My journey in Kaiinos began in a joint workshop conducted in our college. Being a GIS student I used desktop GIS tools for spatial data analysis and feature extraction. I was computing indices like NDVI, NDWI, NDBI. I was also preparing MapLayouts using desktop GIS tools. OpenJUMP was the first FOSS tool that I used. Using this tool helped me to view the source and build my first tool to local police station. This work gave me confidence about my ability to code. My role in KAIINOS improved this ability to code. I had opportunities to work on many sustainability projects and automate some of the routines. In addition to it I also had a chance to experiment statistical analysis like PCA, thresholding and ranking using statistical library R on spatial datasets.
Journey from manual analysis to Machine learning
The satellite images that we access have Digital numbers (DN values) responses . Geotools library is a great FOSS library with good documentation. I use it to read satellite images and apply indices to them. Write algorithms to identify land features is also possible with this library. Some of these outputs have noise.To overcome it we understood the importance of indices. Geotools helps us to fine tune them to suit the problem requirements. Based on context we update these algorithms to extract features based on indices like NDWI, NDVI & etc.Working on some of these indices to understand land use helped me build custom mechanisms for supervised and unsupervised classification.
Journey from file systems to databases
After learning these basic automation tools I started working with team on converting data to relational databases. As we deal with huge volume of data reducing the time to do real time on web analysis was a challenge. We worked on optimizing these spatial datasets. Optimizations that we do are not just for querying but also for storage. By reading through spatial indexing methods we built in house library for optimization.
The world of Web GIS
Conveying information by visualizing data is the major advantage of GIS. So we strive for better visualizations. We use two types of data for visualizations.
- Raster – Satellite image rasters or processed raster or digital elevation models
- Vector – Land features, forests, water bodies etc.
Raster styling is more complicated than vector styling. To address this we had to explore many tools like Open JUMP, QGIS, Geoserver, Geowebcache etc. As a result this helped us to become one of the pioneers in India working on web cartography. And that helped us to conduct workshops all over India. And to make these workshops more effective we collaborate OSGeo Indian chapter.