Time Series (Temporal) of NDVI

Yes!!! I did as you explained and now it works!
The problem was that I was creating the NDVI as independent maps and not as an extra band of the same image…it seems that it makes a difference.
Many thanks for your time and good explanations.

-A final question: what would be the best way to export ONLY de NDVI’s as GeoTIffs ??

(the Sentinel-hub EO-Browser looks fabulous)

glad to hear that it works.

Raster > Data Conversion > Band Select
Here you can define GeoTiff as an output and select the NDVI layer as a GeoTiff. But note that this will lose all metadata required for the time-series plugin.

Understood! Many thanks Andreas.

Thank you, I will try it


I want to do classification of S1 time series images with time series “NDVI values” gathered from EO Browser.

How can I do it?


load all them in SNAP, create a stack, rename the bands according to the date and use this stack as the input for the classifier.

Just make sure you download them in a usable file type (NDVI requires float).

I just have NDVI values in excel format (not as images because it is not a downloadable band as shown below).

Thats why I checked the link below but dont understood how to create csv file format for my NDVI values Since I dont have lattitudes or longitudes but only have mean NDVI values of some polygons.

what do you mean about float?

float = floating point = decimal number [between -1 and +1]

I thought you had downloaded the NDVI rasters for multiple dates. If you have extracted the NDVI value at a specific location I am not sure what would be the best way to use it.

I tried your workflow many times but it didn’t work. Then I created NDVI images as below and tried again but still doesn’t show the time series. It only shows only curser when it is on. (All vector polygons are imported to all NDVI images).

Because of the problem I mentiened above, I extracted all NDVI values manually from EO Browser but I dont know how to integrate those values with my SAR images and classify together. Is there anybody have any suggestions? I have to solve this problem before weekend :frowning:


I don’t know how to explain it any other than above.

As an alternative you can extract the NDVI values along the time series as Pins (in the pin manager), save it as a CSV file and create the graph in MS Excel.

It is demonstrated here: Extracting single pixel information

I need mean NDVI values of polygons instead of a random pixel inside.

Any other options such as QGIS, ArcMap etc. to calculate mean NDVIs? I already have NDVI images as .img format and shape files or kml files as polygons.

I tried in QGIS as below but It doesnt display the spectral profile.

1- I imported all NDVI images as raster layer and all shape files as vector layer
2- Then I added all NDVI images as spectral layers and selected alI vectors as polygons but It doesnt work.

I think I am missing something. I need some help to calculate the my time series of mean NDVI values and download the values.

the QGIS plugin seems to be a good solution (although I don’t know it). I think it is crucial that the data is formatted in a way that the plugin can attach a date to the mean values.

Sorry, I haven’t done this outside SNAP so far.

Of the three date images which were coregistered, I am not getting one date values in the time series plot. Why is it so?

I have a problem to give the same name for all products.

I downloaded all the products (a mix of S2A and S2B in the same area over 1 year). I opened them in SNAP. Right clicked on the product, in Product Explorer window > band maths. The first I could name ‘NDVI’ and create a new band (not virtual expression) using the NDVI calculation. However, it doesn’t work for subsequent products.

The error says: “The band name must be unique within all product scopes”.

Are there some other pre-processing steps to do before using the band maths tool? Coregistration, making a spatial subset, or band subset, or collocation perhaps? My computer memory is also getting overloaded when I try to use the time-series tool.

So far I have the time series graph working for the cursor position with Band 8 and Band 4 separately, but not with them combined as an NDVI.

Each product (=each date) can have only one band called NDVI.
I got it working by resampling and then applying Band Maths on every product.
You then use all products as separate inputs for the time series tool.

I found my naming problem was due to that I still had virtual products of NDVI in the bands for each product. By deleting these products, I could use the raster calculator to make a non-virtual band, which I think avoids the need to resample.

Next problem: I removed then re-added all the products into the time series tool. However, when I try to filter the band and select NDVI, it is not an option.

I clicked on the load configuration option, is it possible to add an NDVI option that way?

Thanks again for your helpful responses.

Oh I missed step 7, to save all products into BEAM-DIMAP format.

Yes, after conversion File > Save Product is required to make the bands permanent.

Sorry for responding slowly, I’m not in the office at the moment

No problem, I am very glad for your responses whenever they come :slight_smile:

I saved a subset (spatial and band, only the new NDVI bands I created) as BEAM-DIMAP, and now I am able to get a time-series graph of NDVI when using the cursor :smiley: :smiley: .

The problem now is to get it to work for an ROI. I added 2 vector data containers, for inside and outside of a landslide area, and made a polygon in each. I select one of the polygons, and click on ‘show averaged ROI’, but then nothing is showing in the graph. Is it because I made the vector data containers in only one product? Ideally I would like to compare the mean values within the 2 polygons on one time-series graph.

hard to tell, to be honest. Usually, having the ROI in one product is sufficient. I didn’t entirely understand why it works only sometimes.