The Effects of Bias Error and System Noise on Parameters Computed from C, T, P and V Profiles

View More View Less
  • 1 Applied Physics Laboratory and Department of Oceanography, University of Washington, Seattle 98105
© Get Permissions
Full access

Abstract

The development of high-resolution profiling instruments has made possible the computation of quantities such as the stability parameter N2 and the Richardson number Ri over scales of less than a meter. Quality control of the data has been done on an ad hoc basis by individual investigators who have used varying degrees of restraint in the claims made for the final results. As the amount of high-resolution data increases and as the parameters computed from it are used in statistical studies a more objective approach is required.

Profiting instruments are subject to a wide variety of difficulties, many of which are particular to the individual systems. Common to all, however, are errors due to uncertainties in the absolute calibrations of temperature T, conductivity C and pressure P, and noise in the sensors or data systems. Linearized equations for salinity S and specific volume α have been found to give good estimates of the errors resulting from typical values of calibration uncertainties and noise in T, C and P. Using these equations expressions have been developed for the evaluation of corresponding uncertainties and noise in N2, Ri and dynamic height anomaly as well as in maps of T and S on potential specific volume surfaces. These estimates represent the best that can be obtained if the instrument system has no other difficulties. Numerical examples for several instrument systems in current use show that significant errors can occur over scales as large as tens of meters.

Abstract

The development of high-resolution profiling instruments has made possible the computation of quantities such as the stability parameter N2 and the Richardson number Ri over scales of less than a meter. Quality control of the data has been done on an ad hoc basis by individual investigators who have used varying degrees of restraint in the claims made for the final results. As the amount of high-resolution data increases and as the parameters computed from it are used in statistical studies a more objective approach is required.

Profiting instruments are subject to a wide variety of difficulties, many of which are particular to the individual systems. Common to all, however, are errors due to uncertainties in the absolute calibrations of temperature T, conductivity C and pressure P, and noise in the sensors or data systems. Linearized equations for salinity S and specific volume α have been found to give good estimates of the errors resulting from typical values of calibration uncertainties and noise in T, C and P. Using these equations expressions have been developed for the evaluation of corresponding uncertainties and noise in N2, Ri and dynamic height anomaly as well as in maps of T and S on potential specific volume surfaces. These estimates represent the best that can be obtained if the instrument system has no other difficulties. Numerical examples for several instrument systems in current use show that significant errors can occur over scales as large as tens of meters.

Save