Fragmentation (GWB_FRAG)#

This module conducts the Fragmentation analysis at a single (or multiple) user-specified observation scale. The result consists of a spatially explicit map, a histogram, and tabular summary statistics. Details on the methodology and input/output options can be found in the Fragmentation product sheet.

Requirements#

A single band (Geo)TIFF image in data format byte and either:

Binary:

  • 0 byte: missing (optional)

  • 1 byte: background

  • 2 byte: foreground (forest)

  • 3 byte: specific background (optional)

  • 4 byte: non-fragmenting background (optional)

Grayscale: (grayt = grayscale threshold in [1,100])

  • [0, grayt-1] byte: background

  • [grayt, 100] byte: foreground

  • 103 byte: specific background (optional)

  • 104 byte: non-fragmenting background (optional)

  • 255 byte: missing (optional)

Processing parameter options are stored in the file input/frag-parameters.txt.

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; GWB_FRAG parameter file:
;; NOTE: do NOT delete or add any lines in this parameter file!
;; Fragmentation analysis at up to 10 user-selected Fixed Observation Scales (FOS):
;; GWB_FRAG will provide one (1) image and summary statistics per observation scale
;;
;; Method_Reporting: choose one of the following 3 methods to analyze the Foreground (FG) pixels:
;;    FAD (FG Area Density); FED (FG Edge Density); FAC (FG Area Clustering):
;; combined with per-pixel reporting in either 5 or 6 fragmentation classes:
;;    FAD/FED/FAC_5/6: per-pixel reporting, color-coded into 5 or 6 fragmentation classes
;;
;; Input map type (byte) and requirements: binary OR grayscale
;; - Binary: 1-background, 2-foreground, optional:
;;       0-missing, 3-special background, 4-non-fragmenting background
;; - Grayscale: [0, grayt-1]-background, [grayt, 100]-foreground (grayt = grayscale threshold in [1,100]), optional:
;;       255-missing, 103-special background, 104-non-fragmenting background
;;
;; Please specify entries at lines 37-43 ONLY using the following syntax:
;; line 36: Method/reporting: FAD_5 (default) or FAD_6, FED_5, FED_6, FAC_5, or FAC_6
;; line 37: Foreground connectivity: 8 (default) or 4
;; line 38: pixel resolution [meters]
;; line 39: up to 10 window sizes (unit: pixels, uneven within [3, 501] ) in increasing order and separated by a single space.
;; line 40: high-precision: 1-float precision  (default)  or 0-rounded byte
;; line 41: statistics: 0 (no statistics - default) or 1 (add summary statistics) or 2 (statistics from FAD-map input)
;; line 42: input map type: Binary (default) or Grayscale grayt (e.g., Grayscale 30)
;;
;; an example parameter file using the default settings on a binary input map:
;; FAD_5
;; 8
;; 100
;; 27
;; 1
;; 0
;; Binary
****************************************************************************
FAC_5
8
100
27
1
1
Binary
****************************************************************************

Example#

The results are stored in the directory output, one directory for each input image accompanied by a log-file providing details on computation time and processing success of each input image.

GWB_FRAG command and listing of results in the directory output:

$ GWB_FRAG -i=$HOME/input -o=$HOME/output
IDL 9.2.0 (linux x86_64 m64).
(c) 2025, NV5 Geospatial Solutions, Inc.

GWB_FRAG using:
dir_input= $HOME/input
dir_output= $HOME/output
% Loaded DLM: TIFF.
Done with: clc3class.tif
Done with: example.tif
Done with: gscinput.tif
Frag finished sucessfully

$ ls -R output/
output/:
clc3class_frag/  example_frag/  frag.log


output/clc3class_frag:
clc3class_fos-fac_5class_27.png  clc3class_fos-fac_5class_27.tif
clc3class_fos-fac_5class.txt
clc3class_fos-fac_5class_27.sav  clc3class_fos-fac_5class.csv

output/example_frag:
example_fos-fac_5class_27.png  example_fos-fac_5class_27.tif
example_fos-fac_5class.txt
example_fos-fac_5class_27.sav  example_fos-fac_5class.csv

Example statistics and spatial result of custom-scale per patch analysis of the input image example.tif, here FAC_5 showing fragmentation color-coded into five categories.

Fragmentation analysis using Fixed Observation Scale (FOS)
(Fragmentation is complementary to Connectivity: Fragmentation = 100% - Connectivity)
Method options: FAD - FG Area Density; FED - FG Edge Density; FAC - FG Area Clustering;
Summary analysis for image:
example.tif
================================================================================
Precision: floating point
FOS parameter settings:
Input type: Binary
Foreground connectivity: 8-conn FG
FOS-type selected: FAC_5
Method: FAC
Reporting style: FAC at pixel level
Number of reporting classes: 5
Pixel resolution [m]: 100.000
Window size [pixels]: 27
Observation scale [(window size * pixel resolution)^2]:
Observation scale:   1
Neighborhood area:   27x27
     [hectare]:     729.00
       [acres]:    1801.40
================================================================================
Proportion [%] of foreground area in foreground cover class:
               Rare (FAC-pixel value within: [0 - 9]):      3.0306
           Patchy (FAC-pixel value within: [10 - 39]):     13.7917
     Transitional (FAC-pixel value within: [40 - 59]):     14.4645
         Dominant (FAC-pixel value within: [60 - 89]):     31.2992
        Interior (FAC-pixel value within: [90 - 100]):     37.4139
================================================================================
================================================================================
A) Image summary:
================================================================================
Reporting unit area [pixels]: 999730
Foreground area [pixels]: 428490
Foreground area [%]: 42.8606
Number of foreground patches: 2850
Average foreground patch size: 150.34737
================================================================================
B) Reporting levels
================================================================================
Foreground (FG) connectivity is available at 4 reporting levels, at:
B1) Pixel-level: method FAD/FED/FAC: check the FG pixel value on the map, or aggregated at
B2) Patch-level: method _APP (Average-Per-Patch): check the FG pixel value on the map
B3) Foreground-level: reference area = all foreground pixels
- Average connectivity at WS [%]:     70.8060
- ECA (Equivalent Connected Area) [pixels]: 221292.76
- COH (Coherence = ECA/ECA_max*100) [%]: 51.644789
B4) Reporting unit-level: reference area = entire reporting unit
- Average connectivity at WS [%]:     30.3479
- COH_ru (ECA/Reporting unit area*100) [%]: 22.135252
================================================================================
================================================================================
Histogram of FG-pixel values rounded to the nearest integer, FGcover[%] at window size:
   Value    WS27
     0     0.0411
     1     0.1351
     2     0.2219
     3     0.2992
     4     0.3230
     5     0.3806
     6     0.3841
     7     0.4196
     8     0.4303
     9     0.3956
    10     0.3701
    11     0.3692
    12     0.3998
    13     0.3701
    14     0.3648
    15     0.3753
    16     0.4215
    17     0.3888
    18     0.4014
    19     0.4189
    20     0.4159
    21     0.4292
    22     0.4273
    23     0.4507
    24     0.4280
    25     0.5064
    27     0.5281
    28     0.5284
    29     0.5057
    30     0.4985
    31     0.4682
    32     0.4899
    33     0.5137
    34     0.5167
    35     0.5230
    36     0.5333
    38     0.5582
    39     0.5613
    40     0.5869
    41     0.6049
    43     0.6535
    44     0.6462
    45     0.6649
    46     0.6686
    47     0.6696
    48     0.6868
    49     0.7029
    50     0.7697
    51     0.7050
    52     0.7547
    53     0.7622
    54     0.7722
    55     0.8087
    56     0.8294
    57     0.8434
    58     0.8532
    59     0.8742
    61     0.8833
    62     0.9027
    63     0.9270
    64     0.9081
    65     0.9405
    66     0.9242
    67     0.9389
    68     0.9568
    69     0.9638
    70     0.9816
    71     0.9888
    72     1.0171
    73     0.9944
    74     1.0215
    75     1.1041
    76     1.0502
    77     1.0733
    78     1.0910
    79     1.0745
    80     1.0672
    81     1.1015
    82     1.0530
    83     1.1281
    84     1.1326
    85     1.1650
    86     1.1758
    87     1.2334
    88     1.2761
    89     1.3202
    90     1.4747
    91     1.6971
    92     1.8117
    93     1.9004
    94     2.0939
    95     2.3226
    96     2.6192
    97     3.3737
    98     4.0479
    99     5.9481
   100    10.1246
../_images/example_fos-fac_5class_27.tif
../_images/example_fos-fac_5class_27.png

Remarks#

  • The result provides additional statistics in txt and csv format.

  • The IDL-specific sav-file can be used in GTB to conduct fragmentation change analysis.

  • The result provides a fragmentation image and a occurrence frequency histogram for each custom observation scale. In the example above, the user selected 1 observation scale with a local neighborhood of 27x27 pixels.

  • This module provides options to use either FAD/FED/FAC, process for multiple observation scales, report at pixel- or patch-level, and to select the number of fragmentation classes (6, 5, 2).

Fragmentation has been used to map and summarize the degree of forest fragmentation in the FAO SOFO2020 report and the Forest Europe State of Europe’s Forest 2020 report with additional technical details in the respective JRC Technical Reports for FAO and FE. The index FAD is adopted in EU policy to report on forest connectivity in the 8th Environmental Action programme and the Nature Restoration Regulation.