Page 130 - Çevre Şehir ve İklim Dergisi İngilizce - Özel Sayı

NATIONAL SCALE LAND COVER
CLASSIFICATION AND MONITORING SYSTEM

Murat ARSLAN - Reyhan ÇAKIR - İsra AKYAZI
Nida KUMBASAR - Ahmet DOĞAN - Emre YAVUZ

ABSTRACT
In our country, many institutions and organizations carry out and plan to
carry out projects within the scope of their duties and responsibilities, which
play an important role in the development of the country. Land cover data
used in these projects are of great importance in the interpretation of the
land, modeling studies, monitoring and evaluation analyses. With the National
Land Cover Classification and Monitoring System (UASIS) project, it is aimed
to establish a national software system that will produce land cover classes
deter-mined in line with the needs of all stakeholders semi-automatically on
a national scale with artificial intelligence and machine learning technologies
and meet monitoring needs.
Sentinel-2 satellite images and auxiliary data received from institutions are
classified semi-automatically with artificial intelligence-based algorithms and
land cover maps are created annually. With the UASIS project, land cover maps
in 5 main classes and 79 subclasses are produced on a national scale. While
32 classes of 79 UASIS subclasses are created automatically, the remaining
47 classes are obtained from auxiliary data. In the UASIS project carried out
in cooperation with the General Directorate of Combating Desertification
and Erosion and the Scientific and Technological Research Council of Turkey
Informatics and Information Security Advanced Technologies Research Center
(TÜBİTAK BİLGEM), the 2021 National Land Cover Map is being produced and
annual land cover map production will continue. In the pilot region studies
conducted in Sakarya, Büyük Menderes and Eastern Black Sea Basins, the
accuracy rate in automatically classified classes are between 71% and 98%.

Keywords: Land Cover, Classification, Remote Sensing, Türkiye

117
Special Issue / 2024