Regime shift in secondary inorganic aerosol formation and nitrogen deposition in the rural US
Data files
Mar 07, 2024 version files 4.12 GB
-
1_site_information.csv
29.75 KB
-
10_monte_carlo_simulations.zip
2.83 GB
-
11_tdep_maps.zip
40.25 MB
-
2_integrated_obs_biweekly.csv
5.07 MB
-
3_simulations_biweekly.csv
32.50 MB
-
4_pH_change_contributions.csv
131.65 KB
-
5_pH_buffer_capacities.csv
207.23 KB
-
6_tdep_trends_emis_hotspots.csv
5.33 KB
-
7_tdep_trends_iasi_hotspots.csv
5.31 KB
-
8_inputs_outputs_isorropia_case_studies.zip
1.19 GB
-
9_sensitivity_simulations.zip
19.03 MB
-
README.md
15.02 KB
Mar 19, 2024 version files 4.23 GB
Abstract
Secondary inorganic aerosols (SIA) play an important role in air pollution and climate change, and their formation modulates atmospheric deposition of reactive nitrogen (Nr; including oxidized and reduced nitrogen), impacting the nitrogen cycle. Large-scale and long-term analyses of SIA formation based on model simulations have significant uncertainties. Here, we improve constraints on SIA formation using decade-long in-situ observations of aerosol composition and gaseous precursors from multiple monitoring networks across the US. We reveal a shift in the formation regime of SIA in the rural US between 2011 – 2020, making rural areas less sensitive to changes in ammonia (NH3) concentrations and shortening the effective atmospheric lifetime of reduced forms of Nr. This leads to potential increases in Nr deposition near NH3 emission hotspots, with ecosystem impacts warranting further investigation. NH3, a critical but not directly regulated precursor of fine particulate matter (PM2.5) in the US, has been increasingly scrutinized to decrease PM2.5. Our findings, however, show controlling NH3 became significantly less effective for mitigating PM2.5 in the rural US. We highlight the need for more collocated aerosol and precursor observations for better characterization of SIA formation in urban areas and regions increasingly impacted by wildfires and dust.
README: Regime Shift in Atmospheric Secondary Inorganic Aerosol Formation in the Rural United States
https://doi.org/10.5061/dryad.zpc866tg3
Versioning
- Version 2 (19 Mar 2024) contains additional files, 12 & 13.
Data files and description
This dataset contains the following data:
1. Integrated observations from several aerosol composition and gaseous precursor monitoring networks and meteorological stations between 2011 and 2020. Observations of the chemical composition and gaseous precursors of aerosols are from: 1) the Clean Air Status and Trends Network (CASTNET), the Interagency Monitoring of Protected Visual Environments (IMPROVE) network, the US Environmental Protection Agency’s (EPA’s) PM2.5 Chemical Speciation Monitoring Network (CSN), and the Ammonia Monitoring Network (AMoN). Meteorological observations are from the NOAA Integrated Surface Database (ISD) and North American Regional Reanalysis (NARR) data.
To integrate the monitoring networks, we first identify the spatial window for collocation determination by comparing observations from CASTNET, IMPROVE, and EPA CSN sites as well as temperature (T) and relative humidity (RH) observations from CASTNET and ISD located within 10, 25, 50, and 100 km of each other. For aerosol observations, no significant difference was found with different spatial windows. However, T and RH from CASTNET and ISD significantly differ when a spatial window of 100 km is used. Therefore, a spatial window of 50 km is selected for observation integration. With this spatial window, we find 68 AMoN sites with at least CASTNET and ISD sites located within 50 km. Combining observations from these three networks provides all the inputs needed for aerosol thermodynamic modeling. All observations are averaged biweekly to match the start and end dates of AMoN observations since it has the lowest sampling frequency.
2. Aerosol thermodynamic simulations based on the integrated observations using the ISORROPIA-II model. ISORROPIA-II is a full thermodynamic model for inorganic aerosol formation, and we use it to simulate aerosol properties and sensitivities of secondary inorganic aerosol (SIA) formation to precursors. The model is run in the “forward mode” to simulate partitionings of total ammonium (NH4T=NH4- + NH3) and total nitrate (NO3T=NO3- + HNO3). Although ISORROPIA-II has been validated with observations from intensive field campaigns, using it with biweekly averaged observations from monitoring networks have not been tested before and require careful evaluation. We conduct nine case studies to investigate the impacts of measurement biases and low temporal resolutions. After determining the preprocessing steps, we investigated aerosol pH trends and buffering effects of different acid-base conjugates. We also simulated SIA sensitivities to 10%, 40%, and 70% reductions in concentrations of sulfate (SO42-), and NO3T, and NH4T. See the Methods section of the article for more details.
3. Time series of reactive nitrogen deposition at different distances to NH3 hotspots. Dividing the contiguous US into four zones according to their distances to the nearest NH3 emission hotspot (<50 km, 50 - 150 km, 150 - 300 km, and >300 km), we analyze the trend of annual Nr total deposition from the Total Deposition Estimates Using the Measurement Model Fusion (TDep MMF) between 2010 and 2019. The TDep MMF combines wet deposition observations from the National Trends Network (NTN), ambient air monitoring data from CASTNET, and simulations from the EPA's Air Quality Time Series (EQUATES) project. The areas of the 95th percentile of NH3 emission rates across the U.S. based on the 2017 EQUATES NH3 emissions are considered NH3 emission hotspots, except for sporadic locations with just one 12 km × 12 km grid. We also considered the hotspots defined as the areas of the 95th percentile of NH3 column amounts from the Infrared Atmospheric Sounding Interferometer (IASI) satellite NH3 observations (v2.2r) across the US.
Description of the data and file structure
This dataset contains 8 csv files, 5 zip files, and 1 xlsx file. Missing values are represented as "NaN" in all files.
CSV files:
"1_site_information.csv": this file contains information about the sites included in this study. Varibles and units are described in the file. AMoN IDs are used to label the sites from different networks that are within 100 km of an AMoN site. Site IDs are the site identification number assigned by the networks. "AMON", "CASTDRYCHEM", "IMPAER", "EPAPM25SD", and "ISD" in the network column stand for AMoN, CASTNET, IMPROVE, CSN, and ISD networks, respectively.
"2_integrated_obs_biweekly.csv": this file contains biweekly integrated observations for all sites. Species and corresponding units are described in the file.
"3_simulations_biweekly.csv": this file contains the ISORROPIA II simulation results with uncertainties. In addition to simulated aerosol chemical composition and aerosol properties (e.g., pH and aerosol water content), this file also includes simulated PM concentrations with 10%, 40%, 70% reductions in SO42-, NO3T, and NH4T (dPM_deltaS_10 - 70, dPM_deltaN_10 - 70, and dPM_deltaA_10 - 70) as well as corresponding sensitivities (dPMdp_deltaS10_10 - 70, dPMdp_deltaN_10 - 70, dPMdp_deltaA_10 - 70). The uncertainties are given as lower and upper bounds of the 95% confidence interval generated using a Monte Carlo method. Variables and corresponding units are described See the Methods section in the manuscript for more details.
"4_pH_change_contributions.csv": this file contains simulated annual pH changes attributed to changes in different precursors and meteorological conditions between 2011 and 2020. Variables and units are described in the file.
"5_pH_buffer_capacities.csv": this file contains simulated buffering capacity of SO42-/HSO4-, NO3-/HNO3, NH3/NH4+, H2O. Variables and units are described in the file.
"6_tdep_trends_emis_hotspots.csv": this file contains reactive nitrogen deposition trends in the four zones defined based on emission hotspots. Variables and units are described in the file.
"7_tdep_trends_iasi_hotspots.csv": this file contains reactive nitrogen deposition trends in the four zones defined based on emission hotspots. Variables and units are described in the file.
"8_inputs_outputs_isorropia_case_studies.zip": this file contains inputs and outputs for 9 ISORROPIA II simulation case studies:
- Case1 (default configuration): Running the model at a time step of three hours to reflect diel patterns of T and RH while chemical composition from CASTNET and AMoN with scaled non-volatile cations (scaling factors derived from IMPROVE and EPA CSN) at each time step are the same as their biweekly average. Observations and simulations are averaged back to a biweekly time step for evaluation. Related files: "case_1_3h_inputs.csv", "case_1_3h_outputs.csv", "case_1_inputs_bw.csv", and "case_1_outputs_bw.csv".
- Case 2: Running the model at a time step of three hours to reflect diel patterns of T and RH while chemical composition from IMPROVE, CSN, and AMoN at each time step are the same as their biweekly average. Observations and simulations are averaged back to a biweekly time step for evaluation. Related files: "case_2_3h_inputs.csv", "case_2_3h_outputs.csv", "case_2_inputs_bw.csv", and "case_2_outputs_bw.csv".
- Case 3: Running the model at a time step of two weeks with chemical composition from CASTNET and AMoN with raw non-volatile cations. Related files: "case_3_inputs_bw.csv" and "case_3_outputs_bw.csv".
- Case 4: Running the model at a time step of three hours to reflect diel patterns of T and RH while chemical composition from CASTNET and AMoN with raw non-volatile cations at each time step are the same as their biweekly average. Observations and simulations are averaged back to a biweekly time step for evaluation. Related files: "case_4_3h_inputs.csv", "case_4_3h_outputs.csv", "case_4_inputs_bw.csv", and "case_4_outputs_bw.csv".
- Case 5: Running the model at a time step of two weeks with chemical composition from CASTNET and AMoN with scaled non-volatile cations (scaling factors derived from IMPROVE and EPA CSN). Related files: "case_5_inputs_bw.csv" and "case_5_outputs_bw.csv".
- Case 6: Running the model at a time step of three hours to reflect diel patterns of T and RH. Chemical composition from CASTNET and AMoN with scaled non-volatile cations (scaling factors derived from IMPROVE and EPA CSN) at each time step are adjusted to be inversely proportional to planetary boundary layer height while maintaining their biweekly average. Observations and simulations are averaged back to a biweekly time step for evaluation. Related files: "case_6_3h_inputs.csv", "case_6_3h_outputs.csv", "case_6_inputs_bw.csv", and "case_6_outputs_bw.csv".
- Case 7: Running the model at a time step of three hours to reflect diel patterns of T and RH while chemical composition from CASTNET and AMoN with scaled non-volatile cations (scaling factors derived from IMPROVE and EPA CSN) at each time step are the same as their biweekly average except for NH4T. Daytime NH4T concentrations are increased by 30% to reflect its diel pattern while maintaining its biweekly concentration the same. Observations and simulations are averaged back to a biweekly time step for evaluation. Related files: "case_7_3h_inputs.csv", "case_7_3h_outputs.csv", "case_7_inputs_bw.csv", and "case_7_outputs_bw.csv".
- Case 8: Running the model at a time step of three hours to reflect diel patterns of T and RH while chemical composition from CASTNET and AMoN with scaled non-volatile cations (scaling factors derived from IMPROVE and EPA CSN) at each time step are the same as their biweekly average. NH3 concentrations are increased by 10% to examine the impacts of NH3 measurement biases. Observations and simulations are averaged back to a biweekly time step for evaluation. Related files: "case_8_3h_inputs.csv", "case_8_3h_outputs.csv", "case_8_inputs_bw.csv", and "case_8_outputs_bw.csv".
- Case 9: Running the model at a time step of three hours to reflect diel patterns of T and RH while chemical composition from CASTNET and AMoN with scaled non-volatile cations (scaling factors derived from IMPROVE and EPA CSN) at each time step are the same as their biweekly average. The fractions of NO3- in NO3T are decreased by 10% to examine the impacts of NH3 measurement biases. Observations and simulations are averaged back to a biweekly time step for evaluation. Related files: "case_9_3h_inputs.csv", "case_9_3h_outputs.csv", "case_9_inputs_bw.csv", and "case_9_outputs_bw.csv".
"9_sensitivity_simulations.zip" this file contains biweekly simulation results of SIA responses to reductions in SO42-, NO3T, or NH4T. The inputs are prepared the same way as in default configuration but with corresponding precursor concentrations reduced by a certain level. The files are named as "delta%P_%d_ouputs.csv". %P is A (ammonium), N (nitrate), or S (sulfate), indicating the precursor. %d is 10, 40, or 70, representing the percentage level of reduction. The columns in each file are:
- TS: time stamp
- site: AMoN site ID
- NH3 (ug/m3): gaseous ammonia concentration in ug/m3
- HNO3 (ug/m3): gaseous nitric acid concentration in ug/m3
- SO42- liquid (ug/m3): liquid phase sulfate ion concentration in ug/m3
- HSO4- liquid (ug/m3): liquid phase bisulfate ion concentration in ug/m3
- NA+ liquid (ug/m3): liquid phase sodium ion concentration in ug/m3
- CA2+ liquid (ug/m3): liquid phase calcium ion concentration in ug/m3
- Cl- liquid (ug/m3): liquid phase chloride ion concentration in ug/m3
- K+ liquid (ug/m3): liquid phase potassium ion concentration in ug/m3
- Mg2+ liquid (ug/m3): liquid phase magnesium ion concentration in ug/m3
- H+ liquid (ug/m3): liquid phase hydrogen ion concentration in ug/m3
- NO3- liquid (ug/m3): liquid phase nitrate ion concentration in ug/m3
- NH4+ liquid (ug/m3): liquid phase ammonium ion concentration in ug/m3
- Aerosol water content (ug/m3): aerosol water content in ug/m3
- Dry mass (ug/m3): aerosol solid phase concentration in ug/m3
- Liquid mass (ug/m3): aerosol liquid phase concentration in ug/m3
- PM (ug/m3): dry particulate matter concentration in ug/m3
- pH: aerosol pH
"10_monte_carlo_simulations.zip": this file contains results of 1000 simulation runs for the default configuration and 500 simulation runs for SIA responses. To reduce the file size, the simulation are stored every 100 simulations. They are named as "MC1.csv" - "MC10.csv" for default simulations and "MC_sens_1.csv" - "MC_sens_05.csv", respectively.
"11_tdep_maps.zip": this zip file contains deposition data stored in the csv files listed below.
- "lats.csv": this file contains latitudes for the 299x459 array.
- "lons.csv": this file contains longitudes for the 299x459 array.
- "dist_to_emis_hotspots.csv": this file describes the distance (in km) of the grid to the closest emission hotspots.
- "dist_to_iasi_hotspots.csv": this file describes the distance (in km) of the grid to the closest iasi NH3 hotspots.
- "Nr_dep_2010.csv" - "Nr_dep_2019": these files contain total reactive nitrogen deposition (in kg N/ha/yr) between 2010 and 2019.
- "NH4T_dep_2010.csv" - "NH4T_dep_2019": these files contain total NH4T deposition (in kg N/ha/yr) between 2010 and 2019.
- "NO3T_dep_2010.csv" - "NO3T_dep_2019": these files contain total NO3T deposition (in kg N/ha/yr) between 2010 and 2019.
- "NH4T_ddep_2010.csv" - "NH4T_ddep_2019": these files contain NH4T dry deposition (in kg N/ha/yr) between 2010 and 2019.
- "NH4T_wdep_2010.csv" - "NH4T_wdep_2019": these files contain NH4T wet deposition (in kg N/ha/yr) between 2010 and 2019.
- "NH3_emis_2010.csv" - "NH3_emis_2019": these files contain NH3 emission (in kg N/ha/yr) between 2010 and 2019.
"12_intercomparison_monitoring_networks.zip": this zip file contains observations from different networks for intercomparison.
- "chem_comparison.csv": this file contains synchronized observations of SO42-, NO3-, NH4+, Ca2+, Cl-, K+, Mg2+, Na+ from CASTNET, EPA CSN, and IMPROVE and distances of the sites to the cluster center.
- "met_comparison_cast_isd.csv": this file contains synchronized temperature and relative humidity observations from CASTNET and ISD and their distance to each other.
- "met_comparison_cast_narr.csv": this file contains synchronized temperature and relative humidity observations from CASTNET and NARR and their distance to each other.
- "met_comparison_isd_narr.csv": this file contains synchronized temperature and relative humidity observations from ISD and NARR and their distance to each other.
"13_Source_Data.xlsx": this file contains source data for the figures presented in the manuscript, extended data, and supplementary inforamtion. See "README" sheet in the file for more details.
Sharing/Access Information
Data was derived from the following sources:
- CASTNET, IMPROVE, EPA CSN, and AMoN observations were obtained from FED - Federal Land Manager Environmental Database (colostate.edu).
- NOAA ISD meteorological observations were obtained from ftp://ftp.ncdc.noaa.gov/pub/data/noaa/
- NARR data were obtained from NCAR RDA Dataset ds608.0 (ucar.edu)
- TDep MMF data were obtained from Index of /castnet/tdep/CURRENT_grids (epa.gov)
- The EPA’s Air Quality Time Series (EQUATES) project datasets are available from EPA managed CMAS Data Warehouse Google Drive: EQUATES - Google Drive
Code/Software
The ISORROPIA II model is available at ISORROPIA Homepage (gatech.edu).
Methods
To integrate the monitoring networks, we first identify the spatial window for collocation determination by comparing observations from CASTNET, IMPROVE, and EPA CSN sites as well as temperature (T) and relative humidity (RH) observations from CASTNET and ISD located within 10, 25, 50, and 100 km of each other. For aerosol observations, no significant difference was found with different spatial windows. However, T and RH from CASTNET and ISD significantly differ when a spatial window of 100 km is used. Therefore, a spatial window of 50 km is selected for observation integration. With this spatial window, we find 68 AMoN sites with at least CASTNET and ISD sites located within 50 km. Combining observations from these three networks provides all the inputs needed for aerosol thermodynamic modeling. All observations are averaged biweekly to match the start and end dates of AMoN observations since it has the lowest sampling frequency.