Contents lists available at ScienceDirect Applied Geography journal homepage: www.elsevier.com/locate/apgeog Global trends analysis of the main vegetation types throughout the past four decades Ghaleb Faour a , Mario Mhawej a,* , Ali Nasrallah a,b a National Center for Remote Sensing, National Council for Scientific Research (CNRS), Riad al Soloh, 1107 2260, Beirut, Lebanon b IRSTEA, University of Montpellier, TETIS, 34090, Montpellier, France ARTICLE INFO Keywords: Change detection NDVI Classification method Global and continental scales Vegetation areas ABSTRACT In remote sensing studies, the photosynthetically active radiation absorbed by chlorophyll in the green leaves of vegetation canopies is measured using Red and Near-Infra Red bands. The Normalized Difference Vegetation Index (NDVI) is one of the most commonly used vegetation indices that are generally obtained from a calculation of the above mentioned bands; it presents a decent surrogate measures of the physiologically functioning surface greenness level. In this study, the latest version of the GIMMS NDVI data set, between the period of January 1982 and December 2015, were used to classify the global vegetation areas into five main categories (i.e. Agriculture Areas, Boreal Forests, Deciduous Forests, Evergreen and Tropical Forests, and Other Vegetation), using a simple and straight-forward method of classification, surnamed Global Vegetation Types Classification (GVTC). The total accuracy of the model reached 90.4% with a kappa value of 87.1%. In each category, a trend analysis has been carried out at both global and continental levels. The objective was to highlight the changes within each category, throughout the past thirty-four years. Results show that Agriculture Areas are increasing worldwide, with a huge upsurge observed since 2011 coinciding with a re- markable decrease in Boreal Forests. Changes in vegetation's classes, between 1982 and 2015, were more pro- nounceable in continents such as Asia, America and Africa; Europe and Oceania showed limited variations throughout this same period. Following these results, regional policies should be reformed and mitigation plans should be established in order to maintain a sustainable development of the global vegetation lands. The GVTC could be implemented with higher spatial resolution imageries for more local-based assessments. 1. Introduction Our world is changing at fast pace. Most notably, global forest area is declining (Orth & Moore, 1983; Verheyen et al., 2016; Wulder, Butson, & White, 2008), sea level is rising (Gardner, Cogley, Moholdt, Wouters, & Wiese, 2015; Meier & Wahr, 2002; Nicholls & Cazenave, 2010), global warming is increasing (Eide, 2008; Fyfe, Gillett, & Zwiers, 2013) and population worldwide is growing (Cincotta, Engelman, & Anastasion, 2003; Sixsmith, 2013). The availability of remote sensing techniques and satellite imageries made it possible to observe and as- sess these changes from space in a time- and resource-effective manner (e.g. Cooper, Chen, Fletcher, & Barbee, 2013; Faour & Mhawej, 2014; Kellner & Hubbell, 2017; Kubanek, Nolte, Taubenböck, Wenzel, & Kappas, 2014; Lillesand, Kiefer, & Chipman, 2014). To describe the physiologically functioning surface greenness level for each picture element and to detect the vegetation trends across the globe, several vegetation indices have been proposed. The most widely used remains the Normalized Difference Vegetation Index (NDVI) (Faour, Mhawej, & Fayad, 2016; Kerr & Ostrovsky, 2003). This index is calculated from the Visible and Near-Infra Red light reflected by ve- getation with values ranging from -1.0 to +1.0 (Tappan, Tyler, Wehde, & Moore, 1992). The usage of NDVI in the literature served different purposes: some authors tried to estimate the Fractional Vegetation Covers (FVC), the Leaf Area Index (LAI) and the surface soil moisture content from NDVI (e.g. Carlson & Ripley, 1997; Carlson, Gillies, & Perry, 1994; Jiang et al., 2006; North, 2002; Wu et al., 2014). Others introduced this index in drought studies while detecting changes in vegetation trends (e.g. Faour, Mhawej, & Abou Najem, 2015; Faour, Mhawej, & Fayad, 2016; Gu, BrownVerdin, & Wardlow, 2007; Liu & Kogan, 1996; Mwaniki & Möller, 2015; Peters et al., 2002; Petropoulos, Griffiths, & Kalivas, 2014; Riva, Daliakopoulos, Eckert, Elias, & Liniger, 2017; Shalaby & Tateishi, 2007; Van Hoek, Jia, Zhou, Zheng, & Menenti, 2016). More- over, the NDVI was used in different discipline, such as forestry and wildfire managements (e.g. Wang, Adiku, Tenhunen, & Granier, 2005; Schrader-Patton, Grulke, & Dressen, 2016; Mhawej, Faour, Abdallah, & https://doi.org/10.1016/j.apgeog.2018.05.020 Received 18 March 2018; Received in revised form 25 April 2018; Accepted 29 May 2018 * Corresponding author. E-mail addresses: gfaour@cnrs.edu.lb (G. Faour), mario.mhawej@gmail.com (M. Mhawej), ali.nasrallah@agroparistech.fr (A. Nasrallah). Applied Geography 97 (2018) 184–195 0143-6228/ © 2018 Elsevier Ltd. All rights reserved. T