The use of satellite data to calibrate ocean wave spectrum partitions are being carried out in recent wave hindcast datasets as a way of correcting historical sea state records. The directional correction method aims at establishing separate fitting coefficients, for windsea and swell partition, as well as for the different sectors from which the waves propagate. This technique is able to identify systematic errors reproduced by the wave models using altimeter data. This work calibrates the significant wave height Hs from wave spectrum partitions in the South Brazil’s coast using a directional multiple linear regression, previously introduced and tested in the New Zealand’s coast. The ERA5 and CAWCR hindcasts are considered and corrected based on the Globwave satellite data. In general, at the boundaries of our study area, both raw wave hindcasts underestimate significant wave height Hs regarding altimeter data, thus requiring correction coefficients greater than one. The scatter index analysis exhibits an improvement in the indices post calibration, as well as shows higher values nearshore with respect to offshore points for both databases before and after calibration. While ERA5 correction slightly overestimates, CAWCR correction underestimates to some extent, with respect to satellite data. The corrections were also further validated using buoy data over the continental shelf. Even in a shallow sea location, where the Rio Grande buoy is, the bulk wave parameter Hs provided by CAWCR shows improvements post calibration. Comparisons pre and post calibration per direction present better results for CAWCR than ERA5. All sectors analysed for the CAWCR hindcast display improvements after of the correction. The CAWCR correction provided good results and was used as boundary conditions for a higher-resolution regional hindcast.