Future risks of cyanobacterial blooms in lakes unveiled by open access data and integrated machine learning models

Future projections of cyanobacterial blooms is of great significance for ecological risks management in lakes under the combined influences of human activities and climate change. However, due to the lack of long-term data on cyanobacterial blooms and the uncertainty of future prediction models, our current understanding of continental-scale cyanobacterial blooms is still insufficient. This study developed an improved Commission Internationale de l´Eclairage (CIE)-based algorithm with multi-mask correction for continental-scale automated detection of cyanobacterial blooms in 185 Chinese lakes from 1984 to 2021 by using 30 m resolution Landsat satellite images, and the future projections of cyanobacterial blooms from 2021 to 2100 were achieved by an integrated machine learning prediction model based on Bayesian Model Averaging (BMA). We found that the high values of the maximum area proportion of cyanobacterial blooms (CB_maxA) were predominantly concentrated in eastern China, and the trajectory of geographic center moved gradually southward and eastward from 1980s to 2010s. Over the past four decades, the CB_maxA exhibited a notable upward trend, with an average annual increase of 0.11%. Climate (42.0%) and human activity (32.3%) variables are primary drivers nationwide, with a 28.6% interactive contribution. In particular, temperature exerts significant positive influences on CB_maxA at the national scale. Under future climate scenarios, the CB_maxA is projected to increase 0.34% per decade, notably in the southern China where with severe cyanobacterial blooms. This study highlights the profound impact of temperature on future cyanobacterial blooms at the national scale and the significant interaction between climate change and human activities. Our approach demonstrates strong applicability and transferability, characterized by accessible input data and robust model, which is crucial for lake ecological risks prevention under changing environments.