打印

作者

Jian Lin;Charles N. Kroll;David J. Nowak

来源

URBAN FORESTRY & URBAN GREENING,Vol.60,Issue1,Article 127062

语言

英文

关键字

Bootstrap;Ecosystem services;Model uncertainty;Monte carlo;Urban forestry

作者单位

Graduate Program in Environmental Science, State University of New York College of Environmental Science and Forestry, Syracuse, NY, 13210, USA;Department of Environmental Resources Engineering, State University of New York College of Environmental Science and Forestry, Syracuse, NY, 13210, USA;USDA Forest Service, Northern Research Station, Syracuse, NY, 13210, USA;Graduate Program in Environmental Science, State University of New York College of Environmental Science and Forestry, Syracuse, NY, 13210, USA;Department of Environmental Resources Engineering, State University of New York College of Environmental Science and Forestry, Syracuse, NY, 13210, USA;USDA Forest Service, Northern Research Station, Syracuse, NY, 13210, USA

摘要

Uncertainty information associated with urban forest models are beneficial for model transparency, model development, effective communication of model output, and decision-making. However, compared with the extensive studies based on the applications of urban forest models, little attention has been paid to the uncertainty of the output from these models. In this study, bootstrap and Monte Carlo simulation were employed to explore the uncertainty of i-Tree Eco. We assess the uncertainties associated with input data, sampling methods and models throughout the processes of urban forest structure and function quantification, and we propagate and aggregate the three sources of uncertainty to derive an estimator of total uncertainty. The uncertainty magnitude is expressed as the coefficient of variation. By applying the uncertainty framework to a network of 15 cities across the United States, we find that the average magnitude of total uncertainty across 15 cities is 12.3 % for leaf area, 13.4 % for carbon storage, 11.1 % for carbon sequestration, 40.7 % for isoprene emissions, and 25.0 % for monoterpene emissions. For leaf and carbon estimators, the total uncertainty is primarily driven by sampling uncertainty; the magnitudes of all three sources of uncertainty are comparable across 15 cities. In contrast, input, sampling, and model uncertainties all contribute to the total uncertainty for isoprene and monoterpene emission estimators, and there are large variations in these three sources of uncertainty across the 15 cities. An analysis of a regression-based approach to estimate input and model error indicated only moderate improvements over using averages across sites when estimating total uncertainty.