1

I'm working on a very large study area. The aim is to create a vegetation mask representing only pines. I'm using Landsat 7 images. Since the recordings were made in the winter it's pretty easy to distinguish the conifers from the rest, because everything else is covered with snow.

Unfortunately I do not get a satisfactory result. What is the best methodology to get good results? I've tried to use a NDVI. Unfortunately, the results are bad.

I have also tried to use a tasseled cap transformation. Unfortunately, I have no experience using TC, so I'm not sure how to interpret/use the results.

I am using Erdas Imagine and ArcGIS.

Do you have any tips on how I can solve the problem?

closed as too broad by PolyGeo, Brad Nesom, Jason Scheirer, Mapperz Jul 27 '14 at 2:54

Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 1
    Could you please elaborate on what "bad" results are? Thanks. – Aaron Jul 26 '14 at 22:18
  • bad means, that I can not clearly see where pines are located. the values ​​do not tell so much and are all positive... – dan_ke Jul 28 '14 at 14:06
1

The task that you are describing is called image classification. There are numerous ways to classify an image--from very basic thresholding to more advanced supervised classification approaches. Image classification using multispectral data like Landsat requires basic radiometric correction, often accomplished using Dark Object Subtraction. Song et al. (2001) wrote a seminal paper to help answer 1) when atmospheric correction is necessary and 2) what methods to use. There is an old, but useful video describing the process here. I also suspect you are experiencing the ETM+ data gap issue, which may be influencing your results. There are numerous methods to address this and a good place to start is NASA.

There is a good description of how to perform image thresholding using NDVI data in ArcGIS available here. Maximum Likelihood classification is a common supervised approach and a good tutorial for Erdas can be found here.

  • Hey, thank you for your answer. I know the process of supervised classification. I just wanted to know what the best method is to classify all the tiles and if I can do that with an index? The problem is that there are a lot of scenes for my (large) area of interest.... – dan_ke Jul 28 '14 at 14:04
  • It is typically best to use individual bands (e.g. R,G,B,NIR) over a single index with a supervised classification. However, it is often helpful to include an index in with the individual bands during the classification. – Aaron Jul 28 '14 at 14:15

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