Deep learning approach for denoising satellite gravimetry measurements using encoder-decoder architecture with skip connections, preserving fine-scale gravity signals while removing instrument and environmental noise.
Random noise with normal distribution. Common in sensor measurements.
Frequency-dependent noise (1/f spectrum). Typical in gravimeters.
Sudden spikes from thruster firings or impacts.
Periodic patterns from orbital effects or thermal drift.