Welcome to Acta Prataculturae Sinica ! Today is Share:

Acta Prataculturae Sinica ›› 2023, Vol. 32 ›› Issue (9): 1-16.DOI: 10.11686/cyxb2022405

   

Spatiotemporal dynamic variation of temperate grassland classes in Inner Mongolia in the last 20 years

Zhi-gui YANG1(), Jian-guo ZHANG1, Jin-rong LI2, Hong-yan YU3, Li CHANG4, Shu-hua YI1, Yan-yan LYU1, Yu-zhuo ZHANG1, Bao-ping MENG1()   

  1. 1.Institute of Fragile Eco-environment,School of Geographic Science,Nantong University,Nantong 226007,China
    2.Yinshanbeilu National Field Research Station of Desert Steppe Eco-hydrological System,China Institute of Water Resources and Hydropower Research,Beijing 100038,China
    3.Qinghai Service and Guarantee Center of Qilian Mountain National Park,Xining 810001,China
    4.College of Urban Environment,Lanzhou City University,Lanzhou 730070,China
  • Received:2022-10-10 Revised:2022-11-28 Online:2023-09-20 Published:2023-07-12
  • Contact: Bao-ping MENG

Abstract:

Grassland classification is essential for rational utilization and effective protection of grassland resources and also crucial for maintaining sustainable development of grassland ecosystems. There has been considerable achievement in documenting spatio-temporal changes in land use types, but few reports on the spatio-temporal dynamic variation of grassland classes at the regional scale. Hence, this study explored the spatio-temporal variation of temperate grassland classes in Inner Mongolia. Several grassland class identification methods were constructed based on remote sensing vegetation index, meteorology, soil, topography, unmanned aerial vehicle (UAV) data and machine learning algorithms. Then the spatio-temporal variation was analyzed based on the optimal classification method. It was found that: 1) Among all the remote sensing classification characteristics and indexes evaluated, the importance values of precipitation and normalized difference vegetation index (NDVI) for grassland classification were higher than those of other indexes in the study area, and the cumulative contribution of the importance values of first 18 classification characteristics and indexes was more than 85%; 2) The random forest (RF) model gave the highest classification accuracy and was superior to other methods in grassland class identification in Inner Mongolia, with an overall accuracy of 82.16% and Kappa coefficient of 0.76; 3) In the past 20 years, the transition between grassland classes in Inner Mongolia has been intense, and has mainly occurred through transition among the classes typical steppe, desert steppe and desert. Compared with the grassland types in the 1980s, the grassland types in 2000-2009 changed from wet to dry, while the grassland types in 2010-2019 changed from dry to wet. The results of this study provide a scientific evaluation of the changes in grassland type in Inner Mongolia under global climate change and as a result of human activities, and provide a theoretical basis and technical support for planning sustainable grassland development in Inner Mongolia.

Key words: grassland classes, characteristic index, machine learning, spatio-temporal variation