We have used joint least-squares reverse time migration (LSRTM), also known as linearized waveform inversion, to jointly image primary and free-surface-multiples data. Although migration of free-surface multiples can provide the benefit of additional illumination, crosstalk artifacts that appear in images pose a challenge for migration-based approaches. Our method formulates the imaging problem as an inversion problem to attenuate these crosstalk noises in the image. We have developed a modified modeling operator that uses the data as an areal source. The modified operator is used in joint LSRTM to account for the free-surface multiples in the data. We have developed 3D synthetic and field data examples to reveal that our proposed method improves imaging in geologically complex areas and attenuated crosstalk artifacts in the image.