The Water Vapor Budget In The Tropical UTLS: Inferences From Intercomparisons Between Large-Scale Model Simulations and In Situ Observations Of Water Vapor At San José, Costa Rica

Ice microphysical processes play a dominant role in the control of the mixing ratio of water vapor at the tropical tropopause and its subsequent entry into the global lower stratosphere. Given the sensitivity of the climate system to cloudiness and water vapor in the upper tropical troposphere and lower stratosphere (UTLS), climate models must be able to simulate both the mean levels and variability of water vapor and relative humidity in this region of the atmosphere. In this context, a particular modeling challenge for atmospheric general circulation models (AGCMs) is to reproduce the super-saturation frequently observed in the cold tropical upper troposphere.  Recently, a two-moment microphysical scheme has been implemented in the NASA GEOS-5 AGCM, and this holds out the promise of a more physically-based representation of clouds and water vapor, including loosening of the model’s limits on supersaturation.  Here we address the model’s performance by comparing it against Ticosonde in situ frostpoint balloon observations made at San José, Costa Rica [10°N, 84°W].  Among in situ data records, Ticosonde stands out by virtue of its more than eleven years of nearly monthly launches as well as four campaigns of intensive observations.  Three of the latter were carried out in concert with NASA research aircraft investigations.  Previous comparisons between Ticosonde data and the MERRA-2 reanalysis have shown that MERRA-2 water vapor is biased high in the upper troposphere (up to 70% at 177 hPa), although this bias virtually disappears at 100 hPa.  This as well as other lines of evidence suggest that the high bias is due in part to inadequacies in the single-moment microphysics.  To examine this hypothesis and to assess the improvements afforded by the two-moment scheme, we conduct two types of simulations with GEOS-5. The first is a set of free-running simulations over the course of the eleven-year Ticosonde data period. We run with both the single-moment microphysical scheme employed in the MERRA-2 reanalysis runs and the new two-moment scheme.  The focus here is to assess the influence of the more detailed microphysical scheme on the model’s relative humidity in the UTLS over the long term.  To investigate higher-frequency variability of relative humidity in the model, we use MERRA-2 Replay simulations of GEOS-5 during the Ticosonde intensive observation periods in early 2006 and the summer of 2007.  (These correspond to the NASA CR-AVE and TC4 airborne campaigns.)  Our intercomparisons between the model output and the in situ frostpoint data shed light on the relative roles of microphysical and cloud processes simulated in the model and provide an important assessment of the ability of the GEOS-5 and similar large-scale models to simulate the influence of UTLS water vapor in our evolving climate system.