|Get on the Grid: Creating a new Raster with MapInfo Pro Advanced |
| ||One of the key new capabilities in MapInfo Pro Advanced is the ability to create raster files from point data using various interpolation methods. This article will help you understand the different interpolation (or gridding) methods that MapInfo Pro Advanced offers and the simple steps required to create a new raster file from your point data. |
What is Interpolation (or Gridding)?
Interpolation, also commonly known as gridding, is the process of predicting or estimating unknown values an array of cells (a raster) from a set of known point observations by applying a mathematical algorithm. Interpolation is used to predict what a value would be for any given cell location covered by the raster from a limited set of input sample observations. It is generally used to estimate new values in locations where no observations exist, by using the information from other observations that are located nearby. Interpolation does this using the assumption that observations located close to one another are spatially correlated. Interpolation can be used to create new raster datasets from scattered point observations and characterize features such as population density, elevation, rainfall, temperature variation, flood, customer patronage, or virtually any measurable information, in an easy to use and visually appealing way.
Different Interpolation Methods available in MapInfo Pro Advanced
MapInfo Pro Advanced supports six interpolation methods. These methods can be accessed from the Create Raster button on the Raster tab.
Inverse Distance Weighting
The Inverse Distance weighting technique leverages the concept of spatial correlation directly. It assumes that the nearer a known sample point is to a cell that is being estimated, the more closely the estimated cells value will resemble the values of the closest points. Input data points that are located within an area of influence (the search radius) of the cell being estimated are individually weighted according to their distance from the cell. The most common weighting model used with this estimation method is a power distance model. But other models such as Linear, Exponential and Gaussian are also supported. MapInfo Pro Advanced provides full control over the weighting models, along with other properties which can influence the geometry and smoothness of the output raster file. This method works best when the input point data is fairly evenly distributed over the area to be interpolated and some degree of smoothing is beneficial in the output.
An example of creating an elevation grid out of point data.
The Triangulation method creates a raster surface by first creating triangles, using a Delaunay triangulation algorithm, from each of the nearest neighbour input data points. It then uses these triangles to estimate the values in each cell in the output raster using linear interpolation. The Triangulation method is most suited situations where you want the input values to be closely honoured in the output raster and the input data points are closely spaced and evenly distributed. Elevation data is probably the best example of this sort of data. The main disadvantage of this method is that the output raster which gets generated is often not smooth and may exhibit a jagged or angular appearance caused by discontinuous or sharp changes in slopes across the triangle edges. It is also not generally suitable for interpolating between large gaps or beyond the boundaries of the input sample points. MapInfo Pro Advanced provides some controls to limit the maximum distance over which triangles will be generated which can help constrain the interpolation and improve the appearance and quality of the output.
An example of creating a grid using triangulation
The Minimum Curvature method is a spline based technique which is widely used in many branches of science. Instead of averaging input values like inverse distance method does, it attempts to fit a flexible continuously smooth surface over the data. Just like trying to stretch a flexible rubber sheet across all the known point values. The algorithm attempts to honour the input data values closely while also attempting to minimize the amount of bending forced required to conform the surface to the data points. The degree of bending is constrained by a tension parameter and this can be specified both within the data area and along the boundaries. MapInfo Pro Advanced provides full control over the various Minimum Curvature algorithm parameters to allow you to create smooth and visually appealing raster files.
An example of creating a minimum curvature grid from elevation data.
The Hotspot Density method is a technique that estimates the proximity (or density) of samples in a given area. The value at each raster cell in the output, is an estimate of the frequency or density of samples which fall within a certain distance radius of that cell. The value at each grid node is determined using a kernel estimation function. The kernel function operates within the influence of the search radius and usually weights each of the input samples using a distance based function similar to the inverse distance method. The Density method supports simple frequency (a count of samples in the search radii) or a number of Kernel Density Estimator (KDE) functions including: Uniform, Triangle, Epanechnikov, Quartic, Triweight, Gaussian and Cosine.
An example of a hotspot density raster.
The Distance method is a spatial analysis operation that creates a raster that records the distance from the closest input data point to each cell in the output raster. It is essentially a raster equivalent of performing a buffer operation. A distance envelope can be specified which will restrict the output raster cell values so they will only be assigned out to a maximum distance defined by the search envelope. Any raster cells that are located further from an input data point than the defined search envelope will be assigned a null value in the output raster. The distance interpolation method can be useful when you need to determine the exact distance a feature is from the closest known point.
An example of a distance grid.
The Stamp method is an advanced method which does not apply any distance based interpolation weights to the input data when assigning values to the output raster cells. Instead the method simply takes the input data points and assigns the values to the corresponding output raster cells in which they fall. Because there is no distance weighting the output raster will only have valid cells where there are corresponding input data points. Depending on the chosen cell size and spatial distribution of the data it is possible that multiple input points could fall within the same output raster cell. MapInfo Pro Advanced offers several different Stamping methods to accommodate this such as: First sample, Last Sample or Average of all samples. The Stamp method is best suited to converting point data which is distributed in a regularly spaced grid pattern (e.g. point data extracted from a previously interpolated dataset) and converting it directly back into a raster data set without applying further interpolation to the input values.
An example taking regularly spaced data points and creating a stamped output raster.
Steps to create a Raster in MapInfo Pro Advanced
The Create Raster function is accessed from the Raster tab on the MapInfo Pro Advanced Ribbon. Click the Create Raster button to show all supported interpolation methods. As described above, some of the methods are more appropriate for equally spaced data with smooth variations between the data points, while other methods are better suited to handling non uniform and disperse points or regular clustered points. So choosing an appropriate method which suites the type and distribution of input data is very important to ensure the correct type of raster output is generated.
Upon choosing the interpolation method the following Create Raster dialog will appear with the properties for the chosen method.
Select the input point data set that you wish to interpolate. The input dataset may be one which is already open in MapInfo Pro (including an active selection of points) or alternatively you may click the Browse button next to the input file selector and directly choose a TAB or a LAS (Lidar LAS file) from your file system.
From the Select columns list choose one or more columns from the input data that contain the values you which to interpolate into a new raster surface. These values represent the measurements at each point (e.g. elevation, flood depth, temperature reading, number of customers etc.). Choosing more than one column of input values will enable you to create multiple output raster's (or a single multi-band raster depending on output format) in a single step.
Step 3. (optional)
Depending on the interpolation method chosen, you can define a Search Radius that will be used to locate valid input points when estimating each of the cells in the output raster. The Search Radius is defined in raster "cells". Or in other words, if the cell size of the output raster is set to 10m (measured in the distance units of the input points coordinate system) then a search radius size of 4 cells in X and Y directions would equate to a distance of 40m in all directions (e.g. an 80m diameter circle). The default search radius is 4 cells which is generally a good choice for most data.
Step 4. (optional)
Adjust the properties for the chosen interpolation method. The properties that are commonly adjusted are provided directly under the method options. Additional, advanced properties, which can be used to fine tune the interpolation algorithm are available under the "More Options" section and can be modified if required.
Step 5. (optional)
Define the cell size for the output raster. By default, the cell size is computed for you automatically. So you don't need to define this explicitly unless you wish to. In automatic mode, MapInfo Pro Advanced will take a sub sample of the input data and analyze its spatial statistics, distribution, clustering etc. and estimate a suitable value for you. If you want to know what the automatic value is before you start the interpolation process you can click the "Suggest" button next to the cell size and the value will be shown. You can fine tune the cell size to suite your needs or the input data more appropriately. For uniformly spaced data the cell size should be approximately half the mean distance between data points. For clustered or irregularly dispersed data, the cell size value requires careful consideration and for the most part should be set manually.
Define the output file path, file name and output format required. If you choose to output the raster in the new MRR format, then you can also define compression settings for the output file. The default compression method is Zip level 3. This is a good compromise between smallish file size and speed. But if you are more interested in speed over file size, then a better choice would be LZ4 compression. If you want to minimize the output file size as much as possible without being too concerned by the speed then Zip level 5 or LZMA level 5 would also be good choices.
Click the "Process" button and MapInfo Pro Advanced will begin a background task to interpolate the raster(s) for you. Should the take some time you will still be able to do other things with MapInfo Pro because the interpolation process runs in the background.
The following raster map demonstrates the capably and scale of data you can process with MapInfo Pro Advanced. This heat map was generated using the Hotspot Density interpolation method on approximately 25,000,000 points which represent the distribution of workers across the entire USA. The entire operation took approximately 5 minutes.
Want to try MapInfo Pro Advanced?
MapInfo Pro Advanced is available for all to try.
If you do not have MapInfo Pro v15.2 you can download the free trial here:
If you have already installed MapInfo Pro v15.2 you can activate a trial of the Advanced version at a time of your choosing. This is done in the Licensing tab of the Backstage area.
| || || |
|Kapil Chaudhary, |
Senior Advisory Engineer
| ||Mahima Relan, |
Associate Software Engineer, QA