Observations, comments and tutorials on the world of open source mapping and geographical information for users and developers.
Comparative Assessment of the Unsupervised Land Use Classification by Using Proprietary GIS and Open Source Software
Mapping and investigating land use land cover (LULC) changes over a particular region is crucial for resource management, sustainability development, and holistic planning. An increasing rate of urban growth and urban sprawl could induce changes in land use as well as land transformation. However, accurate and up-to-date information about LULC is required for providing better understanding and assessing the environmental consequences of such changes.
In this study, the 2017 image from the Sentinel-2A Satellite was utilized to demonstrate the land cover classification analysis in Iskandar Malaysia. Usually, land use classification analysis is conducted through proprietary GIS software. However, this decade shows the advancement in software development, thus the emerging of free/open source software in the geospatial world.
Hence, to execute land cover analysis using the Unsupervised Classification technique, the proprietary GIS software (ArcGIS) and free/open source software (QGIS) were deployed. Then, the examination of accuracy assessment was carried out for the selected software. The sum of 250 random points was established for the assessment purpose.
The results showed the overall accuracy for ArcGIS and QGIS were 82.80% and 80.40% respectively. The kappa coefficient for ArcGIS was 0.7395, while kappa coefficient for QGIS was 0.7094. Besides that, ArcGIS demonstrated better producer's accuracy in the forest and agriculture land covers classification. Meanwhile, QGIS exhibited better producer's accuracy in the built-up and water features classification. To summarize, ArcGIS and QGIS software are reliable to be used in the land cover classification.
This post follows on from my previous three (most recent first in list) linked below. There is no doubt that in just a short time from now we will probably see a 'one-click' installer to by-pass the next steps but a least this method allows you to change 'in/out' the individual libraries if there is are updates using the usual ' sudo apt update && sudo apt upgrade '. http://www.paulshapley.com/2016/08/qgis-server-bringing-map-into-browser.html http://www.paulshapley.com/2016/07/how-to-install-qgis-server-on-ubuntu.html http://www.paulshapley.com/2016/04/how-to-install-postgresql-95-and.html 1. Firstly...Create a new QGIS project, add your layers from PostGIS, Shapefiles or raster (geotiff), style and symbolise those layers, set the projection and ensure you've set up your 'OWS Server' in 'Project Properties'. 2. Let's install the 'Lizmap' Plugin first by going into the 'Plugins' menu then 'Manage
Before we start it is always a good idea to remove and re-install apache2 web server so that we start from the same set up:- To remove Apache2:- $ sudo apt --purge remove apache2 $ sudo apt autoremove To re-install Apache2:- $ sudo apt install apache2 $ sudo /etc/init.d/apache2 restart # or $ sudo service apache2 restart 1. $ sudo apt install apache2 (if not installed already done so as above) 2. $ sudo apt install qgis-server libapache2-mod-fcgid 3. $ sudo a2enmod fcgid 4. $ sudo a2enconf serve-cgi-bin 5. $ sudo service apache2 restart 6. add the following code and don't forget to save the changes:- ScriptAlias /cgi-bin/ /usr/lib/cgi-bin/ <Directory "/usr/lib/cgi-bin/"> Options ExecCGI FollowSymLinks Require all granted AddHandler fcgid-script .fcgi </Directory> into /etc/apache2/sites-available/ 000-default.conf $ sudo gedit /etc/apache2/sites-available/000-default.conf # so that it looks like this:- <VirtualHost *:80>
GRASS GIS is an extremely powerful application for manipulating and modelling large raster datasets such as Landsat 8 tiff bands. As GRASS users we are aware the application does not implement 'on-the-fly-projections' to ease the pain when adding data in a different projection in the way QGIS does. The grass developers insist doing so would introduce artifacts and some distortions when taking measurements off the overlayed data, so we have to deal with a 'one location = one projection/zone/datum/ellipsoid' combination. How do we get our raster images to underlay or overlay our vectors? The golden rule when using GRASS is if in doubt create a new 'location' any time you want to import or use data in a different projection. I mainly work with just two projections, EPSG:27700 (OSGB36) and EPSG:4326 (WGS 84) and In my case I would define two locations and associated mapsets, one for my everyday working projects in (OSGB36 here in the UK) and one for