Statistical info API - Questions regarding history


#1

Hi all,

My questions. For introduction and detail see the rest of the post.

  • Why is the historical NDVI not available before 2016?
  • Why are available data not regular but seem to be randomly distributed in time?
  • Why do those characteristics (historical depths and distribution of date availability) change completely from one place to another ?

Disclaimer : I come from data analysis world, not EO. I apologies in advance for all obviously stupid questions I might ask.

Goal :
I want to extract time series of the NDVI average of some different locations on the world.

What I did :
So I used the Statistical info API for that.
Using different places on Earth, I requested available data, and I was surprised, because :

  • No matter the dates I choose for the request, there is no data available before 2016
  • Available dates are not regularly distributed across the time, but look random
  • History and dates distribution is really different from one place to another

One of the request I made : http://services.sentinel-hub.com/ogc/fis/MyID?LAYER=NDVI&STYLE=INDEX&CRS=EPSG:4326&RESOLUTION=10m&TIME=2000-01-01/2018-12-31&BBOX=-54.8625146271707,-68.6261626271707,-54.8544133728293,-68.6180613728293

The result I get (the circles are when data is available, lines are intervals without data):

Thanks in advance for your hints.


Download NDVI graph values
#2

Hi ibisa,

I feel your pain :slight_smile: Please find below my attempts to answer your questions.

  • Why is the historical NDVI not available before 2016?

Is this a typo? Did You mean before 2015, because the plot shows data before 2016. Anyway, to answer this question properly I would need to know which satellite data are you using: Sentinel-2, Landsat-8, …? If Sentinel-2, then https://www.sentinel-hub.com/develop/documentation/data_sources exaplains why. TL;DR The first out of two Sentinel-2 satellites (Sentinel-2A) was launched in June 2015.

  • Why are available data not regular but seem to be randomly distributed in time?

Excellent question, but difficult to answer. The frequency of available products (images) depends on many factor, such as:

  • latitude (revisit time of a point close to poles is much shorter than of a point at equator). See for example this paper
  • cloud coverage: I’m not sure if you have a threshold on max cloud coverage in your request set. But if you do, then this can throw away acquisitions that are too cloudy (which you should do). Since cloud coverage depends on your region of interest (i.e. there are less clouds in Sahara than in tropical forest near equator) and season (dry season vs. wet).
  • other factors, that I may have missed.
  • Why do those characteristics (historical depths and distribution of date availability) change completely from one place to another ?

Perhaps the part of the answer to this question is already given above. Cloud coverage depends on region of interest and seasonality. The min/max NDVI value depends on type of vegetation in some extent: is at minimum the land still covered with some vegetation, or is the land completely dry.


#4

Indeed, I have found the “MAXCC” parameter, and by modifying this I can have more or less points. I do not know what the default value is (the documentation say 100, but obviously this is not the case). Anyway, this part of the problem is solve. Thank you again.

Regarding the satellite I use, I do not have any idea. How can I know it ? How can I change it ? I have not found any parameter like that in the documentation.

Please check the statistical info api documentation at https://www.sentinel-hub.com/develop/documentation/api/fis-request. This information is hidden in the LAYER parameter. In you example request above you calculate statistics for the LAYER with name NDVI. You can check the properties of this layer (or create a new one) using the configuration utility: https://apps.sentinel-hub.com/configurator/#/


#5

Hi,

Just after putting my last reply, I discovered what you explain and how to change satellite.
This is why I deleted my post, without knowing that you where answering.
Anyway thank you very much for your help.

Now I am facing another problem, but I created a dedicated post for that : Statistical info API + Landsat 8 = Oops!

Thanks again,