SNAP hardware requirements and Tutorials Links toi

Are there any specific hardware requirements for using the SNAP 2.0. I am finding it to be quite slow in post-processing the IW-VV sample data (Sentinel-1,IW,L1 - SLC,Single VV,05/05/2015, ~2.7GB) at the bottom of the following page:

The spec of my machine is Win 7, 64-bit, 32GB RAM, 3.7 GHz Xeon.

I am basically trying to step for calibration given in the the tutorial here:
(outdated & removed, please check here for current tutorials:

Also, could you possibly release tutorials that are supposed to be in the following page:

I am particularly interested in steps that would cover topics for a land cover classification.


The specs of your machine look pretty good. Actually SNAP should run also on machines with just 4GB of RAM. It really depends what you are doing with it and which data you are using. However you computer seems to be well fitted for any application.

We are still working on tutorials and developer guides. Please give us some more time.

Maybe @lveci can say something regarding the speed of the post-processing of the IW-VV sample data.

Okay - thanks. It looks good. I look for the tutorials.

PS I noticed that the SNAP doesn’t release memory that well. So, I worked on some stuff yesterday but have since then closed all the projects however SNAP is still using 20 GB of memory. Could there be a memory leak?

There could be a memory leak but I think the memory is released as soon as memory is required by another process.
Just tried it myself. After closing all products SNAP still used ~2GB. I’ve started another application and used a lot of memory. As soon as the memory reached the limit of my machine the usage of SNAP dropped to 25MB.

Thanks. Any suggestions on any process graphs for detecting land covers?

Help can be given if specify your task.

How big is the scale of your project (continent, country, globe, ice-shelf, region around a city,…)?

Next step is to what are you observing (agriculture, surface coverage patterns, snow coverage, …)?

These basics of remote sensing should give you some ideas to define parameters for processing.

(And for information from the surface the the backscattering coefficient is used, as the sigma0 is roughly said: the response from the surface plane.)

Classification can be performed through the raster analysis analog to the multispectral classification (Landsat, Sentinel-2).

Hi marpet.
Even if memory is released when required from other processes, till that moment PC performances are very lowered. Simple operations like web browsing or windows resizing becomes very painful. It would be greater if, for example, after closing products opened in SNAP, memory would be released. I work with Windows 7 and 8GB RAM

please what can i do

Maybe it helps to move the product into the root directory C:\.
The cause for errors when reading S2 data is often that the path is to long.

emm ok thank u :smile:

hello …i wanted te creat stack …but always the same problem :confused:

have u any idea please

I assume this is a GeoTiff you are opening and it can’t do something with
this specific type. It’s always best to work with the proper EO products
including all metadata and auxiliary files and not just on an image.

Hi @Marpet, what about SNAP 5.0, using a laptop with 2GB of RAM and 2.10 GHz (including the specifications in the attached image?).

if it’s not compatible, any recommendation on how to overcome the problem of DataBuffer indicated in the error message below?

many thanks in advance! :slight_smile:

I would not expect SNAP to work on such a system - certainly not well.

@mengdahl is right. You can start it, yon can open and process some small products something like MERIS reduced resolution. But not much more.
If you want to process something like Sentinel-2 or Sentinel-1 you will have no luck. There is also one way around this databuffer error, buying an new Notebook.
There is one small chance, but it will be slow.
Open the in <snap_dir>/etc
Change the value of snap.jai.defaultTileSize to 50
Set snap.parallelism to 2 or maybe 1 (and remove the leading #)
And you might need to lower the value for snap.jai.tileCacheSize. I don’t know the default on your system. Maybe three-fourths of the default value.
Even if it will run you won’t be happy with the performance.

I have a HP DL380 server with 12GB Ram but SNAP just uses 3GB , I’m trying to run a PCA but its it is taking several days and is very slow.
should I have good VGA Card too?

You can increase the amount of RAM to be made available to SNAP as suggested here:

In your case, setting -xmax to 8G seems reasonable.
But please be a aware that a PCA can take very long even with high performance computers. Please check the discussion here: Normalized Eigenvalues in PCA

How much bands do you use for the PCA and what are their dimensions?