 The SWAT mission will provide high-resolution and two-dimensional measurements of sea surface height, SSH, but despite its unprecedented precision, it still has a significant amount of random noise. To address this issue, researchers have proposed a convolutional neural network, CNN, based noise reduction algorithm. The CNN was trained and tested using simulated data from the North Atlantic and compared to three other noise reduction techniques, a median filter, a Lanxos kernel smoother, and the SWAT denoising algorithm developed by Gomez-Navarro et al. The CNN yielded better results than the other algorithms, with a 2 mm RMSE, a 16 dB variance reduction, and a power spectrum down to 10 to 20 km wavelengths. Additionally, the CNN was found to be more stable and robust than the other algorithms, making it a promising candidate for reducing the noise of flight data from Karen. This article was authored by Inel Trebut, Alisa Carly, Maxim Ballerata, and others. We are article.tv, links in the description below.