Data from: The distribution of tree biomass carbon within the pacific coastal temperate rainforest, a disproportionally carbon dense forest
Data files
Apr 26, 2024 version files 26.30 GB
-
Carbon_Code_TerraUpdate.R
114.17 KB
-
logB.tif
63.95 MB
-
logTS.tif
75.52 MB
-
PerhumidBounds.cpg
5 B
-
PerhumidBounds.dbf
1.38 KB
-
PerhumidBounds.prj
145 B
-
PerhumidBounds.sbn
132 B
-
PerhumidBounds.sbx
116 B
-
PerhumidBounds.shp
6.81 MB
-
PerhumidBounds.shx
108 B
-
PerhumidBounds.xml
7.48 KB
-
qrf_PI_truncated.tif
4.21 GB
-
qrf_PI_unconstrained.tif
4.21 GB
-
qrf_preds_truncated_noVR.tif
4.21 GB
-
qrf_preds_truncated.tif
4.21 GB
-
qrf_preds_unconstrained_noVR.tif
4.21 GB
-
qrf_preds_unconstrained.tif
4.21 GB
-
README.md
4.07 KB
-
slide.tif
811.87 MB
-
wind.tif
58.32 MB
Abstract
Spatially explicit global estimates of forest carbon storage are typically coarsely scaled. While useful, these estimates do not account for the variability and distribution of carbon at management scales. We asked how climate, topography, and disturbance regimes interact across and within geopolitical boundaries to influence tree biomass carbon, using the perhumid region of the Pacific Coastal Temperate Rainforest, an infrequently disturbed carbon dense landscape, as a test case. We leveraged permanent sample plots in southeast Alaska and coastal British Columbia and used multiple quantile regression forests and generalized linear models to estimate tree biomass carbon stocks and the effects of topography, climate, and disturbance regimes. We estimate tree biomass carbon stocks are either 211 (SD = 163) Mg C ha-1 or 218 (SD = 169) Mg C ha-1. Natural disturbance regimes had no correlation with tree biomass but logging decreased tree biomass carbon and the effect diminished with increasing time since logging. Despite accounting for 0.3% of global forest area, this forest stores between 0.63% - 1.07% of global aboveground forest carbon as aboveground live tree biomass. The disparate impact of logging and natural disturbance regimes on tree biomass carbon suggests a mismatch between current forest management and disturbance history.
README: Data from: The distribution of tree biomass carbon within the pacific coastal temperate rainforest, a disproportionally carbon dense forest
https://doi.org/10.5061/dryad.3tx95x6nn
Description of the data and file structure
Spatial layers are presented in Alaska Albers Equal Area Conic (AEA) projection with a 30 x 30 m cell size. NOTE: the quantileRegresionForest::predict() function in R will only work when all spatial layers are provided in lat/long. Original layers found online such as climate, and topography are provided as links below.
The following data are provided:
logB.tif - a binary layer of logging presence/absence
logTS.tif - the time since logging (number of years from 2022)
wind.tif - the disturbance exposure map of wind throw events. Values represent the relative probability of being disturbed.
slide.tif - the disturbance exposure map of landslide events. Values represent the relative probability of being disturbed.
PerhumidBounds.shp - study area extent that encompasses the Tongass National Forest and Great Bear Rainforest (provided in lat/long)
qrf_preds_X.tif - predictions from the quantile regression forest where X indicates whether the data are the estimate from the truncated or unconstrained estimate. Two data layers contain '_noVR' at the end of the file name and are the data layers presented in the supplemental materials. All predictions are clipped to only cover areas with tree cover as determined by the % tree cover layer below.
qrf_PI_X.tif - spatially explicit 80% prediction intervals around the median estimate from the quantile regression forest where X indicates whether the data are the estimate from the truncated or unconstrained estimate. All predictions are clipped to only cover areas with tree cover as determined by the % tree cover layer below.
The predictions of aboveground live tree biomass and associated prediction intervals were generated using the Carbon_Code_TerraUpdate.R file provided. Carbon_Code_TerraUpdate.R uses input files from publicly available US Forest Service FIA data (link below) and CAN Forest Service FAIB data (link below). Data used from the FIA dataset includes (AK_TREE.csv and AK_PLOT.csv; note that when using publicly available FIA data the file name for AK_PLOT.csv is changed to select_Coastal_NBP_RG_PGMBC_PTMaxYear_TBL2SEND.csv). The code is able to accommodate analysis using either the unfuzzed or fuzzed FIA data. The data used from the FAIB dataset include a variety of .xlsx files named TSA__, where the __ denotes two numerical values.
Predictions and associated prediction interval data was derived in part from the logging, wind, landslide, elevation, slope, aspect, climate, forest cover, FIA, and FAIB data. When appropriate we have provided links to datasets not generated by the author team, provided below.
Sharing/Access information
This is a section for linking to other ways to access the data, and for linking to sources the data is derived from, if any.
Links to other publicly accessible locations of the data:
- Digital Elevation Map (used for elevation, slope, aspect): https://asterweb.jpl.nasa.gov/gdem.asp
- Climate (temperature and precipitation): https://www.worldclim.com/version2
- % Forest Cover: https://www.science.org/doi/10.1126/science.1244693
- Forest Inventory and Analysis Data (locations fuzzed): https://www.fia.fs.usda.gov/tools-data/
- Forest Analysis and Inventory Data: https://open.canada.ca/data/dataset/824e684b-4114-4a05-a490-aa56332b57f4
Code/Software
All analysis was done in an R computing environment (v4.1.2). However, for ease of data manipulation for large spatial layers. Some transformations and processing as well as data visualization was conducted in ArcGIS Pro (v2.6.0).