Error in SNAP Histogram tool

snap
#1

Dear SNAP-Team,
I am using SNAP for my teaching at university since years, but I never took the time to report some errors in SNAP which exist since years. I want to change this now and I will try to post existing problems in SNAP during the next weeks while I am teaching my winter classes.

First Problem:
The histogram tool doesn’t refresh properly.
When you open a histogram tool for a single band and change the parameters (#bins, min, max) you should be able to refresh the view (refresh view button) every time when changing at least one of the options (right?). But this doesn’t work. When you change the bins, it very often works and the “refresh view” button becomes available again. But when you just change min or max, it doesn’t work.

Example:
Let`s say you set the bins to 512, the min to 10 and the max to 255. Now you press “refresh view”. Now you get a histogram. Afterwards, you want to change the min to zero. But after changing the number, the “refresh view” is not available again. Even when you use enter after setting the new value for min. This is a big problem, when you try to adjust your histogram to the possible value range of your data (for an export and data comparison) and are still experimenting with the appropriate setting as my students do a lot. My students often get very frustrated, because they have to try out different setups and this programming mistake hinders them.

Furthermore a wish:
The “thinking in bins” is very problematic for simple Integer data - lets say 8bit Landsat 7 data. Bins are not really useful for this type of data. Furthermore, the right setting within the histogram tool is very strange now. If you want to have bin centers that directly match the integer values and if you want to capture all possible values from 0 to 255, then you have to select the following setup:

#bins: 256
min: -0.5
max: 255.5

This is not very intuitive, not a single student understands, why this is so complicated. Even I don’t understand. Can`t you add an option for integer data? I often use the “copy data to clipboard” function and let my students create “band histogram comparisons”, but this is much easier when bot histograms of diffent bands are exported with the same “bins”/categories.

I hope I could explain my concerns/the problems in an understandable way. Please feel free to ask questions if something remains unclear.

Best regards,
Geo

2 Likes
#2

I too have encountered somewhat unfriendly interfaces while practicals using SeaDAS 6, 7, and BEAM. I think some annoyances in SNAP are limitations of the underlying Java packages which may be chosen for their “free” licenses despite the limitations. In any case, decisions of how best to use scarce developer resources are made by the developers themselves.

In my view, many SNAP tools, including histogram, are useful for quick and dirty sanity checks on products, but for presentable and reproducible results statistical software ® or GIS software (QGIS) are needed. Unlike SNAP, which provides capabilities needed by a relatively small remote sensing user community, R and QGIS have orders of magnitude larger user communities united by the need to perform statistical analysis with presentable and reproducible results. In
practice, however, users find R has so many packages with overlapping capabilities it is hard to know which one is a) appropriate to the task, and b) will still be around for the time it takes to complete a PhD reseach project.

#3

gnwiii,
yes you are totally right. There is better software out there for reproducible scientific work. But in most cases there is prior knowledge needed from the students in a class to be able to combine their knowledge in GIS tools, scripting and remote sensing data (and processing) knowledge.

The problem for my classes is, that I can not teach every knowledge just in one single class. And I am teaching first to third term students with little to mostly no knowledge about scripting or other spacial operating software. The topic of my classes is remote sensing in theoretical and practical means. I therefore have to concentrate on the remote sensing part and I would never ever have time to teach R or GIS in the same class. This is why I am using SNAP mostly in combination with Excel. The students do not need any prior knowledge.

SNAP is an awesome software for beginners classes, as everything basic can be done from the GUI. But it is a pain, if little display problems appear. Therefore, I hope my little comments help to improve SNAP further.

Best regards,
Geo

1 Like