qgis-pan-europeo

Pan European Proof of Concept

Quick start
How to use
- Raster configuration
- Resampling methods
About us

This QGIS plugin allows its users to calculate a summary cualitative raster from a set of large rasters; by defining for each one its relative weight, utility attribute function parameters, and optionally, a resampling method for each input and a target resolution for the output.

Because the plugin is intended to process ~30 rasters at 100m pixel resolution (~30xGiB of data) a default resampling method and target precision configuration is provided; Although the plugin can work with any available raster layers. The target resolution can be skipped and the output resolution is set as the one of the first raster.

Basic User Interface	Advanced Options Enabled

Quick start

Install QGIS (latest desktop version on qgis.org)
On the QGIS menu, go to Plugins > Manage and Install Plugins
Enable experimental Plugins on Settings section
All (vertical tab on the left) > Search for “Pan European Proof of Concept” (top horizontal input) > Select the plugin (checkbox) > Click “Install” (bottom right)
The plugin will be available on the “Plugins” section of the toolbar or on the “Plugins” menu, by clicking on the icon:

plugin icon

How to use

Open QGIS
- optional: setup any CRS in meters like EPSG:3857
- optional: open the log panel (View > Panels > Log Messages) to read the plugin’s progress on the “Marraqueta” tab
Load a set of raster layers
[Optional] Select a polygon feature to define the area of study (else the visible area will be used)
Click on the “Pan European Proof of Concept” plugin icon
Configure for each layer/row (see details below)
Configure target raster creation (only for advanced options mode else the first layer resolution is used)
Buttons:
```
Reset: to clear the dialog, load another set of layers
Cancel: to close the dialog and do nothing
Ok: to calculate a new layer
```
After clicking “Ok” calculations will begin, if you feel it’s taking too long, switch to advanced mode.
Then a new, randomly named GTiff raster, will be written into your temporal files.
It can be easily export as a pdf, png or other format by right click on the layer > Export > Save as [Image]

Raster configuration

For each available layer (must be local and written to disk) available configurations are:

Layer enable/disable checkbox
Weight attributes as spinbox & slider (they get adjusted to sum 100 at run time)
[Advanced] Resample method combobox selector (see details below)
Utility function configuration, select between:
a. Min-Max scaling
b. Max-Min scaling, same but inverted
c. Bi-Piecewise-Linear Values, with its two breakpoint setup as data real values
d. Bi-Piecewise-Linear Percentage, with its two breakpoint setup percentage values from real data range (data.max - data.min)
e. Step-Up function, with a single breakpoint setup as data real value
f. Step-Up function, with a single breakpoint setup as percentage
g. Step-Down function, with a single breakpoint setup as data real value
h. Step-Down function, with a single breakpoint setup as percentage

utility functions graph

Notice that for the Bi-Piecewise-Linear functions (c. and d.) crossing the breakpoints will invert the function’s slope.
Also in the case that one of them being zero (or minimun observation) a flat part is removed, e.g., a=0 and b>0:

utility function a=0

You get up-slope and flat. Conversely, if b=0 and a>0, the graph will be reflected the vertically (as c. and d.), getting flat and down-slope.
Finally by one of them being 1 (or maximun observation) instead of 0, you get the other flat part removed.

[Advanced] Target raster creation setup:
- If enabled, by default, creates a HD (1920x1080 pixels) image with each pixel representing an hectare (100x100m) pixel size; this values can be adjusted. With a trade-off on the final raster size and processing time.
- When the zoom level is so high that the shown data is smaller than the configured output raster, the configuration is ignored and the target raster will have the same resolution as the raster in the actual viewport

Known issues

The algorithm can get confused if the shown raster is not in a squared meters projection CRS
Currently different datatypes than Float-32 is untested

Resampling methods

Each target raster is expected to have billion pixels and the resampling method is crucial to the final result in a reasonable time. The following methods are available:

Nearest Neighbor:
    This is the default and fastest algorithm.
    It assigns the value of the closest pixel in the original raster to the corresponding pixel in the output array.
    This method can introduce sharp edges and blockiness in the resampled image.
    Probably the most used method for categorical rasters.

Bilinear:
    This method considers the four nearest neighboring pixels in the original raster.
    It calculates a weighted average of their values based on their distance to the new pixel location in the output array.
    Bilinear interpolation produces smoother results compared to nearest neighbor but may introduce some blurring.
    Probably the most used method in rasters with continuous data.

Cubic:
    This method involves a 4x4 neighborhood of pixels in the original raster.
    It uses a polynomial function to interpolate a new value for the output pixel based on the values of surrounding pixels.
    Cubic interpolation provides smoother results than bilinear but is computationally more expensive.

Cubic Spline:
    This method uses cubic spline interpolation, which is a more advanced technique compared to regular cubic interpolation.
    It offers smoother results but is even more computationally intensive.

Lanczos:
    This method employs a Lanczos filter for interpolation, known for its good preservation of high-frequency details.
    It can be computationally expensive but may be preferred for resampling imagery with sharp edges or fine details.

Average:
    This method calculates the average value of all neighboring pixels in the original raster and assigns it to the output pixel.
    It can be useful for smoothing noisy data but may blur sharp features.

Mode:
    This method assigns the value that appears most frequently among the neighboring pixels in the original raster to the output pixel.
    It can be useful for categorical data but may not be suitable for continuous datasets.

Gauss:
    This method utilizes a Gaussian distribution to weight the values of neighboring pixels in the original raster.
    It can be useful for datasets with continuous variation but is less common than other resampling methods.

About us

Role	Where	Method
Outreach	https://www.fire2a.com	fire2a@fire2a.com
User docs	https://fire2a.github.io/docs/	github-issues “forum”
Algorithms docs	https://fire2a.github.io/fire2a-lib/	Pull Requests
Developer docs	https://www.github.com/fire2a	Pull Requests

Developed by fdobad.82 @ Signal App Branding & testing by Felipe De La Barra felipedelabarra@fire2a.com

This site is open source. Improve this page.