evaluation
Simulations performed with CAM6-chem have been extensively evaluated against satellite, ozonesondes, aircraft and ground observations of key pollutants on a global
and regional scale (e.g.,Lamarque et al.,2012,Tilmes et al.,2015,Val Martin et al.,
2015, Tilmes et al.,2016). Here we focus our evaluation on carbon monoxide (CO)
and ozone (O3) over the Amazon domain [90°–10°W Longitude, 20°S–10°N Latitude],
using aircraft campaigns and CO retrievals from the Measurement of Pollution in the Troposphere (MOPITT), and ozonesondes and long-term surface measurements of O3.
4.3.1
CO Observations
CO is one of the main gases emitted by biomass burning and an important ozone precursor, whose relatively long lifetime (weeks to months) make it an excellent
tracer for fire emissions (Edwards et al., 2006, Yurganov et al., 2010, Gatti et al.,
2014).
We use CO vertical profiles measured from three aircraft campaigns to assess CAMvert versus CAMsurf performance. These aircraft campaigns include the Re- gional Carbon Balance in Amazonia (BARCA Balanco Atmosferico Regional de Car-
bono na Amazonia), theGatti et al.(2014) flights and the South AMerican Biomass
Burning Analyses (SAMBBA). The dataset allows us to evaluate the impact of the smoke plume injection heights in different years and conditions. BARCA emerged as a combination of observations and analysis framework to quantify basin-scale carbon
fluxes with greenhouse gases and aerosols measurements over the Amazon (Andreae et al., 2012). BARCA was divided into two phases that represent the shift in the atmospheric conditions during the two seasons: BARCA–A, at the end of the dry season in 2008 (16 November–2 December), and BARCA–B, at the end of the wet season in 2009 (15–28 May). Flights in BARCA covered an altitude range from the
surface to about 4500 m over most of the Amazon Basin. Gatti et al. (2014) used
CO measurements as part of a sampling programme to observe climate sensitivity of the Amazon carbon pools during 2010 and 2011. The project consisted of bi-weekly
vertical profiles of CO at four sites across the Amazon: Alta Floresta (ALF; 8.80°S,
56.75°W), Rio Branco (RBA; 9.38°S, 67.62°W), Santarem (SAN; 2.86°S, 54.95°W)
and Tabatinga (TAB; 5.96°S, 70.06°W). Aircraft measurements were taken descend-
ing in spiral from approximately 4420 m to 30 m above sea level from 12:00 to 13:00 local time (LT). SAMBBA consisted of a combination of remote sensing, ground- based and aircraft measurements, which aimed to investigate the impacts of biomass
burning pollution over South America (Allan et al.,2014). SAMBBA was conducted
during the dry season of 2012 (14 September–3 October) and included 20 scientific flights sampling the Amazonian atmosphere from the surface up to almost 8 km.
MOPITT is a space-borne instrument aboard the NASA EOS Terra satellite designed to study the distribution, transport, sources, and sinks of CO in the tropo- sphere. It uses gas correlation spectroscopy to retrieve measurements of the emitted and reflected radiance from the Earth in three spectral bands. The amount of en- ergy absorbed into the sensor is correlated to the presence of CO in the atmosphere. MOPITT has a swath of 640 km and a spatial resolution of 22 km at nadir, and provides long record of global CO (since 2000) and global coverage every 3 days (Deeter et al., 2003).
We use daytime monthly joint retrievals (Thermal infra-red TIR and near infra- red NIR product) of the Level 3 Version 7, which provide total CO column and CO
profiles in ten levels, from the surface to 100 hPa (Deeter et al., 2003, 2014, 2018).
lower troposphere, particularly at daytime over land. To compare the MOPITT retrievals to our monthly CO simulations, we interpolate the model outputs to the
ten-level pressure (surface-100 hPa) and horizontal 1°x 1°MOPITT grid. Then,
we apply the a priori and averaging kernels included in the MOPITT retrievals to
account for the sensitivity of the retrieval to the truly observed profiles (Morgenstern
et al., 2012).
4.3.2
Ozone Observations
We use ozonesonde observations compiled by Tilmes et al. (2012) to assess the
performance of the simulated ozone vertical distributions. The dataset includes vertical ozone monthly profiles for 42 stations from 3 networks (World Ozone and Ultraviolet Data Center (WOUDC), the NOAA Earth System Research Laboratory (ESRL) and the Southern Hemisphere ADditional OZonesondes (SHADOZ)) for the period 1995–2011, and also regionally aggregated by similar ozone characteristics for their seasonal median and shape of pdfs. We focus on two regions: Equatorial Americas and Atlantic/Africa, which contain two and three stations, respectively
(Figure B2 in Appendix B; Tilmes et al. (2012)). For the comparison, simulated
ozone monthly outputs were first interpolated to the specific sondes locations and then aggregated and averaged by region and period.
In addition, we assess the ability of our simulation experiments to capture the spatial and temporal distribution of surface ozone. We evaluate our results with ground-based ozone measured at eight stations located within or close to the
Amazon, from the Tropospheric Ozone Assessment Report (TOAR; Schultz et al.
(2017)). These data are publicly available and free to download throughPANGAEA.
We use a minimum of two-years monthly means of the records available and coincid-
ent with the simulations period (2007–2012). (FigureB2 and Table B1). Simulated
monthly ozone mixing ratios at the surface are interpolated at the location of the stations and averaged for the run period.