Reducing Vertical Transport Over Complex Terrain in Photochemical Grid Models

1
Reducing Vertical TransportOver Complex Terrain
inPhotochemical Grid Models

• Chris Emery, Ed Tai, Ralph Morris, Greg Yarwood
• ENVIRON International Corporation
• Novato, California
• 8th Annual CMAS Conference
• Chapel Hill, NC
• October 20, 2009

2
Introduction

• Regional photochemical models over predict
  springtime ozone throughout the inter-mountain
  western U.S.
• CMAQ 2002 WRAP
• CAMx 2005 FCAQTF
• Typically 20 ppb higher than remote measurements
• Results from stratospheric ozone levels in top
  model layer
• Enters CMAQ/CAMx via lateral boundary conditions
  (BCs)
• Derived from output of GEOS-CHEM global chemistry
  model
• Stratospheric ozone is too efficiently
  transported to surface over complex/high terrain
• Rockies, Sierras, Cascades

3
Introduction

• 2002 CMAQ Annual Max Daily 8-hour Ozone

4
Introduction

• WRAP CMAQ and FCAQTF CAMx runs use 19 layers
• Top layer spans 8-15 km
• 3 to 5 layers above PBL
• MM5 run for 34 layers
• WRAP CMAQ and FCAQTF CAMx runs use 2002 BCs
• Ozone in layer 19
• range 100-300 ppb

5
Introduction

• Contributors to high ozone over the Rockies
• High surface altitudes (2-3 km MSL)
• Surface is closer to stratosphere
• Deep PBL mixing and convection (through 4-6 km
  MSL)
• Couple surface to mid-troposphere
• Vigorous resolved vertical circulations (through
  4-8 km MSL)
• Transport layer 19 ozone downward
• Solutions weve identified in this study
• Coarse vertical grid structure (more aloft layers
  help)
• GEOS-CHEM BC interface (improved interpolation
  helps)
• Vertical advection technique (alternative
  approach helps)

6
Approach

• Test bed CAMx 2005 FCAQTF application
• Inert, ozone only, no sources/sinks
• Single 12-km regional grid covering western U.S.
• Track ozone IC/BC over April 2005
• Original IC/BC from 2002 GEOS-CHEM extraction
  (WRAP)
• New IC/BC from 2005 GEOS-CHEM extraction
• Test and evaluate several ideas
• Modify input wind fields (smoothers, filters,
  etc.)
• Improve treatment of CAMx top boundary condition
• Test alternative vertical grid structures/resoluti
  on
• Improve GEOS-CHEM interface technique
• Modify CAMx vertical advection solver

7
Modify Input Winds

• Original Rationale
• Vertical velocity derived from input horizontal
  winds
• CAMx and CMAQ yamo approaches are similar
• Filter strong divergences in input winds to calm
  vertical velocity
• Apply aggressively to upper layers only
• Test on 19-layer structure and compare to
  un-modified case
• Three approaches were investigated
• Smoother-desmoother approach of Yang and Chen
  (2008)
• Divergence minimization from CALMET (Scire et
  al., 2000)
• Mass filter of Rotman et al. (2004)

8
Modify Input Winds

• Results
• Minor (10 ppb) reductions in peak April ozone
• Troubling effects on vertical velocity profiles
  in upper layers
• CAMx surface ozone reductions not caused by
  improved vertical advection
• Instead by artificial dilution of top layer ozone
• Caused by CAMx arbitrary top boundary condition
  (70 ppb)
• CAMx was revised to use zero-gradient top
  boundary conditions for all subsequent tests
• Top BC assigned from top layer concentration (a
  la CMAQ)
• Removes artificial dilution of top layer
• BUT increases surface ozone

9
More Model Layers

• Reprocess input meteorology, no smoothers/filters
• Zero-gradient top boundary condition
• Full 34 MM5 layer structure
• Runs 2x slower than 19 layers, 10-15 ppb ozone
  reduction
• Intermediate 22 layers to improve resolution
  aloft
• Runs 1.1x slower than 19 layers, 10 ppb ozone
  reduction
• 19-layer 22-layer
  34-layer

10
2005 Day-Specific BCs

• New 2005 GEOS-CHEM BCs recently became available
• Zero-gradient top boundary condition
• Much higher stratospheric ozone (occasionally
  1000 ppb)
• Higher surface ozone, different spatial patterns
• NOTE CHANGES TO COLOR SCALE!
• 19-layer 22-layer 34-layer

11
2005 Day-Specific BCs

• Issues found in GEOS-CHEM interface program
• High ozone bias in topmost layers for coarse
  vertical layer structures
• We improved the vertical layer-weighting
  technique
• Figure below shows new ozone profiles

12
Improved 2005 BCs

• Improved vertical weighting technique
• Zero-gradient top boundary condition
• Lower stratospheric ozone, lower surface ozone
• Ozone still higher than with 2002 BCs
• NOTE CHANGES TO COLOR SCALE!
• 19-layer 22-layer 34-layer

13
Revised Vertical Advection

• Revised vertical velocity calculation to remove
  downward bias
• Revised vertical solver to be consistent
• Zero-gradient top BC, improved lateral BC
• 40-70 ppb reduction in April maximum ozone
• 19-layer BASE MODIFIED ADVECTION

14
Revised Vertical Advection

• Comparison of 19, 22, and 34-layer configurations
• NOTE CHANGES TO COLOR SCALE!
• 19-layer 22-layer 34-layer

15
Full Photochemical Run

• Run CAMx on 36/12/4-km FCAQTF grids for April and
  July 2005
• Compare 3 runs
• Original 19-layer, 2002 BCs, original vertical
  advection
• New 22-layer, 2005 BCs, original vertical
  advection
• New 22-layer, 2005 BCs, revised vertical
  advection
• Look at monthly maximum 8-hour ozone fields on 12
  and 4 km grids
• 2002 BCs stratospheric ozone levels removed in
  layer 19

16
Full Photochemical RunApril 2005, 4-km Grid

• 19-layer 22-layer 22-layer
• 2002 BCs 2005 BCs 2005 BCs
• Orig CAMx Orig CAMx Revised CAMx

17
Full Photochemical RunApril 2005, 12-km Grid

• 19-layer 22-layer 22-layer
• 2002 BCs 2005 BCs 2005 BCs
• Orig CAMx Orig CAMx Revised CAMx

18
Full Photochemical RunJuly 2005, 4-km Grid

• 19-layer 22-layer 22-layer
• 2002 BCs 2005 BCs 2005 BCs
• Orig CAMx Orig CAMx Revised CAMx

19
Full Photochemical RunJuly 2005, 12-km Grid

• 19-layer 22-layer 22-layer
• 2002 BCs 2005 BCs 2005 BCs
• Orig CAMx Orig CAMx Revised CAMx

20
On-Going Work

• Additional testing of modified CAMx for full
  photochemical/PM applications
• Complete 2005 FCAQTF Application evaluate ozone
  and PM
• OG projects in Rocky Mountains
• Denver SIP modeling
• CMAQ exhibits similar problems
• EPA/ORD is working on a vertical advection
  modification
• See Young, Pleim, Mathur poster
• Interact with ORD and OAQPS
• Test improvements using a western U.S. CMAQ
  database

21
Acknowledgement

• The authors acknowledge funding support from the
  American Petroleum Institute (API)
• Questions