Broadly speaking, change detection is the process of noting differences in an object by observing it multiple times. In Geographic Information System, or GIS, applications change detection involves comparing two satellite images to see their differences. GIS change detection algorithms compute differences in certain aspects of these images over time, such as the size or location of a feature. Often the time interval extends over an event, to track changes due to the event. For example, GIS satellite imagery could capture a forest before and after a fire. Then, GIS change detection algorithms can be applied to understand the fire’s damage.
GIS change detection algorithms can track trends and seasonal changes in satellite imagery, helping us to understand long-term change. For example, they are used in surveying projects to track deforestation patterns, urban expansion, river flooding, and much more. Understanding trends helps with planning and decision-making, such as wildfire mitigation strategies and monitoring water resource changes to create conservation plans. Change detection is useful for city planning, as well, when tracking rates of growth in certain areas.
Methods of Change Detection in GIS
A GIS marries spatial data with statistical data to reveal important trend information. This spatial data is obtained remotely through a drone or satellite. Once spatial data is gathered over time, it can be analyzed to decipher change. For example, comparing GIS satellite maps of the polar ice cap five years ago to now would reveal ice cap change in the last five years.
The statistical data is usually linked to a location, such as lifestyle data on family size or spending habits in that family’s geographical home. The spending habits of the family unit or person would be tied to where that person lives, and when spending habits for vast numbers of families are compiled, this information can be visually represented on a map. Geographical trends in spending are revealed visually from this map. When assessing change in spending over time for a set location, the spending habits map from one time must be compared with that of a different time.
How to Conduct a Change Detection in ArcGIS
ArcGIS, a leading GIS software, includes change detection tools to help understand trends. ArcGIS’s algorithms can detect change between two rasters. What is a raster? A raster is a data storage mechanism containing numerous pixels, as in a map, but each pixel contains valuable information such as elevation. A raster is a convenient way to store information visually, which is helpful to the eye, and enables advanced statistical and spatial analysis.
An example of a raster would be the visual representation of spending habits for each family, as discussed above. Each pixel in the raster would represent a family unit, and the value for each pixel would be the family’s spending amount. Going along with this example, ArcGIS can detect change between the raster of spending habits for one time period and the raster of spending habits in another.
See how researchers are using data from GIS and people on the ground that offers functional insights for the community.
Compute Change Raster
One of the ArcGIS change detection methods, Compute Change Raster, calculates the difference between two rasters. This difference can be absolute, relative, categorical, or spectral. To calculate the mathematical difference, the first raster dataset would be subtracted from the latter one. This method could work well for the spending habits example.
Spectral Euclidean Distance and Spectral Angle Difference treat each pixel in the image as a vector, where larger values or angles indicate more change. This method relies on the fact that a satellite image is multispectral, or multi-colored, and each color in the spectrum reveals information about the object within the image. Natural materials and man-made materials have different spectral profiles, like “fingerprints,” that allows algorithms to quantitate them.
Categorical difference outputs contain an attribute table with listed transition types, the number of pixels that underwent each transition, and the estimated overall area of each transition type. This kind of analysis can be beneficial for measuring urban expansion, deforestation, and more.
Continuous Change Detection and Classification
Another ArcGIS change detection tool is the Continuous Change Detection and Classification (CCDC). This algorithm evaluates changes in pixel values over time and creates a change analysis raster. Within the change analysis raster, each pixel contains information describing its history over time. This algorithm is meant for detecting observed, visual change, as in charting forest cover differences using satellite imagery. In that case, to ensure accurate results, keep satellite images clear from atmospheric interference such as cloud cover.
The Big Picture
From predicting urban growth to limiting deforestation to making a water conservation plan, Change Detection software is essential in our evolving world. Contact us at Maptelligent to capture a digital copy of your real-world asset to maximize the impact of your change detection software.