River Breezes for Pollutant Dispersion in GoAmazon 2014 / 5

1 The effect of river breezes on pollutant plume dispersion or canalization in the central 2 Amazon was evaluated. A pollution plume changes atmospheric composition downwind of 3 Manaus, a city of 2 million people positioned at the confluence between two wide rivers. Herein, 4 to evaluate the effects of river breezes, two cases were modeled at the mesoscale for March 5 2014. The first case, “with rivers” (wR), simulated the transport and chemistry of the Manaus 6 pollution plume as the rivers were in reality. The second case, “without rivers” (woR), carried 7 out simulations for which all rivers and floodable areas were replaced by forest. The three main 8 conclusions are as follows: (1) Between the two cases, alterations in wind speeds were maximum 9 at local noon, and river breezes influenced horizontal wind fields from surface up to 150 m in 10 altitude, suggesting a capping height of 150 m on most days for the influence of river breezes on 11 pollutant concentrations. In agreement with this modeling result, data sets collected at 500 m by 12 aircraft flights showed no apparent influence of the underlying rivers on plume dispersion. The 13 flights traversed the plume downwind of Manaus during the Observations and Modeling of the 14 Green Ocean Amazon (GoAmazon2014/5) Experiment. (2) Between the wR and wOR cases, 15 changes to downwind concentrations of O3, NOx, and CO pollutants were < 6% as a monthly 16 average at the supersite “T3” of GoAmazon2014/5, which was 70 km downwind of Manaus and 17 located between the two main rivers. As single events at T3, maximum one-hour concentration 18 differences were 39 ppb for O3, 5 ppb for NOx, and 26 ppb for CO. (3) For a focus on the surface 19 layer of the rivers (0 to 150 m in height), river breezes increased the monthly average O3, NOx, 20 and CO surface concentrations by 25%, 25%, and <5%, respectively. In addition, strong 21 canalization occurred 5% of the time based on a difference of 10 ppb in the surface 22 concentrations of at least two of O3, NOx, and CO between the wR and wOR cases. In 23 conclusion, although pollutants dispersed by river breezes could at times be a strong effect on 24 observed pollutant concentrations in the surface river boundary layer, overall most pollution was 25 transported at heights well above the effects of the river breezes and moved downwind along the 26 trajectories of the dominant trade winds. 27 2 Atmos. Chem. Phys. Discuss., https://doi.org/10.5194/acp-2018-347 Manuscript under review for journal Atmos. Chem. Phys. Discussion started: 23 April 2018 c © Author(s) 2018. CC BY 4.0 License.


Introduction
Amazonia represents the single largest hydrographic basin of water volume on Earth (Sioli, 1984).Land coverage by rivers constitutes 5% of the total 7 million square kilometers of the Amazon basin during dry season, while in the wet season the rivers increase in horizontal extent by flooding, reaching a surface coverage of 11% of Amazonia (Hess et al., 2015).An important confluence of wide rivers occurs nearby Manaus, a city of more than 2 million people located at {3.0° S, 60.0° W} in the central Amazon (Figure 1).The Rio Negro ("Black River") flows from the northwest to join the Amazon River, known in Brazil as the Rio Solimões to the west of Manaus and the Rio Amazonas to the east of Manaus.River width around Manaus varies from 2 km in narrow sections to 20 km in broader sections.
Wide rivers such as these can induce important atmospheric processes, among which are river breezes (Oliveira and Fitzjarrald, 1993;Dias et al., 2004;dos Santos et al., 2014).River breezes arise from the unequal heating of land and water bodies.In the morning, land heats faster than water, inducing an ascendancy of air over the land and a corresponding subsidence over the river.In this way, surface winds go from the river toward the land.At an altitude of a few hundred meters, the circulation cell is closed, and the winds go from the land to the river, with subsidence over the central portion of the river.The height of the cell depends on the thermal characteristics of the circulation.At night, the opposite behavior occurs (i.e., the cell reverses) because the river cools more rapidly than land.
These river breeze circulations at day and night can be important for the local weather and pollutant dispersion.For instance, during times of weak trade winds Dias et al. (2004) found that river breeze circulation explained the occurrence of clouds on the eastern bank yet an Oliveira andFitzjarrald (1993, 1994) studied the river breezes in the Manaus region during the Amazon Boundary Layer Experiments (ABLE) (Garstang et al., 1990;Harriss et al., 1990).Based on observations of the meridional component of wind speed, the river breezes were reported as more intense during the dry season than in the wet season, as explained by greater contrast between river and land temperatures given the greater average insolation of the dry season.Simulations further suggested that the river breeze induced by the Rio Negro significantly affected the surrounding daytime surface winds to a distance of 20 km from the rivers (Oliveira and Fitzjarrald, 1994).The modeled distance was further than initially expected based on earlier modeling studies, and the key difference appeared to be an improved representation of the planetary boundary layer (PBL) in the model.
As part of the Large-Scale Biosphere-Atmosphere Experiment in Amazonia-Cooperative LBA Airborne Regional Experiment-2001(LBA-CLAIRE-2001), Trebs et al. (2012) traveled by boat to four locations on the Rio Negro and one on the Solimões River.Daily reversals in surface winds were attributed to river breezes.Measurements were made of NO, NO 2 , and O 3 surface concentrations, and pollution was identified at surface river locations from 10 to 150 km downwind from Manaus.On at least one day, a reversal in wind direction caused by the afternoon influence of the river breeze was associated with a shift in concentrations representative of background and polluted conditions.Manaus pollution directed by the river breezes appeared to be the explanation.The important data sets of this study were, however, overall sparse (i.e., 8 days of July 2001; Manaus population of 1.2 million at that time), and the recommendation by the authors was therefore to implement long-term monitoring stations In 2014 and 2015, the Observations and Modeling of the Green Ocean Amazon (GoAmazon2014/5) Experiment was carried out to study the effects of pollutant outflow from Manaus on atmospheric chemistry, regional climate, and terrestrial ecosystems of an otherwise typically clean background environment (Martin et al., 2016).Under fair-weather conditions, the pollution plume was carried westward by equatorial trade winds (Kuhn et al., 2010;Martin et al., 2017).The GoAmazon2014/5 terrestrial supersite, called "T3", was 70 km to the west of Manaus (Figure 1).
An important question for the GoAmazon2014/5 experiment was to what extent river breezes might disperse or canalize Manaus pollution, thereby possibly influencing the interpretation of data sets collected at the T3 supersite.For a limiting case of full river canalization, no pollution would reach the T3 site.For an opposite limiting case of weak or no river breeze effects, all pollution would follow the stable trade winds when fair-weather conditions prevailed, and air parcels sampled at the T3 site would be interpreted in a fully Lagrangian framework downwind of the Manaus source region.Between these limiting cases, partial dispersion of the Manaus pollution plume would be possible.In the context of these possibilities and in light of the work of Oliveira andFitzjarrald (1993, 1994)

Model Description
The Weather Research and Forecast model coupled with Chemistry (WRF-Chem) is described by Grell et al. (2005).Version 3.6.1 was used for the present study.A two-domain configuration was used (Medeiros et al., 2017).The inner domain represented an area of 302 km × 232 km, had a horizontal resolution of 2 km × 2 km, and had 38 vertical layers from ground to 160 hPa.The outside boundaries of the inner domain were forced by data from an outer domain.
The outer domain represented an area of 1050 km × 800 km, had a resolution of 10 km × 10 km, and had 38 vertical layers from ground to 160 hPa.Both domains were centered on {2.908° S, 60.319° W}.The meteorology of the outside boundary of the outer domain was forced by the Climate Forecast System Reanalysis (CFSv2) product of the National Center for Environment Prediction (NCEP) at a temporal resolution of 6 h and a spatial resolution of 0.5° (Saha et al., 2011).The inputs of surface temperature were also considered based on CFSv2 product.The chemical composition of the outside boundary of the outer domain was forced by the Model for Ozone and Related chemical Tracers (MOZART-4) (Emmons et al., 2010).
Data of the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite at a resolution of 500 m were used for land cover (Channan et al., 2014).These data were used as obtained for the case of "with rivers" (wR).For the case of "without rivers" (woR), the rivers and main flooded areas of MODIS land cover were replaced by forest in the pre-processor of the WRF-Chem model.
The physics parametrizations used in the simulations were described previously (Ying et al., 2009;Misenis and Zhang, 2010;Gupta and Mohan, 2015), including for the study region in the central Amazon (Medeiros et al., 2017).The parametrizations treated the physics of the surface layer (Grell et al., 1994), the land surface (Chen et al., 1997), the boundary layer (Hong et al., 2006), shortwave radiation (Chou and Suarez, 1999), longwave radiation (Mlawer et al., 1997), cloud microphysics (Lin et al., 1983), and cumulus clouds (Grell and Freitas, 2014).At Figure S1, the comparison between observed and simulated temperature, relative humidity and wind speed at "T3" supersite show that the simulations performed herein represent the diurnal cycle of these variables.

For chemical parametrizations, the Model of Emissions of Gases and Aerosols from
Nature (MEGAN, version 2.1) (Guenther et al., 2012) was used for biogenic emissions.
Anthropogenic emissions from transport, power, and industry for Manaus in 2014 were based on the emission inventory of Medeiros et al. (2017).The Regional Acid Deposition Model (RADM2) was used to simulate gas-phase chemistry (Chang et al., 1989).

Model Runs
Simulations of the wR and woR cases were carried out for all days in March 2014.Other characteristics between the two simulations remained the same.This approach aimed to isolate the river breeze effects on the transport of pollutants downwind of Manaus.For time zero, the inner and outer domains were both initialized to CFSv2 and MOZART-4.A spin-up time of 24 h was used, which was sufficient to fully replace the air of the inner domain.After the spin-up period, simulations in lots of 72 h were performed for March 2014 as a balance between conserving computing resources and avoiding excessive numerical drift (Medeiros et al., 2017).

Data Sets
Data sets were collected during the first intensive operating period (IOP1) of the

Height of River Breeze Circulation Cell
The effect of river breezes on horizontal wind speeds was evaluated.As monthly means, the left column of Figure 2 presents the wR case, and the right column shows the differences between the wR and woR cases.The rows represent plots at surface, 100-m altitude, and 500-m altitude.Comparison between columns shows that the river breezes significantly affected mean surface wind speeds but that the effects decreased with altitude.
The change of horizontal wind speeds with altitude is presented in detail in Figure 3 through height cross sections along points A, B, and C above the Rio Negro nearby Manaus (cf. Figure 1).Panels in Figure 3 show wind speed differences between the wR and woR cases for all times as well as for 00:00, 06:00, 12:00, and 18:00 (local time).In the absence of solar radiation (i.e., at 00:00 and 18:00 local), the differences in horizontal wind speeds were relatively small.The strongest differences were at noon corresponding to maximum daily solar irradiance, as expected, because of the largest thermal gradients between land and river at these times (Oliveira andFitzjarrald, 1993, 1994;Dias et al., 2004;Fitzjarrald et al., 2008;de Souza and dos Santos Alvalá, 2014;dos Santos et al., 2014;de Souza et al., 2016).The river breeze effect was confined to less than 150 m, as shown in the monthly average plot of For comparison, aircraft data sets of O 3 , NO x , and CO concentrations from 500-m altitude are plotted in Figure 4 (Martin et al., 2017).Carbon monoxide was mostly inert on the time scales of the simulations, oxides of nitrogen were significantly lost during downwind transport, and ozone was a secondary pollutant rapidly produced within Manaus and over the nearby rivers, quickly reaching steady-state concentrations (Medeiros et al., 2017;Rafee et al., 2017).The panels of the left column of Figure 4 show that the flight paths intercepted the Manaus pollution plume in the planetary boundary layer on March 14 from 10:20 to 11:20 (local time; UTC -4 h).
The panels of the right column show that interception took place on March 21 from 13:00 to 14:00.Below each map, concentrations along the flight tracks are plotted, and the red shading represents times that the aircraft was over a river.The results show that there was no obvious influence of river breezes on the dispersion of the Manaus pollutant plume at 500 m.Although Figure 4 shows only two flights, which were selected for substantial data coverage over both river and land during a single flight, all 16 flights from March 2014 were investigated, and a strong river breeze effect was not apparent in any of them (analysis not shown).The aircraft data sets thus corroborate the tendencies represented in Figures 2 and 3 that the effects of rivers presence on plume dispersion are confined to the surface boundary layer over the rivers, typically below 150 m.

Effects of River Breezes on Downwind Concentrations
The "T3" GoAmazon2014/5 terrestrial supersite was located at {-3.2133 N, -60.winds that passed through Manaus (Figure 1).The panels in the top row of Figures 5, 6, and 7 show O 3 , NO x , and CO concentrations, respectively, at the T3 location for the wR case as the blue line, the woR case as the red line, and their difference as the black line.The pollutant concentrations did not change greatly in the presence or absence of the rivers.Quantitatively, the perturbations caused by the presence of the rivers on the O 3 , NO x , and CO concentrations were less than 6%.Maximum one-hour concentration differences were 39 ppb for O 3 , 5 ppb for NO x , and 26 ppb for CO across the month.The overall implication is that the effects of the trade winds on transport largely dominated over the influence of river breezes in this region when considering the larger part of Manaus pollutant outflow, in agreement with the modeling and observational results of section 4.1.

Surface River Concentrations
Two locations "R1" and "R2" were chosen to evaluate pollutant dispersion and canalization in the river surface layer (0 to 150 m).Based on the prevailing trade winds location R1 at {-3.0699 N, -60.2199E} consistently intercepted the urban outflow (Figure 1).By comparison, location R2 at {-2.9800 N, -60.4901E} was nominally outside of the trajectories of the trade winds that passed over and then downwind of Manaus.For the analysis, the simulated pollutant concentrations at R2 were compared to those at R1 in the wR and woR cases to test the extent of river canalization of the plume, meaning transport of air parcels along the river rather than in the prevailing direction of the synoptic-scale trade winds.
The time series of the simulated O 3 , NO x , and CO concentrations at the R1 and R2 locations are plotted in Figures 5, 6, and 7. Blue lines show the wR case, red lines show the woR case, and black lines represent their difference.At R1, the differences of (wR -woR) were significant, in particular for concentrations of O 3 (+28.4%)and NO x (+26.0%)(Table 1,  the Rio Solimões arrives from the west.The confluence of the two rivers is to the southeast of Manaus, beginning the Amazon River ("Rio Amazonas").Yellow markers show locations of (i) the measurement supersite called "T3", (ii) two river locations "R1" and "R2" considered in the modeling methodology herein, and (iii) points "A", "B", and "C" along a river cross section, also used in the methodology herein.Table 1.Percent change in pollutant concentration X for the case of wR ("with rivers") compared to that of woR ("without rivers"), calculated as (X wR -X woR )/X wR , where X is the monthly mean at a location T3, R1, or R2 for each of O 3 , NO

4
Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2018-347Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 23 April 2018 c Author(s) 2018.CC BY 4.0 License.downwind Manaus and to apply mesoscale modeling to better understand river breeze effects on the dispersion of the Manaus pollution plume.

GoAmazon2014/ 5
project by instrumentation of the G-159 Gulfstream I (G-1) of the ARM Aerial Facility (AAF) of the USA Department of Energy (Schmid et al., 2014; Martin et al., Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2018-347Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 23 April 2018 c Author(s) 2018.CC BY 4.0 License.2017).Concentrations of O 3 (Thermo Scientific Model 49i), NO x (airborne NO x analyzer, Air Quality Devices), and CO (Los Gatos 23r) were measured.The aircraft performed 16 flights during IOP1.Data sets of two flights (March 14 and 21) were chosen for analysis herein based on flight tracks over both river and land while cutting across the Manaus pollution plume at an altitude of approximately 500 m.There were no flights at lower altitude.
Figure 3. Across individual days, the maximum and minimum heights for significant noontime river breeze effects were 300 8 Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2018-347Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 23 April 2018 c Author(s) 2018.CC BY 4.0 License.and 60 m, respectively (results not shown).Overall, the results of Figures 2 and 3 lead to the conclusion that the river breeze effect on wind speeds was confined on most days to below 150 m in altitude, even under high noontime solar irradiance.

Figure 1 .
Figure 1.Satellite image of the Manaus region.The Rio Negro comes from the northwest, and

Figure 3 .
Figure 3. Difference in mean horizontal wind speed for (wR -woR) ("with rivers compared to without rivers").Plots are shown as vertical cross sections along points A, B, and C of the Figure 1 as follows: (a) for all of March 2014, (b) at 00:00, (c) at 06:00, (d) at 12:00, and (e) at 18:00, all in local time (UTC -4 h).

Figure 4 .
Figure 4. Concentrations of O 3 , NO x , and CO measured by instrumentation on board an aircraft during GoAmazon2014/5 at an altitude of approximately 500 m (Martin et al., 2017).Concentrations are plotted in false color, and the legends on the right-hand side of each row show the scaling.Below each main panel, a line plot shows the concentrations marked by points A through H along the flight paths.Red shading demarcates periods when the aircraft was flying over a river.

Figure 5 .
Figure 5.Time series of O 3 concentrations at the T3, R1, and R2 locations.The left column plots

Figure 6 .
Figure 6.Time series of NO x concentrations at the T3, R1, and R2 locations.The left column plots the cases of wR ("with rivers"; blue) and woR ("without rivers"; red).The right column shows the difference in concentrations as (wR -woR).

Figure 7 .
Figure 7. Time series of CO concentrations at the T3, R1, and R2 locations.The left column plots the cases of wR ("with rivers"; blue) and woR ("without rivers"; red).The right column shows the difference in concentrations as (wR -woR).

Figure 8 .
Figure 8. Near-surface concentrations of (a) O 3 , (b) NO x , and (c) CO (March 2, 13:00 local time, UTC -4 h).The first and second columns represent the cases of wR ("with rivers")and woR ("without rivers"), respectively.The vector field in each panel shows the near-surface horizontal winds.The third column shows the difference in concentrations as (wR -woR).For reference, the locations of T3, R1, and R2 are marked.