Comparison of Layer Thickness as Observed by Nimbus E Microwave Spectrometer and by Radiosonde

Abstract

Atmospheric layer thicknesses derived from Nimbus 5 (Nimbus E) Microwave Spectrometer (NEMS) are compared with radiosonde-derived thicknesses for selected short periods, and an average 45 m rms discrepancy is found for the 100?50 kPa layer. The several sources of this discrepancy are quantified in the following ways. Correlation coefficients between pairs of NEMS observations and between pairs of radiosonde observations are each extrapolated to zero separation distance to provide measures of instrument noise; NEMS noise is found to be 16 m rms and radiosonde noise 23 m rms. (This result is substantiated through comparisons of NEMS and radiosonde horizontal layer-mean temperature gradients with independent measures provided by assuming that the smoothed vertical wind profile as measured by rawinsondes is in thermal wind balance.) Correlation coefficient behavior with decreasing separation distance also leads to estimates for those portions of the total discrepancy which are due to different resolution of the sensors (?15 m rms) and real spatial and temporal variation of the atmosphere between observations (?17 m rms). Geometric addition of all contributions yields an estimate (?27 m rms) of NEMS errors (due to basic limitations of a remote sensing system) which vary on longer scales than those previously attributed to noise. Total NEMS error (noise plus more slowly varying error) is then 31 m rms (1.6 K) for the 100?50 kPa layer.