Sparse Matrix Operator Kernel Emissions SMOKE Modeling System

1
Sparse Matrix Operator Kernel Emissions SMOKE
Modeling System

• version 1.3
• TRAINING
• http//envpro.ncsc.org/products/smoke

2
Training Overview

• Emissions processing basics
• SMOKE basics
• SMOKE scripts
• SMOKE programs and options

• Overview lab
• SMOKE inventory lab
• SMOKE monthly lab
• SMOKE daily lab
• Quality assurance lab
• SMOKE problem solving

3
About This Training

• Specific to UNIX
• Run SMOKE from scripts

4
Emissions Data Processing

• Source categories
• Definitions
• Area emissions processing
• Point emissions processing
• Mobile emissions processing
• Biogenic emissions processing
• Merging
• Quality assurance

5
Source Categories (1)

• Area source characteristics
• Country, state, county
• Source category code (SCC)
• Point source characteristics
• Country, state, county, and
• Facility, stack, device, process (EMS-95)
• OR
• Plant, stack, point, segment, SCC (NET/IDA or EPS)

6
Source Categories (2)

• Mobile source characteristics
• Country, state, county
• Road type (e.g., rural interstate, urban local)
• Vehicle type (e.g., light duty gasoline vehicles)
• Link ID (optional)
• Biogenic source characteristics
• Country, state, county
• Landuse type
• OR
• Grid cell
• Landuse type

7
Definitions (1)

• Inventory pollutant A compound or group of
  compounds defined for record keeping and
  regulatory purposes (e.g. CO, NOx, VOC, PM10)
• Species A compound or group of compounds defined
  as part of the estimation of air chemistry in an
  air quality model (e.g., CO, NO, NO2, PAR, ECC)
• Chemical mechanism A set of chemical species and
  their interactions used to represent air
  chemistry (e.g., CB-IV, RADM, SAPRAC)

8
Definitions (2)

• Map projection The mathematical representation
  of the spherical surface of the earth in 2-d
• Model grid A 2-d grid based on a map projection,
  defined by starting coordinates, number of grid
  cells in each direction, and the physical size of
  the grid cells
• Model layers Vertical spatial divisions defined
  by an air quality model because the atmosphere
  has varied characteristics in the vertical
  direction

9
Western 36-km cell domain for training
10
Definitions (3)

• Profile data Factors used for disaggregating
  emissions data as is done during chemical
  speciation or temporal allocation
• Cross-reference A dataset used for matching
  sources in the emissions inventory with profile
  data based on the source characteristics
• Gridding surrogate A dataset developed from data
  at a finer resolution than the emissions, used to
  spatially allocate the emissions to the grid
  cells (e.g., population, housing, agricultural
  regions)

11
Definitions (4)

• Plume rise The rising of the exhaust from point
  sources due to the velocity and temperature of
  the exhaust gases
• Elevated source A point source in which
  emissions are higher than the first model layer
  because of plume rise
• Plume-in-grid A special treatment of elevated
  sources by which the plume rise is modeled with
  extra detail by the air quality model

12
Definitions (5)

• Spatial allocationConvert the source spatial
  extent to the grid cell resolution needed by the
  air quality model
• Chemical speciationConvert the inventory
  pollutant data to the chemical species needed by
  the air quality model
• Temporal allocationConvert the inventory
  temporal resolution to the hourly temporal
  resolution needed by the air quality model

13
Area Emissions Processing

• Import data
• Spatial allocation
• Allocation of county emissions into grid cells
  using spatial surrogate
• Chemical speciation
• Temporal allocation
• Approach also applied to mobile sources road
  dust and non-road mobile sources

14
Point Emissions Processing

• Import, chemical speciation, temporal, plus...
• All emissions for a source in single grid cell
• Use day-specific and hour-specific inventory data
• Determine elevated sources and plume-in-grid
  (PinG) sources
• Special processing and output for elevated and
  PinG sources
• Create 3-d emissions file optional PinG
  emissions files
• OR
• Create special elevated (PinG optional) 2-d
  emissions file
• Meteorology based

15
Mobile Emissions Processing (1)

• Import, chemical speciation, temporal, plus...
• Possibly start with vehicle miles traveled (VMT)
  instead of emissions
• Create emission factors based on meteorology
• Emission factors model such as MOBILE5
• Gridded hourly or average meteorology
• Emissions Emission factor ? VMT
• Spatial allocation may include link sources

16
Mobile Emissions Processing (2)

• Emission factors typically depend on emissions
  process (e.g., exhaust, evaporative, diurnal)
• Temporal allocation and speciation may depend on
  emissions process
• Approach only for on-road mobile sources

17
Biogenic Emissions Processing

• Typically created by BEIS2 emissions model
• About 120 landuse types
• Landuse types have emission factors which are
  adjusted by gridded temperature and solar
  radiation
• Winter and summer emission factors
• If landuse is county total, then use gridding
  surrogate

18
Merging and Formatting

• Combining steps taken for a given source category
  to create model-ready formatted files (e.g.,
  combine import, gridding, speciation, temporal
  allocation steps)
• Combine multiple source categories into a single
  data set (e.g., combine area, biogenics, mobile,
  and point)
• Output correct units, species, time steps, grid
  projection, and file format for the air quality
  model of interest

19
Quality Assurance

• Compare emissions totals from emissions processor
  with inventory totals
• By state, county, SCC, combinations, other
• Compare emissions totals at each stage of the
  processing
• Ensure input file formats are correct
• Ensure no significant errors or warnings in
  processing
• Compare emissions among states and counties
• Compare emission ratios to ambient measurement
  ratios

20
SMOKE Basics

• Capabilities
• Programs
• Dataflows
• Concepts
• Shared details of programs
• Assigns file
• Environment variables for naming files and
  directories

21
CapabilitesData Import

• Formats
• EMS-95 area, mobile (fixed-column or not),
  point, and hour- specific point
• IDA area, mobile, and point
• EPS2 (AFS/AMS) area, point, and period-specific
  point
• BEIS and BEIS2 county and gridded landuse
• Gridded I/O API data as area source
• User-selected pollutants
• 16-character pollutant name limit
• Maximum numbers allowed depends on source
  category
• Area 19
• Mobile 54
• Point 15

22
CapabilitesSpatial Allocation

• Input coordinates Lat-lon or UTM
• Output projections Lat-lon, Lambert, UTM
• Any number and size of cells
• Area apply gridding surrogates or map
  grid cells for pre-gridded data
• Biogenic import gridded or county landuse
• Mobile apply gridding surrogates or map link
  sources to grid
• Point assign point location to grid cell

23
CapabilitesChemical Speciation

• User-selected species (up to 120)
• CB-IV and RADM mechanisms installed by default
• Particulate splits
• 16-character species name limit
• SMOKE outputs all species in profile file that
  match inventory pollutants
• Both mole-based and mass-based speciation matrices

24
CapabilitesTemporal Allocation

• Supports monthly, weekly, and diurnal profiles
• Different diurnal permitted for each day of the
  week
• Can use ozone-season or annual-average inventory
  data from IDA inventory
• Point sources can use day- and hour-specific data
• Biogenic based on gridded meteorology
• Mobile emission factors can be based on gridded
  temperature (to be used with VMT)
• Automatic accounting for holidays

25
CapabilitiesGrowth and Control

• SMOKE imports growth factors to create a growth
  matrix
• Growth matrix is applied to the inventory
• SMOKE imports control factors to create several
  control matrices
• Control matrices are applied during the final
  merge or to the inventory
• Multiplicative, additive, and reactivity controls

26
CapabilitesBiogenic Source Processing

• Released version supports BEIS2 science with 120
  landuse types
• SMOKE-BEIS3 is being released with SMOKE v1.4
• BELD3 data has 230 landuse types
• New emission factors for all 230 landuse types
• Modified light attenuation algorithm for
  calculating ISOP emissions
• Supports CB-IV and RADM2 mechanisms
• NOx and VOC

27
CapabilitesMobile Source Processing

• Can import VMT, other activity data, or emissions
• Can optionally import gridded emissions
• Can drive MOBILE5b for large regions based on
  gridded temperature
• Can use different speeds at different hours
• Temperature used can be at ground, layer-1, or
  1.5 meters
• Can run VMT for CO, NOx, VOC, SO2, NH3, PM10, and
  PM2.5, and toxics
• Can customize road types and vehicle types

28
CapabilitiesPoint Source Processing

• Day-specific and hour-specific data permitted by
  pollutant
• Customizes source definition based on inventory
  type (EMS-95, IDA/NET, EPS)
• Options for elevated sources
• UAM-style with separate emissions file AQM
  computes plume rise
• All sources potentially elevated - SMOKE computes
  plume rise
• Major/Minor sources - SMOKE computes plume rise
  for major sources
• PinG sources - SMOKE outputs special file to
  support PinG for CMAQ and MAQSIP
• Output for UAM-style elevated emissions

29
CapabilitiesQuality Assurance

• Smkreport program
• SMKMerge program optionally outputs state and
  county total emissions
• Mass speciation matrix is used
• Can specify units for output
• Can compare totals from different merges
• Inventory-grid
• Hourly-grid
• Inventory-species-grid
• Built in file format and file quality checks
• PAVE comparisons

30
SMOKE Programs
GRDMAT
SMKMERGE
SMK2EMIS
SMKINVEN
GRWINVEN
TEMPORAL
MRGGRID
SPCMAT
SMKREPORT
CNTLMAT
B
Biogenic only
M
Mobile only
P
Point only
31
SMOKE Dataflows (1)
BiogenicProcessing
Land use Data
HourlyEmissions
MobileProcessing
Model-ReadyEmissions
Meteorology Data
Matrices
MergeProcessing
AreaProcessing
EmissionsReports
PointProcessing
Emissions Inventories
Hourly LayerFractions
32
SMOKE Dataflows (2)
Emissions Inventories
Meteorology Data
HourlyEmissions
TEMPORAL
Profiles andX-refs
SpeciationMatrix
SPCMAT
SMOKEInventory
SMKINVEN
GriddingMatrix
GRDMAT
AQMSpecs
ControlData
ControlMatrix
CNTLMAT
33
Other Processing Paradigms
Model ReadyEmissions
LoadInventory
Growth and Controls
Speciation
TemporalAllocation
Gridding
SMOKE Processing Paradigm
TemporalAllocation
Growth and Controls
Model ReadyEmissions
LoadInventory
Merge
Speciation
Gridding
34
SMOKE for Additional Control Strategies
Model ReadyEmissions
Growth and Controls
Merge
SMOKE for Additional Grids
(for non-mobile sources that do not use VMT)
Model ReadyEmissions
Gridding
Merge
SMOKE for Additional Grids
(for mobile sources that do use VMT)
Model ReadyEmissions
Gridding
Merge
EmissionFactors
TemporalAllocation
35
Shared Details of SMOKE Programs (1)

• Sources must be uniquely defined based on SMOKE
  source characteristics
• Sources are sorted in particular order
• Inventory vectorsAdjustment factors matricesve
  ctors x matrices model-ready emissions
• I/O API
• Library SMOKE uses the I/O API library to create
  intermediate and output files. The files are
  NetCDF files, which means that they are binary,
  direct access, and platform-independent.
• Tools The I/O API tools can be used to
  manipulate I/O API files (e.g., extract a
  window from a gridded dataset).

36
Shared Details of SMOKE Programs (2)

• All programs output log files
• Most programs process multiple source categories
• All programs process only one source category at
  a time (Smkmerge and Mrggrid are exceptions)
• Environment variables are used to name files and
  directories
• Logical file names
• Are environment variables
• e.g., program uses OUTFILE for its output
  file setenv OUTFILE myoutputfile.ncf

37
Shared Details of SMOKE Programs (3)

• Environment variables also control programs
• SMK_SOURCE Set to A, B, M, or P to control
  source category
• LOGFILE Set to a file name to record the
  standard output of informational, error, and
  warning messages
• PROMPTFLAG Y/N to control interactive or batch
  mode
• SMK_MAXERROR Maximum number of errors
• SMK_MAXWARNING Maximum number of warnings

38
Assigns File

• Purpose To set directory names and file names
  for a case
• For running SMOKE from the prompt or from
  scripts, and navigating directories
• Three types of environment variables
• Part of file names or directories
• Correspond to a particular directory
• Correspond to a particular file
• To use gt cd SMKROOT/assigns gt source
  ASSIGNS.train.cb-iv.va12

39
Environment Variables Used inFile and Directory
Names
Description
E.V. Name
Name of the inventory input directory (no version)
INVID
Name of the inventory, including version
INVEN
Name for the inventory output files
INVOP
Name of the scenario/strategy currently being
modeled
ESCEN
Name of the biogenics scenario being used
BSCEN
Name of the meteorology scenario being used
MSCEN
Name of the grid
GRID
Name of chemical speciation type
SPC
Starting date of the emissions files and
simulation
ESDATE
Starting date of the meteorology files
MSDATE
Number of days being simulated per program run
NDAYS
Number of days in the meteorology file
MDAYS
Year of the scenario being simulated
YEAR
40
Environment Variablesfor Controlling Episode
Description
E.V. Name
Julian start date (YYYYDDD)
G_STDATE
Start time in output time zone (HHMMSS)
G_STTIME
Duration (HHMMSS)
G_RUNLEN
Time step (HHMMSS) - only 10000 will work
G_TSTEP
Output time zone number (0-23)
OUTZONE
Emission factor year
EF_YEAR
41
Input Directories
Description
E.V. Name
Example/ Default Value
SMOKE system main directory
SMKROOT
EDSS_ROOT/subsys/smokev1
SMOKE system main directory
SMKDAT
EDSS_ROOT/data/smoke
Inventory input directory
INVDIR
SMKDAT/inventory/INVID
General data directory (e.g., profiles)
GE_DAT
SMKDAT/ge_dat
Meteorology data directory
METDAT
EDSS_ROOT/data/met/MSCEN
Area inventory input data
ARDAT
INVDIR/area
Biogenic input data
BGDAT
INVDIR/biog
Mobile inventory input data
MBDAT
INVDIR/mobile
Point inventory input data
PTDAT
INVDIR/point
Meteorology data often varies at the preference
of the user. This is controlled by users
setting their Assigns file
42
Output Directories
Description
E.V. Name
Example/ Default Value
Inventory output directory
INVOPD
SMKDAT/inventory/INVOP
Scenario-specific, time dep output dir
SCENARIO
SMKDAT/run_ESCEN/ESDATE
Scenario-specific output dir
STATIC
SMKDAT/run_ESCEN/static
Episode-specific, time dep output dir
BASSCN
SMKDAT/run_INVEN/ESDATE
Episode-specific output dir
BASDIR
SMKDAT/run_INVEN/static
Final output directory
OUTPUT
SCENARIO/output/SPC
Reports base directory
REPORTS
SMKDAT/reports
Scenario-specific, time dep reports dir
REPSCEN
REPORTS/ESCEN/ESDATE
Scenario-specific, reports dir
REPSTAT
REPORTS/ESCEN/static
Inventory, time dependent reports dir
REPINVEN
REPORTS/INVOP/ESDATE
Inventory reports dir
REPINVST
REPORTS/INVOP/static
43
Benefits of SMOKE

• Processing is much faster than other systems
• Parallel processing paradigm
• Fast multi-grid processing
• Fast control strategy processing
• Fast multi-chemical mechanism processing
• Fast mutiple formats output (for different AQMs)
• Minimal redundant data storage for decreased file
  size
• Machine-independent binary file formats (IO/API
  NetCDF)
• No third party software
• No grid or inventory size limits
• Processing for ozone and particulate modeling
• BEIS-2 with gridded or county land use data
• Output for CMAQ, MAQSIP, UAM-V, CAMx, modified
  UAM-AERO, REMSAD