Cloudy sky optical paths as derived from ... - Wiley Online Library

3 downloads 0 Views 2MB Size Report
Oct 20, 1998 - measurements, edited by J.A. Pyle, N.R.P. Harris, and G.T. Amana- tidis, Air Pollut. Res. Rep., 56, Eur. Comm., Brussels, Belgium, Bun-.
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 103, NO. D19, PAGES 25,307-25,321,OCTOBER 20, 1998

Cloudy sky optical paths as derived from differential optical absorption spectroscopyobservations T. Wagner,F. Erie, L. Marquard,C. Otten,K. Pfeilsticker,T. Senne, J. Stutz, and U. Platt Institutff•r Umweltphysik,Universityof Heidelberg,Heidelberg,Germany

Abstract. Recently,the influenceof troposphericcloudson zenithsky observations of atmosphericgases(03, NO2, 04, H20, andothers)hasbeenreported[Van Roozendaelet al., 1994; Erle et al., 1995]. For measurements underheavycloudcover,Erle et al. [ 1995] determinedfrom simultaneously measuredabsorptions of differenttropospheric speciesan averageopticalpathfor photonsreachingthe instrumentof > 100 km. This pathlengthwasmuchlargerthanexpected from previousreports[Feigelson,1981]. Thusan importantquestionwaswhetherthe magnitude of the observationwas a typical situationratherthana rare exception.Here we add a largesetof new "cloud event" observationsevaluated from data which were recorded in the Arctic, at midlati-

tudes,andthe tropics.It is shownthatthe observedabsorptionenhancements of 03, H20, and04 perfectlycoincidewith indirectcloudindicators,i.e., changesin the averageintensityandthe colorindexaswell asdirectobservations of cloudpassages detectedby satellite.In particular,we demonstrate that the observedopticalpathshavethe samemagnitudeasthe path lengthsreported by Erle et al. [1995], andthusopticalpathsundercloudyskiesmay indeedbecomemuchlarger thanpreviouslyreported.Besidestheseobservations madeduringspecificmeteorologicalconditionswe alsostudiedopticalpathscontinuously for severalweeksin Arctic winter/springconditions.We deriveda frequencydistributionof pathlengthsfor cloudyskieswith an averagemean opticalpathof 29 km.

ful tool for deriving optical pathsfor the light transmittedto the instrument [Erie et al., 1995; Pfeilsticker et al., 1998]. The Light path enhancement due to tropospheric cloudsis a well- methodto derive opticalpath lengthchangesdue to cloudsrelies known phenomenonin the field of the atmosphericradiative on a comparisonof measuredatmosphericabsorptionsobserved transport[Van de Hulst, 1957, 1980; Fouquart and Lenoble, under cloudy and clear sky conditions,respectively,at the same 1972,' Feigelson, 1981,' Stephens,1978a, b, Stephenset al., solar zenith angle (SZA) for absorbersof known (or independ1978]. In moststudiesthe impactof tropospheric cloudsmodify- ently measured)concentrationssuch as 02, 03, 04, and H20 in ing optical paths is discussedin terms of energy deposition the UV/visible or near-IR spectralrange. (For the radii (r) of [Fouquartet al., 1990, andreferences therein]or photolysis[e.g., cloud particles the Mie parameter 2rtr/X for UV/visible light Luther and Gelinas, 1976, and referencestherein]. Besidespo- rangesfrom =20 to =50. Thus from the Mie theory only a weak tentially changing the atmospheric absorber densities (by wavelengthdependenceis expected[Van de Hulst, 1980]. For (photo-)chemistryor convectivetransport)the effect of clouds polydispersiveparticle size distributionstypical for cloudsit reon spectroscopic measurements is believedto be more important sultsin nearly no wavelengthdependencefor the light scattered in modifyingthe opticalpaths.While the impactof stratospheric inside the cloud (usually clouds appear white). However, for clouds on these measurementswas recognized previously specificmeasurementconditions(broken cloud cover) the sensi[Sarkissianet al., 1991; Fledlet et al., 1993; Pfeilstickerand tivity of our methodshowsa strongwavelengthdependencebePlatt, 1994],the influenceof tropospheric cloudshasso far been causeof the differentRayleighscatteringprobabilityfor photons assumedto be negligible[see,e.g., Sarkissianet al., 1997, and in the UV and visible spectralregion(for detailsseethe lastpart referencestherein]. In contrastto theseearly studies,recentin- of section4.1). Accuratecloudy sky optical path observations(our measurevestigationsfound a stronginfluenceof troposphericcloudson the zenith scatteredlight differentialoptical absorptionspectros- mentsrefer to photonshavingpenetratedthe cloud)are of great copy(ZSL-DOAS) of stratospheric tracegases[Van Roozendael importancefor severalreasons: 1. Measuredoptical pathsmay help to establishmore accuet al., 1994; Erie et al., 1995]. While it is becomingclearthat the impactof troposphericcloudsmustbe consideredin thesemeas- rately the absorption by clouds under heavy cloud cover urements,ZSL-DOAS measurements may also provide a power- [Stephensand Tsay, 1990; Cesset al., 1995,'Ramanathanet al., 1. Introduction

1995; Evans et al., 1995; Liet al., 1995; Pilewskie and Valero, Copyright1998 by the AmericanGeophysicalUnion. Papernumber98JD01021. 0148-0227/98/98JD-01021

$09.00

1995; Imre et al., 1996; Pfeilstickeret al., 1998]. 2. Experimentaldata on cloudy sky optical paths provide a more detailed insight into photolysis rates inside clouds [Madronich, 1987,'Junkermann,1994]. 25,307

25,308

WAGNER ET AL.: CLOUDY SKY OPTICAL PATHS

3. They can be used to test radiative transfermodels,which, for example, are particularlyimportantfor the interpretationof zenith sky observationsof atmospherictracegases[Pfeilstickeret al., this issue]. 4. Once the optical pathsfor cloudy skiesare established,the methodmay also offer a chanceto obtain data on the trace gas abundances insideclouds.In particular,thosemeasurements may also provide information on interstitial 03 in cloudsor the NOx productiondue to thunderstorms. In earlier studiesthe oxygen A-band absorptionmeasuredat low spectralresolution(•1 nm) was frequentlyused to derive optical paths[Feigelson, 1981]. However, becauseof line saturation effects (the optical density (equation2) of individual rotational lines reaches -r'• 1000), oxygen A-band measurements with low resolvingspectrographs suffer from someexperimental shortcomings:First, suchan instrumentalsetupis not very sensitive to a light path enhancementfor cloudy skiessinceeven a large enhancementof the absorptionpath increasesthe integral absorptionof an already saturatedline only weakly. Second,in spectralranges of strong absorbersthe apparentmean optical pathis systematically shorterthanfor weakabsorbers, mainlybecauseof the lower probabilityof photonswith the largestpaths beingobserved[e.g.,Fouquartand Lenoble,1972;Pfeilstickeret al., 1998]. Both reasonscan lead to considerableunderestimation of the opticalpaths[Wagneret al., 1996; Pfeilstickeret al., this issue].In contrast,here we presentoptical path data from observationsof weak (-r' 6.7 or 3.3, respectively.The meanopticalpathlengthfor the transmittedphotonsis a functionof severalcloud parameters [Marshaket al., 1995; Davis et al., 1997]:. ,

SCD 04

0.06

o.04

0.02 o. oo2

0.001

N...

0.000

,

.OE-3

5.0E-4

O.OE+O •



3E+4Intensity 2E+4

cloudy sky ' ' ,clear sky

1E+4

0E*0

,

,• •._. ,

.._.....

,

_,

2 Color-index

=( 1-g)('c)¾0•Mie) !

II

( 11)

Here )•Mieis the mean free path betweenMie scatteringevents, and ¾ is the L6vy index (K. Pfeilsticker,First geometricalpathTime (UT) lengths probability density function derivation of the skylight highly resolvingoxygenA-band observaFigure 8. Simultaneousabsorptionincreaseof severalabsorbers from spectroscopically at Kiruna on March 5, 1994. Additionally, the diurnal variation tions,2, Derivationof the L6vy-indexfor the skylighttransmitted of the color index and the intensityare displayed. by midlatitudeclouds,submittedto Journal of GeophysicalRe0712

i

0936

1200

1424

WAGNER ET AL.' CLOUDY SKY OPTICAL PATHS

25,317 ...

05.03.1994, 12 00 UT :. ::T•, .-=-"- -

?•Z:---:.'?: :_•:_ .-7..• .2, .....•.•,,.-"-•._--7_-•.•.•:•: • ---..:.. . - '•',•:Z