Dataset Open Access

A Convolutional Neural Network classifier identifies tree species in mixed-conifer forest from hyperspectral imagery

Geoffrey A Fricker; Jonathan Daniel Ventura; Jeffrey Wolf; Malcolm P. North; Frank W. Davis; Janet Franklin

Published online: https://www.mdpi.com/2072-4292/11/19/2326

DOI: 10.3390/rs11192326

Abstract:

In this study, we automate tree species classification and mapping using field-based training data, high spatial resolution airborne hyperspectral imagery, and a convolutional neural network classifier (CNN). We tested our methods by identifying seven dominant trees species as well as dead standing trees in a mixed-conifer forest in the Southern Sierra Nevada Mountains, CA (USA) using training, validation, and testing datasets composed of spatially-explicit transects and plots sampled across a single strip of imaging spectroscopy. We also used a three-band ‘Red-Green-Blue’ pseudo true-color subset of the hyperspectral imagery strip to test the classification accuracy of a CNN model without the additional non-visible spectral data provided in the hyperspectral imagery. Our classifier is pixel-based rather than object based, although we use three-dimensional structural information from airborne Light Detection and Ranging (LiDAR) to identify trees (points > 5 m above the ground) and the classifier was applied to image pixels that were thus identified as tree crowns. By training a CNN classifier using field data and hyperspectral imagery, we were able to accurately identify tree species and predict their distribution, as well as the distribution of tree mortality, across the landscape. Using a window size of 15 pixels and eight hidden convolutional layers, a CNN model classified the correct species of 713 individual trees from hyperspectral imagery with an average F-score of 0.87 and F-scores ranging from 0.67–0.95 depending on species. The CNN classification model performance increased from a combined F-score of 0.64 for the Red-Green-Blue model to a combined F-score of 0.87 for the hyperspectral model. The hyperspectral CNN model captures the species composition changes across ~700 meters (1935 to 2630 m) of elevation from a lower-elevation mixed oak conifer forest to a higher-elevation fir-dominated coniferous forest. High resolution tree species maps can support forest ecosystem monitoring and management, and identifying dead trees aids landscape assessment of forest mortality resulting from drought, insects and pathogens. We publicly provide our code to apply deep learning classifiers to tree species identification from geospatial imagery and field training data

Digital Publication of the training data polygons and hyperspectral imagery used in the manuscript "A Convolutional Neural Network classifier identifies tree species in mixed-conifer forest from hyperspectral imagery".

Code is available in a Jupyter Notebook and can be found here: https://github.com/jonathanventura/canopy

National Ecological Observatory Network. 2018. Provisional data downloaded from http://data.neonscience.org on 22 June 2018. Battelle, Boulder, CO, USA

Data to replicate the experiment is available for download in two zipped files: "NEON_D17_TEAK_DP1QA_20170627_181333_RGB_Reflectance.zip" (Imagery) "CNN_LABELS_2019.zip" (Training Label Shapefiles) * Note: The imagery is 5.5 gb (zipped). All code used to run the analysis is located in a repository here: https://github.com/jonathanventura/canopy The only flightline you will need to repeat our results is called "NEON_D17_TEAK_DP1_20170627_181333". If you download your own NEON data, the raw HDF 5 files can be converted to a geotiff using R code found here: http://neonscience.github.io/neon-data-institute-2016//R/open-NEON-hdf5-functions/ Contact the National Ecological Observatory Network (NEON) to download the comparable imagery data files for all sites and collections: https://data.neonscience.org/home.
Files (18.9 GB)
Name Size
CNN_LABELS_2019.zip
md5:cd47bc6b75f674a9fde56536cf4f3996
302.3 kB Download
D17_CHM_all.tfw
md5:1c5c4143ea190be96ccd66b370081ae1
90 Bytes Download
D17_CHM_all.tif
md5:7cd563036f50261f4933fddb1397ce65
959.0 MB Download
D17_CHM_all.tif.aux.xml
md5:f6f84f1a7b1142cc8608a5b8ba1efad6
2.5 kB Download
D17_CHM_all.tif.ovr
md5:7093ed985e579b6c1c488ed4d4355462
325.4 MB Download
D17_CHM_all.zip
md5:d83553422da132670e91d94633ffed3a
299.4 MB Download
Labels_Trimmed_Selective.CPG
md5:ae3b3df9970b49b6523e608759bc957d
5 Bytes Download
Labels_Trimmed_Selective.dbf
md5:154608e331495a1d4b78dbbf6178a336
11.5 kB Download
Labels_Trimmed_Selective.prj
md5:91faad560efe79cfac796ce48b289b45
403 Bytes Download
Labels_Trimmed_Selective.sbn
md5:7396a055c52e0498819501a4eac3ee15
7.2 kB Download
Labels_Trimmed_Selective.sbx
md5:0a30d4dff47b380912b05adb618af470
532 Bytes Download
Labels_Trimmed_Selective.shp
md5:956e78cc3e8e46715cd8b6813627c1ca
462.1 kB Download
Labels_Trimmed_Selective.shp.xml
md5:fe8a57215bd3669ae46b2947eb424253
17.8 kB Download
Labels_Trimmed_Selective.shx
md5:a9fb286779acee2b88816e29fccefd1d
5.8 kB Download
NEON_D17_TEAK_DP1_20170627_181333_reflectance.tif
md5:63ee0d65be0e9acaf7a9d3d4b077231e
9.6 GB Download
NEON_D17_TEAK_DP1_20170627_181333_reflectance.tif.aux.xml
md5:406e22158f789ae3670f004d71a83fe1
7.4 kB Download
NEON_D17_TEAK_DP1_20170627_181333_reflectance.tif.enp
md5:dc0d07b092c3caedb9251657499d1227
639 Bytes Download
NEON_D17_TEAK_DP1_20170627_181333_reflectance.tif.ovr
md5:1a198f7cbc697d13f1e2e8fe320a436e
1.6 GB Download
NEON_D17_TEAK_DP1QA_20170627_181333_reflectance.zip
md5:3a189d128d457e7ddd8bb28b0d131140
5.9 GB Download
NEON_D17_TEAK_DP1QA_20170627_181333_RGB_Reflectance.tif
md5:dd08b5ce3491536cf865e9383d766dcd
135.6 MB Download
NEON_D17_TEAK_DP1QA_20170627_181333_RGB_Reflectance.tif.aux.xml
md5:c269a39838ce83441a09af51e3cf92bc
8.1 kB Download
NEON_D17_TEAK_DP1QA_20170627_181333_RGB_Reflectance.tif.ovr
md5:a0a8aa1b7dad91a5fb60f11ed6d72f0c
13.7 MB Download
NEON_D17_TEAK_DP1QA_20170627_181333_RGB_Reflectance.zip
md5:1b1bf2aa3e618d647491f3c77d0884ae
50.2 MB Download
440
7,775
views
downloads
All versions This version
Views 440324
Downloads 7,7751,882
Data volume 44.8 TB4.5 TB
Unique views 399300
Unique downloads 830274

Share

Cite as