File Download
There are no files associated with this item.
Links for fulltext
(May Require Subscription)
- Publisher Website: 10.1080/01431161.2017.1312619
- Scopus: eid_2-s2.0-85022132466
- WOS: WOS:000401460800002
- Find via
Supplementary
- Citations:
- Appears in Collections:
Article: New land-cover maps of Ghana for 2015 using landsat 8 and three popular classifiers for biodiversity assessment
Title | New land-cover maps of Ghana for 2015 using landsat 8 and three popular classifiers for biodiversity assessment |
---|---|
Authors | |
Issue Date | 2017 |
Citation | International Journal of Remote Sensing, 2017, v. 38, n. 14, p. 4008-4021 How to Cite? |
Abstract | © 2017 Informa UK Limited, trading as Taylor & Francis Group. Remote-sensing products provide opportunities to monitor complex phenomena on the Earth from space for good decision-making. An example is land-cover maps, which are useful for monitoring both human- and naturally induced environmental changes to enhance environmental management for the benefit of the society. Yet there are numerous critical areas of the world such as biodiversity hotspots where the inaccessibility of quality satellite images has hindered progress in our understanding and management of the natural environment. The West African biodiversity hotspot is one such area where persistent thick clouds in freely available satellite images has been a great discouragement to the production of local land-cover maps to monitor the ongoing deforestation and rapid environmental change. Ghana is a country in West Africa where no land-cover map (except global ones) has been published for over a decade although large tracts of land have been converted from their natural states to agricultural land. In this article, we present 30 m land-cover maps of the country using Landsat 8 images and three popular classifiers – Maximum Likelihood Classifier (MLC), Support Vector Machines (SVMs), and random forest (RF) – for the year 2015. We produced these maps for use in our future biodiversity assessment in the region. An overall accuracy (OA) of 85.5% (kappa = 0.77) was achieved for MLC, with similar but slightly lower accuracies for RF and SVM, which indicates that advanced classification algorithms may not have many advantages when applied to process only multispectral optical data. We observe that shrublands, croplands, and orchards are the dominant classes occupying 35.6%, 31.9%, and 18.4% of the country’s land area, respectively. This further indicates that >50.0% of the country’s land area is under agriculture. Thick forests occupy only 8.0% of the land and are almost entirely confined to forest reserves, highlighting the usefulness of these nature reserves. Our maps will provide input for research advancement, policy formulation, and environmental management at both national and subnational levels. |
Persistent Identifier | http://hdl.handle.net/10722/296953 |
ISSN | 2023 Impact Factor: 3.0 2023 SCImago Journal Rankings: 0.776 |
ISI Accession Number ID |
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Hackman, Kwame Oppong | - |
dc.contributor.author | Gong, Peng | - |
dc.contributor.author | Wang, Jie | - |
dc.date.accessioned | 2021-02-25T15:17:02Z | - |
dc.date.available | 2021-02-25T15:17:02Z | - |
dc.date.issued | 2017 | - |
dc.identifier.citation | International Journal of Remote Sensing, 2017, v. 38, n. 14, p. 4008-4021 | - |
dc.identifier.issn | 0143-1161 | - |
dc.identifier.uri | http://hdl.handle.net/10722/296953 | - |
dc.description.abstract | © 2017 Informa UK Limited, trading as Taylor & Francis Group. Remote-sensing products provide opportunities to monitor complex phenomena on the Earth from space for good decision-making. An example is land-cover maps, which are useful for monitoring both human- and naturally induced environmental changes to enhance environmental management for the benefit of the society. Yet there are numerous critical areas of the world such as biodiversity hotspots where the inaccessibility of quality satellite images has hindered progress in our understanding and management of the natural environment. The West African biodiversity hotspot is one such area where persistent thick clouds in freely available satellite images has been a great discouragement to the production of local land-cover maps to monitor the ongoing deforestation and rapid environmental change. Ghana is a country in West Africa where no land-cover map (except global ones) has been published for over a decade although large tracts of land have been converted from their natural states to agricultural land. In this article, we present 30 m land-cover maps of the country using Landsat 8 images and three popular classifiers – Maximum Likelihood Classifier (MLC), Support Vector Machines (SVMs), and random forest (RF) – for the year 2015. We produced these maps for use in our future biodiversity assessment in the region. An overall accuracy (OA) of 85.5% (kappa = 0.77) was achieved for MLC, with similar but slightly lower accuracies for RF and SVM, which indicates that advanced classification algorithms may not have many advantages when applied to process only multispectral optical data. We observe that shrublands, croplands, and orchards are the dominant classes occupying 35.6%, 31.9%, and 18.4% of the country’s land area, respectively. This further indicates that >50.0% of the country’s land area is under agriculture. Thick forests occupy only 8.0% of the land and are almost entirely confined to forest reserves, highlighting the usefulness of these nature reserves. Our maps will provide input for research advancement, policy formulation, and environmental management at both national and subnational levels. | - |
dc.language | eng | - |
dc.relation.ispartof | International Journal of Remote Sensing | - |
dc.title | New land-cover maps of Ghana for 2015 using landsat 8 and three popular classifiers for biodiversity assessment | - |
dc.type | Article | - |
dc.description.nature | link_to_subscribed_fulltext | - |
dc.identifier.doi | 10.1080/01431161.2017.1312619 | - |
dc.identifier.scopus | eid_2-s2.0-85022132466 | - |
dc.identifier.volume | 38 | - |
dc.identifier.issue | 14 | - |
dc.identifier.spage | 4008 | - |
dc.identifier.epage | 4021 | - |
dc.identifier.eissn | 1366-5901 | - |
dc.identifier.isi | WOS:000401460800002 | - |
dc.identifier.issnl | 0143-1161 | - |