There are so many options for satelltie imagery (Sentinel, Landsat, MODIS, Nighttime light and so on) which you can choose from. For the classification steps you mentioned, again there is a plethora of algorithms. What you are asking is to give you a tutorial (step-by-step instructions) on how to do a classification. I think you should find an algorithm you want to use, the data set, the GIS software you prefer, try something yourself and if you have any (specific) issues then edit your question with your focused problem you faced.

In general, for a classification problem you need training samples (to train your classifier). The more samples you have the better. It's a good approach to start with an automatic classification algorithm just to get a rough idea of the spatial distribution of your classes and then move on with a semi-automatic approach (were you have to create training samples for your classes).

Check on this website:

https://earthexplorer.usgs.gov/

Latu Khanikar

Latu Khanikar

1. You can Download either Landsat or Sentinel data. Landsat offers 30m spatial resolution while Sentinel offers 10m spatial resolution images. For Landsat please visit https://earthexplorer.usgs.gov/ and for Sentinel data please visit https://scihub.copernicus.eu/

2. You can download Landsat data from a few decades while Sentinel data are available from 2014

3. There are so many methods to create LULC images using different image classification methods in ArcGIS. It can be supervised or unsupervised

Yes your make and analysis your temporal, spectral and Spatial to determine all the changes of your study area.