely end the practice of deforestation for thousands of years. It should be said that this is a great historical turning point.

In order to develop and improve sloping farmland and return farmland to forest, it is necessary to accurately grasp the distribution and slope of sloping farmland. At present, land management departments all over the country are using the slope data investigated in the detailed land survey. These slope data are obtained manually or semi-manually, so it is difficult to verify the accuracy of the data comprehensively, which will inevitably affect the accuracy of the summary data of provinces, cities and even the whole country. With the improvement of computer data processing ability, the extensive use of automatic measuring instruments and the development of surveying and mapping technology, it is possible to extract the slope and distribution of sloping farmland automatically and accurately by computer. DEM is needed in the process of orthorectification of remote sensing images. In general, the corrected DEM data will be idle. DEM data is hard-won and expensive, so it should be further developed to improve its utilization rate and make it play more roles. This paper discusses how to use the existing advanced technology to realize the automatic extraction of sloping farmland.

2 about DEM of digital elevation model

2. 1 Overview of Digital Elevation Model

In the mid-1950s, C.L.Miller, director of the Photogrammetry Laboratory of Massachusetts Institute of Technology, put forward a general concept: Digital Terrain Model (DTM). Since then, DTM has developed rapidly and has been widely used in many fields including GIS. Digital elevation model (DEM) is a special case of DTM, and both of them are ordered numerical arrays that describe the spatial distribution of ground features. Spatial distribution is described by x, y horizontal coordinate system or latitude and longitude.

Different from DTM, the feature of DEM is the elevation value z, not the attribute value describing soil type, vegetation type and land use. At present, true three-dimensional (3D)GIS is still in the research stage, and DEM is still the main 2.5-dimensional means for GIS to represent 3D terrain. Dems commonly used in GIS include: regular grid and TIN based on 2.5-dimensional representation, and isoline based on 2-dimensional representation. GRID uses a set of grids with the same size to describe the terrain surface, which can fully show the detailed changes of elevation, with simple topological relationship, easy algorithm implementation and convenient manipulation and storage of some spaces. The disadvantage is that it takes up a lot of storage space, and there is inconsistency between irregular ground elements and regular data representation. TIN is composed of a series of disjoint triangles formed by scattered terrain points according to certain rules. Its advantages are high storage efficiency, simple data structure, and coordination with irregular ground objects, which can represent slender functional elements and superimpose arbitrary regional boundaries. However, the implementation of TIN is more complicated and difficult. Commonly used grid generation algorithms are: Inverse Distance Weighted Interpolation (IDW), Bilinear Interpolation, Trend Surface Interpolation, Spline Interpolation, Multilayer Overlay Interpolation Surface Function and Kriging Interpolation. TIN generation algorithm mainly includes: segmentation and merging, point-by-point insertion and step-by-step growth.

2.2 DEM production method

DEM data can generally be obtained or purchased from the competent department of surveying and mapping. If DEM data that meets the projection specification and scale accuracy are obtained, it can be cut according to a certain range of coverage under GIS software and can be used. For example, in the grid module of ARC/INFO software, you can do it with the GRIDclip command.

If you can't get ready-made DEM data, you can also use topographic map to generate it yourself. The steps are as follows:

(1) Digitization and correction of paper topographic map, namely topographic map scanning and geometric correction.

(2) Extraction of elevation information. Comprises the following steps: ① carrying out screen vector tracking on the contour line; (2) Specify elevation values for contour lines; ③ Editing, checking and splicing to generate topological relations.

(3)DEM generation. Comprises the following steps: ① TIN interpolation is carried out on the generated vector map in ARC/INFO software, so that the whole research area contains elevation values; ② Sampling TIN data and converting it into grid data.

(4) Cutting DEM data. The method is the same as before.

2.3 DEM data used in the research project

Data accuracy is a concept closely related to map scale, and different data types must be integrated under a unified accuracy framework. Vector data must be edited, modified and synthesized under the framework of specific projection type and scale; DEM data accuracy is also a concept closely related to scale. The research shows that DEM data generated from1∶ 50,000 topographic map in Zhongshan area can well retain the topographic information of 25 ~ 30 grids, with 20m grids in high mountain areas and 50m grids in low mountain areas.

In this study, the DEM data with the scale of1∶ 50,000 produced by the State Bureau of Surveying and Mapping according to the unified standard are mainly used, and the grid spacing is 25m. The experimental area is located in Longhua County, northern Hebei Province, with an area of 5,492 square kilometers. The lowest altitude is 660m, and the highest altitude is 1244m. The original data format is the Coverage standard format. After coordinate transformation, it is unified to Xi 'an coordinate system 1980. Figure 1 shows the DEM gray scale of the experimental area 1∶50000, and Figure 2 shows the DEM image after shading.

Map 1 DEM gray map

Figure 2 DEM image after shadow processing

3 Slope information extraction

3. 1 Mathematical basis of slope calculation using DEM

On the basis of DEM, the slope and aspect map is compiled. Slope refers to the change rate of pixel elevation values in the grid, and the calculation results are stored in the pixel attributes in the form of degrees, decimals or percentages. Aspect ratio refers to the direction of each pixel surface in the grid, ranging from 0 to 360 degrees. Where 0 degrees represents the north and 90 degrees represents the east. Up to now, the calculation methods of slope and aspect can be summarized into five kinds: four-block method, space vector analysis method, fitting plane method, fitting surface method and direct solution. Practice has proved that the fitting surface method is the best method to solve the slope. The fitting surface method generally adopts quadric surface, that is, 3×3 window (Figure 3):

Fig. 3 Quadratic surface window of fitting surface method

Every point is an elevation point. The formula for solving the slope of G point is as follows:

Collection of Papers on Land Resources Monitoring and Investigation Project [2]

Its slope direction calculation formula is:

Collection of Papers on Land Resources Monitoring and Investigation Project [2]

Where: s is the slope; A is the inclined direction; SWE is the slope of the east-west direction (X axis); SSN is the slope of the north-south direction (y axis). There are four slope algorithms * * * on X axis and Y axis, among which the algorithm with the highest accuracy and calculation efficiency is:

Collection of Papers on Land Resources Monitoring and Investigation Project [2]

Where Δ g is the grid spacing of the grid.

3.2 Generation of Slope Map

According to the above mathematical model of slope calculation, the slope map is generated by computer automatic processing. The slope map should not be generated at any level. Before generating the slope, it is necessary to determine the slope expression level of the slope. In order to keep consistent with the slope grade often used in daily work (that is, the slope grade required to establish a database according to the current land use situation), the slope grade is formulated as follows:

Collection of Papers on Land Resources Monitoring and Investigation Project [2]

According to the grade of slope, the corresponding polygon is automatically extracted from the grid information of DEM by GIS software, and each grade of polygon is composed of different colors, and the grade code of slope is automatically added to the attributes of polygon. The generated slope map style is shown in Figure 4, and different colors represent different slopes.

Fig. 4 Slope vector diagram

3.3 superposition of land use map and slope map (extraction of sloping farmland)

After the above processing, although the different slope grades in the whole map range have been expressed, we don't know where the cultivated land and the non-cultivated land are, so we need to use other technical means to distinguish them. There are generally two ways to solve this problem: manual nesting and computer automatic processing. The manual method is very backward, so I won't go into details here. We only discuss automatic processing methods.

In the process of establishing land use database, the patches of various land types are vectorized. After the vectorized land use map is converted into coordinates and registered with DEM slope map data (the method of coordinate system conversion and registration is not detailed here, please refer to relevant materials), the slope of each cultivated land patch is automatically calculated by using the spatial analysis function of general GIS software. The basic principle of spatial analysis is shown in Figure 5.

Spatial analysis is an operation of superimposing two layers of map elements to create a new element layer. Thus, the original elements are divided, cut and nested, and then new elements are generated. The new element combines the properties of the original two layers of elements. In other words, spatial superposition not only produces new spatial elements, but also links the attributes of input elements to produce new attributes. Spatial superposition can be divided into vector data and raster data. For vector data, the vector superposition method is used to divide, cut and nest the spatial data of the vector, and connect the attributes related to the vector. The superposition result is new vector data and attribute data. For raster data, the grid weighted superposition method is used to add the corresponding elements of two raster files as the corresponding elements of the superposition result.

The spatial analysis of figure 5 automatically assign gradient values to tile.

The attribute data related to vectors, or the attribute relation table obtained by vector superposition, can be further analyzed by attribute statistics, so as to obtain the quantitative relationship between various elements.

In Figure 5, the dark part is a piece of cultivated land, including three different slopes of 1, 2 and 3. The software will automatically carry out weighted average to get the final slope (about 2. 1 in the figure).

Add a "computer slope grade" field to the patch attribute field of vector land use status map (to avoid conflict with the original "slope grade" field), and the slope grade of each patch can be obtained through spatial analysis and automatically assigned to this field. It should be noted that the map of land use status after automatic assignment not only realizes the assignment of farmland patches, but also all patches including woodland, residential area and water area contain slope values, which is an unprecedented representation method and lays a foundation for further application in the future. Figure 6 shows the present situation of land use with different slopes.

Fig. 6 Present situation map of land use with slope.

4. The comparison between the results of automatic extraction of slope by computer and the slope data of original land detailed investigation.

Based on the establishment of the land use database in Longhua County, the slope grade of cultivated land in the original detailed land survey was compared with the slope grade automatically extracted by DEM.

After the new database was built in Longhua County, the total number of map spots was about 35,237, of which 3,934 were marked with slopes according to the initial detailed survey data (there are two reasons for the small number: first, the original detailed survey data was incomplete or lost in the vectorization process of database construction; The second is to label only cultivated land, not other land). We have automatically extracted the slopes of all map points in the county by computer. Now compare the slopes marked with patches with those automatically extracted by computer, and the results are shown in table 1.

In order to verify the accuracy of the results in the table, this study conducted a spot check on the slope while conducting the land renewal survey, and the number of spot checks was about 50. The results show that most of the patches with slope error (phase difference 1 grade) are cultivated land with a slope of about 2 degrees, and 2 degrees is the boundary between 1 and two slope grades, which is normal. Spot checks are also made on the points with gross errors, and the results show that the computer results are correct and the gross errors are caused by human factors.

As can be seen from the table 1, in the initial detailed investigation, due to the limitation of the conditions at that time, about 47% of the estimated slope values were correct, about 47% had errors (the difference of slope grades was 1 grade), and 7% had gross errors. Statistically, these differences are normal, but they only reflect the defects of the treatment methods at that time. These defects were tolerable and helpless at that time, but today, with the rapid development of information technology, they are problems that cannot be ignored. At the same time, the author feels that the classification standard of slope grade is relatively rough at present. Take 1 slope as an example, the slope is 0 ~ 2 degrees. In practical work, we all know that there are essential differences between 0 degree flat land and 2 degree sloping land. If they are classified as one grade, it will inevitably affect their further application in the future.

Table 1 Comparison between original grade and grade automatically extracted by computer

5 conclusion

In the process of building the database of land use status, if we continue to use the slope data in the initial detailed investigation, it will bring two problems: ① the inaccuracy of the original data will affect the further application in the future; ② In the process of building the database, some slope information will be lost, which makes the original inaccurate data even more inaccurate, and it is difficult to check the lack and correctness through pre-inspection. Therefore, it is imperative to re-evaluate sloping farmland by using advanced technical means. Using DEM data, a good slope extraction effect can be obtained, and DEM with the scale of1∶ 50,000 is a more suitable variety.

The current grading standard of slope gradient is too rough, which may bring adverse effects to further application in the future. It is suggested to improve the grading standard of slope gradient. Under the existing conditions, it is suggested that DEM data should be kept in the process of land use database construction, or DEM database should be established to prepare for future application.