Quantifying TOLNet Ozone Lidar Accuracy during the 2014 1 DISCOVER-AQ and FRAPPÉ Campaigns

21 The Tropospheric Ozone Lidar Network (TOLNet) is a unique network of lidar systems that measure high- 22 resolution atmospheric profiles of ozone. The accurate characterization of these lidars is necessary to determine the 23 uniformity of cross-instrument calibration. From July to August 2014, three lidars, the TROPospheric OZone 24 (TROPOZ) lidar, the Tunable Optical Profiler for Aerosol and oZone (TOPAZ) lidar, and the Langley Mobile Ozone 25 Lidar (LMOL), of TOLNet participated in the “Deriving Information on Surface conditions from Column and 26 Vertically Resolved Observations Relevant to Air Quality” (DISCOVER-AQ) mission and the “Front Range Air 27 Pollution and Photochemistry Éxperiment” (FRAPPÉ) to measure ozone variations from the boundary layer to the top 28 of the troposphere. This study presents the analysis of the intercomparison between the TROPOZ, TOPAZ, and LMOL 29 lidars, along with comparisons between the lidars and other in situ ozone instruments including ozonesondes and a P- 30 3B airborne chemiluminescence sensor. In terms of the range-resolving capability, the TOLNet lidars measured 31 vertical ozone structures with an accuracy generally better than ±15% within the troposphere. Larger differences occur 32 at some individual altitudes in both the near-field and far-field range of the lidar systems, largely as expected. In terms 33 of column average, the TOLNet lidars measured ozone with an accuracy better than ±5% for both the intercomparison 34 between the lidars and between the lidars and other instruments. These results indicate very good measurement 35 accuracy for these three TOLNet lidars, making them suitable for use in air quality, satellite validation, and ozone 36 modeling efforts.

142 1993) to calculate differential ozone absorption cross-sections, which are temperature-dependent.

143
The ozone number density profile results from computing the derivative of the logarithm of the on-line to 144 off-line signal ratios. Spatial smoothing is usually necessary to improve the SNR and reduce the statistical errors.  Table 1. This effective resolution determines the capability of the lidars to resolve vertical ozone structure 152 and is not equal to, but is associated with, the fitting window width.

153
All groups applied similar schemes to correct the aerosol interference. These schemes iteratively substitute 154 derived ozone from the DIAL equation into the lidar equation to solve aerosol extinction and backscatter until both to specify a method for all groups because merging is system-dependent and is affected by many factors previously 163 described. Therefore, the three lidar groups merge the ozone profiles at different altitudes optimized for their system 164 and SNR levels such as the example method described by Sullivan et al. (2015). As a result, additional differences 165 between systems can occur due to the non-standardized altitude channel merging.

199
The three TOLNet lidars were deployed next to the BAO tower to take simultaneous measurements before 200 the DISCOVER-AQ/FRAPPÉ campaign. They were only a few hundreds of meters away from each other and were 201 within 5 m of the same elevation (see measurement locations in Table 1).

202
Unlike stratospheric ozone lidars that focus on integrating hours of observations, tropospheric ozone lidars

215
The product of averaged ozone concentration over some specified altitude range can represent the atmospheric ozone 216 abundance and can be also useful for satellite validation. Here, we refer this product as ozone column average with 217 the unit of number density, not to be confused with integrated column ozone often with a unit of the Dobson Unit. The

251
Between July 10 and July 16, a total of 10 ozonesondes were released near the BAO tower and 7 of them 252 were coincident with TOPAZ measurements (3 on July 10, 3 on July 11, and 1 on July 16). TOPAZ mostly agrees 253 with ozonesondes between -5% and 10% (Figure 3 c, d).

269
During the campaigns, the P-3B aircraft measured ozone profiles while doing spirals above the lidar sites.

270
There are 34 coincident profiles between TROPOZ and the P-3B at Fort Collins, 29 between TOPAZ and the P-3B at 271 the BAO tower, and 9 between LMOL and the P-3B at Golden, CO. The distances between the lidar and P-3B spiral 272 center for these paired profiles were less than 11 km. To make coincident pairs between P-3B and lidar data, we 273 interpolate the P-3B data onto the lidar vertical grids with a 15-m vertical resolution. Figure 4 shows the average ozone 274 profiles measured by the lidars and the P-3B as well as their mean relative differences. TROPOZ and the P-3B agree 275 with each other within ±5% between 0.5 to 3.5 km (Figure 4 a, b) with a -0.8% overall average relative difference.

276
TOPAZ agrees with the P-3B within -11% and 3% between 0.5 and 2 km (Figure 4 c, d) with a -2.7% overall average 277 relative difference. TOPAZ underestimates the lower-PBL (<1.5 km) ozone compared to P-3B, but when compared 278 to ozonesondes TOPAZ overestimates ozone at many of these same altitudes (see Figure 3

281
In summary, TOPAZ and LMOL exhibited noticeable negative bias in the PBL compared to the P-3B while 282 TROPOZ measured slightly lower than the P-3B. The differences between the two lidars and the P-3B are not 283 significantly correlated suggesting that the problem was not likely from the P-3B ozone instrument. These differences 284 could at least in part be caused by the lidar systematic errors we mentioned earlier in Section 2.1.5, but could also 285 reflect horizontal ozone variability across the P-3B spirals, which were up to 22 km in diameter. (2) TROPOZ agrees with the P-3B chemiluminescence Instrument below 3.5 km within ±5% with a small column-averaged relative difference of -0.8%. TOPAZ and LMOL exhibit a slightly larger bias mostly between 298 -15% and 5% below 2 km compared to P-3B with a column-averaged difference of -2.7% and -4.9%, respectively.

299
Overall, the TOLNet lidars are capable of capturing high-temporal tropospheric ozone variability and 300 measuring tropospheric ozone with accuracy better than ±15% in terms of their vertical resolving capability and better 301 than ±5% in terms of their column measurement. These lidars have sufficient accuracy for model evaluation and