The Spatial-Temporal Variability Of Antarctic Ross Ice Shelf Polynya Ice Production

Polynyas are isolated areas of open water or thin ice surrounded by sea ice packs in a polar region.There is frequent heat exchange between the warm sea water and the cold air on the surface of latent heat polynyas,along with a large amount of ice and salt production.Latent heat polynyas play an important role in sea ice production and the formation of the Antarctic Bottom Water,and therefore it is of great importance to study polynya dynamics and estimate the ice production accurately.The Ross Ice Shelf Polynya(RISP)is the largest coastal polynya with the highest ice production in the Antarctic Ocean.RISP is located in the Ross Sea,which is one of the main source regions of the Antarctic Bottom Water.Many researchers have estimated the ice production of RISP,but their results vary widely and the trends are inconsistent.Due to the extreme environment in polar region and lack of field measurements,ice production is usually simulated by models.Many factors,however,influence the outputs of ice production models,leading to substantial uncertainty.In this paper,we focus on the spatial-temporal variability of RISP ice production and the uncertainty of ice production estimation.The main contents are listed as follows:1.The spatial-temporal variability of RISP distribution and ice production(a)RISP extent identification based on different methods.We identified daily RISP distribution by thresholds of sea ice concentration(SIC)and thin ice thickness,respectively.Distinct difference is found between two types of RISP extents.The RISP extents based on ice thickness threshold have generally larger area and locate closer to the ice shelf front than extents based on SIC threshold.For the ice thickness threshold method,ice shelf,fast ice or icebergs are easily identified as thin ice by mistake,leading to an exaggerated extent.Precise masks,therefore,are required for accurate extent delineation from thin ice thickness.(b)Spatial variability of RISP ice production.We used RISP area identified by SIC threshold and as well as re-analysis climate data to estimate daily RISP ice production maps during 2003-2017 based on a thermal-dynamic ice production model.The results show that high-rate ice production occurs near the ice shelf front within the distance of 20-30 km,in a shape of a long strip.The farthest ice production reaches to 100 km away from the ice shelf front.Most of the polynya area presents a significant increasing trend of annual ice production,except for specific area.An elongated area within the distance of 6 km from the ice shelf front shows a contrarily significant decreasing trend,and the area between Ross Island and Franklin Island shows slightly decreasing trend.(c)Temporal variability of RISP ice production.Time series of annual and monthly ice production were derived from ice production maps by spatial integrating and averaging.The RISP ice production shows no obvious annual or seasonal trend.Combining the RISP area and ice production time series,we found two extreme events:the largest RISP area occurs in October 2005,with the largest deviation between the area and ice production;and the largest ice production occurs in June 2007.2.Sensitivity analysis of ice production model and heat flux components analysis(a)Sensitivity analysis of ice production model and parameterization.We reviewed the parameterization of previous studies,and selected several parameter values for sensitivity tests.The results show that ice production model is sensitive to bulk transfer coefficients,the latent heat of sea ice fusion,the SIC threshold for the RISP area,and the sea ice density.The bulk transfer coefficients can vary ice production result as much as 87%.By analyzing sensitivity test results,previous studies and theoretical derivation,we recommend bulk transfer coefficients values equal to 0.02,the latent heat of sea ice fusion value equals to 2.79×105 J·kg-1(only for the Ross Sea),the SIC threshold for the RISP area equals to 75%,and the sea ice density value equals to 920 kg·m-3.(b)Heat flux components analysis.To explain the temporal variability of RISP ice production,we divided total ice production into four components according to heat flux sources.The results show that sensible heat flux component accounts for 60.1%of the total ice production,longwave radiation component accounts for 26.9%and latent heat flux component accounts for 20.4%.Solar radiation component accounts for-7.5%,making negative contribution,and it turns into-54.9%in October.The longwave radiation,sensible heat flux and latent heat flux are strongly correlated with RISP area.While solar radiation shows distinct seasonal pattern.It is near zero from April to September as a result of polar night,and sharply increases in October as a result of polar day.(c)The explanation to the extreme event in October 2005.In every October,Vs sharply increases,which is strongly correlated to RISP area.In other words,a high RISP area in October will lead to high negative contribution of Vs,thus ice production will be evidently reduced by Vs instead of an increasing ice production proportional to area.October 2005 is a typical example of this phenomenon,when the largest area leads to the largest inconsistency between area and ice production.3.Impact of environmental factors on RISP ice production(a)The correlation between RISP area and different types of wind speed.According to correlation analysis between six types of wind speeds and RISP area time series,the best correlation was found between the reanalysis wind speed averaged on the whole study area and the polynya area.It performs better than the wind speed measured by two automatic weather stations near the Ross Ice Shelf,or reanalysis wind speed data measured on a single pixel.For the two automatic weather stations,wind speed data from the Vito station has better correlation with the area of polynya than data from the Laurie II station.(b)The mechanism of wind forcing impact.The wind speed is significantly correlated with the RISP area and ice production.The RISP dynamics and ice production during extreme events are obviously related to wind forcing field.We found that persistent offshore strong winds lead to extremely high RISP area and ice production.The effects of wind speed on ice production can be divided into two aspects:the effects on the polynya area,and the effects on heat exchange and phase transformation.The former mechanism has a time-lag of approximately one day.The influences of these two mechanisms are both important,but they exhibit different priority during specific periods.Due to the seasonal variation of solar radiation,strong wind events in different seasons show different impact on the ice production.During April-September,extreme wind events lead to dramatic increases in the RISP area and ice production,while extreme wind events in October also lead to dramatic increases in the polynya area,but slight increases or even decreases in ice production.(c)The impact of icebergs and ice shelf on the spatial distribution of local RISP ice production.The spatial distribution and trends of RISP ice production are related to the activity of icebergs and ice shelf.The stranding and drifting of B-15a and C-16 icebergs near Ross Island lead to a slight decline in local ice production between Ross Island and Franklin Island from 2003 to 2017.The expansion of the front edge of the Ross Ice Shelf lead to a significant decline in ice production within about 6 km from the edge of the ice shelf,while a significant increase in ice production was observed in other part of RISP.