Global maps of twenty-first century forest carbon fluxes

Nature Climate Change, Journal Year: 2021, Volume and Issue: 11(3), P. 234 - 240

Published: Jan. 21, 2021

Language: Английский

---

Amazonia as a carbon source linked to deforestation and climate change

Nature, Journal Year: 2021, Volume and Issue: 595(7867), P. 388 - 393

Published: July 14, 2021

Language: Английский

---

Vegetation fires in the Anthropocene

Nature Reviews Earth & Environment, Journal Year: 2020, Volume and Issue: 1(10), P. 500 - 515

Published: Aug. 18, 2020

Language: Английский

---

Global Carbon Budget 2023

Earth system science data, Journal Year: 2023, Volume and Issue: 15(12), P. 5301 - 5369

Published: Nov. 30, 2023

Abstract. Accurate assessment of anthropogenic carbon dioxide (CO2) emissions and their redistribution among the atmosphere, ocean, terrestrial biosphere in a changing climate is critical to better understand global cycle, support development policies, project future change. Here we describe synthesize data sets methodology quantify five major components budget uncertainties. Fossil CO2 (EFOS) are based on energy statistics cement production data, while from land-use change (ELUC), mainly deforestation, bookkeeping models. Atmospheric concentration measured directly, its growth rate (GATM) computed annual changes concentration. The ocean sink (SOCEAN) estimated with biogeochemistry models observation-based fCO2 products. (SLAND) dynamic vegetation Additional lines evidence land sinks provided by atmospheric inversions, oxygen measurements, Earth system resulting imbalance (BIM), difference between total biosphere, measure imperfect incomplete understanding contemporary cycle. All uncertainties reported as ±1σ. For year 2022, EFOS increased 0.9 % relative 2021, fossil at 9.9±0.5 Gt C yr−1 (10.2±0.5 when carbonation not included), ELUC was 1.2±0.7 yr−1, for emission (including sink) 11.1±0.8 (40.7±3.2 yr−1). Also, GATM 4.6±0.2 (2.18±0.1 ppm yr−1; denotes parts per million), SOCEAN 2.8±0.4 SLAND 3.8±0.8 BIM −0.1 (i.e. sources marginally too low or high). averaged over 2022 reached 417.1±0.1 ppm. Preliminary 2023 suggest an increase +1.1 (0.0 2.1 %) globally reaching 419.3 ppm, 51 above pre-industrial level (around 278 1750). Overall, mean trend consistently period 1959–2022, near-zero overall imbalance, although discrepancies up around 1 persist representation semi-decadal variability fluxes. Comparison estimates multiple approaches observations shows following: (1) persistent large uncertainty estimate emissions, (2) agreement different methods magnitude flux northern extra-tropics, (3) discrepancy strength last decade. This living-data update documents applied this most recent well evolving community presented work available https://doi.org/10.18160/GCP-2023 (Friedlingstein et al., 2023).

Language: Английский

---

Global and Regional Trends and Drivers of Fire Under Climate Change

Reviews of Geophysics, Journal Year: 2022, Volume and Issue: 60(3)

Published: April 11, 2022

Abstract Recent wildfire outbreaks around the world have prompted concern that climate change is increasing fire incidence, threatening human livelihood and biodiversity, perpetuating change. Here, we review current understanding of impacts on weather (weather conditions conducive to ignition spread wildfires) consequences for regional activity as mediated by a range other bioclimatic factors (including vegetation biogeography, productivity lightning) ignition, suppression, land use). Through supplemental analyses, present stocktake trends in burned area (BA) during recent decades, examine how relates its drivers. Fire controls annual timing fires most regions also drives inter‐annual variability BA Mediterranean, Pacific US high latitude forests. Increases frequency extremity been globally pervasive due 1979–2019, meaning landscapes are primed burn more frequently. Correspondingly, increases ∼50% or higher seen some extratropical forest ecoregions including high‐latitude forests 2001–2019, though interannual remains large these regions. Nonetheless, can override relationship between weather. For example, savannahs strongly patterns fuel production fragmentation naturally fire‐prone agriculture. Similarly, tropical relate deforestation rates degradation than changing Overall, has reduced 27% past two part decline African savannahs. According models, prevalence already emerged beyond pre‐industrial Mediterranean change, emergence will become increasingly widespread at additional levels warming. Moreover, several major wildfires experienced years, Australian bushfires 2019/2020, occurred amidst were considerably likely Current models incompletely reproduce observed spatial based their existing representations relationships controls, historical vary across models. Advances observation controlling supporting addition optimization processes exerting upwards pressure intensity weather, this escalate with each increment global Improvements better interactions climate, extremes, humans required predict future mitigate against consequences.

Language: Английский

---

A review of machine learning applications in wildfire science and management

Environmental Reviews, Journal Year: 2020, Volume and Issue: 28(4), P. 478 - 505

Published: July 28, 2020

Artificial intelligence has been applied in wildfire science and management since the 1990s, with early applications including neural networks expert systems. Since then, field rapidly progressed congruently wide adoption of machine learning (ML) methods environmental sciences. Here, we present a scoping review ML management. Our overall objective is to improve awareness among researchers managers, as well illustrate diverse challenging range problems available data scientists. To that end, first an overview popular approaches used date then use broadly categorized into six problem domains, (i) fuels characterization, fire detection, mapping; (ii) weather climate change; (iii) occurrence, susceptibility, risk; (iv) behavior prediction; (v) effects; (vi) Furthermore, discuss advantages limitations various relating size, computational requirements, generalizability, interpretability, identify opportunities for future advances wildfires within context. In total, end 2019, identified 300 relevant publications which most frequently across domains included random forests, MaxEnt, artificial networks, decision trees, support vector machines, genetic algorithms. As such, there exists apply more current — deep agent-based sciences, especially instances involving very large multivariate datasets. We must recognize, however, despite ability models learn on their own, expertise necessary ensure realistic modelling processes multiple scales, while complexity some such requires dedicated sophisticated knowledge application. Finally, stress research communities play active role providing relevant, high-quality, freely by practitioners methods.

Language: Английский

---

Unprecedented burn area of Australian mega forest fires

Nature Climate Change, Journal Year: 2020, Volume and Issue: 10(3), P. 171 - 172

Published: Feb. 24, 2020

Language: Английский

---

Historical background and current developments for mapping burned area from satellite Earth observation

Remote Sensing of Environment, Journal Year: 2019, Volume and Issue: 225, P. 45 - 64

Published: March 5, 2019

Fire has a diverse range of impacts on Earth's physical and social systems. Accurate up to date information areas affected by fire is critical better understand drivers activity, as well its relevance for biogeochemical cycles, climate, air quality, aid management. Mapping burned was traditionally done from field sketches. With the launch first Earth observation satellites, remote sensing quickly became more practical alternative detect areas, they provide timely regional global coverage occurrence. This review paper explores basis area satellite observations, describes historical trends using sensors monitor summarizes most recent approaches map evaluates existing products (both at scales). Finally, it identifies potential future opportunities further improve detection satellites.

Language: Английский

---

Development of a Sentinel-2 burned area algorithm: Generation of a small fire database for sub-Saharan Africa

Remote Sensing of Environment, Journal Year: 2018, Volume and Issue: 222, P. 1 - 17

Published: Dec. 20, 2018

A locally-adapted multitemporal two-phase burned area (BA) algorithm has been developed using as inputs Sentinel-2 MSI reflectance measurements in the short and near infrared wavebands plus active fires detected by Terra Aqua MODIS sensors. An initial map is created first step, from which tile dependent statistics are extracted for second step. The whole Sub-Saharan Africa (around 25 M km2) was processed with this at a spatial resolution of 20 m, January to December 2016. This period covers two half fire seasons on Northern Hemisphere an entire season South. selected existing BA products account it include around 70% global BA. Validation product based two-stage stratified random sampling Landsat images. Higher accuracy values than were observed, Dice coefficient 77% omission commission errors 26.5% 19.3% respectively. standard NASA (MCD64A1 c6) showed similar error (20.4%), but much higher (59.6%), lower (53.6%). over >11,000 images create database that would also small (<100 ha). time continental generated medium sensors (spatial = m), showing their operational potential improving our current understanding impacts. Total estimated 4.9 km2, 80% larger what same (2.7 km2). main differences between found regions where ha) significant proportion total BA, coarse pixel sizes (500 m MCD64A1) unlikely detect them. On negative side, have temporal consequently more affected cloud/cloud shadows less reporting products. derived S2 imagery greatly contribute better impacts regimes, particularly tropical regions, such frequent. named FireCCISFD11 publicly available at: https://www.esa-fire-cci.org/node/262, last accessed November 2018.

Language: Английский

---

Summer warming explains widespread but not uniform greening in the Arctic tundra biome

Nature Communications, Journal Year: 2020, Volume and Issue: 11(1)

Published: Sept. 22, 2020

Arctic warming can influence tundra ecosystem function with consequences for climate feedbacks, wildlife and human communities. Yet ecological change across the biome remains poorly quantified due to field measurement limitations reliance on coarse-resolution satellite data. Here, we assess decadal changes in greenness using time series from 30 m resolution Landsat satellites. From 1985 2016 increased (greening) at ~37.3% of sampling sites decreased (browning) ~4.7% sites. Greening occurred most often warm summer air temperature, soil moisture, while browning cold that cooled dried. Tundra was positively correlated graminoid, shrub, productivity measured Our results support hypothesis stimulated plant much, but not all, during recent decades.

Language: Английский

---