Differing Sensitivities to Fire Disturbance Result in Large Differences Among Remotely Sensed Products of Vegetation Disturbance
Recent advances in high-performance computing (HPC) have promoted the creation of standardized remotely sensed products that map annual vegetation disturbance through two primary methods: (1) conventional approaches that integrate remote sensing-derived vegetation indices with field data and other data on disturbance events reported by public agencies on a year-to-year basis, and (2) “big” data approaches using HPC to automate algorithms and workflows across an entire time series. Given the recent proliferation of these annual products and their potential utility for understanding vegetation dynamics, it is important for product end users (that is, practitioners and researchers in domains other than remote sensing) to understand the differences in their representations of disturbance and the conditions under which they report it. We use fire in California as a case study to compare reported disturbance across three widely used vegetation disturbance products—LANDFIRE (representing the conventional approach), Hansen Global Forest Change (GFC), and North America Forest Dynamics (NAFD), the latter two created from automated approaches. Using Google’s Earth Engine, we compared their total and annual amounts of fire and non-fire disturbance for 2001–2010 and examined the products’ reported disturbance across different environmental and burn conditions. We found that GFC and NAFD reported similar amounts of disturbance that were consistently much lower than LANDFIRE’s reported disturbance across all years, regions, and habitats. We also found that despite the differences in amounts of reported disturbance, the products identified disturbance in similar ranges of bioclimatic conditions and habitat types, and thus, differing environmental conditions in areas reported as disturbed were not the drivers of the difference. Rather, we found that lower sensitivity to fire disturbance for GFC and NAFD, as compared to LANDFIRE, was a key driver of the overall differences in the amounts and locations of reported disturbance; both GFC and NAFD reported much lower amounts of fire disturbance than LANDFIRE across all burn conditions. Furthermore, the difference in reported disturbance between LANDFIRE and GFC/NAFD was greater for fire disturbance than for non-fire disturbance; LANDFIRE reported more than double the total amounts of fire disturbance of GFC and NAFD in the study period. Based on our results, we encourage end users to choose the appropriate disturbance product based not only on spatial extent and habitat but also on the disturbance type of interest (that is, fire and non-fire). Overall, rather than focusing on accuracy, our study quantifies the extent to which the products exhibited differences in the amounts and locations of reported disturbance to provide insight into these products’ representations of disturbance and help end users evaluate and choose the most appropriate product for their needs.
Keywordsvegetation disturbance fire Hansen Global Forest Change LANDFIRE North American Forest Dynamics (NAFD)
The authors would like to thank the Geospatial Innovation Facility (GIF), Matthew Potts and Iryna Dronova at the University of California, Berkeley.
- Anderegg WRL, Martinez-Vilalta J, Cailleret M, Camarero JJ, Ewers BE, Galbraith D, Gessler A, Grote R, Huang C-Y, Levick SR, Powell TL, Rowland L, Sánchez-Salguero R, Trotsiuk V. 2016. When a tree dies in the forest: scaling climate-driven tree mortality to ecosystem water and carbon fluxes. Ecosystems 19:1133–47.CrossRefGoogle Scholar
- Arneth A, Sitch S, Pongratz J, Stocker BD, Ciais P, Poulter B, Bayer AD, Bondeau A, Calle L, Chini LP, Gasser T, Fader M, Friedlingstein P, Kato E, Li W, Lindeskog M, Nabel JEMS, Pugh TAM, Robertson E, Viovy N, Yue C, Zaehle S. 2017. Historical carbon dioxide emissions caused by land-use changes are possibly larger than assumed. Nat Geosci 10:79.CrossRefGoogle Scholar
- Cohen WB, Healey SP, Yang Z, Stehman SV, Brewer CK, Brooks EB, Gorelick N, Huang C, Hughes MJ, Kennedy RE, Loveland TR, Moisen GG, Schroeder TA, Vogelmann JE, Woodcock CE, Yang L, Zhu Z. 2017. How similar are forest disturbance maps derived from different Landsat time series algorithms? For Trees Livelihoods 8:98.Google Scholar
- FRAP. 2018. FRAP fire perimeters product metadata. http://frap.fire.ca.gov/data/frapgisdata-sw-fireperimeters_download. Last accessed 20 Nov 2018.
- Gorelick N, Hancher M, Dixon M, Ilyushchenko S, Thau D, Moore R. 2017. Google earth engine: planetary-scale geospatial analysis for everyone. Remote Sens Environ. http://www.sciencedirect.com/science/article/pii/S0034425717302900.
- Goward SN, Huang C, Zhao F, Schleeweis K, Rishmawi K, Lindsey M, Dungan JL, Michaelis A. 2016. NACP NAFD project: forest disturbance history from Landsat, 1986–2010. ORNL DAAC, Oak Ridge, Tennessee, USA. http://daac.ornl.gov/cgi-bin/dsviewer.pl?ds_id=1290. Last accessed 11 Feb 2017.
- Goward SN, Masek JG, Cohen W, Moisen G, Collatz GJ, Healey S, Houghton RA, Huang C, Kennedy R, Law B, Powell S. 2008. Forest disturbance and North American carbon flux. Eos Trans Am Geophys Union 89(11):105–6.Google Scholar
- Healey SP, Cohen WB, Yang Z, Kenneth Brewer C, Brooks EB, Gorelick N, Hernandez AJ, Huang C, Joseph Hughes M, Kennedy RE, Loveland TR, Moisen GG, Schroeder TA, Stehman SV, Vogelmann JE, Woodcock CE, Yang L, Zhu Z. 2017. Mapping forest change using stacked generalization: An ensemble approach. Remote Sens Environ. http://www.sciencedirect.com/science/article/pii/S0034425717304418.
- Hyde J, Strand EK, Hudak AT, Hamilton D. 2015. A case study comparison of landfire fuel loading and emissions generation on a mixed conifer forest in northern Idaho, USA. Fire Ecol 11. https://www.fs.fed.us/rm/pubs_journals/2015/rmrs_2015_hyde_j001.pdf.
- Kalluri S, Gundy J, Haman B, Paullin A, Van Rompay P, Vititoe D, Weiner A. 2015. A High performance remote sensing product generation system based on a service oriented architecture for the next generation of geostationary operational environmental satellites. Remote Sens 7:10385–99.CrossRefGoogle Scholar
- Kennedy RE, Yang Z, Cohen WB. 2010. Detecting trends in forest disturbance and recovery using yearly Landsat time series: 1. LandTrendr—temporal segmentation algorithms. Remote Sens Environ. http://www.sciencedirect.com/science/article/pii/S0034425710002245.
- Kumar U, Ganguly S, Nemani RR, Raja KS, Milesi C, Sinha R, Michaelis A, Votava P, Hashimoto H, Li S, Wang W, Kalia S, Gayaka S. 2017. Exploring subpixel learning algorithms for estimating global land cover fractions from satellite data using high performance computing. Remote Sens 9:1105.CrossRefGoogle Scholar
- LANDFIRE disturbance product metadata. https://landfire.cr.usgs.gov/distmeta/servlet/gov.usgs.edc.MetaBuilder?TYPE=HTML&DATASET=FA9. Last accessed 6 Jan 2017.
- LANDFIRE. 2018a. About web page. https://www.landfire.gov/about.php. Last accessed on 20 Nov 2018.
- LANDFIRE. 2018b. Guidelines for evaluating LANDFIRE fuel data. https://www.landfire.gov/documents/EvaluatingLANDFIREFuelsData.pdf. Last accessed on 20 Nov 2018.
- McKerrow A, Dewitz J, Long DG, Nelson K, Connot JA. 2016. A comparison of NLCD 2011 and LANDFIRE EVT 2010: regional and national summaries. https://pubs.er.usgs.gov/publication/70177839.
- Moritz MA, Stephens SL. 2008. Fire and sustainability: considerations for California’s altered future climate. Clim Change. http://www.springerlink.com/index/5411702235mx5432.pdf.
- USGS. 2016. Earth Resources Observation and Science (EROS) User Services. Email correspondence on October 13, 2016.Google Scholar
- Zhao F, Huang C, Goward SN, Schleeweis K, Rishmawi K, Lindsey MA, Denning E, Keddell L, Cohen WB, Yang Z, Dungan JL, Michaelis A. 2018. Development of Landsat-based annual US forest disturbance history maps (1986–2010) in support of the North American Carbon Program (NACP). Remote Sens Environ 209:312–26.CrossRefGoogle Scholar