Foliar trait model application overview

Detailing code and datasets used in mapping foliar traits from 2018 NEON AOP survey over Crested Butte, CO.

This project contains a repository that handles the management and merging of NEON AOP remote sensing data with field/lab-based trait data to develop PLSR models, as used for trait mapping that was conducted in the East River, CO from data collected in June, 2018. The code leverages a plsr ensembling routine (external to this repository) to create trait models and assess the model performance. This code was created as part of an effort to generate foliar trait maps throughout the Upper East River watersheds, home to the Rocky Mountian Biological Laboratory and DOE's Watershed Function Scientific Focus Area site in Crested Butte, CO in association with NEON's Assignable Asset program. Here we document this repository, associated code repositories required for AOP data processing, and sources for both AOP and field based data used in this study.

A full description of this study can be found in the following publication, and use of this code should cite this manuscript:

K. Dana Chadwick, Philip Brodrick, Kathleen Grant, Tristan Goulden, Amanda Henderson, Nicola Falco, Haruko Wainwright, Kenneth H. Williams, Markus Bill, Ian Breckheimer, Eoin L. Brodie, Heidi Steltzer, C. F. Rick Williams, Benjamin Blonder, Jiancong Chen, Baptiste Dafflon, Joan Damerow, Matt Hancher, Aizah Khurram, Jack Lamb, Corey Lawrence, Maeve McCormick. John Musinsky, Samuel Pierce, Alexander Polussa, Maceo Hastings Porro, Andea Scott, Hans Wu Singh, Patrick O. Sorensen, Charuleka Varadharajan, Bizuayehu Whitney, Katharine Maher. Integrating airborne remote sensing and field campaigns for ecology and Earth system science. Methods in Ecology and Evolution, 2020. doi:10.1111/2041‐210X.13463

The code in this repository was used merge, subset, and organize data for application of PLSR ensemble code, as well as plot and assess model quality. It also includes code to apply PLSR equations to reflectance datasets.
DOI

Trait Map Availability

All trait maps that were generated as part of this project are available as assets in Earth Engine. There is also code for visualizing, mosaicing, and downloading subsets of these data avaliable here. In addition, these data are available for download and are archived on ESS-DIVE with additional documentation. When using these data, please cite them:


Chadwick, K. D., Brodrick, P. G., Grant, K., Henderson, A. N., Bill, M., Breckheimer, I., … Maher, K. (2020). NEON AOP foliar trait maps, maps of model uncertainty estimates, and conifer map. A Multiscale Approach to Modeling Carbon and Nitrogen Cycling within a High Elevation Watershed. doi:10.15485/1618133

Please visit our documentation below for links to airborne and field datasets and additional access information. All data and codes assocaited with this project are publicaly availiable. If you need assistance in accessing these data, please get in touch.

Associated Repositories

We discuss the application of these repositories in the associated manuscript, but list them here for easy reference. These repositories are maintained by Phil Brodrick, Research Technologist at the Jet Propulsion Laboratory

Atmospheric Correction Wrapper: Contains code to run ACORN on kernels of a flightline to utilize localized observational input information as well as to estimate visibility.


Shade Ray Tracing: Contains code to perform a ray trace between the ground, sun, and sensor in order to determine if there was any intercept to induce shading. Can utilize either a digital surface model or a combination of a digital elevation model and a tree canopy height map.


Conifer Model: Code to train and apply a deep learning conifer model from VSWIR data.


PLSR Ensembling: Code utilized for PLSR ensembling, previously published (Martin et al., 2018)


Associated Data Packages

Remote Sensing Data

There are several different remote sensing data packages that were used in this work. They are described in greater detail in the associated manuscript, but they are listed and linked to here for easy reference:




Radiance & Observational Data DOI
Documentation

LiDAR Point Cloud & Rasters DOI
Digital Terrain Model GEE Asset
Digital Surface Model GEE Asset
Tree Canopy Height GEE Asset

Custom Processed Reflectance Data, Shade Masks, & Canopy Water Content DOI
Reflectance Data GEE Asset
Shade Masks: DSM GEE Asset, DTM + TCH GEE Asset
Canopy Water Content GEE Asset
Atmospheric Water Vapor GEE Asset
Observational Data GEE Asset

Field Data

There are several different field data packages used in this work. They are described in greater detail in the associated manuscript, but they are listed and linked to here for easy reference:


Site Description and Species Cover
Foliar Carbon, Nitrogen, and d13C
Leaf Mass per Area and Water Content