W10: Data Classification Methods

Data classification definition

Data classification involves using a classification method and a number of classes for aggregating data and map features. A GIS package typically offers different data classification methods.

Data classification types:

• Equal Intervals

• Quantiles

Equal Interval definition

This classification method divides the range of data values into equal intervals. For example, if you specify three classes for a field with values ranging from 0 to 300, it creates three classes with ranges of 0–100, 101–200, and 201–300.

The equal interval classification is best applied to familiar data ranges, such as percentages and temperature.

Quantile definition

Also called equal frequency/count, this classification method divides the total number of data values by the number of classes and ensures that each class contains the same number of data values.

A quantile classification splits the data into classes that each contain the same number of observations.

Comparing data distributions or when equal representation of categories is needed.

In GIS classification, “quantile classification” usually means you choose a number of classes and each class gets the same count of features; if you choose 5 classes, you are using quintiles

Natural Break (Jenks)

): Also called the Jenks optimization method, this classification method optimises the grouping of data values. The founder of the Natural Breaks data classification method was a cartographer named George Frederick Jenks.

The method uses a computing algorithm to minimise differences between data values in the same class and to maximise differences between classes. The first thing to remember about the Natural Breaks (Jenks) classification is that it is an optimisation method for choropleth maps.

In short, it arranges each group, so each class has less variation.

Best For: Data with uneven distributions or natural clusters.

User-defined (data classification) explained

User Defined: The user-defined (manual) classification method allows the user to set custom class ranges for data categorisation in GIS. Unlike automated classification methods (e.g., Natural Breaks, Quantile, or Equal Interval), which rely on statistical algorithms to define class boundaries, the manual method gives full control to the user, enabling them to tailor the classification to specific requirements, datasets, or study contexts.

Manual classification allows for incorporating expert knowledge, real-world conditions, or domain-specific thresholds.

Example: For NDVI, the user might set specific thresholds for land cover types based on vegetation studies (e.g., -1.0 to 0 for water bodies; 0.1 to 0.22 for built-up areas).

Land Surface Temperature (LST)

Land Surface Temperature (LST) is a critical measure in the field of Earth sciences, representing the temperature of the Earth’s surface as it is measured from space by satellites equipped with remote sensing technology.

Unlike air temperature, which is commonly reported in weather forecasts, LST reflects the ‘skin temperature’ of the Earth, essentially how hot the surface would feel to the touch in a particular location.

It helps in assessing urban heat island effects, monitoring agricultural droughts, studying water cycles, and understanding global climate change patterns.

LST - Measurement Methods

LST is primarily measured using thermal infrared radiation emitted by the Earth’s surface, which is detected by remote sensing instruments.

Different surfaces - such as urban areas, water bodies, vegetation, and bare soil - emit varying amounts of thermal radiation based on their material properties and temperature.

These variations are captured by sensors aboard satellites like Landsat or MODIS, which can then be used to derive accurate surface temperature readings.

LST data is invaluable for a wide range of applications, including meteorology, climatology, hydrology, and environmental studies.