Boreal forest land covermapping in Iceland and Finland using Sentinel-1A

The complete Sentinel-1A (S1A) data set since autumn 2014 until September 2015 over two test sites of the EU FP7 project NorthState has been collected: Hallormsstaour in the north-east of Iceland of 50×50 km2 and Hyytiälä in southern Finland of 200×200 km2. The dense 20m-resolution dual-polarization...

Full description

Bibliographic Details
Main Authors: Haarpaintner, J., Davids, C., Storvold, R., Johansen, K. S., Ãrnason, K., Rauste, Yrjö, Mutanen, Teemu
Other Authors: Ouwehand, L.
Format: Other Non-Article Part of Journal/Newspaper
Language:English
Published: European Space Agency (ESA) 2016
Subjects:
Online Access:https://cris.vtt.fi/en/publications/4caa046f-0d60-41d6-bbb0-4646f39ae218
http://www.scopus.com/inward/record.url?scp=84988515694&partnerID=8YFLogxK
Description
Summary:The complete Sentinel-1A (S1A) data set since autumn 2014 until September 2015 over two test sites of the EU FP7 project NorthState has been collected: Hallormsstaour in the north-east of Iceland of 50×50 km2 and Hyytiälä in southern Finland of 200×200 km2. The dense 20m-resolution dual-polarization S1A time series allow for a new level of forest land cover monitoring capabilities compared to ESA's former Envisat A(dvanced)SAR sensor. Temporal filtered dual-polarized mosaics clearly show different land covers that can be classified into forest land cover (FLC) products. Even single agricultural fields can be distinguished from these mosaics. The S1A data set also allowed the construction of a precise water mask for these sites. Results of S1A-based FLC based on two different approaches are compared to 2010 ALOS PALSAR derived FLC results and very high resolution (VHR) NDVI aerial photos. The forest land cover classes extracted are forest, disturbed forest, peatland, grassland, bare land, and settlements.