Empirical nitrogen and sulfur critical loads of U.S. tree species and their uncertainties with machine learning

• DOI: 10.1016/j.scitotenv.2022.159252
• Publication Date: 2022-10-08T01:19:58Z
• Authors: Nathan R. Pavlovic; Shih Ying Chang; Jiaoyan Huang; Kenneth Craig; Christopher Clark; Kevin Horn; Charles T. Driscoll
• Keywords: Tree (set theory); CLs upper limits; Deposition

Critical loads (CLs) of atmospheric deposition for nitrogen (N) and sulfur (S) are used to support decision making related air regulation land management. Frequently, CLs calculated using empirical methods, the certainty results depends on accurate representation underlying ecological processes. Machine learning (ML) models perform well in modeling processes with non-linear characteristics significant variable interactions. We bootstrap ensemble ML methods develop CL estimates assess uncertainties growth survival 108 tree species conterminous United States. trained predict characterize relationship between response. Using four statistical we quantified uncertainty 95 % confidence intervals (CI). At lower bound estimate, 80 or more have been impacted by exceeding a over >50 range, while at upper percentage is much (<20 across >60 range). Our analysis shows that can be effectively quantify critical their uncertainties. The range sufficiently large warrant consideration management regulatory respect deposition.