Portal de Periódicos da CAPES

Comment on “On the steric and mass‐induced contributions to the annual sea level variations in the Mediterranean Sea” by David García et al.

2007; American Geophysical Union; Volume: 112; Issue: C12 Linguagem: Inglês

10.1029/2007jc004196

2156-2202

Luciana Fenoglio-Marc, Jürgen Kusche, M. Becker, Ichiro Fukumori,

GNSS positioning and interference

[1] García et al. [2006] (hereafter referred to as GCDVG) present an analysis of mass-induced sea level variation (SLVmass) of the Mediterranean Sea based on satellite altimetry, ocean modeling and Gravity Recovery and Climate Experiment (GRACE) gravity measurements. In particular, GCDVG find consistency between altimeter sea level variations corrected for steric effects and changes in oceanic mass derived from Gravity Recovery and Climate Experiment (GRACE). GCDVG show that the annual cycle of the sea level variability (SLVtotal) is dominated by changes in its steric part (SLVsteric) and that the basin average of mass (SLVmass) is maximum in February. They conclude that, when the sea level is rising (falling) the Mediterranean Sea is actually losing (gaining) mass. [2] GCDVG appear to use an unrealistically large estimate of steric sea level variability (SLVsteric) and neglect changes in continental water storage that can significantly affect GRACE measurements over the Mediterranean Sea. These issues are examined in turn below by reviewing relevant individual basin averages of the Mediterranean Sea in comparison to results of other studies. In particular, while also finding consistency between altimetry and GRACE measurements, Fenoglio-Marc et al. [2006] (hereafter referred to as FKB), in a similar study of the Mediterranean Sea, find steric effects only accounting for about 60% of the total annual sea level variability and the annual cycle of SLVmass having a maximum in November. [3] The amplitude and phase of the relevant basin averages of GCDVG are summarized and compared with those of FKB in Table 1. There is good agreement for total sea level variability (SLVtotal) from altimetry, while results differ for most of the other components. [4] The total sea level estimates are comparable between the two studies. However, although the phase is consistent with each other, the two studies differ significantly in the amplitude of SLVsteric. The results differ in spite of both studies utilizing temperature and salinity estimates from the same ocean model analysis to estimate SLVsteric (ECCO/JPL [Fukumori et al., 1999]). The GCDVG estimate of 94 mm amplitude appears to be excessively large in comparison to other published results. For instance, García-Lafuente et al. [2002] and Fukumori et al. [2007] estimate SLVsteric amplitudes of 50 mm and 45 mm, respectively, comparable to that of the FKB estimate based on climatological hydrography (Medar/Medatlas and WOA01) and the same ECCO model forced solely by diabatic forcing (Table 1). Others [Bouzinac et al., 2003; García-Lafuente et al., 2004; Vignudelli et al., 2003] also estimate similar steric changes that are about 50% of the total sea level change in contrast to GCDVG's estimate of SLVsteric dominating SLVtotal. [5] The two studies also differ significantly in spatial variability. The GCDVG estimate of SLVsteric (their Figure 3b) has a very large variability across the Mediterranean basin that varies from approximately 50 mm to 160 mm, a factor of 3 with a range of 110 mm out of a basin mean of 94 mm. In comparison, that of FKB is much more uniform, consistent with the large-scale nature of seasonal heating and cooling, with a range of 40 to 70 mm (a 30-mm range out of a basin mean of 50 mm) (Table 1). [6] The climatological hydrography data from Ishii et al. [2006] and a reexamination of the ECCO results using an updated version of the ECCO model and a slightly different integration procedure (Table 1), confirm a basin mean of annual amplitude between 45 and 60 mm with variations in local amplitudes of about 30 mm. With the reexamined ECCO results the agreement between the GRACE and the steric-corrected altimetry estimates of seawater mass change SLAmass (correlation 0.8 and root mean square differences 29 mm) is higher than in FKB (Table 1). As both GCDVG and FKB studies use estimates from the same ocean model, the differences in SLVsteric must be due to how steric height is evaluated from the model's temperature and salinity. [7] Differences of SLVmass between the two studies can be ascribed to differences in GRACE data processing. Both studies employ similar corrections to the published GRACE estimates including effects of nontidal barotropic oceanic variability as well as long-wavelength secular variations and atmospheric effects based on satellite laser ranging (SLR). However, the two estimates differ in length scales that are retained. GCDVG employ GRACE coefficients up to degree 15 and smooth the resulting estimate by a 1000-km-radius Gaussian filter [Wahr et al., 2004] so as to isolate the Mediterranean Sea area. In comparison, FKB use the GRACE coefficients up to degree 90 and a Gaussian filter with a 400-km radius. This can account partly for the difference in amplitude and phase between the filtered basin average from GRACE. Increasing the filter radius results in a decrease in amplitude and an increase in phase of the annual harmonic. These differences are about 10 mm in amplitude and 20 degrees in phase for a filter radius of 400 km with coefficients up to degree 90 and a filter radius of 1000 km with coefficients up to degree 15 (Table 2). The 400- and 1000-km filters retain 50% of the signal at degrees 21 and 8, respectively. [8] More importantly, FKB account for leakage of land hydrologic variations into observed changes of mass over the Mediterranean Sea that GCDVG do not take into consideration. Because of the effective large-scale smoothed averaging by the spherical harmonic analysis employed in the GRACE estimates, any mass variation over land will contribute to estimates over the ocean within the smoothing radius. Owing to the small dimension of the Mediterranean Sea, the coastal area cannot be masked as in work by Chambers [2006]. [9] The leakage of land hydrology has a sizeable effect on the oceanic mass estimated from GRACE, as the amplitude of the land hydrology signal is much larger. Using continental water storage estimates from the Land Dynamics Model (LAD) [Milly and Shmakin, 2002], FKB estimate the annual amplitude of this leakage to be 29 mm (43-degree phase) which reduces the estimate of the annual change of net oceanic mass in the Mediterranean Sea to 28 mm with a phase of 327 degrees (Table 1). The annual amplitude and phase of the leaked signal is about the same for filters and truncations used in GCDVG and FKB (Table 2). [10] Finally, the resulting GRACE mass estimate represents a large-scale smoothed average over the surface of the Earth whereas the sea level estimate does not. GCDVG do not account for this discrepancy. In comparison, FKB consider the reduction of amplitude in the former due to this averaging [Velicogna and Wahr, 2006; Klees et al., 2007], yielding a 52-mm amplitude (327-degree phase) GRACE-based annual estimate of seawater mass change over the Mediterranean Sea (Table 1). [12] A comparison between estimates of annual mass change variation in the Mediterranean Sea by GCDVG and FKB has revealed differences both in methodology and in intermediate results obtained by approximately the same input data. One of the differences is the amplitude of the annual component of steric sea level change. While other published steric height estimates are comparable to that of FKB, it is not clear why the GCDVG estimate differs. [13] A second difference results from the method used to estimate changes in mass of the Mediterranean Sea from GRACE gravity field coefficients. GCDVG neglect leakage of continental hydrology and therefore include in their solution an apparent change in water mass that can amount to 40% of the total signal based on hydrological model output. [14] GCDVG conclude that the low accuracy in GRACE data is an impediment to the study. FKB show that GRACE is able to detect water mass variation in the Mediterranean Sea, but that the error of the GRACE estimate is still larger than the error of the estimate derived from the steric-corrected altimeter data. [15] However, an examination of additional temperature and salinity data results in an even closer agreement between GRACE estimates and altimetry measurements than reported by FKB. The quality of the GRACE data has meanwhile improved and a further reduction of the errors in the water mass estimation from GRACE can be expected. [16] We acknowledge F. Raicich, S. Zerbini, M. Marcos, M. Tsimplis, C. K. Shum, C. Y. Kuo, R. Schmidt, S. Vignudelli, E. Stanev, and S. Grayek for discussions. The study was partially funded by DFG (SPP1257/STREMP). Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.