ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume VI-3/W1-2020
ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., VI-3/W1-2020, 83–90, 2020
https://doi.org/10.5194/isprs-annals-VI-3-W1-2020-83-2020
ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., VI-3/W1-2020, 83–90, 2020
https://doi.org/10.5194/isprs-annals-VI-3-W1-2020-83-2020

  17 Nov 2020

17 Nov 2020

LAND USE/LAND COVER CHANGES USING MULTI-TEMPORAL SATELLITE

H. T. T. Nguyen1, Q. T. N. Chau1, A. T. Pham2, H. T. Phan1, P. T. X. Tran1, H. T. Cao1, T. Q. Le3, and D. T. H. Nguyen4 H. T. T. Nguyen et al.
  • 1Dept. of Forest Resource and Environment management, Tay Nguyen University, Vietnam
  • 2Department of Agriculture and Rural Development Dak Nong, Vietnam
  • 3Space Technology Institute, Vietnam
  • 4Can Tho University, Vietnam

Keywords: Landsat images, land use/land cover change (LULCC), accuracy assessment, forest cover changes, Random Forest algorithm

Abstract. Producing the map of land use land cover change (LULCC) at the local extent is fundamental for a variety of applications such as vegetation, forest covers, soil degradation, and global warming. Understanding the directions and spread trend of LULCC plays significant role in obtaining useful data for the local authorities in making land-use policies under the context of climate change. Dak Nong is located in the Central Highlands of Vietnam having the largest tropical forest. Over the past decades, the natural forest in the region has significantly declined due to the pressure of population growth and social-economic development. The current study analyzed the LULCC in the province over the four periods: 2005–2018, 2005–2010, 2010–2015, and 2015–2018. Information from Landsat satellite imagery captured in 2005, 2010, 2015, and 2018 was utilized to create the LULC maps and detect the land-use changes. The Random Forest (RF) was employed to categorize the images into nine different LULC classes. The study showed that classification accuracy was achieved from 72.49% to 84.55% with a kappa coefficient of 0.69 to 0.81. The findings revealed a significant decrease in the natural forest over time from 53.1% to 42.7%, 36.8%, and 34.6% in 2005, 2010, 2015, and 2018, respectively. Meanwhile, the other types of LULC showed an increase in the area over the periods, especially croplands. It was noticeable that the continuous decrease in the forest area over the years has put pressure on the natural environmental resources and generated the risk of climate change.