Coupled NMM-CALMET Meteorology Development for the CALPUFF Air Dispersion Modelling in Complex Terrain and Shoreline Settings Presented at: European Geoscience General Assembly Vienna, Austria - April, 2011 Authors: Zivorad Radonjic,* Dr. Douglas B. Chambers, Bosko Telenta and Dr. Zavisa Janjic * Contact: Zivorad Radonjic,

Overview Introduction Study Background Approach To Simulations Approach To Validation Comparison of WRF-NMM Model Versions and Effects of Horizontal Resolutions Validation with Local Observations One Year vs. Five Year Meteorological Datasets Conclusions

Introduction A study was undertaken to prepare and validate high- resolution three-dimensional meteorology suitable for use as input into the CALPUFF/CALMET air dispersion model system in complex terrain with a shoreline. The main goal was to demonstrate the good performance of CALMET in a setting that involves both complex terrain and a shoreline (land-water interface). Improvements in CALMET performance possible using fine resolution meso-scale model inputs were also demonstrated.

Study Background Meteorology required for CALPUFF modelling of industrial site in complex terrain with shoreline – A challenging meteorological environment Site-specific meteorology required for both long- and short-range modelling No local observational data available for time period in question

Approach To Simulations Weather Research and Forecasting - Nonhydrostatic Mesoscale Model (WRF-NMM) used as meso-scale model WRF-NMM modelled on 6 km and 2 km horizontal resolution WRF-NMM used to initialize CALMET modelled on 2 km and 250 m horizontal resolution 2009 meteorology modelled Example Large Domain (6 km x 6 km resolution) for WRF-NMM Model Example Small Domain (2 km x 2 km resolution) for WRF-NMM – CALMET Model

Approach To Validation Simulations validated by: – Comparison of two versions of WRF-NMM – Comparison between models run of different horizontal resolutions – Comparison of simulations with observational data Wind Rose Comparisons Descriptive Statistics All validations at 10 agl Select Meteorological Stations Used for Model Validation

Comparison of WRF-NMM Model Versions WRF-NMM Versions and compared – Versions yielded similar results – Good predictions of wind direction – Wind speeds over-predicted due to averaging / surface – Roughness over water causes higher wind speed predictions Wind Rose Comparison for Large WRF-NMM Domain (6 km resolution) - Charlo Airport

Effects of Horizontal Resolutions - 1 Both WRF-NMM simulations over-predict wind speed due to averaging / under-prediction of surface roughness over water / land interface – Improved accuracy with finer (2 km) resolution WRF-NMM – CALMET simulation provides better prediction of wind speed due to better representation of surface characteristics and vertical resolution No significant improvement in agreement with wind direction between the three models. Wind Rose Comparison for Gaspe Airport

Effects of Horizontal Resolutions - 2 Wind speed predictions improve with finer resolution (250 m) WRF- NMM- CALMET model – Slight under-prediction versus over-prediction for other models No significant change in agreement with wind direction between the four models. Wind Rose Comparison for Bathurst Airport

Effects of Horizontal Resolutions - 3 Descriptive statistics used to evaluate simulation results Wind Speed Summary Statistics – Bathurst Station MODEL – Horizontal Resolution BiasMAERMSE WRF-NMM - 6 km WRF-NMM - 2 km WRF-NMM/CALMET - 2 km WRF-NMM/CALMET m Q-Q Plot of Observed vs. Modelled Wind Speed – Bathurst Station

Effects of Horizontal Resolutions - 4 Finer resolution improves: – BIAS (average of all the differences between forecast and observation) – Mean Average Error (the average magnitude of errors between forecast and observation) – Root Mean Square Error (measures the average magnitude of the error in a set of forecasts, with increase weighting on larger errors) Descriptive statistics consistent with wind rose data – Over-prediction of wind speeds at larger horizontal resolutions – Good agreement of wind speeds at smaller horizontal resolutions – More accurate predictions with higher horizontal resolutions – WRF-NMM/CALMET on 250 m resolution most accurate simulation

Validation with Local Observations - 1 WRF-NMM/CALMET on 250 m horizontal resolution run for 2009 meteorology Observations from local (2.5 km away) meteorological station available for 1999 to 2003 Good agreement noted between 2009 generated meteorology and observational data X - Site X – Nearby Site

Validation with Local Observations - 2 CALMET Derived Site Wind Rose 2009 vs. Nearby Site Observations

One Year vs. Five Year Meteorological Datasets WRF-NMM/CALMET on 250 m horizontal resolution run for 2009 meteorology compared to 2004 to 2008 meteorology Differences between simulated meteorology for a 5-year period and a 1-year period are inconsequential CALMET Derived Site Wind Rose 2009 vs. Nearby Site Observations For well simulated meteorological data sets a single year of the derived three-dimensional meteorology can, for most purposes, be considered to provide a reasonable description of on-site observations.

Conclusions 1.Improvements in wind speeds predictions are achieved with use of finer resolution meso-scale meteorological modelling in shoreline complex terrain situations 2.Wind direction is not sensitive to effects of finer resolution modelling in these situations 3.Coupled WRF-NMM/CALMET system provides a sound alternate to costly and time consuming on-site data collection 4.One year of simulated meteorology corresponds closely to longer periods of record and can used satisfactorily in environmental assessments / air dispersion modelling if generated meteorology sufficiently represents on-site observations.