Oxygen lines in solar granulation I. Testing 3D models against new observations with high spatial and spectral resolution

Aims. We seek to provide additional tests of the line formation of theoretical 3D solar photosphere models. In particular, we set out to test the spatially-resolved line formation at several viewing angles, from the solar disk-centre to the limb and focusing on atomic oxygen lines. The purpose of these tests is to provide additional information on whether the 3D model is suitable to derive the solar oxygen abundance. We also aim to empirically constrain the NLTE recipes for neutral hydrogen collisions, using the spatially-resolved observations of the O I 777 nm lines. Methods. Using the Swedish 1-m Solar Telescope we obtained high-spatial-resolution observations of five atomic oxygen lines (as well as several lines for other species, mainly Fe I) for five positions on the solar disk. These observations have a high spatial (sub-arcsecond) and spectral resolution, and a continuum intensity contrast up to 9% at 615 nm. The theoretical line profiles were computed using the 3D model, with a full 3D NLTE treatment for oxygen and LTE for the other lines. Results. At disk-centre we find an excellent agreement between predicted and observed line shifts, strengths, FWHM and asymmetries. At other viewing angles the agreement is also good, but the smaller continuum intensity contrast makes a quantitative comparison harder. We use the disk-centre observations we constrain S H, the scaling factor for the efficiency of collisions with neutral hydrogen. We find that S (H) = 1 provides the best match to the observations, although this method is not as robust as the centre-to-limb line variations to constrain S (H). Conclusions. Overall there is a very good agreement between predicted and observed line properties over the solar granulation. This further reinforces the view that the 3D model is realistic and a reliable tool to derive the solar oxygen abundance.