In the end of November 2018 the InSight mission landed on the Martian surface deploying a six axis broadband and short period seismometer, in order record marsquakes and ambient noise, which can be used to constrain the deeper interior structures of the planet. In this study we process the three components of the broad band seismometer and apply autocorrelation methods to more than a year of ambient noise data and to 250 selected marsquake waveforms, to retrieve the empirical Greens functions (EGFs), that are related to the subsurface impedance discontinuities below the lander. The most prominent impedance discontinuity is commonly represented by the crust-mantle boundary (Moho), which yet needs to be confirmed on Mars. A lot of care is taken to attenuate the spurious signals and the wind induced lander resonance, that contaminate the data, before the autocorrelation analysis. We further apply attribute filtering and reject data windows that exceed wind speed thresholds or reveal too large amplitudes at later lags, which are often related to residuals of artefacts or lander resonance. To further improve the signal to noise ratio of the EGFs we apply phase-weighted stacking. We observe a good agreement in the autocorrelograms of the ambient noise and the marsquakes and determine several stable arrivals on the vertical component, in particular at 10.8 and 12.7 s, that we interpret as trapped P-waves, which are reflected between the surface and the potential Moho or other layer interfaces. Using available velocity constraints we convert the arrival times to possible Moho depths of 19-28 and 22-33 km, respectively. In addition, we find that large part of the seismic energy during the quiet night period of Mars is related to the 2.4 Hz mode and that removing it from the autocorrelations the distinct arrivals can no longer be identified.