Application of mesoscale atmospheric models to Mars missions

Within the last decade, there has been a rapid increase in the number of mesoscale atmospheric models, originally developed for Earth, that have been adapted for application to the atmosphere of Mars. Aside from the numerous scientific investigations that can be addressed with these codes, the models have also become an integral and invaluable tool for predicting the atmospheric environment through which landed spacecraft must descend and then operate.

Unlike most scientific investigations where models are used to explore underlying processes or advance the physical understanding of dynamical systems, the application of mesoscale models for missions is focused almost exclusively on characterization of the state and variability of the atmosphere and characterization of the uncertainty of the models—a purely engineering approach. The atmospheric characterization output from models is used in two significant ways. The first is in landing site selection. The second is in engineering performance studies for entry, descent, and landing. Multiple, sometimes dozens, of potentially interesting landing sites are proposed for a mission based on science. These sites must be evaluated against expected engineering performance—a process that includes identification of atmospheric conditions that exceed design specifications of the system. During site selection activities for the Mars Exploration Rovers, numerous high science priority landing sites were rejected due to model predictions of high winds, or in some cases, due to large uncertainties in the model predicted wind fields. Other sites were retained because the predicted environment plus the uncertainty were found to be within engineering performance tolerance.

Inaccurate model predictions or failure to bound predictions with sufficient uncertainty can have a major, if not disastrous, impact on the mission. From a scientific standpoint, perfectly good and safe landing sites may be rejected based on erroneous model output indicating the contrary. Likewise, the model might indicate a site is within engineering design, when in reality it is not. In the former instance, there is a potential loss of science. In the latter situation, there is the potential loss of an entire mission, representing a huge economic loss (often greater than $1B), loss of all science, and possibly a program-hobbling public relations fiasco.

Providing model output with quantitative measures of uncertainty is particularly challenging at Mars where suitable data for validation of the models is meager or completely lacking, especially with regard to winds. Furthermore, initialization and time-dependent boundary condition information, such as the distribution of atmospheric dust, polar cap boundaries and properties, and subsurface regolith properties are poorly constrained. The standard practice to bridge the gap between these modeling challenges and the need to provide quantitative information has been to use multiple independent models and to adjust tunable parameters over a reasonable range of values so as to produce an ensemble of simulations that bound all probable realizations. Data assimilation techniques are just now making an appearance to help with improving model predictions and estimating uncertainty in global circulation models, but the use and impact of such techniques on mesoscale modeling has gone largely unexplored.

Specific examples from the Mars Exploration Rovers, Phoenix Scout Mission, and the upcoming Mars Science Laboratory missions will be presented to illustrate the basic approach to providing mission support, identify some of the mission-specific challenges, and frame a discussion of which advances in Mars meteorology and modeling are likely to yield the greatest benefit in support of future Mars missions.