Grid Data of GIS Learning

Grid Data uses a regular grid to describe the location and value of spatial symptom features corresponding to the location of each grid unit. In terms of concept, changes in spatial phenomena are reflected by changes in grid unit values. Many data in the geographic information system are expressed in a grid format. Grid Data is complementary to vector data in many aspects. Combining the two types of data is a common feature of GIS projects.

I. Elements of the Raster Data Model:

The Grid Data Model is also known as grid, raster map, surface cover, or image in GIS. A grid consists of rows, columns, and grid units. Rows and columns start from the upper left corner of the grid. In a two-dimensional coordinate system, rows are taken as Y coordinates and columns as X coordinates. At this point, it is a bit similar to latitude as Y coordinate and longitude as X coordinate.

Grid Data uses a single grid unit to represent a point, a series of adjacent grid units to represent a line, and a set of adjacent grids to represent a plane. Each grid unit in a grid has a value, an integer or floating point type. The value of an integer grid element usually represents category data. For example, the land type commonly uses 1 to represent the city land, and 2 to represent the forest land. Floating point grid unit values often represent continuous data. For example, the precipitation model may have precipitation values such as 20, 15, 12, and 23. Floating Point grids require more computer storage resources than integer grids. This is an important factor that must be considered in a wide range of GIS projects. In addition, the data query and display of floating point grids should be based on 12.0 ~ The value range 19.0 is not a single value.

Because the resolution of the Raster Data Model is affected by the size of its grid unit, there is a disadvantage in expressing the precise location of spatial elements. In algorithm, a grid can be regarded as a matrix of rows and columns. Its unit value is a two-dimensional array, which makes it easier to operate, set, and analyze data than vector data.

Ii. Raster Data Type:

1. Satellite Images: remote sensing satellite images are recorded in a grid format. Satellite image elements represent the light energy that is reflected or emitted from the Earth's surface. By analyzing the Image Element value, the image processing system can extract various topics from satellite images, such as land use, hydrology, water quality, and soil erosion.

2. Digital Elevation Model: The Digital Elevation Model (DEM) is composed of the arrangement of the same interval elevation data. DEM is based on points, but it is easy to convert the elevation point to the grid unit center to a grid data.

3. digital positive image (doq): a digital image produced by aerial photography or other remote sensing data, the displacement caused by camera lens Skew and terrain fluctuations has been eliminated. Numbers are projected as references to geographical coordinates and can be registered with topographic maps and other maps.

4. Binary scan file: it is a scanned image containing a value of 1 or a value of 0.

5. digital raster: a scanned image of the USGS topographic map.

6. Graphic files: JPG, Tiff, GIF, etc;

7. Raster data for specific geographic information system software:

3. Grid Data Structure and compression:

The raster data structure refers to the storage of raster data, so that they can be used and processed by computers. Common structures include unit encoding, chain code, block code, and Region quad-tree (For details, refer to geographic information principles and introduction to geographic information).

Raster image compression is usually divided into lossy compression and lossless compression.

Lossless Compression refers to Data Compression Based on statistical redundancy. This ensures that image information is not lost or distorted during data compression and restoration. Common examples: Travel Length Encoding (RLE), namely, travel encoding, incremental modulation encoding (DM), and Hoffmann encoding (LZW );

Lossy compression: the use of Human Visual errors, the use of efficient limited distortion data compression algorithm, allowing the compression process to lose certain information. Common JPEG compression standards, with a maximum compression ratio of 50: 1;

JPEG compression standards are static image data compression standards developed by ISO and IEC. It has two basic algorithms: one is a lossy compression algorithm based on discrete cosine transformation. A Lossless Compression Algorithm Based on prediction technology.

Iv. Projection and geometric transformation of raster data:

Geometric transformations of satellite images are often referred to as geographic coordinate references in image processing. The following two methods are commonly used for geographic coordinate reference:

1. Affine Transformation)

It uses rotation, movement, and proportional transformation for geographic coordinate reference. The transformation equation is as follows:

X' = AX + by + c

Y' = dx + ey + F

X and Y represent the number of columns and number of rows, and the coefficient E is negative. Because the image and coordinate system have different origins, the image's coordinate origins are in the upper right corner, and the coordinate system's origins are in the lower left corner.

2. polynomial equation: polynomial equation provides a mathematical model for differential proportional transformation and image rotation. The degree of complexity of the model is expressed by the degree of polynomial, which ranges from 2 to 5. For example, second-order polynomials are converted using the following equations:

 

 

3. re-sampling: Fill the unit values in the new grid with the unit values in the original grid.

Three Common Sampling Methods:

Nearest Neighbor Method: Enter the nearest cell value in the original grid into each cell in the new grid.

Bilinear interpolation: Fill in the weighted average values of the four nearest cells in the original grid into each cell in the new grid.

Cubic convolution: Fill in the weighted average values of 16 nearest neighbor units in the original mesh into each unit in the new mesh.