GNN-SPP 0.0.1

Modeling tree species composition with GNN.

Sam Gordon
2023-02-27

Predictor Variables

Environmental Variables

nbr, tcb, tcg, tcw, delta_nbr, delta_tcb, delta_tcg, delta_tcw, twi, slope, chm, mag, dem
yod, precip, tmax, tmin, aspect, lcpri, lcsec, wetlands, ecozones, tax_2019, dec_ecozones

Species Matrix

Basal Area per plot of 131 tree species

Bray-Curtis Dissimilarity Evaluation

Imputing species frequency matrix with GNN

Imputing Proportional Species Frequency (n/total)

Imputing geographically closest plot among 5 nearest neighbors in CCA ordination

Imputing random plots from FIA training set

Overall Dissimilarity of FIA Plots

Species Presence/Absence

Binary Presence/ Absence Accuracy for top 50 tree species (by basal area)

Species n_test n_pred Accuracy PPV TPR TNR
red.maple 274 282 0.686 0.77 0.792 0.435
sugar.maple 236 220 0.676 0.75 0.699 0.641
American.beech 209 192 0.689 0.729 0.67 0.711
eastern.hemlock 122 119 0.638 0.42 0.41 0.742
white.ash 161 159 0.635 0.56 0.553 0.693
yellow.birch 143 161 0.722 0.609 0.685 0.744
eastern.white.pine 91 95 0.712 0.389 0.407 0.805
balsam.fir 70 59 0.889 0.729 0.614 0.95
red.spruce 95 93 0.877 0.753 0.737 0.922
black.cherry 133 140 0.638 0.471 0.496 0.711
northern.red.oak 89 82 0.797 0.561 0.517 0.88
quaking.aspen 47 65 0.784 0.215 0.298 0.851
paper.birch 54 39 0.853 0.462 0.333 0.937
eastern.hophornbeam 88 82 0.733 0.402 0.375 0.837
striped.maple 69 58 0.828 0.517 0.435 0.912
American.elm 61 65 0.82 0.431 0.459 0.887
sweet.birch 61 47 0.83 0.447 0.344 0.921
Norway.spruce 4 8 0.979 0.25 0.5 0.984
apple.spp. 20 26 0.897 0.115 0.15 0.938
American.basswood 59 43 0.82 0.372 0.271 0.918
green.ash 19 24 0.925 0.292 0.368 0.954
red.pine 16 9 0.956 0.444 0.25 0.987
northern.white.cedar 20 18 0.938 0.389 0.35 0.97
hawthorn.spp. 19 30 0.9 0.167 0.263 0.932
chestnut.oak 22 20 0.943 0.5 0.455 0.973
Scotch.pine 5 11 0.969 0.182 0.4 0.977
serviceberry.spp. 30 48 0.846 0.188 0.3 0.891
white.oak 31 36 0.874 0.25 0.29 0.925
bigtooth.aspen 23 34 0.884 0.176 0.261 0.923
American.hornbeam..musclewood 28 27 0.879 0.148 0.143 0.936
bitternut.hickory 23 24 0.915 0.292 0.304 0.954
black.ash 16 19 0.925 0.158 0.188 0.957
gray.birch 17 16 0.936 0.25 0.235 0.968
shagbark.hickory 25 31 0.877 0.129 0.16 0.926
pitch.pine 2 5 0.982 0 0 0.987
black.oak 18 20 0.918 0.15 0.167 0.954
pignut.hickory 14 20 0.928 0.15 0.214 0.955
white.spruce 3 5 0.979 0 0 0.987
scarlet.oak 8 9 0.961 0.111 0.125 0.979
eastern.redcedar 8 5 0.977 0.4 0.25 0.992
tamarack 8 12 0.959 0.167 0.25 0.974
silver.maple 7 7 0.969 0.143 0.143 0.984
pin.cherry 12 6 0.954 0 0 0.984
larch.spp. 1 2 0.992 0 0 0.995
black.locust 7 6 0.967 0 0 0.984
blackgum 5 6 0.977 0.167 0.2 0.987
swamp.white.oak 6 3 0.982 0.333 0.167 0.995
black.spruce 5 4 0.977 0 0 0.99
boxelder 1 3 0.99 0 0 0.992
black.walnut 8 6 0.964 0 0 0.984

Corrections

If you see mistakes or want to suggest changes, please create an issue on the source repository.

Citation

For attribution, please cite this work as

Gordon (2023, March 5). CAFRI Labs: GNN-SPP 0.0.1. Retrieved from https://cafri-labs.github.io/acceptable-growing-stock/posts/gnn-spp-001/

BibTeX citation

@misc{gordon2023gnn-spp,
  author = {Gordon, Sam},
  title = {CAFRI Labs: GNN-SPP 0.0.1},
  url = {https://cafri-labs.github.io/acceptable-growing-stock/posts/gnn-spp-001/},
  year = {2023}
}