0

I am attempting to create Supervised Land Cover classification maps using the classes (Mangrove, Non-Mangrove, Barren, Urban, and Water) for four period of interests (1988, 2000, 2017 and 2020) using ArcMap 10.8 Maximum Likelihood Classifier (MLC). I am using Landsat images acquired from USGS Earth Explorer. The objective is to evaluate the extent of changes in each of the classes for each period.

I have been successful at classifying the images for 1988 and 2000 using the 1988 signature file. However, when attempting to run the MLC via a Supervised Classifier on the remaining images I get the following error messages pictured in the screenshot below.

enter image description here

Improve this question
4
  • 1
    For questions that involve error messages we ask that you show us them in full using text rather than a picture so that they can be searched and also read easily on all devices. There is an edit button beneath your question which will enable you to do that and " " or {} buttons that enables you to format any highlighted text nicely.
    – PolyGeo ♦
    Commented Mar 10, 2021 at 3:16
  • 1
    Try to cleanup workspace
    – FelixIP
    Commented Mar 10, 2021 at 3:44
  • 1
    Please do not include chit chat like greetings, statements of appreciation/thanks, etc - see meta.stackexchange.com/questions/2950/…
    – PolyGeo ♦
    Commented Mar 10, 2021 at 5:11
  • 1
    Please note that, based on what you are describing, it is critical that your data be radiometrically calibrated across scenes. This is the case for single-sensor data or, even more importantly, for multi-sensor data (eg., mixtures TM5, ETM+7 and OLI). This corrects for radiometric differences due to scene angle/illumination and different dynamic ranges across sensors. Calibration can be done via a simple linear regression or using a pseudo-invariant features. Unfortunately, I am not aware of a way to do this in ArcGIS. Your error indicates a mismatch between the signature file and raster data.
    – Jeffrey Evans
    Commented Mar 10, 2021 at 20:43

2 Answers 2

Reset to default
1

You might want to start by focusing on Error 010108 that says you have a mismatch in number of bands in the signature file and raster dataset. Double check your inputs, and check if the rasters for all four years have the same number of bands. More bands have been added in recent Landsat satellites, so your 2017 and 2020 images likely have more bands than your 1988 and 2000 images do.

Improve this answer
0

suggest you ensure that the raster file upon which you created training samples is the same as the input raster in the Maximum Likelihood Classifier tool.

ML Classifier screenshot

Using the above image as a example, ensure the raster file for training samples (blue square) is the same raster you input in the Maximum Likelihood Classifier tool (Red square). From your error message, the raster xxxx_compo_clip_2017 is different from the classifier raster. Try fixing this by following the steps I've outlined here.

After series of trial and error, this was a precise solution to the problem for me. I hope it helps.

Improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.