Hi,
I use the same Data right now. did you find out the answer?
Hello,
I’ve got a question regarding the time series analysis tool. Is it possible to create a graph for more than one ROI in one Image, to compare the different ROI’s over time?
no, currently only multiple points are supported
Although it is already some months ago, I tested this again and it works out (the show at cursor position option must be active as well)
a new tutorial on time-series analysis has been published: Time-series analysis with Sentinel-1
3 Likes
Excellent material! Thank you very much.
It will be very useful for me.
happy to hear that, thank you.
Dear Andreas, this tutorial is awesome. Very clear, although I have a few doubts.
- Can I use it over the ocean (As this is an example of a scene over terrain)?
- Can you please explain what information does the temporal stack of avg, min, std dev give? As you have mentioned
temporal statistics are not identical to the mathematical average and standard deviation, because these are calculated based on linearly scaled data?
Can you please explain what does this mean? How do I interpret the output as?
Thanks in advance
as long as you don’t remove the ocean areas during the terrain correction, you can use it on any data. If it is open sea area, I recommend the Ellipsoid correction instead.
They compute the mean, min/max and standard deviation of all values along the time series of one pixel. The arithmetic mean is just not 100% correct when the data was converted into log-scale (dB), because here other laws apply (https://www.reddit.com/r/explainlikeimfive/comments/6d49o6/eli5_why_shouldnt_i_average_logarithmic_values/)
2 Likes
Whoa, I just forgot to mention that I was not using terrain correction and used Ellipsoid correction instead since the scene is completely over the ocean, which I realised through this forum.
Glad for the confirmation.
And thanks for the link on log values. So, I assume mean, min/max & std dev ultimately means how on an avg that particular pixel pinned is varying?
Well, in the end, min/max are still mathematically valid, and the mean still gives you a good representation of how all images in the stack look at average. Same applies for the standard deviation (which areas are constant, which are changing over time). So you can use them but the absolute values have to be treated with care.
1 Like
Thanks a ton for the response. Will use them carefully.
Hello! ABraun, why can’t I find the time series icon you referred?
OK, I find it. Thanks a lot!
Another question,when I export the time series data. It shows like this, 
the value seems not the same as radar dB value as in the graph.
probably MS Excel did not interpret these values correctly because of missing information on the column separator. Please check here how to load them while controlling these parameters: How to open CSV files with the correct delimiter/separator – Affinity Help Center
Thanks! I’ll have a try!
Hi ABraun
Are there ever times when you would not want to transform the data to dB and rather use a linear scale? Can it affect land cover classification in the latter processing steps?
There is no general answer on this, it depends on the classes you aim at and how well these are separable under the given contrasts, also the classifiers might perform differently with linear and dB input. You’ll have to try. Here are some related discussions:
Hello! I have a problem with the Time series tool. The chart is not displayed correctly for the option Show Average ROI. When I move the click in the ROI the value of NDVI it’a about 0,25, but for the Average ROI it’s about 0,45. Please help,I need help urgently!
Please help!!
I need help urgently!
Thank you!