Harvard Forest

Data Archive

HF450

Quantifying Forest Edge Area in the Northeastern USA 2016

Related Publications

Data

• hf450-01: forest edge proportions (preview)
• hf450-02: land cover (preview)
• hf450-03: script to calculate forest area and edge area from land cover products
• hf450-04: script to calculate forest area and edge area from land cover imagery
• hf450-05: script to combine outputs from edge area VHR data script
• hf450-06: script to calculate forest edge proportion from the US national forest inventory (FIA)
• hf450-07: script to process and combine outputs from analyses to create tables of forest edge area and proportion

Overview

• Lead: Luca Morreale, Jonathan Thompson, Lucy Hutyra
• Investigators: Valerie Pasquarella
• Contact: Information Manager
• Start date: 2014
• End date: 2016
• Status: completed
• Location: Northeastern States
• Latitude: +36.0 to +49.4 degrees
• Longitude: -96.6 to -66.9 degrees
• Elevation: 0 to 2000 meter
• Datum: WGS84
• Taxa:
• Release date: 2024
• Language: English
• EML file: knb-lter-hfr.450.1
• DOI: digital object identifier
• EDI: data package
• DataONE: data package
• Related links:
• USDA FIA DataMart
• Study type: long-term measurement
• Research topic: forest-atmosphere exchange; regional studies
• LTER core area: primary production, land use and land cover change, human-environment interactions
• Keywords: aboveground production, biomass, forest dynamics, inventories, land cover, landscape
• Abstract:

  Temperate forests are the most fragmented forest biome, yet current understanding of fragmentation effects on ecosystem processes, such as carbon cycling, is rooted in tropical forest research. In the associated manuscript, we review the effects of persistent fragmentation on temperate forest ecosystem processes and quantify the extent to which the US national forest inventory and land-cover maps represent forest edge area. We find a systematic underrepresentation of forest edges across all methods. Compared with very high resolution (1 m) maps, conventional 30 m resolution forest cover maps underestimate forest edge area by 16.4%, on average. Accounting for all forest edge area and edge effects on forest structure and growth results in a 14.8% median increase in aboveground forest carbon estimates with 23.8% and 74.2% increases in agriculturally and urban dominated counties, respectively. We conclude by proposing improvements to forest inventories, maps, and models to better represent the fragmented temperate forest landscape.

  We provide Google Earth Engine scripts (Gorelick et al. 2017; written in JavaScript) to calculate forest edge area and forest cover from commonly-used land cover maps, including the 2016 National Land Cover Database (NLCD), the 2016 Land Change Monitoring, Assessment, and Projection annual product (LCMAP), and the 2016 MODIS Land Cover IGBP annual product (Yang et al. 2018; Sulla-Menashe et al. 2019; Brown et al. 2020). We also include equivalent scripts to process a very-high resolution (1-m pixel size; VHR) land cover map of the Chesapeake Bay Watershed in 2014 (Pallai and Wesson 2017). We provide an R script to calculate forest edge proportion from the US national forest inventory (USDA FIA), following methods to identify inventory plots containing forest edge as described in Morreale et al. (2021) and using the R library rFIA to access the FIA data (Stanke et al. 2020).

  We also provide a data table containing the estimated forest area and forest edge proportion for each county within the separate from each of the land cover maps (NLCD, LCMAP, MODIS), the VHR land cover maps, and a separate data table with estimates of edge proportion from the FIA for each county within the study area. We include an R script that processes the intermediate outputs from Google Earth Engine and combines estimates of forest area and forest edge proportion from all sources into a single table.

• Methods:

  We assessed the capacity of commonly-used land cover maps and the US national forest inventory (NFI) to accurately estimate the extent of forest fragmentation by leveraging the increasing availability and computational feasibility of very high resolution (1 m pixel size; hereafter VHR) land-cover products to map forest edge area in selected regions. VHR maps have been shown to greatly increase the ability to characterize landscape heterogeneity, including urban canopy excluded from definitions of forest used in traditional land-cover maps (Wickham and Riitters 2019). For our baseline estimation of forest area we used VHR maps of the 260,000 km2 Chesapeake Bay Watershed in 2014 and the 27,000 km2 Commonwealth of Massachusetts (Fig S1) in 2016, which map tree canopy with up to 98% accuracy (Pallai and Wesson 2017). This established a baseline for quantifying bias in forest area and forest edges relative to the other maps. We performed robustness tests on our estimates of forest cover and forest edge area from the VHR data using a 100 m2 minimum area threshold. We report minimal differences at the county level.

  We applied methods described in Morreale et al. (2021) to estimate forest edge prevalence within the U.S. NFI and present our results for the NFI as the percent of stems categorized as forest edge. We used Google Earth Engine (Gorelick et al. 2017) to calculate forest cover and forest edge area across four satellite-based forest maps. In the US, counties represent a coherent political boundary within which land-use zoning is typically consistent, and they therefore can function as independent samples for the purposes of this analysis. We aggregate forest cover data from the 2016 National Land Cover Database (NLCD), the 2016 Land Change Monitoring, Assessment, and Projection annual product (LCMAP), and the 2016 MODIS Land Cover IGBP annual product to the county level (Yang et al. 2018; Sulla-Menashe et al. 2019; Brown et al. 2020). The spatial resolution of these datasets are 30 m, 30 m, and 500 m, respectively, and we condense the legend of available forest classes for comparability. This is a representative sample of products with differing definitions of and methods for identifying forests that are commonly used in forest area analyses and forest dynamics modeling efforts.

  Brown JF, Tollerud HJ, Barber CP, et al. 2020. Lessons learned implementing an operational continuous United States national land change monitoring capability: The Land Change Monitoring, Assessment, and Projection (LCMAP) approach. Remote Sensing of Environment 238: 111356.

  Gorelick N, Hancher M, Dixon M, et al. 2017. Google Earth Engine: Planetary-scale geospatial analysis for everyone. Remote Sensing of Environment 202: 18–27.

  Morreale LL, Thompson JR, Tang X, et al. 2021. Elevated growth and biomass along temperate forest edges. Nature Communications 12: 7181.

  Pallai C and Wesson K. 2017. Chesapeake Bay Program Partnership High-Resolution Land Cover Classification Accuracy Assessment Methodology.

  Stanke H, Finley AO, Weed AS, et al. 2020. rFIA: An R package for estimation of forest attributes with the US Forest Inventory and Analysis database. Environmental Modelling and Software 104664.

  Sulla-Menashe D, Gray JM, Abercrombie SP, and Friedl MA. 2019. Hierarchical mapping of annual global land cover 2001 to present: The MODIS Collection 6 Land Cover product. Remote Sensing of Environment 222: 183–94.

  Wickham J and Riitters KH. 2019. Influence of high-resolution data on the assessment of forest fragmentation. Landscape Ecology 34: 2169–82.

  Yang L, Jin S, Danielson P, et al. 2018. A new generation of the United States National Land Cover Database: Requirements, research priorities, design, and implementation strategies. ISPRS Journal of Photogrammetry and Remote Sensing 146: 108–23.

• Organization: Harvard Forest. 324 North Main Street, Petersham, MA 01366, USA. Phone (978) 724-3302. Fax (978) 724-3595.

• Project: The Harvard Forest Long-Term Ecological Research (LTER) program examines ecological dynamics in the New England region resulting from natural disturbances, environmental change, and human impacts. (ROR).

• Funding: National Science Foundation LTER grants: DEB-8811764, DEB-9411975, DEB-0080592, DEB-0620443, DEB-1237491, DEB-1832210. Other funding: United States Department of Agriculture National Institute of Food and Agriculture Award 2017-67003-26487; National Science Foundation Research Traineeship (NRT) grant to Boston University DGE 1735087; Fellowship from Rafiki B. Hariri Institute at Boston University

• Use: This dataset is released to the public under Creative Commons CC0 1.0 (No Rights Reserved). Please keep the dataset creators informed of any plans to use the dataset. Consultation with the original investigators is strongly encouraged. Publications and data products that make use of the dataset should include proper acknowledgement.

• License: Creative Commons Zero v1.0 Universal (CC0-1.0)

• Citation: Morreale L, Thompson J, Hutyra L. 2024. Quantifying Forest Edge Area in the Northeastern USA 2016. Harvard Forest Data Archive: HF450 (v.1). Environmental Data Initiative: https://doi.org/10.6073/pasta/1dacd88e847defa27ec978317b4e33a7.

Detailed Metadata

hf450-01: forest edge proportions

1. GEOID: unique county identifier
2. EdgeProp: proportion of forest classified as forest edge in the county (unit: dimensionless / missing value: NA)
3. Source: data source for forest edge and area estimates

hf450-02: land cover

1. Source: data source for forest edge and area estimates
2. State: unique state identifier (FIPS code)
3. CountyFP: county identifier within the state (FIPS code)
4. GEOID: unique county identifier
5. CountyN: character name of county
6. Land_Area: area of land in the county in m2 (unit: meterSquared / missing value: NA)
7. LULC: land-use/land cover designation
8. Pixel_Count: number of pixels of the designated land-cover in the county (unit: dimensionless / missing value: NA)
9. Area_km2: area in km2 of the designated land cover in the county (unit: kilometerSquared / missing value: NA)
10. Land_Area_km2: area of land in the county in km2 (unit: kilometerSquared / missing value: NA)

hf450-03: script to calculate forest area and edge area from land cover products

• Compression: none
• Format: JavaScript
• Type: script

hf450-04: script to calculate forest area and edge area from land cover imagery

• Compression: none
• Format: JavaScript
• Type: script

hf450-05: script to combine outputs from edge area VHR data script

• Compression: none
• Format: JavaScript
• Type: script

hf450-06: script to calculate forest edge proportion from the US national forest inventory (FIA)

• Compression: none
• Format: R script
• Type: script

hf450-07: script to process and combine outputs from analyses to create tables of forest edge area and proportion

• Compression: none
• Format: R script
• Type: script