Monitoring rubber plantation expansion using Landsat data time series and a Shapelet-based approach

Abstract The expansion of tree plantations in tropical forests for commercial rubber cultivation threatens biodiversity which may affect ecosystem services, and hinders ecosystem productivity, causing net carbon emission. Numerous studies refer to the challenge of reliably distinguishing rubber plantations from natural forest, using satellite data, due to their similar spectral signatures, even when phenology is incorporated into an analysis. This study presents a novel approach for monitoring the establishment and expansion of rubber plantations in Seima Protection Forest (SPF), Cambodia (1995–2015), by detecting and analyzing the ‘shapelet’ structure in a Landsat-NDVI time series. This paper introduces a new classification procedure consisting of two steps: (1) an exhaustive-searching algorithm to detect shapelets that represent a period for relatively low NDVI values within an image time series; and (2) a t-test used to determine if NDVI values of detected shapelets are significantly different than their non-shapelet trend, thereby indicating the presence of rubber plantations. Using this approach, historical rubber plantation events were mapped over the twenty-year timespan. The shapelet algorithm produced two types of information: (1) year of rubber plantation establishment; and (2) pre-conversion land-cover type (i.e., agriculture, or natural forest). The overall accuracy of the rubber plantation map for the year of 2015 was 89%. The multi-temporal map products reveal that more than half of the rubber planting activity (57%) took place in 2010 and 2011, following the granting of numerous rubber concessions two years prior. Seventy-three percent of the rubber plantations were converted from natural forest and twenty-three percent were established on non-forest land-cover. The shapelet approach developed here can be used reliably to improve our understanding of the expansion of rubber production beyond Seima Protection Forest of Cambodia, and likely elsewhere in the tropics.

[1]  Zhiqiang Yang,et al.  Detecting trends in forest disturbance and recovery using yearly Landsat time series: 1. LandTrendr — Temporal segmentation algorithms , 2010 .

[2]  J. Gerber.  Conflicts over industrial tree plantations in the South: Who, how and why? , 2011 .

[3]  Zheng Zhang,et al.  Phenology-Based Vegetation Index Differencing for Mapping of Rubber Plantations Using Landsat OLI Data , 2015, Remote. Sens..

[4]  Francis E. Putz,et al.  Critical need for new definitions of “forest” and “forest degradation” in global climate change agreements , 2009 .

[5]  C. Tucker.  Red and photographic infrared linear combinations for monitoring vegetation , 1979 .

[6]  G. Foody,et al.  Estimating tropical forest biomass with a combination of SAR image texture and Landsat TM data: An assessment of predictions between regions , 2012 .

[7]  Eamonn J. Keogh,et al.  Time series shapelets: a novel technique that allows accurate, interpretable and fast classification , 2010, Data Mining and Knowledge Discovery.

[8]  Zhe Li,et al.  Integrating Mahalanobis typicalities with a neural network for rubber distribution mapping , 2011 .

[9]  W. Cohen,et al.  North American forest disturbance mapped from a decadal Landsat record , 2008 .

[10]  Pol Coppin,et al.  Review ArticleDigital change detection methods in ecosystem monitoring: a review , 2004 .

[11]  Martha C. Anderson,et al.  Landsat-8: Science and Product Vision for Terrestrial Global Change Research , 2014 .

[12]  B. Minasny,et al.  Mapping and identifying basal stem rot disease in oil palms in North Sumatra with QuickBird imagery , 2011, Precision Agriculture.

[13]  Clinton N. Jenkins,et al.  The Impacts of Oil Palm on Recent Deforestation and Biodiversity Loss , 2016, PloS one.

[14]  Jason Lines,et al.  Classification of time series by shapelet transformation , 2013, Data Mining and Knowledge Discovery.

[15]  S. Goward,et al.  An automated approach for reconstructing recent forest disturbance history using dense Landsat time series stacks , 2010 .

[16]  D. Lobell,et al.  Landsat-based classification in the cloud: An opportunity for a paradigm shift in land cover monitoring , 2017 .

[17]  B. Griscom,et al.  Biomass estimations and carbon stock calculations in the oil palm plantations of African derived savannas using IKONOS data , 2004 .

[18]  A. Angelsen.  Agricultural expansion and deforestation: modelling the impact of population , 1999 .

[19]  Preesan Rakwatin,et al.  Oil Palm Tree Detection with High Resolution Multi-Spectral Satellite Imagery , 2014, Remote. Sens..

[20]  David E. Knapp,et al.  Automated mapping of tropical deforestation and forest degradation: CLASlite , 2009 .

[21]  Subhash Bhatia,et al.  Palm oil: Addressing issues and towards sustainable development , 2009 .

[22]  L. Bremer,et al.  Does plantation forestry restore biodiversity or create green deserts? A synthesis of the effects of land-use transitions on plant species richness , 2010, Biodiversity and Conservation.

[23]  J. Rogan,et al.  Modelling dry season deciduousness in Mexican Yucatán forest using MODIS EVI data (2000–2011) , 2013 .

[24]  C. Justice,et al.  High-Resolution Global Maps of 21st-Century Forest Cover Change , 2013, Science.

[25]  Dirk Pflugmacher,et al.  Cross-border forest disturbance and the role of natural rubber in mainland Southeast Asia using annual Landsat time series , 2015 .

[26]  Nicholas C. Coops,et al.  Estimating urban vegetation fraction across 25 cities in pan-Pacific using Landsat time series data , 2017 .

[27]  Maggi Kelly,et al.  Improving the Prediction of African Savanna Vegetation Variables Using Time Series of MODIS Products. , 2017, ISPRS journal of photogrammetry and remote sensing : official publication of the International Society for Photogrammetry and Remote Sensing.

[28]  Petr Keil,et al.  Comment on “High-resolution global maps of 21st-century forest cover change” , 2014, Science.

[29]  Lars Laestadius,et al.  When is a forest a forest? Forest concepts and definitions in the era of forest and landscape restoration , 2016, Ambio.

[30]  T. Stone,et al.  Using multi-temporal satellite data to evaluate selective logging in Para, Brazil , 1998 .

[31]  Zhiming Feng,et al.  Mapping rubber tree plantations using a Landsat-based phenological algorithm in Xishuangbanna, southwest China , 2015 .

[32]  Deborah Lawrence,et al.  Committed carbon emissions, deforestation, and community land conversion from oil palm plantation expansion in West Kalimantan, Indonesia , 2012, Proceedings of the National Academy of Sciences.

[33]  Gregory P. Asner,et al.  Selective logging changes forest phenology in the Brazilian Amazon: Evidence from MODIS image time series analysis , 2009 .

[34]  Onisimo Mutanga,et al.  Commercial tree species discrimination using airborne AISA Eagle hyperspectral imagery and partial least squares discriminant analysis (PLS-DA) in KwaZulu-Natal, South Africa , 2013 .

[35]  Zhe Zhu,et al.  Change detection using landsat time series: A review of frequencies, preprocessing, algorithms, and applications , 2017 .

[36]  David Edwards,et al.  Increasing Demand for Natural Rubber Necessitates a Robust Sustainability Initiative to Mitigate Impacts on Tropical Biodiversity , 2015 .

[37]  M. Konik,et al.  Object-oriented approach to oil spill detection using ENVISAT ASAR images , 2016 .

[38]  P. Beckschäfer.  Obtaining rubber plantation age information from very dense Landsat TM & ETM + time series data and pixel-based image compositing , 2017 .

[39]  A. Ziegler,et al.  Trends, drivers and impacts of changes in swidden cultivation in tropical forest-agriculture frontiers: A global assessment , 2012 .

[40]  O. Mertz,et al.  The carbon sequestration potential of tree crop plantations , 2013, Mitigation and Adaptation Strategies for Global Change.

[41]  Jiaguo Qi,et al.  Monitoring selective logging in tropical evergreen forests using Landsat: Multitemporal regional analyses in Mato Grosso, Brazil , 2005 .

[42]  J. Rogan,et al.  Remote sensing technology for mapping and monitoring land-cover and land-use change , 2004 .

[43]  Bangqian Chen,et al.  Mapping deciduous rubber plantations through integration of PALSAR and multi-temporal Landsat imagery , 2013 .

[44]  Joanne C. White,et al.  An integrated Landsat time series protocol for change detection and generation of annual gap-free surface reflectance composites , 2015 .

[45]  Hongmei Li,et al.  Past, present and future land-use in Xishuangbanna, China and the implications for carbon dynamics , 2008 .

[46]  Jinwei Dong,et al.  Mapping tropical forests and rubber plantations in complex landscapes by integrating PALSAR and MODIS imagery , 2012 .

[47]  M. Herold,et al.  Monitoring forest cover loss using multiple data streams, a case study of a tropical dry forest in Bolivia , 2015 .

[48]  E. Veldkamp,et al.  Conversion of lowland tropical forests to tree cash crop plantations loses up to one-half of stored soil organic carbon , 2015, Proceedings of the National Academy of Sciences.

[49]  Jefferson Fox,et al.  Rubber Tree Distribution Mapping in Northeast Thailand , 2011 .

[50]  Jianchu Xu,et al.  The Rubber Juggernaut , 2009, Science.

[51]  Michael A. Wulder,et al.  Landsat continuity: Issues and opportunities for land cover monitoring , 2008 .