Prediction of future Alaskan lake methane emissions using a small-lake model coupled to a regional climate model

Abstract

Methane emissions from lakes will increase with climate warming. However, these emissions are not accounted for in the land surface schemes of Global Climate Models (GCMs). Because climate projections depend on future atmospheric CH4 concentrations, the positive feedback loop between CH4 emissions from lakes and climate warming is not simulated. To address this issue, our objective was to develop a modelling approach where an arctic-lake CH4 emission model was forced directly with GCM output (no downscaling) and formulated with parameters that are generally available for lakes within a GCM framework. This will enable future lake-model to GCM coupling. The model was hindcast for 1976-2005 and forecast for 2071-2100. Using observed meteorological forcing, the hindcasts had a cold bias (-0.15 to -0.63 °C) and root-mean-square error (RMSE) of 0.38 to 0.90 °C, relative to observations. The GCM-forced hindcasts had a warm bias (+0.96 to +3.13°C) and RMSE of 1.03 to 3.50 °C. Our CH4 diffusion parameterization was transferable between four Alaskan lakes, after local adjustment of wind drag, but different ebullition parameterizations were required for two deeper lakes versus two shallower lakes. Under three climate scenarios, we simulated lake-bottom water to warm by up to 2.24°C, increasing the simulated CH4 surface flux by 38-129%. However, the limited availability of observed CH4 data renders our results poorly validated. Therefore, our model should be considered as a proof-of-concept pathway toward direct coupling of lake-models to GCMs. Rigorous validation would require additional timeseries observations of areal free-surface diffusive and ebullitive CH4 fluxes from lakes.