Multiple-model based simulation of urban atmospheric methane concentration and the attributions to its seasonal variations: A case study in Hangzhou megacity, China

Abstract

Cities are treated as global methane (CH4) emission hotspots and the monitoring of atmospheric CH4 concentration in cities is necessary to evaluate anthropogenic CH4 emissions. However, the continuous and in-situ observation sites within cities are still sparsely distributed in the largest CH4 emitter as of China, and although obvious seasonal variations of atmospheric CH4 concentrations have been observed in cities worldwide, questions regarding the drivers for their temporal variations still have not been well addressed. Therefore, to quantify the contributions to seasonal variations of atmospheric CH4 concentrations, year-round CH4 concentration observations from 1st December 2020 to 30th November 2021 were conducted in Hangzhou megacity, China, and three models were chosen to simulate urban atmospheric CH4 concentration and partition its drivers including machine learning based Random Forest (RF) model, atmospheric transport processes based numerical model (WRF-STILT), and regression analysis based Multiple Linear Regression (MLR) model. The findings are as follows: (1) the atmospheric CH4 concentration showed obvious seasonal variations and were different with previous observations in other cities, the seasonality were 5.8 ppb, 21.1 ppb, and 50.1 ppb between spring-winter, summer-winter and autumn-winter, respectively, where the CH4 background contributed by −8.1 ppb, −44.6 ppb, and −1.0 ppb, respectively, and the CH4 enhancements contributed by 13.9 ppb, 65.7 ppb, and 51.1 ppb. (2) The RF model showed the highest accuracy in simulating CH4 concentrations, followed by MLR model and WRF-STILT model. (3) We further partition contributions from different factors, results showed the largest contribution was from temperature-induced increase in microbial process based CH4 emissions including waste treatment and wetland, which ranged from 38.1 to 76.3 ppb when comparing different seasons with winter. The second largest contribution was from seasonal boundary layer height (BLH) variations, which ranged from −13.4 to −6.3 ppb. And the temperature induced seasonal CH4 emission and enhancement variations were overwhelming BLH changes and other meteorological parameters.