How good is TanDEM-X 50 m forest/non-forest map? Product validation using temporally corrected geo-browser supplied imagery through Collect Earth

Abstract

TanDEM-X Forest/Non-Forest (FNF) map(s) have been one such data focusing on the status of global forest coverage, which has played an essential role in combating climate change. Although the producers have carried out verification and comparison analyses, the need for accuracy assessments in a broader sense creates uncertainties for the users to approve the new data. For this purpose, TanDEM-X 50 m FNF maps were exclusively examined visually through 66,000 test grids within 30 geocells selected from temperate, boreal, and tropical forest zones. Thus, it was aimed to provide product accuracy utilizing visual inspections to the end users of TanDEM-X FNF maps. In addition, Collect Earth (CE) software was used to evaluate the dataset visually, and its advantages or disadvantages were compared with similarly designed studies. Consequently, even though the producers’ data sets were found to have an accuracy of around 85%, it was observed that there were some issues, especially in the definition of the “non-forest” class. CE software was found to be helpful in such studies. However, the dependence of the analyses on geo-browser supplied imagery had some limitations in estimating the accuracy of a new dataset.

Keywords:

• TanDEM-X FNF
• Collect Earth
• accuracy assessment
• land cover
• Google Earth Engine

Acknowledgement

Authors thank the EOC Geoservice of the Earth Observation Center (EOC) of the German Aerospace Center (DLR) for freely disseminating the TanDEM-X Forest/Non-Forest maps.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data and codes availability statement

The data and codes used to support the findings of the study are available at the following link: https://doi.org/10.6084/m9.figshare.22118468. This resource contains examples and demonstrations that illustrate the results of the study and can be accessed for example analysis and verification of the study’s conclusions.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Notes on contributors

Emre Akturk

Emre Akturk is a current visiting scientist in the Department of Ecology and Conservation Biology at Texas A&M University, where he focuses on research related to forestry applications of geospatial computing and remote sensing, spaceborne laser-based systems, and land cover/use changes. As a key contributor to this article, He played a role in designing the study methodology, interpreting visual data, and preparing and writing the original draft. Email: [email protected] and [email protected].

Arif Oguz Altunel

Arif Oguz Altunel is an Associate Professor at Kastamonu University’s Faculty of Forestry, with research interests including forest operations, geoscience applications, hazard monitoring, and topography. He played a key role in bringing attention to the importance of the study’s topic and contributed to analyzing visual data, and writing and editing the original draft. E-mail: [email protected].

Ayhan Atesoglu

Ayhan Atesoglu is a Professor in the Department of Forest Engineering at Bartın University, Türkiye. His research interests are remote sensing-applications in forestry, geographic information systems, and land use/cover classification and change detection. As a key contributor to this article, he played a role in designing the study methodology, interpreting visual data, and writing the original draft. Email: [email protected].

Mehmet Seki

Mehmet Seki currently works at the Department of Forest Engineering, Faculty of Forestry, Karabuk University as an Assistant Professor. His fields of specialization include forest biometrics, forest mensuration, forest modeling, and forest growth and yield. As a co-author of this article, he played a role in interpreting visual data, analysis and preparing the original draft. E-mail: [email protected].

Serdar Erpay

Serdar Erpay is a freelance forest engineer with an MSc in forest engineering degree from Bartın University. He specializes in the remote sensing applications regarding forest resources, particularly, land-use land-cover change detection, disturbance monitoring/evaluation through geoscience capabilities. He configured all aspects of the GEE engine used in this study and helped analyze the data. Email [email protected].