In a climate change era, improved information on ice sheets and ices shelves are required in order to monitor theirs changes and the possible contribution to sea level rise over the 21st century and beyond ([1], WCRP). In recent years activities were promoted by SCAR, CliC (WCRP), IASC, to improve models for predicting the future evolution of ice sheets and significant progress has been reported [2]. Nevertheless, there are still large discrepancies between simulations [3, 4, 5] and large uncer- tainties still remain related to our limited quantitative understanding of many of the processes that control the mass balance and stability of ice sheets, and the inability to characterize such processes from space using existing or planned sensors. Foremost among them is the vertical temperature profile within the ice, which is sparsely measured in situ, but is a primary control on the rate at which ice flows by deformation. With the launch of L-band radiometers (SMOS, Aquarius, SMAP), which are capable of observing ice sheet in depth, a new methodology was proposed to infer information by using satellite data and glaciological models [6]. More recently, the use of microwave radiometry in a low frequencies range (i.e., a wide range 0.50–2.00 GHz) was suggested to go deeper in the ice and retrieve the ice sheet and shelf temperature profile (i.e., from the top to the bottom) through the use of techniques similar to those used by radiometric atmospheric sounders [7] and proposals for space mission development are in preparation in Europe and U.S. The methodology was already successfully tested in airborne campaigns in both Greenland and Antarctica [8, 9]. With the purpose of developing an operational ice sheet/shelf temperature retrieval methodology appliable to spaceborne data, a model analysis started to better understand the contribution of observed brightness temperature at the different frequencies as a function of depth. The analysis was conducted by using a microwave emission model [10], able to simulate the brightness temperature, which use as inputs the ice temperature and density profiles based on glaciological models [11, 12]. Ice sheet is discretized by 2-m thick layers between surface and 50m in depth, then 50-m layers down to the bedrock. This allows to a better characterization of the upper part of the ice sheet where the most rapid ice density changes happened. As the horizontal advection is neglected in this model, the validity area is limited where the ice velocity [13] is lower than 10m yr-1. Four geophysical parameters (i.e., ice thickness, surface temperature, snow accumulation, and geothermal heat flux) are needed to run models. Analysis was performed over all Antarctica and information on these parameters are derived from literature [14, 15, 16, 17]. For each considered 25×25km pixel, the nominal values for ice thickness (H), surface temperature (Tsurf), snow accumulation (Acc), and geothermal heat flux (GHF) were used to compute the TB. Moreover, for the three last parameters, theirs uncertainty was considered as follows: Tsurf±1.5K, and Acc±0.5 Acc. Regarding GHF, its value was transformed in a basal temperature (Tbasal) by means of Robin model and the uncertainties were considered as Tbasal±4.5K. In addition to TB, and to provide detailed information about the contribution of each ice sheet layer to the observed TB, the weighting function has been computed [18, 19]. Considering that: ̃T ice B = ∫ H 0 T (z) W (z) dz (1) where T(z) is the temperature of the ice layer, the weighting function (W(z)) can be defined W is the weight of the considered layer, ke is the extinction coefficient, and z is the depth. Based on the weight computation, it is possible to identify for each frequency the ice sheet part proving the maximum contribution to the signal. For that, the minimum thick of the normalized weight representing the 90% of the total weight has been calculated. The portion of the function that subtend this percentage has been identified using a rolling integral. The upper (ztop) and lower (zbottom) limits of these intervals have been calculated together with their differences (∆z). These values were then normalized to the thickness in order to compare values within the continent and, finally, the uncertainty (σ∆z) was calculated by considering the range of variability for all the parameters. Sensitivity to single parameters was also computed and it was demonstrated that TB is function of H and Tsurf, but also Acc and GHF could influence TB under specific site conditions. Obtained Results are summarized in Fig. 1 where ztop, zbottom, ∆z, and σ∆z maps of Antarctica are represented at 0.50, 0.95, 1.40, and 1.85 GHz respectively. At 0.50 GHz the lower 50% of the ice column contribute to the 90% of the simulated TB with the zbottom always located at the ice/bedrock interface. Consequently, the first half of the ice column, including the surface, shows a limited contribute to the total emission. The uncertainty to different parameters is then limited. Increasing the frequency, ztop and zbottom rise in the ice column up to the surface, ∆z decreases and uncertainty increases at 0.95 and 1.40 GHz. At 1.85 GHz uncertainty is low because it is less sensitive to GHF and accumulation variability. These results confirmed that different frequencies are sensitive to different portion of the ice column and then can be used to better estimate different geophysical parameters and constrain the ice temperature profile. Moreover, lowest and highest frequencies could be used to constrain geothermal heat flux and surface temperature, respectively, with a relatively small uncertainty whereas frequencies in the 0.95–1.40 GHz range could be potentially used to constrain snow accumulation even if with a higher error. In addition, simulations show that in the inner part of Antarctica ice sheet TB decreases with frequency while close to the coast the TB is almost frequency independent. Also, this information could be used to constraints the retrieval depending on location. Detailed analysis of the obtained results and the advanced retrieval method will be presented at the conference.
On the retrieval of ice sheet temperature profile by means of low frequency wide band radiometry: sensitivity to the physical parameters
CERRATO R.;
2023-01-01
Abstract
In a climate change era, improved information on ice sheets and ices shelves are required in order to monitor theirs changes and the possible contribution to sea level rise over the 21st century and beyond ([1], WCRP). In recent years activities were promoted by SCAR, CliC (WCRP), IASC, to improve models for predicting the future evolution of ice sheets and significant progress has been reported [2]. Nevertheless, there are still large discrepancies between simulations [3, 4, 5] and large uncer- tainties still remain related to our limited quantitative understanding of many of the processes that control the mass balance and stability of ice sheets, and the inability to characterize such processes from space using existing or planned sensors. Foremost among them is the vertical temperature profile within the ice, which is sparsely measured in situ, but is a primary control on the rate at which ice flows by deformation. With the launch of L-band radiometers (SMOS, Aquarius, SMAP), which are capable of observing ice sheet in depth, a new methodology was proposed to infer information by using satellite data and glaciological models [6]. More recently, the use of microwave radiometry in a low frequencies range (i.e., a wide range 0.50–2.00 GHz) was suggested to go deeper in the ice and retrieve the ice sheet and shelf temperature profile (i.e., from the top to the bottom) through the use of techniques similar to those used by radiometric atmospheric sounders [7] and proposals for space mission development are in preparation in Europe and U.S. The methodology was already successfully tested in airborne campaigns in both Greenland and Antarctica [8, 9]. With the purpose of developing an operational ice sheet/shelf temperature retrieval methodology appliable to spaceborne data, a model analysis started to better understand the contribution of observed brightness temperature at the different frequencies as a function of depth. The analysis was conducted by using a microwave emission model [10], able to simulate the brightness temperature, which use as inputs the ice temperature and density profiles based on glaciological models [11, 12]. Ice sheet is discretized by 2-m thick layers between surface and 50m in depth, then 50-m layers down to the bedrock. This allows to a better characterization of the upper part of the ice sheet where the most rapid ice density changes happened. As the horizontal advection is neglected in this model, the validity area is limited where the ice velocity [13] is lower than 10m yr-1. Four geophysical parameters (i.e., ice thickness, surface temperature, snow accumulation, and geothermal heat flux) are needed to run models. Analysis was performed over all Antarctica and information on these parameters are derived from literature [14, 15, 16, 17]. For each considered 25×25km pixel, the nominal values for ice thickness (H), surface temperature (Tsurf), snow accumulation (Acc), and geothermal heat flux (GHF) were used to compute the TB. Moreover, for the three last parameters, theirs uncertainty was considered as follows: Tsurf±1.5K, and Acc±0.5 Acc. Regarding GHF, its value was transformed in a basal temperature (Tbasal) by means of Robin model and the uncertainties were considered as Tbasal±4.5K. In addition to TB, and to provide detailed information about the contribution of each ice sheet layer to the observed TB, the weighting function has been computed [18, 19]. Considering that: ̃T ice B = ∫ H 0 T (z) W (z) dz (1) where T(z) is the temperature of the ice layer, the weighting function (W(z)) can be defined W is the weight of the considered layer, ke is the extinction coefficient, and z is the depth. Based on the weight computation, it is possible to identify for each frequency the ice sheet part proving the maximum contribution to the signal. For that, the minimum thick of the normalized weight representing the 90% of the total weight has been calculated. The portion of the function that subtend this percentage has been identified using a rolling integral. The upper (ztop) and lower (zbottom) limits of these intervals have been calculated together with their differences (∆z). These values were then normalized to the thickness in order to compare values within the continent and, finally, the uncertainty (σ∆z) was calculated by considering the range of variability for all the parameters. Sensitivity to single parameters was also computed and it was demonstrated that TB is function of H and Tsurf, but also Acc and GHF could influence TB under specific site conditions. Obtained Results are summarized in Fig. 1 where ztop, zbottom, ∆z, and σ∆z maps of Antarctica are represented at 0.50, 0.95, 1.40, and 1.85 GHz respectively. At 0.50 GHz the lower 50% of the ice column contribute to the 90% of the simulated TB with the zbottom always located at the ice/bedrock interface. Consequently, the first half of the ice column, including the surface, shows a limited contribute to the total emission. The uncertainty to different parameters is then limited. Increasing the frequency, ztop and zbottom rise in the ice column up to the surface, ∆z decreases and uncertainty increases at 0.95 and 1.40 GHz. At 1.85 GHz uncertainty is low because it is less sensitive to GHF and accumulation variability. These results confirmed that different frequencies are sensitive to different portion of the ice column and then can be used to better estimate different geophysical parameters and constrain the ice temperature profile. Moreover, lowest and highest frequencies could be used to constrain geothermal heat flux and surface temperature, respectively, with a relatively small uncertainty whereas frequencies in the 0.95–1.40 GHz range could be potentially used to constrain snow accumulation even if with a higher error. In addition, simulations show that in the inner part of Antarctica ice sheet TB decreases with frequency while close to the coast the TB is almost frequency independent. Also, this information could be used to constraints the retrieval depending on location. Detailed analysis of the obtained results and the advanced retrieval method will be presented at the conference.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
