We have investigated Lorentz violation through analyzing tides-subtracted gravity data measured by superconducting gravimeters. At the level of precision of superconducting gravimeters, we have brought up and resolved an existing issue of accuracy due to unaccounted local tidal effects in previous solid-earth tidal model used. Specifically, we have taken local tides into account with a brand new first-principles tidal model with ocean tides included, as well as removed potential bias from local tides by using a worldwide array of 12 superconducting gravimeters. Compared with previous test with local gravimeters, a more accurate and competitive bound on space-space components of gravitational Lorentz violation has been achieved up to the order of 10 −10 . Einstein's equivalence principle is the foundation of general relativity. It is based on the universality of free fall, local Lorentz invariance and, local position invariance [1,2]. The universality of free fall has been tested up to accuracies of 10 −13 [3][4][5][6]. Local position invariance has been tested, e.g., by gravitational red shift measurement with atom interferometers or clocks [7,8]. In comparison, testing local Lorentz invariance (LLI) is a broad field, as violations of LLI might manifest themselves in the gravity sector itself, or in the matter sectors as well as their coupling [9,10].In the simplest case, violations of LLI in the gravity sector manifest themselves as a dependence of the force of gravity between two objects on the direction of their separation. Competitive bounds in this sector have been established by various experiments and observations [24,25], such as gravimetry [11][12][13], lunar laser ranging [14][15][16] and astrophysics observations [17,18]. Among these, local gravimetry is the one of the easy-to-access and very precise ground-based method. The underlying idea is simple: if the force of gravity is anisotropic, then the local acceleration of free fall on the rotating earth should exhibit a modulation correlated with the earth's rotation. In analyzing such tests, the influence of the sun, the moon and the planets have to be taken out, which is done by subtracting a "tidal model" describing of these influences.However, a persisting problem has been pointed out in previous works [11][12][13] whether a simple first-principles solid-earth tidal model or a more sophisticated empirical model should be used. The simple first-principles solidearth tidal model does not include any Lorentz violation signal. But it's not accurate enough beyond 10 −10 g with- * E-mail:zkhu@mail.hust.edu.cn † E-mail:hm@berkeley.edu out including local tidal effects like ocean tides. At the precision of superconducting gravimeters, it may produce fake Lorentz violating signals [13]. Sophisticated empirical models are a lot more accurate, but it's based on fitting of gravity measurement which itself may contain Lorentz-violating signals. In this work, we have reconciled the conflict with a worldwide network of gravimeters analyzed with first-principle ti...