Abstract
The altimeter radar backscatter cross-section is known to be related to the ocean surface wave mean square slope statistics, linked to the mean surface acceleration variance according to the surface wave dispersion relationship. Since altimeter measurements also provide significant wave height estimates, the precedent reasoning was used to derive empirical altimeter wave period models by combining both significant wave height and radar backscatter cross-section measurements. This article follows such attempts to propose new algorithms to derive an altimeter mean wave period parameter using neural networks method. Two versions depending on the required inputs are presented. The first one makes use of Ku-band measurements only as done in previous studies, and the second one exploits the dual-frequency capability of modern altimeters to better account for local environmental conditions. Comparison with in situ measurements show high correlations which give confidence in the derived altimeter wave period parameter. It is further shown that improved mean wave characteristics can be obtained at global and local scales by using an objective interpolation scheme to handle relatively coarse altimeter sampling and that TOPEX/Poseidon and Jason-1 altimeters can be merged to provide altimeter mean wave period fields with a better resolution. Finally, altimeter mean wave period estimates are compared with the WaveWatch-III numerical wave model to illustrate their usefulness for wave models tuning and validation.