[SoC] Project proposal - surface reflectance mapper tool for SoC

[Helena Burriel]

Hi everyone,

I am a first time poster so please excuse me if I don't follow proper
etiquette (also let me know it, I want to improve!).

I am writing because I am interested in developing a tool I find could be
useful for the Remote Sensing community. I also am willing to know your
opinion about it.

Here is the situation. I am currently optimizing an satellite simulator,
and most of the inputs for it come in the form of user map inputs in a
variety of formats the user has to provide (CSV, ASCII, and other simple
text input formats), and then they have to define manually the maps in a
text format and later with matlab or some other program interpolate the
values to the size of the whole scene, with as you may guess, the
interpolation problems in the adjacency between two different surfaces, the
lack of variety inside a single class, etc. From my point of view, I could
ignore that part since is not my job to provide inputs for the simulator,
but from the point of view of the potential user is really a problem.

I am thinking of a graphic tool where you could define a canvas size and
pixel resolution, then select from a surface reflectance palette the type
of surface you want and paint it in x,y coordinates inside the canvas of
the size of the map. It would be also selectable the type of variation in
one class, defining ranges and the type of statistical distribution for the
area of the class, e.g. reflectance values from 0-300 following a laplacian
distribution. It would be also possible to generate the outputs on the form
of geotiff, netcdf or other geolocated format.

For the part of having a reflectance palette I am thinking at first import
some current surface library such as Corine Land Cover (one of the simplest
versions at the beginning), and also define an influence area in pixels for
transitioning from one class to another to simulate adyacency between
classes.

>From the point of view of technology, I am thinking on using OSSIM, CImg,
for geometry and visualization, so C++14 or maybe C++ as backend and Python
or HTML5 as frontend. I am fluent in C++, Python not so much but I can
manage well, and for the web technologies I have no problem. I am also
thinking on whether using a database from the beginning or just use binary
netcdf as support until the collection of surface reflectances demand it. I
would like to hear your opinion on all that, so please and thank you!

I also think maybe this project is too big for a SoC, so I proposed a first
approach to define the core of the application that would leave building
from it once the SoC is over, and also allow to grow a community around it.

I am willing to see if this kind of tool would spark some interest in the
Remote Sensing community and find more people interested in using it. I am
open to discussion or collaboration outside SoC, as well.

Thanks,

hburriel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/soc/attachments/20200408/64fda382/attachment.html>