Abstract
Fire use for land management is widespread in natural tropical and plantation forests, causing major environmental and economic damage. Recent studies combining active fire alerts with annual forest-cover loss information identified fire-related forest-cover loss areas well, but do not provide detailed understanding on how fires and forest-cover loss are temporally related. Here, we combine Sentinel-1-based, near real-time forest cover information with Visible Infrared Imaging Radiometer Suite (VIIRS) active fire alerts, and for the first time, characterize the temporal relationship between fires and tropical forest-cover loss at high temporal detail and medium spatial scale. We quantify fire-related forest-cover loss and separate fires that predate, coincide with, and postdate forest-cover loss. For the Province of Riau, Indonesia, dense Sentinel-1 C-band Synthetic Aperture Radar data with guaranteed observations of at least every 12 days allowed for confident and timely forest-cover-loss detection in natural and plantation forest with user’s and producer’s accuracy above 95%. Forest-cover loss was detected and confirmed within 22 days in natural forest and within 15 days in plantation forest. This difference can primarily be related to different change processes and dynamics in natural and plantation forest. For the period between 1 January 2016 and 30 June 2017, fire-related forest-cover loss accounted for about one third of the natural forest-cover loss, while in plantation forest, less than ten percent of the forest-cover loss was fire-related. We found clear spatial patterns of fires predating, coinciding with, or postdating forest-cover loss. Only the minority of fires in natural and plantation forest temporally coincided with forest-cover loss (13% and 16%) and can thus be confidently attributed as direct cause of forest-cover loss. The majority of the fires predated (64% and 58%) or postdated forest-cover loss (23% and 26%), and should be attributed to other key land management practices. Detailed and timely information on how fires and forest cover loss are temporally related can support tropical forest management, policy development, and law enforcement to reduce unsustainable and illegal fire use in the tropics.
Highlights
Indonesia’s forest-cover-loss rates are among the highest globally [1,2] and are driven mainly by expansion of and conversion to industrial forest plantations [3]
Results for the Province of Riau, Indonesia show that dense Sentinel-1 C-band Synthetic Aperture Radar (SAR) time series provide confident, timely, and gap-free forest-cover-loss detection in both natural and plantation forest
Having such dense and regular information on forest-cover loss available enabled an assessment of the temporal relation between active fires and tropical forest-cover loss at high temporal detail
Summary
Indonesia’s forest-cover-loss rates are among the highest globally [1,2] and are driven mainly by expansion of and conversion to industrial forest plantations [3]. Consistent information on the temporal relationship between fires and forest-cover loss are currently missing, but can help to better understand fire-related management practices in tropical natural and plantation forest [5,12,15,16]. In our recent study [33], we used Sentinel-1 data for near real-time forest cover loss detection at a dry tropical forest site in Bolivia dominated by large-scale industrial logging. We study the relationship between Sentinel-1-based, near real-time forest-cover loss information and daily VIIRS active fires alerts to characterize fire-related forest-cover loss in the Province of Riau, Indonesia. We first evaluate how timely and accurate forest-cover loss in natural and plantation forest can be detected with confidence using dense Sentinel-1 C-band SAR data. In 2015, remaining nfoarttuhirsasltufdoyr.eIsnt2c0o15v,erreemdain~i1n.g6n7amturialllifornesht aco(v1e8re.d5%~1.o67f mRiilaliuon’shala(n18d.5%aroefaR),iawu’hs lialendpalraena)t,ation forest covered ~3.w58hilme pilllainotnatihona f(o4r1e.s8t %co)v.erAedd~e3t.5a8ilmedillidoenshcari(4p1t.i8o%n).oAndtehtaeilmededtehsocrdiputiosnedontothge emneethroadteutshede forest land cover maptIaonndgodenneaesricaatl’seafstohsreedsfteodsreecfsritnipiltaitnoidnon[c3o5cv]aearnndmbcaeopnfosainuddenradarcielnaasss[w3di5teh]s.clerWispsettihofnaonlcl0ao.n2w5bheIanfaodnudonndleesissnitah[3a’5sn].f3o0W%reecsfatonldolopewyfinition [35] and considceorvaerreasasnown‐iftohrelset.ss than 0.25 ha and less than 30% canopy cover as non-forest
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
