OMBNN3 SSM/I Algorithm

    The OMBNN3 algorithm is based on the neural network (NN) technique.  It performs multi-parameter retrievals, it simultaneously produces four parameters: wind speed, columnar water vapor, columnar liquid water and sea surface temperature.  The new OMBNN3 algorithm takes into account the interdependence of these physically-related atmospheric and oceanic parameters and preserves proper physical relationships among these parameters.  It utilizes five SSM/I brightness temperature channels.
Fig. 1.  The OMBNN3 algorithm architecture.

The algorithm has extended the range of wind speeds over which useful retrievals can be obtained.  It not only improves the accuracy of the wind speed retrievals, especially at high wind speeds (without bias correction), but makes available three additional  fields.  Table 1 indicates the importance of the inclusion of water vapor, liquid water, and SST in  retrieval algorithms on the accuracy of wind speed.  The GSW is the original linear algorithm, and the GSWP algorithm contains the water vapor correction suggested by Petty (1993).  The RMS error statistics of the OMBNN3 algorithm, which takes into account also the liquid water and SST influences, are lower than those of the GSWP algorithm over  all wind speeds, and especially at wind speeds > 15 m/s.

Table 1.  Error budget (m/s) for different SSM/I wind speed algorithms (over about 15,000 buoy/SSMI matchups); for clear and clear+cloudy (in parentheses) conditions.
Algorithm Algorithm RMSE Total RMSE W > 15 m/s RMSE
Multiple Linear Regression & Linear Approximation (GSW) 1.4 (1.8) 1.8 (2.1)  (2.7)
Multiple Linear Regression & Nonlinear Approximation (GSWP)  1.3 (1.6) 1.7 (1.9)  (2.6)
Multiple Nonlinear Regression & Nonlinear Approximation (GS) 1.4 (2.2) 1.8 (2.5)   (2.7)
Physically-Based (PB) 1.3 (1.8)  1.7 (2.1)  (2.6)
Neural Network (OMBNN3) 1.0 (1.3) 1.5 (1.7) (2.3)
GSW - Goodberlet et al., 1989; GSWP - Petty, 1993;  GS - Goodberlet and Swift, 1992; PB - Wentz, 1997;
OMBNN3 - Krasnopolsky et al., 1996
Fig. 2.  Multi-parameter vs. single-parameter retrievals.  Errors in wind speed as functions of V, L, and SST (over about 13,000 matchups).  Blue - GSW, green - GSWP, and red - OMBNN3

Fig. 2 shows the wind speed difference (buoy minus satellite wind speeds) characteristics (retrieval errors in wind speed) as functions of three other parameters: columnar water vapor, columnar liquid water and sea surface temperature for three algorithms GSW, GSWP and OMBNN3.  Including nonlinear water vapor correction in GSWP reduced the bias and its dependence on the water vapor concentration  (and partly on SST which is closely related to water vapor); however, it did not reduce its dependence on the liquid water concentration. This correction  also did not significantly improve the standard deviation of the differences.  The OMBNN3 algorithm, with its simultaneous multi-parameter retrievals, reduced the bias and its dependence on all three other parameters together with a significant improving the standard deviation of the differences.

A multi-parameter empirical ocean algorithm for SSM/I retrievals
by: Krasnopolsky, V.M., W.H. Gemmill, and L.C. Breaker, 1998,Technical Note, OMB Contribution No. 154, NOAA/NCEP/EMC, Canadian Journal of Remote Sensing, in press

A New Transfer Function for SSM/I Based on an Expanded Neural Network Architecture
by: Krasnopolsky VM, WH Gemmill & LC Breaker, 1996, NOAA/NCEP/EMC technical note, OMB Contribution No. 137 , Camp Springs, MD, 38 pp.

Return to SSM/I Products Description Page
Return to SSM/I Ocean Products Page

Return to Marine Modeling and Analysis Branch Main Page

Please send comments and questions to Vladimir Krasnopolsky

Last changed 25 November 1998