Study of the hydrogen escape rate at Mars during martian years 28 and 29 from comparisons between SPICAM/Mars express observations and GCM-LMD simulations

We have simulated the 3D atomic hydrogen density in the Martian upper atmosphere and associated Jeans escape rate during Martian years 28 and 29. The coronal Lyman-alpha brightness is computed using a 3D radiative transfer model which accounts for the monthly average hydrogen density for these two years and is compared to a large set of observations by Mars Express/SPICAM. The simulated brightness is generally in good agreement with the observations for Ls <230 degrees and Ls >330 degrees for Martian year 28 and Ls < 270 degrees, Ls > 340 degrees for Martian year 29, but the model strongly underestimated the brightness for 230 < Ls < 330 degrees for Martian year 28 and 270 < Ls < 340 degrees for Martian year 29. In these simulations the transport of water vapor contributes to the production of hydrogen at high altitudes during southern summer. A possible explanation for the model discrepancy is an underestimate of this water transport, associated with an underestimate of the hygmpause altitude and/or an underestimate of the supersaturation of the mesosphere. Considering this discrepancy, we estimate the hydrogen escape rate during these two Martian years to vary by almost two orders of magnitude, between similar to 10(25) to 6 x 10(26) s(-1) (equivalent to a global layer of water similar to 33 to 2000 mm deep every billion years), in agreement with the seasonal variations estimated directly from the fit of the SPICAM observations during the Martian year 28 by Chaffin et al. (2014). Our analysis suggests that episodic dust storms and associated enhancements at high altitude near perihelion are a major factor in the H escape estimates averaged over one martian year or longer periods, but the accumulated water lost at this rate for 4 billion years is much lower than the amount of water needed to form the flow channels observed on Mars.