3.3.2.1 Image characteristics and interpretation of land use types
Aerial photo interpretation is the process of extracting the required thematic information from remote sensing images and other related data according to the classification system . When interpreting, we must not only start from the direct characteristics of land use categories such as special image colors, geometric patterns, shapes, sizes, etc., but also combine the distribution characteristics and mutual relationships of different categories in topography and landforms. For the indirect signs, the annotations on topographic maps of the same scale should also be considered to facilitate logical reasoning and comprehensive analysis. If necessary, on-site surveys and visits will be required to finally establish interpretation signs. For example, the interpretation signs of various land use types in this mining area are as follows.
141. Ditch land: It is an important cultivated land in the valley. Under a stereoscope, it can be observed that it is in the valley, including ditch strip land and ditch platform land. The plot is smaller than the flat dry land. The image has irregular edges and blocks of light yellow, goose yellow, and light green tones.
142. Terraced dry land: Generally constructed artificially, the texture is uniform. There are obvious steps in the stereoscope. There are yellow spots and dark red alternating grass or shrub-like image features at the connection between the upper and lower steps. The field ridges are in the shape of regular arc strips arranged like concentric circles, and the colors are light yellow and light green. Alternately distributed in mountains and areas with large slope changes.
143. Sloping dry land: The slope can be observed under a stereoscope, with no obvious ridges. The texture is smooth and irregular, with fine texture and high contrast. Generally, sunny slopes are light yellowish white, and shady slopes are light yellowish white. Light blue-green in color.
144. Sichuan Dry Land: The terrain is relatively flat and open, with a large area. The image is in the shape of relatively regular large strips, with high brightness. The color tone is uneven light yellow and light green, and can be seen in the fields. There are criss-crossing white paths.
15. Vegetable plot: The image of the vegetable plot is clearly visible. The plot is small and rectangular, with a checkered carpet-like structure inside. The image is vermilion with a small amount of variegated colors, and the color is brighter. Mostly distributed near residential areas, ponds, and ditches.
21. Orchard: There are often regular rows and plant spacing, and the surroundings are clear. The images are reflected as regularly arranged small dark red round spots and shadows.
211. Apple orchard: distributed in red or magenta regular dots on the image.
212. Persimmon Garden: The shape of the plot is similar to that of dry land, and the area is smaller. It is dark red in color and has rough texture.
213. Other orchards: red or magenta, with clear stripes and obvious boundaries with the surrounding areas.
22. Other gardens: sparsely dotted, with uneven stripes, dark red to red. There is a clear boundary with the surrounding area.
31. Woodland: The trees grow densely, with tall trunks, in a continuous large shape. Usually the tree crowns are connected into pieces, in the shape of purple-red strips, with obvious but irregular boundaries, and are mostly distributed in mountainous areas. and within residential areas.
311. Coniferous forest: dark red with dot-like texture. Under a stereoscope, the tree crown can be seen as pointed, and pointed shadows can be seen.
312. Broad-leaved forest: The crown is red, forming a grape bunch-like texture, and the shadow is oval.
32. Shrubs: brown or maroon. The shadow pattern structure is rough, the pattern shape is irregular, and it appears spherical when viewed under a stereoscope.
33. Sparse forest: The image is in the shape of light red or light purple flakes, and the internal structure is finely divided and irregularly distributed.
36. Nursery: Generally distributed at the edge of residential areas, the perimeter is regular in the image, the spacing between rows and plants is clear, the tree crown is small and has bright red stripes, the plot is rectangular or square, and the interior is grid-shaped.
4. Grassland: It has a velvety structure, uniform and smooth shadow patterns, and is yellow-green with red spots in between.
41. Natural grassland: light gray or brownish red in color, with blurred borders, mostly distributed in mountainous areas.
42. Artificial grass: regular in shape, red in color, distributed in lowlands and around institutions.
51. Towns: There are many streets in urban land, which are arranged regularly and clearly, and are dark blue in color. There are neat and dense residential areas, located next to the main traffic lines, and the area is large.
52. Rural residential areas: The outline in the image is irregular, the houses are densely packed and appear blue-grey, the internal roads are curved, there are many trees inside, the area is small, and there are rural roads connecting it.
53. Independent industrial and mining land: Its boundaries are generally fenced, and the internal houses are tall and neat. There are usually material storage yards, connected by dedicated railways or roads, and good greening conditions.
531. Land for industrial and mining facilities: blue-gray in color, with various shapes, and obvious locations for large equipment.
532. Stockyard land: mostly cyan or blue-gray in color, in a regular stacked shape, with batches of ground objects (such as wood, steel pipes, etc.) visible, and with certain facilities, such as Elevated driving vehicles, winch conveying machinery, waste transport lanes in waste rock mountains, etc. are connected by railways or highways, and the stockyards in large mining areas are connected by dedicated railway lines.
533. Industrial and mining residential land: In large-scale factories and mines, the houses are arranged regularly and are blue-gray in color. They are mostly isolated from production areas or stockyards. There are occasionally dark red and alternate patterns (trees) in the front and back of the house.
534. Land for industrial and mining public facilities: It has a clear outline and a certain geometric shape, such as square or rectangular. It generally has a wider site, tall and neat houses, and good greening around it.
535. Small independent industrial and mining land: It is an independent whole, with coal gangue mountains or chimneys, etc., and a relatively small stockyard, connected by simple roads, with clear images, such as chimneys and kilns. There are few houses around the buildings, which are generally small coal kilns, iron smelting plants, brick and tile factories and other industrial enterprises.
55. Special land: This is a type of land that is difficult to identify. Because it is less distributed, it can be mapped and delineated on the spot.
61. Railway land: The railway is in the shape of a narrow and long streamline in blue. The dedicated railway line in the mining area leads directly to the stockpile yard in the mining area. The arc is large and even when turning.
62. Highway land: The asphalt pavement is blue-gray, with dark red road protection forests on the roadside; the cement road or gravel road is light gray or gray-white; the curve of the road is large, and the traffic on the road is Vehicles are clearly visible.
63. Rural roads: gray-white, network-shaped, narrow, with large curvatures, connected to villages and fields.
71. Rivers: Rivers are generally curved and streamlined black strips. There are bridges on them when roads pass by. The edges of the river bed are irregular and the width of the river surface is uneven. There are usually vegetable fields nearby.
72. Reservoir: The water surface is black, the area is large, there are dams, and most of them are gourd-shaped.
73. Pit ponds: generally have regular borders and are round, oval or rectangular. The water surface is mostly black, with a small area and easy to distinguish.
78. Hydraulic structures: in the form of white strips, located downstream of the reservoir.
81. Wild grassland: gray or brownish red, irregularly distributed in large areas, with a relatively uniform texture and no three-dimensional effect.
86. Bare rock or gravel: white, irregular in shape, with gray stripes in the middle, mostly distributed in mountains.
87. Field ridge: Curved linear shape, gray or brown, shaded, located between sloping dry land and terraced dry land.
Based on the above image interpretation marks, each color infrared photo is analyzed and interpreted one by one. The interpretation results can be drawn on a 0.05mm thick transparent polyester covered on the color infrared aerial photo. On the film, in order to carry out on-site sampling inspection of the recognition rate, at the same time, the boundaries of the type spots on the image that cannot be directly interpreted according to the logo are drawn, and their category attributes are adjusted on the spot.
3.3.2.2 Verification of land use interpretation results in mining areas
In order to improve the accuracy of the category attributes of land use maps, field verification of the interpretation results must be carried out. The verification adopts the sampling method, and the number of samples is determined based on the sampling theory and the number of interpreted patterns.
(1) Sampling scale
First of all, it is necessary to determine the allowable error in the total area of ??the land use type map. Because in the statistics of land use type areas, category interpretation errors have the greatest impact on classification area errors. Only by improving the interpretation accuracy can the accuracy of area statistics be ensured. According to the "Technical Regulations for Land Use Survey", the interpretation accuracy of map spots needs to be above 95%. Therefore, α = 0.05 is taken to determine the number of random samples within the 95 confidence level.
Assuming that the sampling land type patch events are normally distributed, the number of sampling samples is given by the following formula
Research on environmental dynamic monitoring and analysis in industrial and mining areas
In the formula, n is the number of sampling samples; P is the ratio of the number of correctly interpreted patterns to the total number of patterns in the population; Q is the ratio of the number of incorrectly interpreted patterns to the total number of patterns in the population; d is the allowable error; N is the total number of spots; t is the multiple of the standard deviation.
Since the values ??of P and Q are unknown, their estimators p and q are used instead during actual sampling.
When the value of N is very large, the above formula can be approximately expressed as:
Research on dynamic monitoring and analysis of environmental conditions in industrial and mining areas
When n0/N can be ignored Timing, n=n0.
When n0/N cannot be ignored,
Research on dynamic monitoring and analysis of environmental conditions in industrial and mining areas
When the confidence level is 95, set t=2, p can be obtained based on land use survey data in the mining area.
(2) Sampling inspection results
During sampling verification, the field categories and the interpretation categories on the images are compared, the interpretation result matrix is ??statistically calculated, and the accuracy rate is calculated accordingly.
3.3.2.3 Compilation method of land use map
The map compilation procedure is as follows:
(1) First, take aerial photos indoors (1:10000 ), divide the various patterns of land use types, and then cover the aerial photos with polyester film. Within the scope of the use area, use a pencil to draw the outline of the type to make a sketch. Then conduct field verification, focusing on categories that are prone to interpretation errors.
(2) Use the topographic map as the working base map, and use an aerial photo converter to accurately fit the photos of different scales with the topographic map. On the boundary of the effective area of ??each interpretation piece, find 5 very obvious feature points corresponding to the topographic map (flat areas), or find control points in blocks (mountainous undulating terrain), adjust the scale and use instruments and base map movement and other operations, and then transfer the land use map patches to the topographic map to form a land use map sketch.
(3) Put the land use sketch together with the border and coordinates of the topographic map on the digital board. With the support of the geographical information system, complete the input, editing and drawing of the graphics, and automatically count the land use types. area, automatically draw results maps and perform multi-information comprehensive analysis.
With the help of various commands provided by GIS, the functions of graphic input, topological relationship generation, graphic editing, geometric transformation, graphic splicing, statistical calculation, and automatic cartography are used to complete automatic drawing results (figure omitted) and Statistics of land use area. The flow chart is shown in Figure 3-4.
Figure 3-4: The process of preparing land use maps using GIS
3.3.2.4 Land use area statistics
The conventional land area statistics method generally uses quadrature The instrument measures the area of ??various plots on the land use map and then performs the adjustment. The workload is large and the measurement accuracy varies with the instrument and personnel used. When using a geographic information system to operate the attribute table, since the land use map uses the actual map contour Gaussian coordinates to correct each map patch and match the map contour, its errors have been automatically distributed, so there is no need for adjustment. The main reason affecting the area accuracy is the error in the digitization process, which is negligible. In addition, general mining areas include multiple mines. Therefore, while digitizing, the well field boundaries of each mine are input in order to calculate the land use area of ??each mine.