Predicting Martian Dune Characteristics Using Global and Mesoscale MarsWRF Output
Claire Newmanworking with
Nick Lancaster, Dave Rubin and Mark Richardson
Acknowledgements: funding from NASA’s MFR program, and use of NASA’s HEC facility right here at Ames.
Overview of talk
• Motivation (why dunes?)
• Dune features we can compare with
• Some dune theory
• Modeling approach
• Preliminary results: Global
• Preliminary results: Gale Crater
Motivation: how do dunes provide insight into recent climate change on Mars?
• Higher obliquity should produce stronger circulations & surface stresses, hence might expect more saltation and dune formation
• Features (i.e., dune orientations, migration directions, etc.) in disagreement with predictions for current wind regime may indicate inactive dunes formed in past orbital epochs
Also –
• In absence of near-surface meteorological monitoring, predicting characteristics of active dunes can help us confirm that we – have the current wind regime right – understand dune formation processes
Dune features we can compare with
Dune features we can compare with• Locations
Bourke and Goudie, 2009
• Locations• Bedform (crest) orientations
Images: NASA/JPL/University of Arizona
Dune features we can compare with
• Locations• Bedform (crest) orientations
Images: NASA/JPL/University of Arizona
Dune features we can compare with
• Locations• Bedform (crest) orientations• Inferred migration directions (crater dunes)
Hayward et al., 2009
Dune features we can compare with
• Locations– Use a numerical model to predict saltation over a Mars
year, assuming a range of saltation thresholds
• Bedform (crest) orientations– Apply ‘Gross Bedform-Normal Transport’ theory
• Inferred migration directions (crater dunes)– Assume correlated with resultant (net) transport direction
Dune features we can compare with
Some dune theory
2 key issues
• Dunes form in the long-term wind field
• Dune orientations are not determined by net transport
Gross Bedform-Normal Transport
A simple example to illustrate a point
Dune crest
Rubin and Hunter, 1987
A simple example to illustrate a point
Dune crest
First wind direction
Gross Bedform-Normal Transport
A simple example to illustrate a point
Dune crest
First wind direction
Second wind direction
Gross Bedform-Normal Transport
A simple example to illustrate a point
Net transport = 0
Dune crest
First wind direction
Second wind direction
Gross Bedform-Normal Transport
A simple example to illustrate a point
Net transport = 0
But both wind directions shown cause the bedform to build
Dune crest
First wind direction
Second wind direction
Gross Bedform-Normal Transport
A simple example to illustrate a point
Net transport = 0
But both wind directions shown cause the bedform to build
=> rather than net transport, we are interested in gross transport perpendicular to the bedform, regardless of the ‘sense’ of the wind (i.e., N-S versus S-N)
Dune crest
First wind direction
Second wind direction
Gross Bedform-Normal Transport
Bedform orientation – the theory
A = NET TRANSPORT OF SAND
A
Bedform orientation – the theory
Bedform orientation – the theory
B + C = GROSS BEDFORM-NORMAL SAND TRANSPORT
B
C
Bedform orientation – the theory
• Basic concept: dunes form due to sand transport in both directions across bedform
• Bedforms align such that total transport across dune crest is maximum in a given wind field
• where total transport = Gross Bedform-Normal Transport [Rubin and Hunter, 1987]
B + C = GROSS BEDFORM-NORMAL SAND TRANSPORT
B
C
Bedform orientation – the theory
Modeling approach (1)
• Run numerical model to predict near-surface winds at all times for a long time period (at least 1yr) to capture the long-term dune-forming wind field
• Choose a saltation threshold and calculate sand fluxes in all directions [0, 1, …359° from N]
• Consider all possible bedform orientations [0, 1, …179° from N]
• Sum the gross sand flux perpendicular to each orientation over the entire time period
• Find the orientation for which the total gross flux is maximum
• NB: secondary maxima indicate secondary bedform orientations
Bedform orientation – the approach
The numerical model: MarsWRF
• Mars version of planetWRF (available at www.planetwrf.com)
• Developed from Weather Research and Forecasting [WRF] model widely used for terrestrial meteorology
• Multi-scale 3D model capable of:– Large Eddy Simulations– Standalone mesoscale– Global– Global with nesting
• Using global and global with nesting for these studies
Version of MarsWRF used here includes:
• Seasonal and diurnal cycle of solar heating, using correlated-k radiative transfer scheme (provides good fit to results produced using line-by-line code)
• CO2 cycle (condensation and sublimation)
• Vertical mixing of heat, dust and momentum according to atmospheric stability
• Sub-surface diffusion of heat
• Prescribed seasonally-varying atmospheric dust (to mimic e.g. a dust storm year or a year with no major storms) or fully interactive dust (with parameterized dust injection)
• Ability to place high-resolution nests over regions of interest
Results
Present day global dune results comparing with Mars Global Digital Dune Database (MGD3, e.g. Hayward et al., 2009):
1. Dune centroid azimuth
– compare with GCM-predicted resultant transport direction
Present day global dune results comparing with Mars Global Digital Dune Database (MGD3, e.g. Hayward et al., 2009):
1. Dune centroid azimuth
– compare with GCM-predicted resultant transport direction
2. Slipface orientation
– compare with normal to GCM- predicted bedform orientation
Present day global dune results comparing with Mars Global Digital Dune Database (MGD3, e.g. Hayward et al., 2009):
1. Dune centroid azimuth
– compare with GCM-predicted resultant transport direction –agreement => within 45° of dune centroid azimuth direction
2. Slipface orientation
– compare with normal to GCM-predicted bedform orientation – agreement => within 45° of normal to slipface
Comparison of predicted and inferred migration direction for present day using saltation threshold=0
Green => agreement between predicted and inferred migration directionBlue => no agreementRed => no comparison possible
Comparison of predicted and inferred migration direction for present day using saltation threshold=0.007N/m2
Green => agreement between predicted and inferred migration directionBlue => no agreementRed => no comparison possible
Comparison of predicted and inferred migration direction for present day using saltation threshold=0.021N/m2
Green => agreement between predicted and inferred migration directionBlue => no agreementRed => no comparison possible
Comparison of predicted and measured bedform orientation direction for present day using saltation threshold=0
Green => agreement between predicted and inferred bedform orientationBlue => no agreementRed => no comparison possible
Comparison of predicted and measured bedform orientation direction for present day using saltation threshold=0.007N/m2
Green => agreement between predicted and inferred bedform orientationBlue => no agreementRed => no comparison possible
Green => agreement between predicted and inferred bedform orientationBlue => no agreementRed => no comparison possible
Comparison of predicted and measured bedform orientation direction for present day using saltation threshold=0.021N/m2
Comparison of predicted and inferred migration direction for obliquity 35° using saltation threshold=0.007N/m2
Green => agreement between predicted and inferred migration directionBlue => no agreementRed => no comparison possible
Comparison of predicted and inferred migration direction for present day using saltation threshold=0.007N/m2
Green => agreement between predicted and inferred migration directionBlue => no agreementRed => no comparison possible
Modeling approach (2)
One or more high-resolution ‘nests’, placed only over regions in which increased resolution is desired
Nesting in MarsWRF
20°E 25°E 30°E 35°E
5°N
5°S
0°
15°E
Mother [global] domain (only a portion shown)
Domain 2(5 x resolution of domain 1)
Domain 3(5 x resolution
of domain 2; 25 x resolution
of domain 1)
20°E 25°E 30°E 35°E
5°N
5°S
0°
15°E
Mother [global] domain
Domain 2
Domain 3
2-way nesting => feedbacks between domains
1-way nesting => parent forces child only
Sample results for Gale at Ls ~ 0° using mesoscale nesting in global MarsWRF
Gale dune studies – early results (~4km resolution)
Gale Crater: predicted
(a) resultant transport direction [black arrows]
(b) dune orientations [white lines]
for saltation threshold = 0
Gale dune studies – early results (~4km resolution)
Gale Crater: predicted
(a) resultant transport direction [black arrows]
(b) dune orientations [white lines]
for saltation threshold = 0.007N/m2
Gale dune studies – early results (~4km resolution)
Gale Crater: predicted
(a) resultant transport direction [black arrows]
(b) dune orientations [white lines]
for saltation threshold = 0.021N/m2
Gale dune studies – present day MarsNo saltation threshold Saltation threshold = 0.007N/m2
No saltation threshold Saltation threshold = 0.007N/m2
Gale dune studies – present day Mars
No saltation threshold Saltation threshold = 0.007N/m2
Gale dune studies – present day Mars
From Hobbs et al., 2010
Gale dune studies – present day Mars
• Simply changing the assumed saltation threshold greatly improves the match to observed bedform orientations
• Effect of large dust storms not yet examined, but likely will also impact winds hence orientations
• Orbital changes and impact on circulation widen parameter space even further!
Gale dune studies – obliquity 35°
Gale Crater: predicted
(a) resultant transport direction [black arrows]
(b) dune orientations [white lines]
for saltation threshold = 0
Gale dune studies – obliquity 35°
Gale Crater: predicted
(a) resultant transport direction [black arrows]
(b) dune orientations [white lines]
for saltation threshold = 0.007N/m2
Gale dune studies – obliquity 35°
Gale Crater: predicted
(a) resultant transport direction [black arrows]
(b) dune orientations [white lines]
for saltation threshold = 0.021N/m2
Gale dune studies – obliquity 35°No saltation threshold Threshold = 0.007N/m2 Threshold = 0.021N/m2
Conclusions (1)
• Bedform orientations and dune migration directions may be related (roughly) to ‘gross’ and ‘net’ sand transport by long-term (~multi-annual) wind regime
• Typical global model resolutions (2°-5°) are too low to resolve topographic features influencing dunes
• Mesoscale numerical modeling spanning entire year can be used to predict long-term wind regime and sand transport
• Rubin and Hunter [1987] Gross Bedform-Normal Transport theory provides clean predictions of bedform orientations
Conclusions (2)
• Recent studies (e.g. Bridges et al., Nature, in press) indicate currently active dune fields that can be used to validate our approach for present day Mars
• Possibly inactive regions with strong discrepancies not explainable by varying saltation threshold, dust loading, etc. may imply formation during past orbital epoch
• Future study regions include Proctor Crater, a North Polar dune field, …
• Lots more to do!