## 1. Introduction

[2] In gravimetry the technique known as “stripping” has been used whenever a part of the Earth's subsurface mass density distribution was known (represented by a model produced as a result of other geoscientific investigations), in order to unmask the remaining gravitational signal of the unknown (and sought) anomalous subsurface density distribution. The strongest signal in gravity data is due to topographic relief onshore and ocean bottom relief offshore. When these two surfaces are known, the gravitational effects of the reference (e.g., constant average) topographic masses onshore and seawater density contrast offshore can be removed from the gravity data by means of topographic and bathymetric corrections respectively. The next strongest signal in the gravity data is due to the crustal/lithospheric thickness and density composition (as a result of the combination of its isostatic and tectonophysical states). An isostatic compensation scheme may be adopted to compute an isostatic correction to gravity data, or a crustal/lithospheric model is adopted or produced to compute the crustal or lithospheric stripping correction. In this latter step various approaches may be taken depending on the purpose of the study [cf., e.g., *Kaban et al.*, 1999, 2003, 2004; *Kaban and Schwintzer*, 2001] for global studies and [e.g., *Bielik*, 1988; *Artemjev and Kaban*, 1994; *Artemjev et al.*, 1994; *West et al.*, 1995; *Kaban*, 2001, 2002; *Zeyen et al.*, 2002; *Bielik et al.*, 2004; *Dérerová et al.*, 2006; *Braun et al.*, 2007; *Tassara et al.*, 2007; *Tesauro et al.*, 2007; *Alvey et al.*, 2008; *Jiménez-Munt et al.*, 2008, and references therein] for large regional investigations. In regional studies the stripped gravity data are typically interpreted by integrated forward modeling with use of all possible geophysical constraints. For global studies the best currently available global crustal model is CRUST 2.0 [*Bassin et al.*, 2000], which is an upgrade of the CRUST 5.1 model [*Mooney et al.*, 1998]. The publically available CRUST 2.0 model contains information on subsurface spatial distribution and density of the following global components: ice, soft and hard sediments, upper, middle and lower (consolidated) crust.

[3] When the gravimetric inverse problem is formulated in terms of attraction, the anomalous gravity data required as observables in the inversion/interpretation become, by definition, the gravity disturbances [e.g., *Vajda et al.*, 2006, 2007]. The use of gravity disturbances in global and regional studies eliminates the need of the “geophysical indirect effect” correction [e.g., *Hackney and Featherstone*, 2003; *Hinze et al.*, 2005; *Vajda et al.*, 2006, and references therein]. The normal gravity is subtracted from the actual observed gravity in the definition of the gravity disturbance. This has two implications [e.g., *Vajda et al.*, 2006, 2008]. First, the surface of the reference ellipsoid (not the geoid) is the bottom interface of topographic masses globally, as well as the upper interface of all density contrasts defining the stripping corrections. Second, a model normal mass density distribution inside the reference ellipsoid generating the normal gravity field is the background density distribution model against which the density contrasts used in stripping corrections are defined. These two conditions have to be satisfied in order to keep the equivalence between decomposing the real Earth's subsurface density distribution and the observed gravity disturbances [cf. *Vajda et al.*, 2008].

[4] Our aim here is to evaluate on a global scale the gravity disturbances corrected for the attraction of the topography (ellipsoid-referenced topographic correction), the ocean density contrast (ellipsoid-referenced bathymetric stripping correction), and the crustal density contrast (down to the Moho interface). We take into account the global distribution of ice, sediments, and consolidated crustal components based on the CRUST 2.0 model (crustal stripping correction). The crustal stripping correction is computed and applied in several consecutive steps: (1) stripping the attractions of the density contrasts (relative to the constant reference crustal density of 2670 kg/m^{3}) of the ice, and the soft and hard sediments; (2) stripping the attractions of the density contrasts (relative to the constant reference crustal density of 2670 kg/m^{3}) of the upper, middle, and lower consolidated crust of the CRUST 2.0 model; and (3) stripping the attraction of the density contrast of the entire crust (volumetric domain between the reference ellipsoid and the Moho interface) of constant density of 2670 kg/m^{3} relative to a constant reference density of the encompassing mantle. The reason for the stepwise compilation of the crustal stripping correction and of the complete crust-stripped gravity disturbances is the following. The application of the topographic and stripping corrections of steps 1 and 2 removes the topographic masses above the reference ellipsoid and transforms the volumetric domain between the reference ellipsoid and the Moho interface globally (disregarding the heterogeneities within topography other than sediments and ice, and disregarding the crustal heterogeneities not accounted for by the CRUST 2.0 density model) into a model crust of a constant 2670 kg/m^{3} density. The gravity disturbance respective to this stripping stage is respective to a model Earth of no topography, constant crust down to the Moho interface, and real density below the Moho interface. The strongest signal in such a gravity disturbance is the attraction of the density contrast of the Moho interface relative to the mantle. We expect that this type of gravity disturbance is best suited for refining the geometry of the Moho interface by means of gravimetric interpretation. The final step, step 3, transforms the constant 2670 kg/m^{3} density model crust into mantle, removing the signal of the Moho interface density contrast from gravity data. The complete crust-stripped gravity disturbances ideally contain only the gravitational signal of the sub-Moho density inhomogeneities. They are therefore best suited for studying the upper mantle (mantle lithosphere) and deeper mantle.

[5] The gravity disturbances with the individual corrections applied are mapped in section 3, and the data sets are made available to the scientific community for geophysical studies in Data Sets S1–S5 in the auxiliary material. We refer to the bathymetrically stripped and topographically corrected gravity disturbances briefly as “BT gravity disturbances.” To decompose the complete crust-stripped gravity disturbances into individual contributions (attractions) of the crust density heterogeneities and interfaces and to interpret/invert the individual signals is a nonunique and altogether complex matter that needs to be approached with the help of all possible additional geoscientific constraints. The approach taken depends on the objective of the study. We aim here only at providing the stepwise complete crust-stripped gravity data to the geophysical community for further global studies.

[6] The gravity disturbances and the topographic and bathymetric corrections are computed globally at the Earth's surface in a spectral form up to degree 180 of the spherical harmonics (which represents roughly 100 km in terms of half wavelength). The ice stripping correction is computed with a spectral resolution complete to degree 90. All the corrections and the stepwise corrected gravity disturbances are computed at the Earth's surface on a 1 × 1 arc degree grid of the geocentric spherical coordinates. We shall follow a sign convention, whereby an attraction to be corrected for is subtracted from gravity data, while a correction (negative attraction) is to be added to the gravity data.