Using high-resolution imagery and deep learning to classify land-use following deforestation: a case study in Ethiopia

被引:13
作者
Masolele, Robert N. [1 ]
De Sy, Veronique [1 ]
Marcos, Diego [1 ]
Verbesselt, Jan [1 ]
Gieseke, Fabian [2 ]
Mulatu, Kalkidan Ayele [3 ]
Moges, Yitebitu [4 ]
Sebrala, Heiru [4 ]
Martius, Christopher [5 ]
Herold, Martin [1 ]
机构
[1] Wageningen Univ Res, Lab Geoinformat Sci & Remote Sensing, Pb Wageningen, Netherlands
[2] Univ Munster, Dept Informat Syst, Leonardo Campus 3, Munster, Germany
[3] Int Ctr Trop Agr CIAT, Addis Ababa, Ethiopia
[4] Environm Forest & Climate Change Commiss, Natl REDD Secretariat, Addis Ababa, Ethiopia
[5] Ctr Int Forestry Res CIFOR Germany gGmbH, Bonn, Germany
关键词
Attention U-Net; deep learning; Planet-NICFI; Land-use following deforestation; deforestation drivers; remote sensing; FOREST; SATELLITE; ATTENTION; NETWORKS; DRIVERS; COVER;
D O I
10.1080/15481603.2022.2115619
中图分类号
P9 [自然地理学];
学科分类号
0705 ; 070501 ;
摘要
National-scale assessments of post-deforestation land-use are crucial for decreasing deforestation and forest degradation-related emissions. In this research, we assess the potential of different satellite data modalities (single-date, multi-date, multi-resolution, and an ensemble of multi-sensor images) for classifying land-use following deforestation in Ethiopia using the U-Net deep neural network architecture enhanced with attention. We performed the analysis on satellite image data retrieved across Ethiopia from freely available Landsat-8, Sentinel-2 and Planet-NICFI satellite data. The experiments aimed at an analysis of (a) single-date images from individual sensors to account for the differences in spatial resolution between image sensors in detecting land-uses, (b) ensembles of multiple images from different sensors (Planet-NICFI/Sentinel-2/Landsat-8) with different spatial resolutions, (c) the use of multi-date data to account for the contribution of temporal information in detecting land-uses, and, finally, (d) the identification of regional differences in terms of land-use following deforestation in Ethiopia. We hypothesize that choosing the right satellite imagery (sensor) type is crucial for the task. Based on a comprehensive visually interpreted reference dataset of 11 types of post-deforestation land-uses, we find that either detailed spatial patterns (single-date Planet-NICFI) or detailed temporal patterns (multi-date Sentinel-2, Landsat-8) are required for identifying land-use following deforestation, while medium-resolution single-date imagery is not sufficient to achieve high classification accuracy. We also find that adding soft-attention to the standard U-Net improved the classification accuracy, especially for small-scale land-uses. The models and products presented in this work can be used as a powerful data resource for governmental and forest monitoring agencies to design and monitor deforestation mitigation measures and data-driven land-use policy.
引用
收藏
页码:1446 / 1472
页数:27
相关论文
共 84 条
[61]   Recent Advances in Forest Observation with Visual Interpretation of Very High-Resolution Imagery [J].
Schepaschenko, Dmitry ;
See, Linda ;
Lesiv, Myroslava ;
Bastin, Jean-Francois ;
Mollicone, Danilo ;
Tsendbazar, Nandin-Erdene ;
Bastin, Lucy ;
McCallum, Ian ;
Bayas, Juan Carlos Laso ;
Baklanov, Artem ;
Perger, Christoph ;
Duerauer, Martina ;
Fritz, Steffen .
SURVEYS IN GEOPHYSICS, 2019, 40 (04) :839-862
[62]   Attention gated networks: Learning to leverage salient regions in medical images [J].
Schlemper, Jo ;
Oktay, Ozan ;
Schaap, Michiel ;
Heinrich, Mattias ;
Kainz, Bernhard ;
Glocker, Ben ;
Rueckert, Daniel .
MEDICAL IMAGE ANALYSIS, 2019, 53 :197-207
[63]   Landsat-Based Land Use Change Assessment in the Brazilian Atlantic Forest: Forest Transition and Sugarcane Expansion [J].
Silva, Alindomar Lacerda ;
Alves, Diogenes Salas ;
Ferreira, Matheus Pinheiro .
REMOTE SENSING, 2018, 10 (07)
[64]   Assessment of land use land cover dynamics and its drivers in Bechet Watershed Upper Blue Nile Basin, Ethiopia [J].
Sisay, Getahun ;
Gitima, Ginjo ;
Mersha, Mengistie ;
Alemu, Woubet G. .
REMOTE SENSING APPLICATIONS-SOCIETY AND ENVIRONMENT, 2021, 24
[65]   A third of the tropical African flora is potentially threatened with extinction [J].
Stevart, T. ;
Dauby, G. ;
Lowry, P. P., II ;
Blach-Overgaard, A. ;
Droissart, V. ;
Harris, D. J. ;
Mackinder, B. A. ;
Schatz, G. E. ;
Sonke, B. ;
Sosef, M. S. M. ;
Svenning, J-C. ;
Wieringa, J. J. ;
Couvreur, T. L. P. .
SCIENCE ADVANCES, 2019, 5 (11)
[66]  
Stocker TF, 2014, CLIMATE CHANGE 2013: THE PHYSICAL SCIENCE BASIS, P1, DOI 10.1017/cbo9781107415324
[67]  
Summary for Policymakers, 2001, CLIMATE CHANGE 2001, P2
[68]  
Tadese Semegnew, 2021, ScientificWorldJournal, V2021, P6685045, DOI 10.1155/2021/6685045
[69]   Assessing land use and land cover change detection using remote sensing in the Lake Tana Basin, Northwest Ethiopia [J].
Tewabe, Dires ;
Fentahun, Temesgen .
COGENT ENVIRONMENTAL SCIENCE, 2020, 6 (01)
[70]  
Tewolde Berhan G. E., 1991, Plant genetic resources of Ethiopia., P75