GNGTS 2019 - Atti del 38° Convegno Nazionale

GNGTS 2019 S essione 1.4 215 a uniqueness theorem is proved as a theoretical background for the proper application of the method. The method effectiveness has been tested using simulated gravity data. In particular, a simulated Moho is computed and two density models are assumed to compute the direct gravity effect to be used in the inversion procedures. These tests gave relevant insight in the devised inversion method. Reliable results have been obtained when density models are consistent with gravity simulated data. It can be thus assumed that, under unbiased conditions, the method is stable and can be applied to Moho depth estimate. Also, the impact of depth information proved to be relevant. Collocation allows easily the integration of different data types and this can improve the solution. Particularly, this applies to the bias in the mean depth that is reduced if depth data are considered. On the contrary, the solution based on gravity only cannot compensate for this bias. Another important remark is related to density information. It was clearly proved that density plays an important role in the inversion. If proper density information are available, the iterative inversion procedure based on collocation can give reliable results. On the other hand, when assuming inconsistency between density and simulated gravity data, we obtained remarkable distortions in the estimated depths. Furthermore, in this case, depth information is not effective in reducing the bias and even on the given depth data a significant discrepancy is present. Thus, it can be concluded that the collocation-based procedure can be applied for getting Moho depth estimates from gravity and depth information (from e.g. seismic). However, as it is for any other inversion method, improper density information can induce serious biases in the estimates even if depth data are included in the inversion. Two possible evolution lines of the collocation inversion method can be devised. One line is to investigate its behavior in local areas where high frequency Moho patterns are present. Being collocation quite a stable and regularizing technique (Barzaghi et al. 1992), this doesn’t seem to be a too critical issue. The second possible application of this methodology is on global Moho estimates based on gravity data coming from the gravity dedicated satellite missions. Since proper spherical model equations can be quite easily derived and linearized (see e.g. Eshagh et al. 2011), global applications of the collocation method seem to be feasible. In this context, additional experimental tests on the Iranian area have been carried on. Collocation method (Ebadi et al. 2019) as well as two approaches based on isostasy principle presented by Sjöberg (Sjöberg 2009) and Jeffrey (Jeffrey 1976] for estimating the regional Moho in Iran using gravity observations. For this purpose, data of the GOCO03S satellite only global geopotential model have been considered. They were subsequently reduced by topography/ bathymetry, sediment and crystalline crust data effect by using the SRTM30_PLUS DTM and the CRUST1.0 model. The three different gravimetric approaches gave coherent estimates. The estimated Moho depths obtained using collocation, Jeffrey and Sjöberg’s approaches proved to be statistically equivalent when compared to seismic derived values. The collocation method has been applied iteratively in three steps and the numerical computations showed that an iterative process in collocation method could not change the results significantly even though the Moho estimate based on collocation method in the third step contains more high frequency details. Furthermore, the collocation solution proved to be less biased than those based on Jeffrey and Sjöberg’s methods when considering discrepancies with respect to seismic Moho estimates. To evaluate our results, we have compiled a 277 points collection of local seismic estimations in this area. The overall standard deviation of the differences between the results of the collocation, Sjöberg and Jeffrey’s methods and the seismic estimates is around 6.0 km. The minimum RMS of differences is between collocation estimates and point-wise seismic data since, as mentioned, collocation led to less biased discrepancies with the considered seismic values. Although the application of sediment and consolidated crust corrections in our solutions provides a reasonable agreement with point-wise seismic data over most of continental areas like central Zagros, Sanandaj-Sirjan, Kopeh-Dagh and Alborz Mountains, this leads to unrealistic estimates under the Makran subduction zone, Oman Sea, Persian Gulf and Caspian Sea. This