Observing Weather over Oceans from SSM/I Using Neural Networks

Using a neural network (NN) technique on SSM/I satellite brightness temperature data, accurate and detailed fields of ocean surface winds, columnar water vapor and liquid water over the ocean has been derived. This technique provides, in particular, high quality wind speed retrievals over areas of moderate moisture and at high wind speeds (> 18 m/s). Retrievals of wind speed data in moist areas and at high wind speeds are of special meteorological interest, as those areas are associated with storms and active fronts. It is precisely under these situations that simple linear regression algorithms used in operational data processing fail. A neural network technique as a transfer function to retrieve wind speeds from SSM/I brightness temperatures is designed because of its ability to deal with nonlinear problems without the prior specification of a functional form. The current version of neural networks has been designed to retrieve simultaneously four geophysical parameters: ocean surface wind speed, columnar water vapor, columnar liquid water, and sea surface temperature. It is the simultaneous retrieval that is unique about the new algorithm, allowing the information from one variable to contribute to the improvement of the other variables (e.g., improved accuracy of wind speed retrievals at high wind speed).

The NN data (surface wind speeds, columnar water vapor & columnar liquid water) along with sea level pressure analyses, ship and buoy data, and also wind vector data from the ERS-2 satellite provide a comprehensive and complimentary description of ocean weather. Several cases will be presented to demonstrate this. We show that: 1) The NN algorithm successfully separates wind speed, columnar water vapor, and columnar liquid water signals contained in the SSM/I brightness temperatures. 2) The new algorithm generates high wind speeds (>18 m/s) in areas where these events are well supported by other data. These winds when compared to sea level pressure analyses are in close agreement. Regions of large pressure gradients match well with high wind speeds. 3) High gradients of the columnar water vapor are related to the position of ocean surface fronts. On the other hand, the structure of the water vapor field is very different from both wind speed and liquid water, and its high values are related to a moist atmosphere which originates from subtropical sources. 4) Higher amounts of columnar liquid water are related to areas of water vapor convergence which are closely associated with active storms and frontal locations and with convective clouds.

These data can be seen at http:/polar.wwb.noaa.gov/winds/