Haifei Liu

Haifei Liu


Research Focus & Metrics

Academic Career Length

1Year (2026-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
18 November 2026 in Land
Verified
Journal Article
Open access
PDF file
RQRui Qin
Rui Qin
SCShi Cao
Shi Cao
HLHaifei Liu
Haifei Liu
XLXia Lu
Xia Lu
HWHuiChao Wu
HuiChao Wu
LLLuo Liu
Luo Liu
Annual monitoring of cropland abandonment supports targeted cropland management. Consistent event definitions, denominators and spatial support enable comparisons across regions and periods. This study aims to characterize annual land-cover transitions that serve as a proxy for new cropland abandonment in Yiyang, China, and to compare their spatial associations between the plain and hilly regions. Random forest classification provides annual 30 m Landsat land use and land cover (LULC) maps for 2010–2025. An operational three-year rule identifies cropland in the preceding year followed by forest or grassland in two consecutive years. Annual mapped event area is divided by preceding-year cropland area. Long-term grid responses average valid annual rates, and spatially blocked XGBoost–SHAP models retain 6302 grids and 19 predictors. During the period of 2012–2024, mean annual mapped event area is 104.60 km2, and the mean annual mapped event rate is 2.31%. Mean mapped event rates are 6.00% in the hilly region and 1.41% in the plain region, with respective peaks in 2015 and 2020. Citywide, plain-region and hilly-region holdout R2 values are 0.744, 0.672 and 0.589, and AUCs for relatively high-rate grids are 0.957, 0.913 and 0.880. Elevation accounts for 38.06% of citywide relative SHAP importance; mean cropland patch area and fragmentation index together account for 35.39% in the hilly region. A same-cohort 2021 comparison shows response-period sensitivity in predictor rankings. These mapped transitions highlight region-specific associations that can guide field checks; land-use records are needed to distinguish the cessation of cultivation from intentional woody-cover conversion.