Revealing post-megafire spectral and compositional recovery in the Siberian boreal forest using Landsat time series and regression-based unmixing approach
Suri G. Bao , Wen J. Wang , Zhihua Liu , Hankui K. Zhang , Lei Wang , Jun Ma , Hongchao Sun , Shengjie Ba , Yeqiao Wang , Hong S. He
{"title":"Revealing post-megafire spectral and compositional recovery in the Siberian boreal forest using Landsat time series and regression-based unmixing approach","authors":"Suri G. Bao , Wen J. Wang , Zhihua Liu , Hankui K. Zhang , Lei Wang , Jun Ma , Hongchao Sun , Shengjie Ba , Yeqiao Wang , Hong S. He","doi":"10.1016/j.rse.2024.114307","DOIUrl":null,"url":null,"abstract":"<div><p>Megafires trigger abrupt ecosystem changes and may lead to alternative successional pathways, thereby undermining the recovery and resilience of forests. However, comprehensive multi-perspective assessments of post-megafire forest recovery remain limited. In this study, we assessed short- and long-term spectral and compositional recovery following the 1987 Black Dragon Megafire in the Siberian boreal forest of Northeast China. We first disentangled continuous fractions of five forest cover types for eight time periods between 1987 and 2022 utilizing Landsat time series and regression-based unmixing approach. Machine learning regression models (support vector regression (SVR), random forest regression (RFR), and light gradient boosting machine regression (LGBMR)) were trained using synthetically mixed datasets generated from the Landsat spectral library. Independent accuracy assessment indicated that LGBMR outperformed SVR and RFR, with mean absolute errors ranging from 6.87% to 10.04% across five cover types, suggesting the effectiveness of the unmixing approach for mapping forest cover type fractions in the Siberian boreal forest. We then assessed post-megafire spectral and compositional recovery of forests using forest cover type fractions and spectral indices across different levels of burn severity. The short-term assessment indicated rapid spectral recovery rates dominated by shrubs and deciduous broadleaved forests, especially in high severity areas. Over the long-term scale, the spectral recovery required 15–16 years to return to pre-fire baselines, surpassing the rate of compositional recovery. Recovery rates were significantly faster for broadleaved forests compared to needleleaved forests, resulting in mixed forests in low-severity areas and a predominance of broadleaved forests in high-severity areas. Our findings indicated that megafires catalyzed a shift from needleleaved to broadleaved forests in the Siberian boreal forest, suggesting lower resilience of these forests to high-severity megafires. This study offers valuable insights into the profound and enduring effects of catastrophic events such as megafires on the recovery and resilience of boreal forests, and informs restoration and adaptation efforts thereafter.</p></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":null,"pages":null},"PeriodicalIF":11.1000,"publicationDate":"2024-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Remote Sensing of Environment","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0034425724003250","RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ENVIRONMENTAL SCIENCES","Score":null,"Total":0}
引用次数: 0
Abstract
Megafires trigger abrupt ecosystem changes and may lead to alternative successional pathways, thereby undermining the recovery and resilience of forests. However, comprehensive multi-perspective assessments of post-megafire forest recovery remain limited. In this study, we assessed short- and long-term spectral and compositional recovery following the 1987 Black Dragon Megafire in the Siberian boreal forest of Northeast China. We first disentangled continuous fractions of five forest cover types for eight time periods between 1987 and 2022 utilizing Landsat time series and regression-based unmixing approach. Machine learning regression models (support vector regression (SVR), random forest regression (RFR), and light gradient boosting machine regression (LGBMR)) were trained using synthetically mixed datasets generated from the Landsat spectral library. Independent accuracy assessment indicated that LGBMR outperformed SVR and RFR, with mean absolute errors ranging from 6.87% to 10.04% across five cover types, suggesting the effectiveness of the unmixing approach for mapping forest cover type fractions in the Siberian boreal forest. We then assessed post-megafire spectral and compositional recovery of forests using forest cover type fractions and spectral indices across different levels of burn severity. The short-term assessment indicated rapid spectral recovery rates dominated by shrubs and deciduous broadleaved forests, especially in high severity areas. Over the long-term scale, the spectral recovery required 15–16 years to return to pre-fire baselines, surpassing the rate of compositional recovery. Recovery rates were significantly faster for broadleaved forests compared to needleleaved forests, resulting in mixed forests in low-severity areas and a predominance of broadleaved forests in high-severity areas. Our findings indicated that megafires catalyzed a shift from needleleaved to broadleaved forests in the Siberian boreal forest, suggesting lower resilience of these forests to high-severity megafires. This study offers valuable insights into the profound and enduring effects of catastrophic events such as megafires on the recovery and resilience of boreal forests, and informs restoration and adaptation efforts thereafter.
期刊介绍:
Remote Sensing of Environment (RSE) serves the Earth observation community by disseminating results on the theory, science, applications, and technology that contribute to advancing the field of remote sensing. With a thoroughly interdisciplinary approach, RSE encompasses terrestrial, oceanic, and atmospheric sensing.
The journal emphasizes biophysical and quantitative approaches to remote sensing at local to global scales, covering a diverse range of applications and techniques.
RSE serves as a vital platform for the exchange of knowledge and advancements in the dynamic field of remote sensing.