JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 101,NO. D14, PAGES 19,337-19,342, AUGUST 27, 1996 Spatial resolution of a transport inversion technique Luis A. de P. Vasconcelos Department of Chemistry, Washington University,St. Louis,Missouri Jonathan D. W. Kahl andDesong Liu Department of Geosciences, University of Wisconsin, Milwaukee Edward S. Macias and Warren H. White Department of Chemistry, Washington University,St. Louis, Missouri Abstract. A spatial calibration for conditional frequency analysis, the inversion technique applied to routine back trajectories to study theassociation between geographical regions and conditions at a receptor, is discussed for trajectories arriving at Hopi Point, Arizona. Several 0.5 øx 0.5 ø cells wereselected at three different distances fromHopi Point. These cells were used to mimic sources, with air residing over them considered contaminated by imaginary emissions. The distribution of back trajectory segment endpoints overa latitude-longitude grid wasthenexamined as a function of the trajectories sojourn overeach imaginary source cell on theirway to thereceptor. The areal extent of significant associations, evaluated for different seasons and thevarious idealsource positions, suggests that this inversion technique canresolve the direction of potential sources but haslimited resolution for their radial dis- tance from the receptor. introduction Back trajectory analysis is instrumental in establishing the spatial domainof sources contributing to the composition of air parcelsarriving at selected receptorsites [National Re- search Council (NRC), 1993]. A group of methods known collectively as residence time analysisuses statistical tech- niques to extract patterns of empirical association between the trajectories andthe concentrations of arriving air [White et al., 1994]. The type of analysis examined herewas developed by Ashbaugh et al. [1985] to identify source areas influencing sulfurconcentrations at GrandCanyon NationalPark. Bresch et al. [1987] and Gebhart and Malm [1991] extended the analysis to other national parks and carbonaceous species concentrations.The results presented are an extension of a previously developed calibration [Vasconcelos et al., this is- sue] anddocument the method's spatial resolution. Vasconcelos et al. [this issue] described conditional fre- quency analysis as an investigation of the spatial distribution of backtrajectory segment endpoints on a grid superimposed overthe region containing the receptor. The resolving power of the method is limitedby the variability of the trajectories arrivingat the receptor. If air passing over a source cell al- ways passes overits neighbor andvice versa, it is not possible to determine which cell contains the source based on the characteristics of the air mass measured at the receptor. In this paper we studythis limitationby examining the case when the criterion of selection of the air masses is their so- journover specific cells before arriving at thereceptor. These cellsmimic sources in the sense that any air parcel passing Copyright1996 by the American Geophysical Union. Papernumber96JD01659. 0148-0227/96/96JD-01659509.00 over them is considered contaminated by their imaginary emissions. Trajectories The Air Resources Laboratories atmospheric transport and dispersion (ARL-ATAD) model [Heffter, 1980] was usedto estimate 3-day backward trajectories arriving4 times daily at Hopi Point, Arizona (36ø07'N-112ø15'W;2,150 meters above mean sealevel (msl)). Input to the model consists of vertical profilesof wind, temperature, and pressure measured at Na- tional Weather Servicerawinsonde stations. Air parceltrans- port is simulated astwo-dimensional advection usinghorizon- tal winds averaged throughout the mixed layer, which is de- termined using the temperature and pressure profiles. The mixed layer is variable and may reach up to 3000 m in depth. The main output of the modelis a setof 24 latitude-longitude coordinates that give the estimated position of an air parcel every3 hours during the 72 hours prior to its arrivalat the re- ceptor. Occasionally, trajectories terminate prematurely or are missing entirelydue to gaps in the input meteorological data set or to advection out of the computational domain. In this study, incomplete trajectories are considered invalid and dis- carded. Inversion Method The methodology followed here has been described in de- tail by Vasconcelos et al. [thisissue]. A grid with 0.5 ø longi- tude by 0.5ø latitudecells is superimposed over the region defined by 130øW-95øW and 25øN-50øN. The ATAD trajec- tories useddescribe parcels arrivingin intervals of 6 hours, at 0000, 0600, 1200, and 1800 UT. The position of a parcel is logged every 3 hours by a "timestamp," which represents one 19,337