Datasets and code for: Carbon sink conservation: Cost-effective spatial priorities and feasible management policies for China

This repository contains the datasets and code associated with the paper "Carbon sink conservation: Cost-effective spatial priorities and feasible management policies for China." 

Suggested Citation:Liu, J., Zhang, M., Xia, Y., Wu, L., & Chen, C. (2026) Carbon sink conservation: Cost-effective spatial priorities and feasible management policies for China. Land Use Policy, 165, 107974. https://doi.org/10.1016/j.landusepol.2026.107974.

All spatial datasets are GeoTIFF (.tif) files at a 5-km spatial resolution covering China’s land area.

1. Carbon_sink_2001-2015

Historical carbon sink capacity, indicated by Net Biome Productivity (NBP) in gC m⁻² d⁻¹, from 2001–2015, based on an ensemble of eight widely recognized datasets.

2. Carbon_sink_projection_SSPs_2020-2100

Projected carbon sink capacity (gC m-2 d-1) under different Shared Socioeconomic Pathway (SSP) scenarios for the period 2020–2100 (ten-year intervals), based on an ensemble mean of XGBoost and LightGBM model outputs.

3. Carbon_sink_priority

Cost-effective spatial priority levels for carbon sink conservation under different SSP scenarios and future periods, corresponding to Fig. 5 and Fig. 6a of the paper.

4. Carbon_sink-biodiversity_synergy

Scenario- and period-mean priorities for carbon sink conservation (Fig. 6a), priorities for biodiversity conservation, and their spatial synergy and mismatch (Fig. 7c).

5. Code_for_machine_learning

Python scripts for training machine learning models (ANN, RF, XGBoost, and LightGBM) and for generating predictions for carbon sink capacity. The code is provided in Jupyter Notebook, developed in Python 3.8 and executed on Ubuntu 20.04 with CUDA 11.8 support.

For additional details regarding the datasets, please refer to the main article and its supplementary materials. Questions may be directed to Dr. Jingyi Liu ([e-mail]).