Understanding how our planet's land cover changes over time is crucial for tackling climate change. However, getting detailed, long-term data has been tough. A latest released land-cover dataset offers an unprecedented glimpse into global land-cover dynamics from 1985 to 2022 at a remarkable 30-meter resolution. The study was published in Earth System Science Data. The dataset is available in CASEarth thematic data system as well as Zenodo. This dataset is called GLC_FCS30D and is developed by a research team led by Prof. LIU Liangyun from the Aerospace Information Research Institute (AIR) of the Chinese Academy of Sciences (CAS). Developed using dense-time-series of Landsat imagery, this dataset contains 35 land-cover subcategories, and provides 26 time steps, which is updated every five years before 2000 and annually after.
Researchers proposed a novel method by combining the advantages of continuous change detection algorithm, local adaptive modeling and Landsat satellite imagery, which helps to spot the landscape changes over time. Then, they used temporal-consistency optimization algorithms to classify the changing areas and make sure the data is accurate.
Researchers quantified GLC_FCS30D with over 84,000 validation samples from around the world, and found it to be impressively accurate, especially in identifying forests and croplands. They also compared it with other datasets, and found it to be consistent and reliable over time.
"The findings of the dataset showed that over the past 37 years, alterations in forest and cropland have been the primary drivers for global land-cover changes. There has been a significant decline of forest cover, totaling around 2.5 million square kilometers, while the expanses of cropland have surged, increased by roughly 1.3 million square kilometers,” said Prof. LIU.
