AGU2012-GC31A963: Model Estimates of Pan-Arctic Lake and Wetland Methane Emissions X.Chen 1, T.J.Bohn 1, M. Glagolev 2, S.Maksyutov 3, and D. P. Lettenmaier 1 1 University of Washington, Seattle, Washington, USA ; 2 Moscow State University, Moscow, Russia; 3 National Institute for Environmental Studies, Tsukuba, Japan San Francisco, December 5, 2012 Contact: 3. Model Calibration – over West Siberia 5. Simulated Atmospheric Concentrations over W. Siberia Abstract Lakes and wetlands are important sources of the greenhouse gases CO2 and CH4, whose emission rates are sensitive to climate. The northern high latitudes, which are especially susceptible to climate change, contain about 50% of the world’s lakes and wetlands. With the predicted changes in the regional climate for this area within the next century, there is concern about a possible positive feedback resulting from greenhouse gas emissions (especially of methane) from the region’s wetlands and lakes. To study the climate response to emissions from northern hemisphere lakes and wetlands, we have coupled a large-scale hydrology and carbon cycling model (University of Washington’s Variable Infiltration Capacity model; VIC) with the atmospheric chemistry and transport model (CTM) of Japan’s National Institute for Environmental Studies and have applied this modeling framework over the Pan-Arctic region. In particular, the VIC model simulates the land surface hydrology and carbon cycling across a dynamic lake-wetland continuum. The model includes a distributed wetland water table that accounts for microtopography and simulates variations in inundated area that are calibrated to match a passive microwave based inundation product. Per-unit-area carbon uptake and methane emissions have been calibrated using extensive in situ observations. In this paper, the atmospheric methane concentrations from a coupled run of VIC and CTM are calibrated and verified for the Pan-Arctic region with satellite observations from Aqua’s Atmospheric Infrared Sounder (AIRS) and Envisat’s Scanning Imaging Absorption Spectrometer for Atmospheric Cartography (SCIAMACHY) instruments. We examine relative emissions from lakes and wetlands, as well as their net greenhouse warming potential, over the last half-century across the Pan-Arctic domain. We also assess relative uncertainties in emissions from each of the sources. 4. Wetland Methane Emissions over West Siberia 2. Modeling Framework VIC’s dynamic lake-wetland continuum 3. Linkages among models 1.Simulate land surface hydrology and carbon cycle using Variable Infiltration Capacity (VIC) model 2.VIC’s dynamic lake-wetland model impounds surface water, allowing lakes to expand and flood wetlands; within exposed wetlands, water table is distributed according to microtopography 3.Distributed water table is input to Walter-Heimann (2000) wetland methane emissions model; total grid cell methane fluxes are input to NIES Atmospheric Chemistry Tracer Transport Model (CTM) Simulated emissions computed in section 4 (above) are input to the CTM model As an example, simulations using 3 different lake CH4 emission rates are shown below AIRS satellite [CH4] observations are sensitive primarily to middle-upper troposphere [CH4] To compare with AIRS, our simulated [CH4] has been convolved with the AIRS sensitivity kernel Resulting mid-upper-troposphere [CH4]: CTM, Lake CH4 = 10 mg CH4/m2/day CTM, Lake CH4 = 250 mg CH4/m2/day CTM, Lake CH4 = 500 mg CH4/m2/day AIRS [CH4], JJA Because CTM is poorly constrained at these altitudes, and because CTM’s resolution is so coarse (2.5 degree), it is difficult for us to reproduce the AIRS spatial pattern in upper troposphere with the CTM simulations. Is there a way to focus on near-surface [CH4]? 1. Research Domain Lakes and wetlands are the world’s largest natural source of CH4, a powerful greenhouse gas Their CH4 emissions are sensitive to climate factors such as temperature and precipitation Nearly 50% of the world’s lakes and wetlands occur in the northern high latitudes, North of 45 N The high latitudes have experienced pronounced climate change over the last half-century, and are projected to continue doing so through the next century There is concern that future changes in lake and wetland CH4 emissions could produce a substantial feedback to climate change Uncertainties are large, due to sparse observations Can we use a combination of remote sensing and process-based models to monitor CH4 emissions and reduce uncertainty? Global Lake and Wetland Database (GLWD), Lehner and Döll (2004) Glagolev et al. (2011): 3.9 +/- 1.4 Tg CH4/y Our Estimate: 3.6 Tg CH4/y ( Tg CH4/y) Annual average methane emissions, Our modeling framework matches both the total magnitude and approximate spatial distribution of the estimate of Glagolev et al (2011). Inundated extent calibrated to match the daily AMSR-E-based (passive microwave) product of Schroeder et al (2010) Saturated extent calibrated to match classifications of ALOS/PALSAR (synthetic aperture radar) imagery (courtesy of NASA/JPL) CH4 fluxes calibrated to match in situ observations of Glagolev et al (2011) Northern High Latitudes: 45N-80N, 180W-180E 6. Surface Methane Concentration 7. Conclusions SCIAMACHY kernel: sensitive to whole atmospheric methane concentration, but most sensitive to near surface concentration AIRS kernel: sensitive to upper layer atmospheric methane concentration SCIA_kernel Observation Difference of two kernels on model output shows correctly the high emission areas at West Siberia and northern Canada, as expected The difference of observations (SCIAMACHY and AIRS) reflects the high-emission areas (north America, central Eurasia), but does not match the model output well. WHY? 1. VIC model successfully reconstructed lake and wetlands topography, and gave reasonable variation of water tables over the whole Pan-Arctic region  Daily variation of atmospheric methane concentration. Two satellite observations are at different times of the day  Inaccuracy of satellite observation data compared to actual atmospheric methane concentration. Lower resolution of AIRS observation would make the difference more biased. Possible solution: downscaling of AIRS observation data 2. CTM model reproduced the lower atmospheric methane concentration patterns well. Coupling of hydrological model (VIC) and atmospheric model (CTM) could show atmospheric methane concentration well where hydrological cycle and carbon cycle are highly integrated 3. Future work: 1) Analysis of satellite observation of methane, finding a better way of deriving near surface methane concentration from observations; 2) Conduct CTM runs using surface methane emissions from all possible source from VIC output (i.e. West Siberia and northern Canada) Note: for current simulation, wetlands methane emission is mainly from West Siberia.  Lack of emission from northern Canada. AIRS_kernel DIFFERENCE Simulation