The Qinghai Tibet Plateau has multiple vegetation types, ranging from broad-leaved and coniferous forests in low and medium altitude areas to high-altitude and arid alpine grasslands. Accurately and detailedly drawing vegetation distribution maps on the Qinghai Tibet Plateau is crucial for better understanding the impact of climate change on terrestrial ecosystems. However, existing land cover datasets on the Qinghai Tibet Plateau either have low spatial resolution or insufficient vegetation types to characterize certain unique ecosystems, such as high-altitude gravel. Using state-of-the-art remote sensing methods, including Sentinel-1 and Sentinel-2 images, environmental and terrain datasets, and machine learning models using the Google Earth Engine platform, a 10 m resolution TP land cover map for the year 2022 (referred to as TP_LC10-2022) was created, which includes 12 vegetation categories and 3 non vegetation categories.
The data sources are Sentinel-1 and Sentinel-2 remote sensing data, as well as environmental and terrain datasets.
Using state-of-the-art remote sensing methods, including Sentinel-1 and Sentinel-2 images, environmental and terrain datasets, and machine learning models using the Google Earth Engine platform, a 10 m resolution TP land cover map for the year 2022 (referred to as TP_LC10-2022) was created, which includes 12 vegetation categories and 3 non vegetation categories.
The overall classification accuracy of dataset TP_LC10-2022 is 86.5%, with a Kappa coefficient of 0.854. Compared with four existing land cover products worldwide, TP_LC10-2022 shows significant improvement in reflecting local scale vertical changes in the southeastern region of the Qinghai Tibet Plateau. In addition, we found that high-altitude gravel accounts for 13.99% of the Qinghai Tibet Plateau region, which has been overlooked in existing land cover datasets. The shrub area accounts for 4.63% of the Qinghai Tibet Plateau region, characterized by the different shapes of deciduous and evergreen shrubs, mainly determined by terrain, which have been overlooked in existing land cover datasets.
This work is licensed under a
CC BY 4.0.
CSTR:11738.11.NCDC.ZENODO.DB6442.2024
10.5281/zenodo.8228112
