Network Design for a Cost-Effective Atmospheric Methane Measurement Network over India

Studies have shown that the uncertainty of methane emission estimates over India is as high as 40-60%, largely due to the lack of observations. In India, measurements are limited to a few locations, with the majority of them being flask measurement stations. The observational constraint of the measurements could be greatly improved with the development of a network of continuous measurement stations at well-chosen locations. For this study, we have designed an atmospheric methane measurement network for India using transport modeling techniques and a scaling-factor-based inversion approach. A network optimization algorithm selects the combination of observation locations that gives the most uncertainty reduction in the estimates of posterior methane emission fluxes over India. The backbone of this study is a simple analytical inversion setup that utilizes the STILT (Stochastic Time Inverted Lagrangian Transport) model, a sectorial emission model based on EDGAR, as well as fluxes from wetlands and biomass burning. The state space of the inversion consists of monthly emissions, separated by sector, aggregated spatially to the level of political states.

The challenge in network design is to formulate an appropriate target quantity, which the network will be optimized to constrain. Using the annual total emissions as the single target results in a network that will optimally constrain the largest sources, irrespective of their spatial location or the seasonality of the source. Thus, we also included other targets, such as political-state-level emissions, sectoral emissions, and seasonality. For the study, we used a base network of existing stations (“base”) and added further stations from a candidate set (“extended”) on the basis of the incremental uncertainty reduction they provide. We found that more measurement stations along the Indo-Gangetic Plains and North-Eastern India are required. An optimized network was also designed from scratch using the same strategy and it was found to yield similar uncertainty reduction compared to the “base” + “extended” network despite having fewer stations. The effectiveness of the optimal network and the base network in reducing the uncertainties of the different emission categories is assessed and discussed.