| Dust events exhibit significant influence on all aspects of human life,health,global climate and marine ecosystems by reducing the atmospheric visibility,changing the radiation budget balance and transporting organic matter.The prevention and management of dust storm disasters have always been the hot topic concerned by related countries’ government.The detection and identification of dust aerosol based on satellite images make it possible to synchronously monitor dust evolution on a global scale,which also provide the technical basis for dust related research.The definition of thresholds in the threshold-based dust detection algorithms always change with the variability of study area and periods,which make the process of dust detection more complex.Thus,a more reliable dust detection algorithm without definition of thresholds is needed.Analyzing the influence mechanisms of the multi-scale change of dust storm offered the theoretical basis for the forecast of dust weather and measures on disaster prevention and alleviation.Numerous factors have been pointed out to influence the dust storm,however,their relative importance to dust variation in recent 40 years over the global dust belt were unclear.In the context of global warming,the effect of ENSO on the anomalies of global atmospheric circulation,temperature,and precipitation experienced an interdecadal transition,while its interdecadal impact on dust activities as well as their influence mechanism have not been well documented.Thus,based on the remote sensing and reanalysis data,this study proposed an SVM-based dust detection algorithm at first,then analyzed the relative importance of the influence factors on the interannual change of dust emission over each dust source in the global dust belt,and next,further investigated the interdecadal change in the relationship between dust activities and ENSO as well as their influence mechanism.The main contents and findings are as follows:(1)Comparing the clustering characteristic of the combination of visible and thermal infrared bands in six different objects,including dust storms,desert,gobi,green land,thin cloud,and thick cloud,to get the optimal band combination that can distinguish dust aerosols from other features,and then make this band combination as the feature vectors of SVM to identify the dust pixels.Results indicated that the optimal band combination was(B7-B3)/(B7+B3),B20-B31,and B31/B32.The proposed SVM-based dust detection algorithm was higher in accuracy compared with the traditional methods.On the other hand,it avoided the uncertainty and complexity that was generated by repeatedly defining the thresholds in the traditional methods.In addition,the SVM-based method was of relatively high generalization ability that can be applied to different cases.(2)Analyzing the primary drivers that influence the interannual change of dust emission(DUEM)over dust sources in the dust belt in recent 40 years based on the MERRA-2 dust emission data and Lindeman-Merenda-Gold(LMG)method.Comparing the abovementioned primary drivers and corresponding DUEM under two different warming phases(one is the accelerated warming period during 1980-1996,the other is warming hiatus during 2001-2012),to investigate the possible association between dust emission trend and global warming.Results indicated that the wind regimes,such as the nocturnal low-level jet,were key factors to the wintertime DUEM over the Central Sahel.The springtime dust storms related to cold air and cyclones primarily occurred in the southern coast of the Mediterranean and northwestern China.The cold high and heat low were typical mechanisms for the summertime DUEM,which frequently formed in western North Africa,the Middle East,and northwestern Indian subcontinent.Whereas the land cover and drought conditions play significant roles in the relatively wetter regions,i.e.,the southern coast of the Mediterranean,the Ustyurt Plateau,the northwest coast of Indian Ocean,the Thar desert and the Taklimakan desert.As for the association between global warming and dust variation trend,the dust emissions over the inland dust source regions showed a decreasing trend,while a rising trend was observed over the dust sources in the low-lying coastal areas on the foot of high mountains.(3)Analyzing the interdecadal change of the relationship between ENSO and dust activities over dust sources in the dust belt based on the MERRA-2 aerosol product and statical regression model.Results indicated that the impact of ENSO on DUEM and surface dust mass concentration(DUSMASS)over the dust source regions in southern Middle East and South Asia showed significant interdecadal change,in addition,those two dust source regions showed similar change.For DUSMASS,its association with ENSO during 2000-2014(R=-0.67,P<0.01)was significantly stronger than that during1982-1996(R=-0.52,P>0.01)and this change was more remarkable in the South Asian dust source.The change of Atlantic sea surface temperature anomaly(SSTA)pattern weakened the impact of ENSO on DUSMASS,while that of Indian Ocean SSTA pattern and PDO tended to strengthen ENSO’s effect.Both the Atlantic and Indian Ocean SSTA patterns were modulated by the duration of ENSO events(i.e.,continuing and emerging ENSO).The Eurasian continent and Indian Ocean thermal contrast was less likely to cause the shift of ENSO-DUSMASS relationship.For DUEM,its association with ENSO shifted from the significant negative relation during 1982-1996(R=-0.68,P<0.01)to insignificant relation during 2000-2014(R=-0.19,P>0.05)and the correlation during 1982-1996 was more obvious in the southern Middle East dust source.The change of Atlantic SSTA pattern strengthen the impact of ENSO on DUEM and that of Indian Ocean SSTA pattern weakened ENSO’s effect,while PDO showed no significant influence on shift of ENSO-DUEM relationship.(4)According to the abovementioned interdecadal change of relationship between ENSO and dust activities,this study focused on two dust sources that showed significant difference in the relationship between two different warming phases,they were western Sahel and southern Middle East.The common atmospheric influence mechanisms that led to this difference was further explored.The multi-year sliding correlation analysis showed that the relationship between ENSO and DUEM and DUSMASS over the western Sahel exhibited a see-saw pattern under the two warming phases,however,the relation was not significant in either phase.Therefore,the dust column concentration(DUCMASS)was taken into account and window of sliding correlation analysis was extended in this part.Results indicated that the winter ENSO index presented a weakly relation(R=0.11,P>0.05)with DUCMASS over the western Sahel during the accelerated warming period(1980-1996),while exhibited a significant positive correlation(R=0.60,P<0.01)during the warming hiatus stage(1998-2014).On the contrary,the relationship between ENSO and DUCMASS over the Rub’ al Khali desert was significant in the accelerated warming period(R=-0.65,P<0.01),which disappeared in the warming hiatus stage(R=-0.27,P>0.05).The shift in the zonal location of the ENSO induced temperature anomalies in the subtropical Afro-Asian region changed the wind field over the Sahel and the Middle East,which led to the contrast in dust transport over those two regions under different warming phases.On the other hand,the move in the location of sea level pressure(SLP)center influenced the ENSO impact on DUCMASS over the western Sahel and the Rub’ al Khali desert by changing the convective precipitation and atmospheric stability.All of those effects led to the stronger relationship between ENSO and DUCMASS over the Rub’ al Khali desert that appeared in 1980-1996,as well as the more remarkable impact of ENSO on DUCMASS over the western Sahel that occurred in 1998-2014. |
PDF Full Text Request