TITLE:
Comparing Eight Computing Algorithms and Four Consensus Methods to Analyze Relationship between Land Use Pattern and Driving Forces
AUTHORS:
Xinliang Liu, Yi Wang, Yong Li, Feng Liu, Jianlin Shen, Liang Ou, Juan Wang, Runlin Xiao, Jinshui Wu
KEYWORDS:
Land Use Pattern, Spatial Scales, Consensus Methods, Computing Algorithms
JOURNAL NAME:
International Journal of Geosciences,
Vol.10 No.1,
January
24,
2019
ABSTRACT: Although many computing algorithms have been developed to analyze the relationship between land use pattern and driving forces (RLPDF), little has been done to assess and reduce the uncertainty of predictions. In this study, we investigated RLPDF based on 1990, 2005 and 2012 datasets at two spatial scales using eight state-of-the-art single computing algorithms and four consensus methods in Jinjing rive catchment in Hunan Province, China. At the entire catchment scale, the mean AUC values were between 0.715 (ANN) and 0.948 (RF) for the single-algorithms, and from 0.764 to 0.962 for the consensus methods. At the subcatchment scale, the mean AUC values between 0.624 (CTA) and 0.972 (RF) for the single-algorithms, and from 0.758 to 0.979 for the consensus
methods. At the subcatchment scale, the mean AUC values were between 0.624 (CTA) and 0.972 (RF) for the single-algorithms, and from 0.758 to 0.979 for the consensus methods. The result suggested that among the eight single computing algorithms, RF performed the best overall for woodland and paddy field; consensus method showed higher predictive performance for woodland and paddy field models than the single computing algorithms. We compared the simulation results of the best - and worst-performing algorithms for the entire catchment in 2012, and found that approximately 72.5% of woodland and 72.4% of paddy field had probabilities of occurrence of less than 0.1, and 3.6% of woodland and 14.5% of paddy field had probabilities of occurrence of more than 0.5. In other words, the simulation errors associated with using different computing algorithms can be up to 14.5% if a probability level of 0.5 is set as the threshold. The
results of this study showed that the choice of modeling approaches can greatly affect the accuracy of RLPDF prediction. The computing algorithms for specific RLPDF tasks in specific regions have to be localized and optimized.