Hi Dannica,

NDVI and NDWI can be derived by using Landsat bands.

LANDSAT NDVI/NDWI products are also available to download (you may use Google Earth Engine or USGS Earth Explorer). e.g. https://developers.google.com/earth-engine/datasets/catalog/LANDSAT_LT04_C01_T1_8DAY_NDVI

https://developers.google.com/earth-engine/datasets/catalog/LANDSAT_LT04_C01_T1_8DAY_NDWI

ASTER and SRTM DEMS are available to download, unless you have any other local data source depending on the required spatial resolution.

Indishe Senanayake thank you very much for your response. However, I already have these ancillary data and I’ll be having six data layers—4 Landsat bands and 2 ancillary data. I‘m looking for a particular ArcGis tool/s which will allow me to do a multi-source classification. How will I be able to combine Landsat bands with the ancillary data? From what I understand, the composite tool is used for combining satellite images in order to create RGB band combinations. What about if I want to incoporate ancillary data? Any suggestions will be highly appreciated, thanks!

Dear Dannica Rose Aquino , Indishe Senanayake briefly outlined the essence.

Here It is not clear what kind of research you are going to do, and what kind of the study area you work with.

I'd like to add the following:

NDVI and NDWI - ARE derived from the Landsat band data directly. Thus, there is nothing to add exactly to the indices you mention. There are some enhancement methods like that outlined in the papers attached below, but its application is limited. Note that in many cases the NDVI and NDWI are not a best choice. Using the ready-for -analysis products, you need to check the data quality carefully. These indices are very sensitive to background, local atmospheric conditions, which may be not taken into account by the general processing algorithms (especially, when there is not enough in-situ data), etc. If you still decide to use the NDVI and NDWI Any case, it is up to you to decide which method to use.

Generally, I'd recommend using the SWIR bands. They are extremely useful in classification of water (and wetness) and plant cover. There are many methods to choose for the certain research.

The usability of DEM depends on the DEM source, its accuracy, your understanding the study area, computational resources available for work.... For applying DEM in the classification protocols it is possible to check the following approaches (see the references below): Article Topography-driven satellite imagery analysis for landslide mapping

this is the last, but not the least, just the starting point.

Dannica Rose Aquino I'm not sure whether you can consider NDVI/NDWI as ancillary layers in your case, when they are derived from Landsat bands themselves. However, in ArcGIS you can make a composite image using multiple bands (not only RGB bands). You can give it a try by making a composite with your chosen bands and run a classification.

Dannica Rose Aquino, although I would not call the RS derived indices ancillary data, but I understand that you want to create a composite image having the 4 Landsat bands, and NDWI + NDVI layers. This can be done either with a custom code (which can easily be implemented using Google Earth Engine, for instance). Nevertheless, you can also use remote sensing software such ERDAS, ENVI, etc, to create a "Layer Stack" composed of all your 6 layers, and then the finally obtained layer stack can be classified using one of the above mentioned software or programming languages. Other options would be to apply techniques such as Analytic Hierarchical Process (AHP), and then have every single image classified independent and then overlay them latter (using for instance the Raster Calculator tool in ArcGIS software) in accordance with the obtained empirical relationship from AHP technique. However, I would not recommend very much the last option, since this takes into account the weights assigned to each band and hence, it is better useful when looking for a particular feature for which you know/informed about how the different bands contribute to presence, absence or distribution of such a feature of interest.

All the best,

Maurice