Adjoint sensitivity analysis in an online weather forecast model: Developing the Adjoint of Black Carbon in WRFPLUS


Black carbon is a primary absorbing aerosol with an impact on climate that is potentially large, and still highly uncertain. Sparse in situ aerosol measurements and complex interactions between dynamics, microphysics, and radiation motivate development of models to predict climate impacts of black carbon sources, such as anthropogenic emissions and biomass burning. Regional modeling is a valuable means of determining the relative influences of emissions, wet and dry removal, and subgrid-scale vertical transport on aerosol distribution. In order to isolate the impact of each of these mechanisms, traditional sensitivity studies employ multiple forward model runs. In contrast, adjoint sensitivities of a scalar model response metric with respect to all parameters, such as spatially and temporally resolved emission rates and initial conditions, can be found with a single forward and reverse model run. Previous CTM adjoint sensitivity studies of black carbon use reanalysis meteorology, and thus decouple important feedbacks. Here we introduce recent developments in the inclusion of black carbon into the WRFPLUS adjoint and tangent linear models. Utilizing WRF online dynamics will enable sensitivity studies of regional weather phenomena to the presence of absorbing aerosol. This talk will present new adjoint sensitivity analysis of the relationship between the vertical distribution of black carbon and hourly anthropogenic and biomass burning emission inventories. These efforts lay the groundwork for the online 4D-Var assimilation of black carbon aerosol, and, eventually, a full suite of chemically speciated observations with the adjoint of WRF-Chem.

Aug 20, 2013 3:30 PM — 4:30 PM
Bechtel Collaboratory, Discovery Learning Center
Engineering Center, University of Colorado at Boulder, Boulder, CO 80309

University of Colorado Boulder