Analysis of the controlling factors of dust low frequency variability

Mineral dust is interacting with most components of the Earth's system affecting climate from different processes. Based on published studies, transport models seem to represent satisfactorily the daily and seasonal variability of dust, while failing reproducing low frequency variability. To understand the underlying causes, we examined seven CMIP5 models, which have included interactive dust emission against satellite derived dust optical depth. We found that multi-model mean largely captures global spatial pattern and zonal mean of Dust Optical Depth (DOD) for present-day climatology. But inter-annual variations of DOD are neither captured by most of models nor by multi-model mean, as models do not capture the observed connections between DOD and local controlling factors such as surface wind speed, bareness, precipitation, and landuse.

The implementation of a dust emission within the dynamic land model of the NOAA Geophysical Fluid Dynamics Laboratory (GFDL) climate model is used to better represent key controlling factors of dust emission: surface winds, soil moisture, frozen or snow cover soil, vegetation and landuse changes. The results are analyzed over two major arid regions producing large amount of dust (West Africa and Australia) to contrast their differences and similarities in term of emission. We will show the importance of land use changes to explain dust low frequency variability, and how it may affect climate through dust interaction with radiation.