TY - JOUR T1 - Ecosystem Service-Based Sensitivity Analyses of i-Tree Eco JF - Arboriculture & Urban Forestry (AUF) SP - 287 LP - 306 DO - 10.48044/jauf.2020.021 VL - 46 IS - 4 AU - Jian Lin AU - Charles N. Kroll AU - David J. Nowak Y1 - 2020/07/01 UR - http://auf.isa-arbor.com/content/46/4/287.abstract N2 - Trees are known to provide various ecosystem services and disservices to urban communities, which can be quantified using models based on field and environmental data. It is often uncertain how tree structure and environmental variables impact model output. Here we perform a sensitivity analysis (SA) of i-Tree Eco, a common urban forest model, to analyze the relative impact of different model inputs on three module outputs: biogenic volatile organic compound (BVOC)(isoprene and monoterpenes) emissions, carbon storage and sequestration, and dry deposition of nitrogen dioxide, sulfur dioxide, and ozone. The SA methods included novel applications of the Morris one-at-a-time method and a variance-based decomposition method, which integrates Monte Carlo simulation with Latin hypercube sampling and Iman Conover analysis. A case study was performed in New York City, New York, USA, with field plot data collected in 2013. Genus has the largest influence on BVOC emissions by determining base emission rates and its high interactions with other input factors, and BVOC emissions are sensitive to leaf biomass in a concave manner and temperature in a convex manner, while isoprene emissions show a strong linear relationship with photosynthetically active radiation (PAR). Diameter at breast height plays the most important role for both carbon storage and sequestration estimators; crown light exposure and tree condition are also important for carbon sequestration. Dry deposition velocity is sensitive to leaf area index and relative humidity in a nearly linear way, while sensitive to temperature and PAR in a concave manner. The results provide guidance to facilitate future field plot campaigns and model development. The knowledge revealed by the SA is also beneficial for model uncertainty reduction, which in turn facilitates more effective urban forest management and decision-making. ER -