Utilizing satellite data to monitor and visualize terrain changes in a map application

Abstract

The aim of this project was to improve route planning for forest machines by using up to date and accurate satellite data. Traditional maps are not always up to date, are too general and do not contain enough information about the carrying capacity of the terrain or current changes. Sub optimal routes can for example increase fuel consumption and soil erosion. Satellite data can be used to update and improve maps top reflect near real time conditions.

In the project, a Python based system was implemented that automatically retrieves and analyses the satellite data from Sentinel Hub and creates KML files based on them. The KML files can be used to visualize the images on a Google Earth Pro. OpenCV tools like findContours were used to analyze the satellite images and to identify visual changes in the data. Additionally, an interface was build to allow users to to retrieve KML files and images to a map service such as Google Earth.

The system's performance was tested using automated HTTP calls to test the API's response times in different situations. The results were analyzed using SPSS software and two way ANOVA test. The purpose of the test was to determine if the size of the map regions and the number of map layers have an effect on the response times. The results showed that the number of map layers selected by the user had the highest impact on response times. The size of the area had a significant impact to the response times when the user revisited the same area where the satellite data had already been retrieved and analyzed. The test showed that that the repeated use is significantly faster that retrieving and analyzing new data. This shows that the system is best suited to use case where the same areas are repeatedly viewed.