Photochemical grid model estimates of lateral boundary contributions to ozone and particulate - PowerPoint Presentation

Photochemical grid model estimates of lateral boundary contributions to ozone and particulate matter across the continental United States Kirk BakerU.S. Environmental Protection AgencyResearch Triangle Park, NCJanuary 6, 2016 1

Outline Regulatory modelingSource attribution: apportionment & sensitivity Lateral boundary inflow attribution projectReferences 2

Regulatory Modeling Use regional to local scale photochemical transport models (CMAQ & CAMx ) Typically use 12 km grid res., sometimes 4 km, not coarser than 12 km for a regulatory assessment 2011 NEI based emissions O3 & PM2.5 NAAQS review cycle Interstate transport rules: NOX SIP Call, CAIR, CSASPR, etc. NESHAP sector rules such as Mercury & Air Toxics (MATS)New Source Review/Prevention of Significant Deterioration: single source permit modeling for O3 & secondary PMState/local agencies: NAAQS attainment, Regional Haze rule progressMobile source sector rulesOther types of assessments using “regulatory” quality modeling but not necessarily for rulemakings:National Air Toxics Assessment 2011 3

Source Sensitivity & Apportionment Modeling ApproachesHow will the modeled concentrations change based on changes to emissions? Source sensitivity approaches Brute force zero out or emissions perturbations Decoupled Direct Method (DDM) What are the various contributors to modeled concentrations? Source apportionment approachesOzone and PM source apportionment (OSAT, PSAT, ISAM) (Kwok et al 2013; Kwok et al, 2015; ENVRION, 2015)Tracers: inert or reactive*All techniques have strengths and limitations4

5 Source Sensitivity & Apportionment Examples Source groups may be single sources, groups of sources (e.g. sector, biogenics , lateral boundary inflow), entire Counties, entire States, entire Countries… Baker and Kelly, 2014

Lateral boundary attribution: motivation Increasing interest in characterizing the contribution from chemical lateral boundary inflow (Dolwick et al, 2015)Compare chemically reactive and non-reactive tracer approaches for estimating lateral boundary inflow contribution to O3 and PM2.5Illustrate the strengths and weaknesses of the various approaches Are any techniques efficient enough to be part of routine model application More project details available in Baker et al, 2015 6

Background All assessments 12km annual 2011 CAMx platform CB6r2 gas chemistry; ISORROPIA inorganic chemistryGEOS-CHEM chemical inflow Surface to 50 mb with 25 layers O3 boundary contribution estimated using multiple techniques Reactive tracers: Ozone Source Apportionment Technology (OSAT) with stratified boundaries (west, north, east, south, top) uses reactive tracers through all chemical and physical processes in the model Reactive tracers: RTRAC with stratified boundaries (west, north, east, south) with the west and north boundaries further stratified by layers: 1 to 14, 15 to 22, and 23 to 25Non-reactive tracers: boundary condition only run (no chemistry)77/8/2015

Background All assessments 12km annual 2011 CAMx platform CB6r2 gas chemistry; ISORROPIA inorganic chemistryGEOS-CHEM chemical inflow Surface to 50 mb with 25 layers PM2.5 boundary contribution to PM2.5 sulfate, nitrate, ammonium, EC, primary component of OC, and other primarily emitted PM2.5 Reactive tracers: Particulate Source Apportionment Technology (PSAT) with stratified boundaries (west, north, east, south, top)Non-reactive tracers: boundary condition only run (no emissions or chemistry)87/8/2015

Reactive Tracers: OSAT, RTRAC/RTCMC The CAMx reactive tracer (RTRAC) probing tool provides a flexible approach for introducing gas and particulate matter tracers within CAMx simulations; can not be run at the same time as OSAT/PSAT Each RTRAC tracer is influenced by boundary conditions, advection, diffusion, emissions and dry deposition . Gas-phase tracers can also undergo chemical destruction and/or production using either a simpler (RTRAC) or more complex (RTCMC) chemistry interface. The RTRAC Chemical Mechanism Compiler (RTCMC) allows the user to externally define a full chemistry mechanism with no limits on complexity (within available computer resources). 97/8/2015

RTCMC Template for CB6r2 provided by ENVIRON Example input chemistry control file for 2 sets of extra O3 destruction reactions for the boundary tracking simulation The configuration for this project does not account for NO titration A total of 8 additional sets of tracers were used to track 3 separate vertical layers on the west and north boundaries and full faces east and south Additional RTCMC input is a second ICON and BCON file that only contains tracer concentrations (e.g. O3A, O3B, etc.) Fortran program to manipulate ICBC input files for RTCMC provided by ENVIRONNo attempt to apply RTCMC for PM boundary contributions107/8/2015

O3 Contribution Monthly average O 3 contribution from the west lateral boundary using the OSAT approach . Surface level. 11

O3 Contribution Monthly average O 3 contribution from the north lateral boundary using the OSAT approach . Surface level. 12

Method Comparison Monthly average O 3 contribution from all lateral boundaries using OSAT (left panels), the difference in monthly average O 3 contribution using inert tracers (middle panels) and the RTRAC approach (right panels). Surface level. Cool colors in the difference plots indicate OSAT estimates are higher and warm colors indicate the alternative approach estimates are higher. Inert and RTRAC tend to have larger lateral boundary O3 contribution than OSAT reactive tracer approach13

Method Comparison Scatter density plots showing hourly model estimated lateral boundary contribution methods compared at CASTNET monitor locations: OSAT and inert tracers (top left), OSAT and RTRAC (top right). Hourly model estimated bulk O 3 compared with estimated lateral boundary contribution from the inert tracers (bottom left) and OSAT (bottom right) approaches at CASTNET locations. Colors represent the percentage of points falling at each location on the plot so warm colors indicate areas with a large amount of values. 14

Western boundary inflow (RTRAC) Layers 1-14 (left); 15-22 (mid); 23-25 (right) 15 Northern boundary inflow (RTRAC) Layers 1-14 (left); 15-22 (mid); 23-25 (right) *results shown above are surface level

PM2.5 Contribution Monthly average PM2.5 contribution from all lateral boundaries and the model top using the PSAT approach. Surface level. Contribution tracked from each lateral face, just shown in aggregate here for brevity. 16

B ias (model estimate – measured estimate) paired in time and space with modeled contribution from lateral boundary inflow using the PSAT approach. Only IMPROVE sites shown. IMPROVE PM2.5 17

CASTNET O3 Hourly bias (model estimate – measured estimate) paired in time and space with modeled contribution from lateral boundary inflow using the OSAT approach. Only model estimates of ozone where the lateral boundary contribution is greater than 90% of the bulk modeled O 3 are shown. Bias greater than zero indicates a model over-prediction of baseline ozone and below zero indicates a model under-prediction of baseline ozone. Colors represent the percentage of points falling at each location on the plot so warm colors indicate areas with a large amount of values.No obvious spatial patterns in bias18

Concluding Remarks Inert tracers do not provide a physically realistic contribution estimate for ozoneBetter ways of evaluating the boundary inflow? This type of assessment misses the situations where “observed” BCON influence is not captured due to mischaracterized meteorology OSAT more computationally efficient than RTRAC approach Not clear any approach efficient enough for routine application 19

References Baker, K.R., Emery, C., Dolwick, P., Yarwood, G., 2015. Photochemical grid model estimates of lateral boundary contributions to ozone and particulate matter across the continental United States. Atmospheric Environment 123, 49-62 . Dolwick, P., Akhtar, F., Baker, K.R., Possiel, N., Simon, H., Tonnesen, G., 2015. Comparison of background ozone estimates over the western United States based on two separate model methodologies. Atmospheric Environment 109, 282-296 . Kwok, R., Baker, K.R., Napelenok, S., Tonnesen, G., 2015. Photochemical grid model implementation of VOC, NO x, and O 3 source apportionment. Geoscientific Model Development 8, 99-114.Baker, K.R., Kelly, J.T., 2014. Single source impacts estimated with photochemical model source sensitivity and apportionment approaches. Atmospheric Environment 96, 266-274.Kwok, R., Napelenok, S., Baker, K.R., 2013. Implementation and evaluation of PM2.5 source contribution analysis in a photochemical model. Atmospheric Environment 80, 398-407.ENVIRON, 2015. CAMx Users Manual. www.camx.com. 20