ACCURACY OF SPECTRAL UV MODEL-CALCULATIONS .1. CONSIDERATION OF UNCERTAINTIES IN INPUT PARAMETERS

Authors
Citation
P. Weihs et Ar. Webb, ACCURACY OF SPECTRAL UV MODEL-CALCULATIONS .1. CONSIDERATION OF UNCERTAINTIES IN INPUT PARAMETERS, JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES, 102(D1), 1997, pp. 1541-1550
Citations number
72
Categorie Soggetti
Metereology & Atmospheric Sciences
Volume
102
Issue
D1
Year of publication
1997
Pages
1541 - 1550
Database
ISI
SICI code
Abstract
Even in idealized, clear sky conditions, spectral UV models need well- specified input parameters such as solar zenith angle, pressure, ozone and aerosol optical depth to produce realistic results. UV models bas ed on exact analytical methods are widely considered to be, absolutely accurate with errors in results widely attributed to inaccurate input parameters. The uncertainties in the standard determinations of each input parameter were determined for the following methods of specifyin g each parameter: (1) ozone from satellites or ground-based Dobson spe ctrometer measurements; (2) aerosol optical depth from sunphotometer m easurements: (3) single-scattering albedo; three different procedures using ground-based spectrometer measurements and radiative transfer ca lculations were studied; (4) ground albedo; the determination of the g round albedo is achieved using satellite measurements, winch have a re solution of some 100 km; ground-based measurements may give some infor mation about the albedo of the ground, but the albedo needed for radia tive transfer calculations is a weighted average of the surrounding al bedos. The resulting inaccuracies in UV calculations from errors in in put parameters were simulated with the discrete ordinate method model by Stamnes et al. [1988] using the uncertainties specified above for t he different input parameters. The simulations of tile errors of the s pectral UV calculations were carried out for each input parameter at 0 degrees, 50 degrees, and 80 degrees zenith angle and analyzed at 305 and 380 mn. The overall maximum error may be determined by combining t he different errors.