Pergamon Atmospheric Environment Vol. 31, No. 4, pp. 609 620, 1997 Copyright © 1996 Elsevier Science Ltd Printed in Great Britain. All rights reserved PII: S1352--2310(96)00199--9 1352-2310/97 $17.00 + 0.130 SENSITIVITY ANALYSIS OF A UV RADIATION TRANSFER MODEL AND EXPERIMENTAL PHOTOLYSIS RATES OF NO2 IN THE ATMOSPHERE OF MEXICO CITY T. CASTRO,* L. G. RUIZ-SUAREZ,*t J. C. RUIZ-SUAREZ,~" M. J. MOLINA:~ and M. MONTERO* * Centro de Ciencias de la Atm6sfera, UNAM. Circuito Exterior de Cd. Universitaria, M6xico, D.F. 04510; t Departamento de Fisiea, CINVESTAV-IPN, Unidad M6rida, AP 73, CORDEMEX 97310, M6rida Yuc., M6xico; and ~ Chemistry Department, MIT, Cambridge, MA 02139, U.S.A. (First received 19 December 1995 and in final form 19 June 1996) Abstract--Photolysis of key species such as nitrogen dioxide, ozone and aldehydes, are the elementary reactions leading to the formation of photochemical smog. Calculations and experimental measurements (in particular, nitrogen dioxide) of these key reactions are reported, as well as, the sensitivity of them to the ozone column, local albedo and properties of the urban aerosol layer in Mexico City. A radiation transfer model (RTM), based on the delta-Eddington approximation, was used in the calculations. The results show the importartce of providing local resolution for photolysis rates by considering different local conditions within the modeling domain of air quality models. Copyright © 1996 Elsevier Science Ltd Key word index: Nitrogen dioxide, ozone, formaldehyde, photolysis rates, Mexico City. ]INTRODUCTION Air quality models require among other parameters, of the specification of photolysis rate constants (the so-called "J value~¢') of reactions occurring in the troposphere. The most important reactions are: 0 3 "~- hv --* 0 2 -~- O(1D) (R1) NOz + hv ---, NO + O (R2) HCHO + hv --* CO + H2 (R3) HCHO + hv --* HCO + HO2. (R4) These reactions take place in several regions of the solar spectra, mainly between 280 and 420 nm. The constants of dissociation of reactions (R1) and (R2) have been studied experimentally under various at- mospheric conditions. In such experiments a gas mix- ture, placed inside a quartz tube, is exposed to UV radiation. The change of gas concentration after the reaction and the products of secondary reactions are measured following the exposure to solar radiation. The rate of reaction is determined by the time of exposure, the initial concentration of the gas, the concentration of secondary products and by the stan- dard chemical kinetics considerations. In this way, the rates of reactions (R1) and (R2) have been measured in the past, under several atmospheric conditions by several workers (Jackson et al., 1975; Harvey et al., 1977; Zafonte et al., 1977; Sickles et al., 1978; Dickerson et al., 1982; Parrish et al., 1983; Mad- ronich et al., 1983; Junkermann et al., 1989; Shetter et al., 1992; Brauers and Hofzumahaus, 1992). On the other hand, the calculation of J's by radiation transfer models requires major simplifications which usually introduce uncertainties as large as 20% (Madronich, 1987). In these models the transfer equation is solved and the actinic flux is thus obtained (Demerjian et al., 1980; Ruggager et at., 1993, 1994). Recently, prelimi- nary calculations of solar irradianee and photolysis rates for NO2, 03 and HCHO under Mexico City conditions were reported for the first time (Ruiz- Sufirez et al., 1993 a, b). This work has three aims: (a) perform a sensitivity analysis of calculated J's to changes in albedo, to load and refraction index of aerosols, and ozone column. This analysis is used to select possible scenarios for the experiments and to understand the underlying behavior of rate constants when different atmospheric conditions are present, (b) carry out an experimental program to obtain the nitrogen dioxide photolysis rates around the city and (e) adjust our model by comparing the theoretical results with these experimental measurements. In this way, more realis- tic routine calculations of J's can be done for Mexico City. Furthermore, an additional comparison is done against two models reported by (Madronieh, 1987): J = C[f -1 + 2AL]E (1) 609