Diversity Indexes

This repo store codes used to process raster and vectorial data aiming at computing/testing few diversity indices.

Synthesis

What for?

These focus on showing diversity of classes of land-cover in a raster layer. That diversity may speak out in terms of raw number of classes, but also may be thought as diversity of spatial distribution (number of patches).

Example of a parcel (RPG Complété 2020 - ODR) overlaying a very diverse amount of land-cover pixels (OSO 2021 - CESBIO)

For who?

It may be useful for landscape ecologists or any other kind of domains aiming at assessing the diversity of raster cells contained in polygon data.

How?

Despite the code had been parameterized, it still needs installs and little skills in python/tech (using conda for install, editing .json file...)

Appended fields :

Between parenthesis : the field name used

Basics infos (computed for each polygons individually):

• total cells number (cells_n)
• total patch number (patch defined as a contiguous set of cells belonging to the same class) (patchs_n)
• class number (class_n)

Diversity indexes

For more infos about diversity indexes and their meaning look at Fragstat docs

Basics indexes:

• Class diversity (class_d)
• Shannon diversity (shannon_d)
• Simpson diversity (simpson_d)
• Jaccard diversity (jaccard_d)
• Shannon evenness (shannon_e)
• Dominance index (dominance_i)
• Contagion index (contag_i)

Patch based indexes:

• Landscape Division Index (LDI) (ldi)
• Landscape Shape Index (LSI) (lsi)
• Patch Richness Index (PRI) (pri)
• Interspersion end Juxtaposition Index (IJI) (iji)

Edges analysis

• Edge diversity index (e_div_i)
• Edge mode (e_mode)
• Edge class number (e_class_n)
• Difference in class number between edge and entire polygon (e_p_class_n)

Negative buffer edges analysis (-15 meters)

• Negative buffer edge diversity index (nbe_div_i)
• Negative buffer edge mode (most represented lc class) (e_mode)
• Negative buffer edge class number (nbe_class_n)
• Difference in class number between negative buffer edge and entire polygon (nbe_p_classn)

Usage

This tool take two inputs :

• one one-band raster layer made of integers
• one vectorial layer (format read by geopandas/fiona see below) made of polygons

Installs :

• DL this repo
• Using console get to the repo directory and type :

conda create --name <env> --file requirements.txt

Parameters :

The file param.json allows you to precise the name of your inputs which nevertheless must be stored in the "data" directory. You can precise the name of your output. If the directory "output" doesn't exists it will be created.

List of supported vector format:

'DXF', 'CSV', 'OpenFileGDB', 'ESRIJSON', 'ESRI Shapefile', 'FlatGeobuf', 'GeoJSON', 'GeoJSONSeq', 'GPKG', 'GML', 'OGR_GMT', 'GPX', 'Idrisi', 'MapInfo File', 'DGN', 'PCIDSK', 'OGR_PDS', 'S57', 'SQLite', 'TopoJSON'