The impact of surface dust source exhaustion on the martian dust cycle, dust storms and interannual variability, as simulated by the MarsWRF General Circulation Model

Observations of albedo on Mars suggest a largely invariant long-term mean surface dust distribution, but also reveal variations on shorter (seasonal to annual) timescales, particularly associated with major dust storms. We study the impact of finite surface dust availability on the dust cycle in the MarsWRF General Circulation Model (GCM), which uses radiatively active dust with parameterized 'dust devil' and wind stress dust lifting to enable the spontaneous production of dust storms, and tracks budgets of dust lifting, deposition, and total surface dust inventory. We seek a self-consistent, long-term 'steady state' dust cycle for present day Mars, consisting of (a) a surface dust distribution that varies from year to year but is constant longer-term and in balance with current dust redistribution processes, and (b) a fixed set of dust lifting parameters that continue to produce major storms for this distribution of surface dust. We relax the GCM's surface dust inventory toward this steady state using an iterative process, in which dust lifting rate parameters are increased as progressively more surface sites are exhausted of dust. Late in the equilibration process, the GCM exhibits quasi-steady state behavior in which few new surface grid points are exhausted during a 60 year period with constant dust lifting parameters. Complex regional-scale dust redistribution occurs on time-scales from less than seasonal to decadal, and the GCM generates regional to global dust storms with many realistic features. These include merging regional storms, cross-equatorial storms, and the timing and location of several storm types, though very early major storms and large amounts of late storm activity are not reproduced. Surface dust availability in key onset and growth source regions appears vital for 'early' major storms, with replenishment of these regions required before another large storm can occur, whereas 'late' major storms appear primarily dependent on atmospheric variability. For the parameter space explored, no simulation achieves a steady state with continuing major storms lasting longer than 60 years when a constant wind stress lifting threshold is used. However, such a long-term steady state is achieved when a variable threshold is introduced, in which the threshold increases as dust is removed. This negative feedback on lifting slows it sufficiently for a balance to be produced between dust removal and re-deposition, even in key source regions for major storms. One concern is that the long-term surface dust distributions produced in these simulations show significant differences to the observed northern hemisphere albedo map, in particular predicting Tharsis and NE Arabia to be relatively dust-free. Although some observed high albedo regions may not have significant mobile dust, others likely have a dust cover several meters thick. The mismatches may reflect deficiencies in the GCM or the iterative process used, or the existence of ancient deep dust deposits formed during a past climate epoch.