Ensemble Kalman filter data assimilation of Thermal Emission Spectrometer temperature retrievals into a Mars GCM

Abstract

Thermal Emission Spectrometer (TES) retrieved temperature profiles are assimilated into the GFDL Mars Global Climate Model (MGCM) using the Local Ensemble Transform Kalman Filter (LETKF) to produce synoptic maps of temperature, winds, and surface pressure and their uncertainties over the course of a Martian year. Short‐term (0.25 sol) forecasts compared to independent observations show reduced root mean square error (to 3–4 K global RMSE for a 30‐sol evaluation period during the northern hemisphere autumn) and bias compared to a free running model. Several enhanced techniques result in further performance gains. A 4D‐LETKF considers observations at their correct hour of occurrence rather than every 6 h. Spatially varying adaptive inflation and varying the dust distribution among ensemble members refine estimates of analysis uncertainty through the ensemble spread. Enhancing dust and water ice aerosol schemes and the application of empirical bias correction using time mean analysis increments help account for model biases. Full‐year experiments using prescribed dust opacities and observed TES dust opacities show that while realistic dust distributions are essential to match observed temperatures with a free run simulation, analyses from data assimilation are more robust with respect to imperfections in aerosol distribution. The data assimilation system described here is being used to generate a new reanalysis of Mars weather and climate, which will have many scientific and engineering applications.