Matthew G. Hethcoat
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Matthew G. Hethcoat has a total of 45 co-authors. Here are the top co-authors by number of shared publications:
- Joao M. B. Carreiras
School of GeoSciences, University of Edinburgh, Edinburgh, UK
4 shared publications - Robert G. Bryant
Geography Department, University of Sheffield, Sheffield, S3 7ND, United Kingdom
4 shared publications - David P. Edwards
Department of Plant Sciences and Centre for Global Wood Security, University of Cambridge, Cambridge, UK
3 shared publications Anna D. Chalfoun
Department of Zoology & Physiology U.S. Geological Survey Wyoming Cooperative Fish and Wildlife Research Unit University of Wyoming 1000 E. University Ave. Laramie WY 82071 USA
2 shared publicationsDavid P. Edwards
Department of Animal and Plant Sciences University of Sheffield Sheffield UK
2 shared publications- Filipe Machado Franca
School of Biological Sciences University of Bristol Bristol UK
2 shared publications
Latest Publications
- Open Access
- 11 July 2024
- Published by Springer Nature in Scientific Data
- Vol. 11 (1) , 1-11
The Canadian Fire Spread Dataset
Satellite data are effective for mapping wildfires, particularly in remote locations where monitoring is rare. Geolocated fire detections can be used for enhanced fire management and fire modelling through daily fire progression mapping. Here we present the Canadian Fire Spread Dataset (CFSDS), encompassing interpolated progressions for fires >1,000 ha in Canada from 2002–2021, representing the day-of-burning and 50 environmental covariates for every pixel. Day-of-burning was calculated by ordinary kriging of active fire detections from the Moderate Resolution Imaging Spectroradiometer and the Visible Infrared Imaging Radiometer Suite, enabling a substantial improvement in coverage and resolution over existing datasets. Day of burning at each pixel was used to identify environmental conditions of burning such as daily weather, derived weather metrics, topography, and forest fuels characteristics. This dataset can be used in a broad range of research and management applications, such as retrospective analysis of fire spread, as a benchmark dataset for validating statistical or machine-learning models, and for forecasting the effects of climate change on fire activity.
- 11 July 2024
- Published by Springer Nature in Scientific Data
- Vol. 11 (1) , 1-11
Open AccessThe Canadian Fire Spread Dataset
Satellite data are effective for mapping wildfires, particularly in remote locations where monitoring is rare. Geolocated fire detections can be used for enhanced fire management and fire modelling through daily fire progression mapping. Here we present the Canadian Fire Spread Dataset (CFSDS), encompassing interpolated progressions for fires >1,000 ha in Canada from 2002–2021, representing the day-of-burning and 50 environmental covariates for every pixel. Day-of-burning was calculated by ordinary kriging of active fire detections from the Moderate Resolution Imaging Spectroradiometer and the Visible Infrared Imaging Radiometer Suite, enabling a substantial improvement in coverage and resolution over existing datasets. Day of burning at each pixel was used to identify environmental conditions of burning such as daily weather, derived weather metrics, topography, and forest fuels characteristics. This dataset can be used in a broad range of research and management applications, such as retrospective analysis of fire spread, as a benchmark dataset for validating statistical or machine-learning models, and for forecasting the effects of climate change on fire activity.
- Open Access
- 6 January 2024
- Published by MDPI AG in Remote Sensing
- Vol. 16 (2) , 230
Unrecorded Tundra Fires in Canada, 1986–2022
Climate-driven changes in fire regimes are expected across the pan-Arctic region. Trends in arctic fires are thought to be generally increasing; however, fire mapping across the region is far from comprehensive or systematic. We developed a new detection workflow and built a dataset of unrecorded tundra fires in Canada using Landsat data. We built a reference dataset of spectral indices from previously mapped fires in northern Canada to train a Random Forest model for detecting new fires between 1986 and 2022. In addition, we used time series information for each pixel to reduce false positives and narrow the large search space down to a finite set of regions that had experienced changes. We found 209 previously undetected fires in the Arctic and sub-Arctic regions, increasing the mapped burned area by approximately 30%. The median fire size was small, with roughly 3/4 of the fires being <100 ha in size. The majority of newly detected fires (69%) did not have satellite-derived hotspots associated with them. The dataset presented here is commission error-free and can be viewed as a reference dataset for future analyses. Moreover, future improvements and updates will leverage these data to improve the detection workflow outlined here, particularly for small and low-severity fires. These data can facilitate broader analyses that examine trends and environmental drivers of fire across the Arctic region. Such analyses could begin to untangle the mechanisms driving heterogeneous fire responses to climate observed across regions of the Circumpolar North.
- 6 January 2024
- Published by MDPI AG in Remote Sensing
- Vol. 16 (2) , 230
Open AccessUnrecorded Tundra Fires in Canada, 1986–2022
Climate-driven changes in fire regimes are expected across the pan-Arctic region. Trends in arctic fires are thought to be generally increasing; however, fire mapping across the region is far from comprehensive or systematic. We developed a new detection workflow and built a dataset of unrecorded tundra fires in Canada using Landsat data. We built a reference dataset of spectral indices from previously mapped fires in northern Canada to train a Random Forest model for detecting new fires between 1986 and 2022. In addition, we used time series information for each pixel to reduce false positives and narrow the large search space down to a finite set of regions that had experienced changes. We found 209 previously undetected fires in the Arctic and sub-Arctic regions, increasing the mapped burned area by approximately 30%. The median fire size was small, with roughly 3/4 of the fires being <100 ha in size. The majority of newly detected fires (69%) did not have satellite-derived hotspots associated with them. The dataset presented here is commission error-free and can be viewed as a reference dataset for future analyses. Moreover, future improvements and updates will leverage these data to improve the detection workflow outlined here, particularly for small and low-severity fires. These data can facilitate broader analyses that examine trends and environmental drivers of fire across the Arctic region. Such analyses could begin to untangle the mechanisms driving heterogeneous fire responses to climate observed across regions of the Circumpolar North.
- Open Access
- 1 January 2022
- Published by MDPI AG in Remote Sensing
- Vol. 14 (1) , 179
Combining Sentinel-1 and Landsat 8 Does Not Improve Classification Accuracy of Tropical Selective Logging
Tropical forests play a key role in the global carbon and hydrological cycles, maintaining biological diversity, slowing climate change, and supporting the global economy and local livelihoods. Yet, rapidly growing populations are driving continued degradation of tropical forests to supply wood products. The United Nations (UN) has developed the Reducing Emissions from Deforestation and Forest Degradation (REDD+) programme to mitigate climate impacts and biodiversity losses through improved forest management. Consistent and reliable systems are still needed to monitor tropical forests at large scales, however, degradation has largely been left out of most REDD+ reporting given the lack of effective monitoring and countries mainly focus on deforestation. Recent advances in combining optical data and Synthetic Aperture Radar (SAR) data have shown promise for improved ability to monitor forest losses, but it remains unclear if similar improvements could be made in detecting and mapping forest degradation. We used detailed selective logging records from three lowland tropical forest regions in the Brazilian Amazon to test the effectiveness of combining Landsat 8 and Sentinel-1 for selective logging detection. We built Random Forest models to classify pixel-based differences in logged and unlogged regions to understand if combining optical and SAR improved the detection capabilities over optical data alone. We found that the classification accuracy of models utilizing optical data from Landsat 8 alone were slightly higher than models that combined Sentinel-1 and Landsat 8. In general, detection of selective logging was high with both optical only and optical-SAR combined models, but our results show that the optical data was dominating the predictive performance and adding SAR data introduced noise, lowering the detection of selective logging. While we have shown limited capabilities with C-band SAR, the anticipated opening of the ALOS-PALSAR archives and the anticipated launch of NISAR and BIOMASS in 2023 should stimulate research investigating similar methods to understand if longer wavelength SAR might improve classification of areas affected by selective logging when combined with optical data.
- 1 January 2022
- Published by MDPI AG in Remote Sensing
- Vol. 14 (1) , 179
Open AccessCombining Sentinel-1 and Landsat 8 Does Not Improve Classification Accuracy of Tropical Selective Logging
Tropical forests play a key role in the global carbon and hydrological cycles, maintaining biological diversity, slowing climate change, and supporting the global economy and local livelihoods. Yet, rapidly growing populations are driving continued degradation of tropical forests to supply wood products. The United Nations (UN) has developed the Reducing Emissions from Deforestation and Forest Degradation (REDD+) programme to mitigate climate impacts and biodiversity losses through improved forest management. Consistent and reliable systems are still needed to monitor tropical forests at large scales, however, degradation has largely been left out of most REDD+ reporting given the lack of effective monitoring and countries mainly focus on deforestation. Recent advances in combining optical data and Synthetic Aperture Radar (SAR) data have shown promise for improved ability to monitor forest losses, but it remains unclear if similar improvements could be made in detecting and mapping forest degradation. We used detailed selective logging records from three lowland tropical forest regions in the Brazilian Amazon to test the effectiveness of combining Landsat 8 and Sentinel-1 for selective logging detection. We built Random Forest models to classify pixel-based differences in logged and unlogged regions to understand if combining optical and SAR improved the detection capabilities over optical data alone. We found that the classification accuracy of models utilizing optical data from Landsat 8 alone were slightly higher than models that combined Sentinel-1 and Landsat 8. In general, detection of selective logging was high with both optical only and optical-SAR combined models, but our results show that the optical data was dominating the predictive performance and adding SAR data introduced noise, lowering the detection of selective logging. While we have shown limited capabilities with C-band SAR, the anticipated opening of the ALOS-PALSAR archives and the anticipated launch of NISAR and BIOMASS in 2023 should stimulate research investigating similar methods to understand if longer wavelength SAR might improve classification of areas affected by selective logging when combined with optical data.
- Open Access
- 25 August 2020
- Published by IOP Publishing in Environmental Research Letters
- Vol. 15 (9) , 094057
Mapping pervasive selective logging in the south-west Brazilian Amazon 2000–2019
Tropical forests harbour the highest biodiversity on the planet and are essential to human livelihoods and the global economy. However continued loss and degradation of forested landscapes, coupled with a rapidly rising global population, is placing incredible pressure on forests globally. The United Nations has developed the Reducing Emissions from Deforestation and forest Degradation (REDD +) programme in response to the challenges facing tropical forests and in recognition of the role they can play in climate mitigation. REDD + requires consistent and reliable monitoring of forests, however, national-level methodologies for measuring degradation are often bespoke and, because of an inability to track degradation effectively, the majority of countries combine reporting for deforestation and forest degradation into a single value. Here, we extend a recent analysis that enabled the detection of selective logging at the scale of a logging concession to a regional-scale estimation of selective logging activities. We utilized logging records from across Brazil to train a supervised classification algorithm for detecting logged pixels in Landsat imagery then predicted the extent of logging over a 20 year period throughout Rondônia, Brazil. Approximately one-quarter of the forested lands in Rondônia were cleared between 2000 and 2019. We estimate that 11.0% of the forest area present in 2000 had been selectively logged by 2019, comprising >11 500 km2 of forest. In general, rates of selective logging were twice as high in the first decade relative to the last decade of the period. Our approach is a considerable advance in developing an operationalized selective logging monitoring system capable of detecting subtle forest disturbances over large spatial scales.
- 25 August 2020
- Published by IOP Publishing in Environmental Research Letters
- Vol. 15 (9) , 094057
Open AccessMapping pervasive selective logging in the south-west Brazilian Amazon 2000–2019
Tropical forests harbour the highest biodiversity on the planet and are essential to human livelihoods and the global economy. However continued loss and degradation of forested landscapes, coupled with a rapidly rising global population, is placing incredible pressure on forests globally. The United Nations has developed the Reducing Emissions from Deforestation and forest Degradation (REDD +) programme in response to the challenges facing tropical forests and in recognition of the role they can play in climate mitigation. REDD + requires consistent and reliable monitoring of forests, however, national-level methodologies for measuring degradation are often bespoke and, because of an inability to track degradation effectively, the majority of countries combine reporting for deforestation and forest degradation into a single value. Here, we extend a recent analysis that enabled the detection of selective logging at the scale of a logging concession to a regional-scale estimation of selective logging activities. We utilized logging records from across Brazil to train a supervised classification algorithm for detecting logged pixels in Landsat imagery then predicted the extent of logging over a 20 year period throughout Rondônia, Brazil. Approximately one-quarter of the forested lands in Rondônia were cleared between 2000 and 2019. We estimate that 11.0% of the forest area present in 2000 had been selectively logged by 2019, comprising >11 500 km2 of forest. In general, rates of selective logging were twice as high in the first decade relative to the last decade of the period. Our approach is a considerable advance in developing an operationalized selective logging monitoring system capable of detecting subtle forest disturbances over large spatial scales.
- Open Access
- 29 February 2020
- Published by Wiley in Ecology and Evolution
- Vol. 10 (6) , 2803-2812
Mass–abundance scaling in avian communities is maintained after tropical selective logging
Selective logging dominates forested landscapes across the tropics. Despite the structural damage incurred, selectively logged forests typically retain more biodiversity than other forest disturbances. Most logging impact studies consider conventional metrics, like species richness, but these can conceal subtle biodiversity impacts. The mass–abundance relationship is an integral feature of ecological communities, describing the negative relationship between body mass and population abundance, where, in a system without anthropogenic influence, larger species are less abundant due to higher energy requirements. Changes in this relationship can indicate community structure and function changes. We investigated the impacts of selective logging on the mass–abundance scaling of avian communities by conducting a meta-analysis to examine its pantropical trend. We divide our analysis between studies using mist netting, sampling the understory avian community, and point counts, sampling the entire community. Across 19 mist-netting studies, we found no consistent effects of selective logging on mass–abundance scaling relative to primary forests, except for the omnivore guild where there were fewer larger-bodied species after logging. In eleven point-count studies, we found a more negative relationship in the whole community after logging, likely driven by the frugivore guild, showing a similar pattern. Limited effects of logging on mass–abundance scaling may suggest high species turnover in logged communities, with like-for-like replacement of lost species with similar-sized species. The increased negative mass–abundance relationship found in some logged communities could result from resource depletion, density compensation, or increased hunting; potentially indicating downstream impacts on ecosystem functions. Synthesis and applications. Our results suggest that size distributions of avian communities in logged forests are relatively robust to disturbance, potentially maintaining ecosystem processes in these forests, thus underscoring the high conservation value of logged tropical forests, indicating an urgent need to focus on their protection from further degradation and deforestation.
- 29 February 2020
- Published by Wiley in Ecology and Evolution
- Vol. 10 (6) , 2803-2812
Open AccessMass–abundance scaling in avian communities is maintained after tropical selective logging
Selective logging dominates forested landscapes across the tropics. Despite the structural damage incurred, selectively logged forests typically retain more biodiversity than other forest disturbances. Most logging impact studies consider conventional metrics, like species richness, but these can conceal subtle biodiversity impacts. The mass–abundance relationship is an integral feature of ecological communities, describing the negative relationship between body mass and population abundance, where, in a system without anthropogenic influence, larger species are less abundant due to higher energy requirements. Changes in this relationship can indicate community structure and function changes. We investigated the impacts of selective logging on the mass–abundance scaling of avian communities by conducting a meta-analysis to examine its pantropical trend. We divide our analysis between studies using mist netting, sampling the understory avian community, and point counts, sampling the entire community. Across 19 mist-netting studies, we found no consistent effects of selective logging on mass–abundance scaling relative to primary forests, except for the omnivore guild where there were fewer larger-bodied species after logging. In eleven point-count studies, we found a more negative relationship in the whole community after logging, likely driven by the frugivore guild, showing a similar pattern. Limited effects of logging on mass–abundance scaling may suggest high species turnover in logged communities, with like-for-like replacement of lost species with similar-sized species. The increased negative mass–abundance relationship found in some logged communities could result from resource depletion, density compensation, or increased hunting; potentially indicating downstream impacts on ecosystem functions. Synthesis and applications. Our results suggest that size distributions of avian communities in logged forests are relatively robust to disturbance, potentially maintaining ecosystem processes in these forests, thus underscoring the high conservation value of logged tropical forests, indicating an urgent need to focus on their protection from further degradation and deforestation.
- Open Access
- 11 February 2020
- Published by Wiley in Ecological Applications
- Vol. 30 (4) , e02091
Burning savanna for avian species richness and functional diversity
Prescribed fire is used throughout fire‐prone landscapes to conserve biodiversity. Current best practice in managing savanna systems advocates methods based on the assumption that increased fire‐mediated landscape heterogeneity (pyrodiversity) will promote biodiversity. However, considerable knowledge gaps remain in our understanding of how savanna wildlife responds to the composition and configuration of pyrodiverse landscapes. The effects of pyrodiversity on functional diversity have rarely been quantified and assessing this relationship at a landscape scale that is commensurate with fire management is important for understanding mechanisms underlying ecosystem resilience. Here, we assess the impact of spatiotemporal variation in a long‐term fire regime on avian diversity in North West Province, South Africa. We examined the relationship between (1) species richness, (2) three indices of functional diversity (i.e., functional richness, functional evenness, and functional dispersion) and four measures of pyrodiversity, the spatial extents of fire age classes, and habitat type at the landscape scale. We then used null models to assess differences between observed and expected functional diversity. We found that the proportion of newly burned (<1‐yr post‐fire), old, unburned (≥10 yr post‐fire), and woodland habitat on the landscape predicted species and functional richness. Species richness also increased with the degree of edge contrast between patches of varying fire age, while functional dispersion increased with the degree of patch shape complexity. Lower than expected levels of functional richness suggest that habitat filtering is occurring, resulting in functional redundancy across our study sites. We demonstrate that evaluating functional diversity and redundancy is an important component of conservation planning as they may contribute to previously reported fire resilience. Our findings suggest that it is the type and configuration, rather than the diversity, of fire patches on the landscape that promote avian diversity and conserve ecological functions. A management approach is needed that includes significant coverage of adjacent newly burned and older, unburned savanna habitat; the latter, in particular, is inadequately represented under current burning practices.
- 11 February 2020
- Published by Wiley in Ecological Applications
- Vol. 30 (4) , e02091
Open AccessBurning savanna for avian species richness and functional diversity
Prescribed fire is used throughout fire‐prone landscapes to conserve biodiversity. Current best practice in managing savanna systems advocates methods based on the assumption that increased fire‐mediated landscape heterogeneity (pyrodiversity) will promote biodiversity. However, considerable knowledge gaps remain in our understanding of how savanna wildlife responds to the composition and configuration of pyrodiverse landscapes. The effects of pyrodiversity on functional diversity have rarely been quantified and assessing this relationship at a landscape scale that is commensurate with fire management is important for understanding mechanisms underlying ecosystem resilience. Here, we assess the impact of spatiotemporal variation in a long‐term fire regime on avian diversity in North West Province, South Africa. We examined the relationship between (1) species richness, (2) three indices of functional diversity (i.e., functional richness, functional evenness, and functional dispersion) and four measures of pyrodiversity, the spatial extents of fire age classes, and habitat type at the landscape scale. We then used null models to assess differences between observed and expected functional diversity. We found that the proportion of newly burned (<1‐yr post‐fire), old, unburned (≥10 yr post‐fire), and woodland habitat on the landscape predicted species and functional richness. Species richness also increased with the degree of edge contrast between patches of varying fire age, while functional dispersion increased with the degree of patch shape complexity. Lower than expected levels of functional richness suggest that habitat filtering is occurring, resulting in functional redundancy across our study sites. We demonstrate that evaluating functional diversity and redundancy is an important component of conservation planning as they may contribute to previously reported fire resilience. Our findings suggest that it is the type and configuration, rather than the diversity, of fire patches on the landscape that promote avian diversity and conserve ecological functions. A management approach is needed that includes significant coverage of adjacent newly burned and older, unburned savanna habitat; the latter, in particular, is inadequately represented under current burning practices.
- Open Access
- 4 September 2019
- Published by Springer Nature in Oecologia
- Vol. 191 (2) , 475-482
The impact of secondary forest regeneration on ground-dwelling ant communities in the Tropical Andes
Natural regeneration of abandoned farmland provides an important opportunity to contribute to global reforestation targets, including the Bonn Challenge. Of particular importance are the montane tropics, where a long history of farming, frequently on marginal soils, has rendered many ecosystems highly degraded and hotspots of extinction risk. Ants play crucial roles in ecosystem functioning, and a key question is how time since abandonment and elevation (and inherent temperature gradients therein) affect patterns of ant recovery within secondary forest systems. Focusing on the Colombian Andes across a 1300 m altitudinal gradient and secondary forest (2–30 years) recovering on abandoned cattle pastures, we find that over time ant community composition and species richness recovered towards that of primary forest. However, these relationships are strongly dependent on elevation with the more open and warmer pasturelands supporting more ants than either primary or secondary forest at a particular elevation. The loss of species richness and change in species composition with elevation is less severe in pasture than forests, suggesting that conditions within pasture and its remaining scattered trees, hedgerows and forest fragments, are more favourable for some species, which are likely in or near thermal debt. Promoting and protecting natural regenerating forests over the long term in the montane tropics will likely offer significant potential for returning ant communities towards primary forest levels.
- 4 September 2019
- Published by Springer Nature in Oecologia
- Vol. 191 (2) , 475-482
Open AccessThe impact of secondary forest regeneration on ground-dwelling ant communities in the Tropical Andes
Natural regeneration of abandoned farmland provides an important opportunity to contribute to global reforestation targets, including the Bonn Challenge. Of particular importance are the montane tropics, where a long history of farming, frequently on marginal soils, has rendered many ecosystems highly degraded and hotspots of extinction risk. Ants play crucial roles in ecosystem functioning, and a key question is how time since abandonment and elevation (and inherent temperature gradients therein) affect patterns of ant recovery within secondary forest systems. Focusing on the Colombian Andes across a 1300 m altitudinal gradient and secondary forest (2–30 years) recovering on abandoned cattle pastures, we find that over time ant community composition and species richness recovered towards that of primary forest. However, these relationships are strongly dependent on elevation with the more open and warmer pasturelands supporting more ants than either primary or secondary forest at a particular elevation. The loss of species richness and change in species composition with elevation is less severe in pasture than forests, suggesting that conditions within pasture and its remaining scattered trees, hedgerows and forest fragments, are more favourable for some species, which are likely in or near thermal debt. Promoting and protecting natural regenerating forests over the long term in the montane tropics will likely offer significant potential for returning ant communities towards primary forest levels.
- Open Access
- 1 February 2019
- Published by Elsevier in Remote Sensing of Environment
- Vol. 221, 569-582
A machine learning approach to map tropical selective logging
- 1 February 2019
- Published by Elsevier in Remote Sensing of Environment
- Vol. 221, 569-582
Open AccessA machine learning approach to map tropical selective logging
- Open Access
A Machine Learning Approach to Map Tropical Selective Logging
Hundreds of millions of hectares of tropical forest have been selectively logged, either legally or illegally. Methods for detecting and monitoring tropical selective logging using satellite data are at an early stage, with current methods only able to detect more intensive timber harvest (>20 m 3 ha −1 ). The spatial resolution of widely available datasets, like Landsat, have previously been considered too coarse to measure the subtle changes in forests associated with less intensive selective logging, yet most present-day logging is at low intensity. We utilized a detailed selective logging dataset from over 11,000 ha of forest in Rondônia, southern Brazilian Amazon, to develop a Random Forest machine-learning algorithm for detecting low-intensity selective logging (< 15 m 3 ha −1 ). We show that Landsat imagery acquired before the cessation of logging activities (i.e. the final cloud-free image of the dry season during logging) was better at detecting selective logging than imagery acquired at the start of the following dry season (i.e. the first cloud-free image of the next dry season). Within our study area the detection rate of logged pixels was approximately 90% (with roughly 20% commission and 8% omission error rates) and approximately 40% of the area inside low-intensity selective logging tracts were labelled as logged. Application of the algorithm to 6152 ha of selectively logged forest at a second site in Pará, northeast Brazilian Amazon, resulted in the detection of 2316 ha (38%) of selective logging (with 20% commission and 7% omission error rates). This suggests that our method can detect low-intensity selective logging across large areas of the Amazon. It is thus an important step forward in developing systems for detecting selective logging pan-tropically with freely available data sets, and has key implications for monitoring logging and implementing carbon-based payments for ecosystem service schemes.
Open AccessA Machine Learning Approach to Map Tropical Selective Logging
Hundreds of millions of hectares of tropical forest have been selectively logged, either legally or illegally. Methods for detecting and monitoring tropical selective logging using satellite data are at an early stage, with current methods only able to detect more intensive timber harvest (>20 m 3 ha −1 ). The spatial resolution of widely available datasets, like Landsat, have previously been considered too coarse to measure the subtle changes in forests associated with less intensive selective logging, yet most present-day logging is at low intensity. We utilized a detailed selective logging dataset from over 11,000 ha of forest in Rondônia, southern Brazilian Amazon, to develop a Random Forest machine-learning algorithm for detecting low-intensity selective logging (< 15 m 3 ha −1 ). We show that Landsat imagery acquired before the cessation of logging activities (i.e. the final cloud-free image of the dry season during logging) was better at detecting selective logging than imagery acquired at the start of the following dry season (i.e. the first cloud-free image of the next dry season). Within our study area the detection rate of logged pixels was approximately 90% (with roughly 20% commission and 8% omission error rates) and approximately 40% of the area inside low-intensity selective logging tracts were labelled as logged. Application of the algorithm to 6152 ha of selectively logged forest at a second site in Pará, northeast Brazilian Amazon, resulted in the detection of 2316 ha (38%) of selective logging (with 20% commission and 7% omission error rates). This suggests that our method can detect low-intensity selective logging across large areas of the Amazon. It is thus an important step forward in developing systems for detecting selective logging pan-tropically with freely available data sets, and has key implications for monitoring logging and implementing carbon-based payments for ecosystem service schemes.
- 1 September 2018
- Published by Elsevier in Biological Conservation
- Vol. 225, 1-9
Impacts of selective logging management on butterflies in the Amazon
- 1 September 2018
- Published by Elsevier in Biological Conservation
- Vol. 225, 1-9
Impacts of selective logging management on butterflies in the Amazon