Raster data & analysis : Raster data & analysis Saravana Kumar G
M.Tech RS & Gis
Geographic Matrix : Geographic Matrix The quest for a logical way to represent the real world has for centuries been one of the central themes of the science of geography.
The geographic matrix proposed by Berry(1964) was a major milestone in the development of data organization methods in geographic studies.
Geographic Data : Geographic Data In database terminology , the ways of representing data are known as data models.
A finite grid of square or rectangular cells, field based data bases are generally tabled as the raster data model.
In general there are three types of geographic data
Vector
Raster
Surface
Surface data are most commonly represented by the raster model.
Topography can be conveniently represented either by the vector model or by the raster model.
Raster Data. : Raster Data. Raster model is one of the variants of the field-based model of geographic data representation.
Raster model represents geographical phenomena that are continuous over a large area.
The field-based model is also commonly referred to as a tessellation( geometric arrangements of figures that completely cover a flat surface ) model.
The spatial data unit can be of regular shapes( regular tessellation ) or irregular shapes( irregular tessellation ).
Raster Model. : Raster Model. There are three basic forms of regular tessellation
triangular
square
hexagonal
All these tessellations have been used as the basis of spatial data models.
The regular square or rectangular tessellation is called as the raster model.
WHY RASTER? : WHY RASTER? It has been most widely used for the following practical reasons
It is compatible with different types of hardware devices for data capture and output( image scanners, computer monitors and electrostatic plotters).
It is compatible with the concepts and methods of bit-mapped images in computer graphics.
It is compatible with grid-oriented co-ordinate systems, such as plane rectangular co-ordinate system.
Raster data nature & characteristics : Raster data nature & characteristics Raster data model is characterized by subdividing a geographic space into grids cells
The linear dimensions of each cell define the spatial resolution of the data( determined by the size of the smallest object in the geographic space to be represented)
The size is also known as the MMU.
Each grid cell must contain a value. The value can be an integer, a floating point number, or a character( a code value)
These grid cell values can be used directly for computation( interpolation of contours, isotherms and isohyets)
How it is stored… : How it is stored… Raster data files are stored in different file formats. The differences between these file formats are mainly due to the different algorithms used to compress the raster data files.
In order to minimize the data storage requirements, raster data are often stored in a compressed form.
The data are decompressed “on-the fly” when they are used by an application program.
There are now more than ten industry standard raster data formats in use.
File formats used commonly. : File formats used commonly. BMP – Bit Map Graphics.
TIFF – Tagged Image File Format.
GIF – Graphic Interchange Format.
JPEG – Joint Photographic Experts group.
PNG – Portable Network Graphics.
MrSID – Multi-resolution Seamless Image Database.
GEO TIFF.
GRID.
PCX.
PRINCIPLES OF RASTER DATA COMPRESSION. : PRINCIPLES OF RASTER DATA COMPRESSION. A single raster data file contains several million grids.
In a black & white line map measuring 50cm 50cm will produce 400 million pixels( picture elements) when it is scanned at a resolution of 25micro mts( approximately 1000 dpi).
This lead to a very large data file.
The actual size of the file is dependent on the bit depth( 8bit, 16bit and 32bit).
A few raster files can easily consume all the space on the hard drive of a small computer.
Data compression is therefore an important feature of digital representation of raster data.
Raster – Based GIS Data Analysis. : Raster – Based GIS Data Analysis. The data analysis of a GIS includes a variety of data processing functions that aim to derive spatial relationships, patterns, and trends that are implicit in the source data.
The results of data analysis may be used immediately for spatial problem solving and decision making or as a input for further spatial analysis and modeling.
Raster – Based GIS Analysis Techniques. : Raster – Based GIS Analysis Techniques. Raster – based GIS analysis techniques can be classified in many ways. For example, Giordano et al.(1994) identified six categories of geographic data analysis operations.
They are,
Logical Operations: uses logical operators( i.e., “AND”, “OR”, and “XOR”).
Arithmetic Operations: uses arithmetic operators( i.e., addition, subtraction, multiplication, division, and assignment ).
Overlay Operations: are processes that merge attribute values from two or more layers, using logical operators or arithmetic operators.
Geometric Property Operations: processes that compute indices that describe the geometric properties pertaining to spatial features on layer( shape, size, angle and topological relations).
Slide 13: Cont…
Geometric Transformation Operations: processes that modify the spatial properties of features on a layer by applying linear transformations ( scale change and transformations).
Geometric Derivation Operations: processes that create new features from existing features on a layer( generalization, triangulation, filtering and surface interpolation).
Another common method of classifying raster data processing operations is based on the ways of using raster cells in the operations. On a raster layer, the smallest addressable unit is a point, which is represented by a cell. Each point can be addressed as a part of a neighborhood of surrounding values. When all neighboring points having the same attribute value are grouped together, a region is identified.
In raster-based data processing, some processes use the values of individual cells only, and others rely on neighborhood relationships or regional associations.
logical basis for classifying raster data analysis. : logical basis for classifying raster data analysis. Local operations
Neighborhood Operations
Extended Neighborhood Operations
Regional Operations
Local Operations : Local Operations They are the processes that create an output layer on which the value of each cell is a function of the cell at the same location on the input layer.
Local operations perform raster-based data analysis on the point-by-point or cell-by-cell basis. There are two important groups of local operations.
Reclassifying OR recoding
Overlay analysis
Reclassification : Reclassification The purpose of reclassification is to create a new raster layer by changing the attribute values of the cells of the input layer. This usually use either logical or arithmetic operations in the forms given below.
Assigning a new value to each value on the input map layer with the purpose of developing a binary (O and 1) mask for use in subsequent GIS analysis, it is process known as binary masking.
Assigning new values to classes or ranges of old values with the purpose of reducing the number of classes in the original input layer or to group values into categories in a new classification scheme.
Assigning ranks to unique values or categories of values found on the input layer.
Assigning ranks or weighting to a qualitative map layer to generate a quantitative map layer.
Overlay Analysis : Overlay Analysis It is regarded as one of the oldest map analysis methods. Map users have a long history of putting one map on the top of the another in order to detect the association between their contents( Goodchild 1992).
The manual method of overlay analysis was time-consuming and error-prone because it was often necessary to redraft the maps before they could be overlaid.
There was also a limit to the number of maps that could be analyzed by overlaying. To find ways that would automate overlay analysis was one of the major objectives of developing GIS.
Overlay Analysis : Overlay Analysis
Neighborhood Operations : Neighborhood Operations They are processes that create an output layer on which the value of each cell is a function of the cells neighboring the cell at the same location on the input layer.
They are also known as context or focal operations, make use of the topological relationship of adjacency between cells in the input raster layer to create a new raster layer.
The neighborhood is defined by using a “window” that is moved over each cell on the input layer to extract new cells values for the output raster layer. Local neighborhood operations are developed on the assumption that the value of a particular cell in a raster layer is often affected by the values of its neighboring cells.
Slide 21: Neighborhood Operation.
Neighborhoods can be defined by rectangles, circles, wedges, doughnut shapes (annulus) etc.
Extended Neighborhood Operations : Extended Neighborhood Operations They are the processes that create an output layer on which the value of each cell is a function of the cells neighboring and beyond the immediate neighborhood of the cell at the same location on the input layer.
Operations on extended neighborhoods process a set of points that extend over a much larger area than the local neighborhoods and hence involve a greater number of cells. A new raster layer is produced as a result.
There are five important groups of operations under this category
Statistical analysis
Determination of distance, proximity, and connectivity
Geometric transformation of raster layers
Buffering
Viewshed analysis.
Statistical Analysis Of Raster Layers : Statistical Analysis Of Raster Layers Geographic analysis and modeling often start with statistics that describe the characteristics of the entire raster layer or selected parts of its contents.
Statistical analysis is based on the use of attribute values rather than the location of the cells on a rater layer.
The resulting descriptive statistics pertaining to a single raster layer include the mean, the median and the most common value, frequency, spatial autocorrelation, among many other statistical indices, calculated for all the raster cells on the entire layer.
When two raster layers are compared, a variety of descriptive statistics can be generated to establish the correlation between the themes of data that they represent.
Determination of distance, proximity, and connectivity. : Determination of distance, proximity, and connectivity. In raster-based geographic data processing, distances are normally measured as straight-line or Euclidean distances between two cells.
If the two cells have the same row number, their distance is computed by using the distance between their column number and their row number times the spatial resolution.
If they have different row and column numbers, the distance is computed by using Pythagorean theorem.
When the distances between a particular cell and all the cells on the raster layer are computed, the resulting raster layer will be made up of cells representing concentric distance values from that cell. From these distance values, concentric equidistant zones can be developed by reclassification, leading to the creation of a proximity map.
Rotation, Translation, and Scaling. : Rotation, Translation, and Scaling. These are data-preprocessing functions that will ensure the compatibility between the spatial and geometric qualities of a raster layer and the objectives of a particular task in raster-based geographic data processing.
The aim of these functions is to spatially position a raster layer, using the method of image-to-image registration, or to produce a new raster layer, using the method of image-to-map rectification, so that reliable measurements can be made in raster-based GIS analysis.
Buffering : Buffering In the context of geographic data processing, a buffer is an area that is created around a particular spatial feature. This feature can be a point, a line, or an area.
In the raster data model, the concept of spatial feature does not apply because individual features are not represented as independently identifiable entities but as a collection of contiguous raster cells having the same attribute values.
A buffer is defined as the raster cells that are at a specific distance from a particular cell, a linear array of cells, or a cluster of cells. The process of creating buffers, or buffering is a very important function in GIS applications.
In raster-based geographic data processing, the primary function of buffering is to identify all those cells that are adjacent to, or affected by, selected cells having a particular value on a raster layer.
Viewshed Analysis. : Viewshed Analysis. Raster-based operation that makes use of terrain elevation data to determine all areas visible to an observer located at a specified point in a space.
It is capable of determining the indivisibility between any two points on a raster layer of elevation values.
It requires two input layers, a raster layer showing the locations of one or more viewpoint cells, which can be points, lines, or polygons and a DEM layer.
To determine the viewshed of a particular view point cell, its elevation value is first obtained from the DEM by an overlay operation. Then, this elevation value is compared with the values of all the cells on the DEM.
These cells will be classified as visible or invisible according to three factors, their elevation values relative to the value of the viewpoint cell, their positions along the line of sight, and whether they are higher or lower than the highest point between them and the viewpoint cell.
Regional Operations : Regional Operations They are processes that create an output layer by identifying cells that intersect with or fall within each region on the input layer.
On a raster layer, a region is a collection of cells that exhibits the same attribute characteristic.
Also referred to as a zone, and operations on regions are called zonal operations.
In geography, a region is an extended area with homogeneous characteristics with different shapes.
There are three major categories of zonal operations.
Identification of regions and reclassification
Category-wide overlay
Calculation of area, perimeter, and shape.
Identification Of Regions and Reclassification. : Identification Of Regions and Reclassification. A region is identifiable visually by differentiating its cell values.
A region is made up of only one single cluster or clump of cells but it is common for a region to include many individual clusters are scattered all over the layer.
The zonal operation that does this function in raster-based data processing is known as parceling. This operation identifies clusters of spatially contiguous cells that have the same values and reclassifies them by giving each identified cluster a unique value.
The output from this particular zonal operation produces a new raster layer on which individual clusters of cells, which originally have identical values, will be uniquely identified by different values.
Category-wide Overlay. : Category-wide Overlay. Category-wide overlay operations make use of two input raster layers to create a new raster layer.
However, unlike the location-specific analysis operations that combine the input layers on a cell-by-cell basis.
Category-wide overlay operations summarize the spatial coincidence or intersection of the entire category of values represented on one layer with the values contained on the other layer.
Overlaying raster layers in this way is different from location specific overlaying.
Whereas location-specific overlay operations generate a new layer by combining existing layers.
Category-wide overlaying operations simply use the boundaries represented on one raster layer to extract cell values from other layers.
Calculation Of Area, Perimeter, and Shape. : Calculation Of Area, Perimeter, and Shape. Obtaining information about the area, perimeter, and shape of a region often constitutes the first step in many GIS applications.
On a raster layer, area is determined by counting the number of grid cells within the boundary of a region, while perimeter is determined by adding the number of exterior cell edges of the region.
Using area and perimeter measurements, it is possible to calculate the perimeter/area ratio that is commonly used to describe the shape of a region.
References: : References: Concepts and techniques of GIS
-ALBERT K.W. YEUNG.
THANK YOU : THANK YOU