Multispectral analysis of a tropical radiance set from the TIROS operational vertical sounder

Abstract

Analysis of the state of the atmosphere is constrained by the quantity and quality of the observations. In data sparse areas, meteorological analysis relies heavily on remotely sensed data, especially those collected by earth orbiting satellites. This research examines the information content of the TIROS-N Operational Vertical Sounder (TOVS) during 20-29 January 1979 over the tropical Pacific Ocean. Vertical, horizontal, and temporal statistical characteristics are examined. The TOVS channels are highly correlated (|r| [greater than or equal to]0.6) except for the infrared water vapor channels and the microwave window and 300 mb channels. The horizontal structure varies according to spectral channel (absorbing constituent and effective evaluation), geography, and synoptic condition. Horizontal correlation is particularly sensitive to water vapor and cloud amount. In equatorial sectors, moisture channels have higher correlations and larger length scales than thermal sensing channels; in the subtropics, the opposite is true. Temporal variation is largest in the water vapor and microwave window channels, and in synoptically active regions with brightness temperature variances typically ten times larger than in synoptically quiescent regions. Attempts to augment TOVS data in cloudy regions and for missing passes using a full quadratic response surface regression model were only marginally successful. At some locations, thermal infrared channels had brightness temperature RMS errors of 0.5 to 2.5 standard deviations from the mean, and the water vapor and microwave window channels had RMS errors of 2.5 to 5.5 standard deviations from mean values. The TOVS channels with peak energy contribution from below 90 mb were synthesized using principal components analysis. Over 94 percent of the areal variance at any particular time is represented by the first 5 eigenfunctions. The first two eigenfunctions, vertical mean thermal and mid-tropospheric moisture representations respectively, typically account for 65% and 15% of the observed spatial variance...