260 likes | 443 Views
BlueSky Team Meeting K. Anderson (Natural Resources Canada), J Cochrane (Parks Canada) D. Lyder (Alberta Environment), A. Pankratz (Environment Canada), R . Stull, R. Schigas (University of BC), S Sakiyama (BC MoE ). Predictive Performance.
E N D
BlueSky Team Meeting K. Anderson (Natural Resources Canada), J Cochrane (Parks Canada) D. Lyder (Alberta Environment), A. Pankratz (Environment Canada), R. Stull, R. Schigas(University of BC), S Sakiyama (BC MoE)
Predictive Performance • Forecast comparisons: timing, location (spatial coverage), magnitude Time and space comparisons the most challenging • Timing perfect, but location off • Location perfect, concentrations good, but timing off • Multiple Models - Uncertainty • Comparisons to Observations: AQ monitors, satellite • Ground level AQ Stations PM2.5: issues (local influences, vertical and horizontal spatial representation, hourly vs 24 hour, saturation) • Satellite (HMS analysis): issues (column integrated, cloud interference, daytime only)
Performance Evaluations • STI: Pilot System August 2009 wildfires • All components functioning properly • Good job at predicting regional smoke coverage • Smoke from fires outside the domain (BC Alberta) can be important. • Yao, J. et al, (2013): 35 days during 2010 wildfire season, BlueSky forecasts compared to: • B.C. PM2.5 monitoring data (IOA, r, NRMSE, FB) • HMS smoke analysis (Figure of Merit in Space) • Population Health Indicators (Asthma Medication dispensations, Asthma-related physician visits)
Results • Global Analysis (all forecasted and measured values at any time and location were included) r = .4 • Modest agreement (HMS and measured values) • BlueSky: Generally smaller smoke plumes than HMS • Better agreement during intense fire periods • Over prediction bias (predicted high peaks not seen in measurements) • Similar performance to other smoke forecasting systems (European, NOAA) • Significant associations between BlueSky forecasts and respiratory health outcomes
Evaluations Cont’d • Klikach, V et al. (2012) comparisons: 2010, 2011 with PM2.5 measures and satellite imagery • good qualitative agreement • challenges in comparing with PM2.5 measurements (other sources vs smoke) • Carry over smoke important • Ho and Lyder (2013) • Several US BlueSky evaluations: Latest Strand et al (2012) • 2007 S. California fires (did not predict high measured PM2.5), • 2009 N. California fires (predictions performed well)
Needs for Further Work • Challenges in comparing to observations • Sometimes operational issues create problems • Performance in timing and location of plume patterns is reasonable • Comparisons to PM2.5 measurements: further work needed to improve magnitudes of concentrations • Evaluation Studies Point to Needs and Further Work: • HLSPLIT blasts (improved – better source treatment, new Hysplit versions), • carry over smoke (improved - puffs retained for 72 hours), • inclusion of smoke outside domain (improved but not North America wide)
Further Needs and Work Cont’d • Latest implementation of the Canadian BlueSky needs to be evaluated (various improvements since 2010) • Produce only relative PM2.5 (high, med, low) or probabilities of high, med, low? • Focus on fuel loading, plume rise, fire timing (U.S. Joint Fire Science Council) • US NWS (NOAA) Smoke Forecasts have automated plots of Figure of Merit in Space and Measure of Effectiveness for every forecast based on the HMS analysis • Comprehensive (not just one area) • Continuous – look for trends, spot issues and respond, assess overall performance
Qualitative Evaluation of the Eastern Canada Forecasts HMS Smoke Analysis (NOAA) Eastern BlueSky Forecast Forecast smoke pattern at the surface, compared to satellite image analysis of smoke 0622, 18 00 UTC June 22 2013 Smoke from US wildfires (not accounted for)
Forecast Produced June 14 Forecast Date and Time in Image: Jun 15 0900 EDT
BlueSky East • Only qualitative evaluation • Ontario MNR was satisfied that system was producing forecasts that made sense based on reports of smoke (location and timing) • Ontario Environment reported some consistency with air quality measurements
References Yao, J. et al Brauer, M. Henderson, S.B., 2013 Evaluation of a Wildfire Smoke Forecasting System as a Tool for Public Health Protection. Environ Health Perspect. 121(10): 1142–1147. Klikach, V., Lyder, D, Cheng, L., Sakiyama, S. Hicks, G., and Anderson K. (2012) Development and Evaluation of a BlueSky Wildfire Smoke Forecasting System for Western Canada. AWMA Annual Conference . Paper 2012-A-507-AWMA. June 2012 Ho, V and Lyder, D. (2013) An Evaluation of the Western Canadian BlueSky Smoke Forecasting System for 2012. Poster International Smoke Symposium, University of Maryland University College Strand et al. (2012) Analysis of BlueSky Gateway PM2.5 predictions during the 2007 sorthern California and 2009 northern California Fires. JGR Vol 117, Issue D17, 16, Sept 2012
Playground Update • Early 2013, Beta Testers group (16): 3 provided feedback on their experience • Summarized feedback: created a laundry list of improvements • BC MoE 20 K. • Prioritized list into the following improvements by STI
Playground Update: Improvements • STI Contract: • Transfer all parts of Canadian Playground to UBC (installed on dedicated server at UBC) • Upgrade to the latest version of the BlueSky framework (v 3.5.1) that includes HYSPLIT v4.9 (MPI) and improved dispersion results with KML output • Previously MM5 as met forecast model - incorporate WRF meteorological domains and clarify domain areas for users • National coverage • Address system load, response times, and ease of use, coverage
Playground Update: Improvements • Incorporate a pathway for wildfire emissions and dispersion scenarios : ability to forecast smoke from individual wildfires • Testing current prototype: • Retrospective Analysis: AESRD (D. Schroeder) (previous burns), Parks Canada (J. Cochrane) • Planning Tool: AESRD (D. Schroeder) • Planned Burns: Parks Canada (J. Large), Saskatchewan • Improved system up and running completely at UBC – May?
Future Plans/Needs • Include Pile Burning • Feedback from current testers, update laundry list of improvements and prioritize: user interface, outputs (24 hrave, health risk for 1 hour, source characterization (burn start and end times, diurnal profile output, etc.), new plume rise model • Provincial contributions to fund (Sask, BC) • Mobile App (Playground and other Burn information: regulations, ventilation index, local weather) • BC FERIC $$ for development • Need a Playground Lead
Outreach: CSSP Charter • 4. Objective(s) d. Webinars and Presentations: an outreach program intended on making agencies aware of the BlueSky—Canada products and provide an opportunity for their feedback.
Primary Audience • Environmental Agencies (Air Quality, Weather) • Wildfire Response Agencies • Transportation • Media • Emergency Response • Health Agencies (Provincial, Federal) • First Nations • Tourism • Public • ?
How? • Establish/build relationships with primary audience: • Associations: Forest (?), Health(?), Environmental (AWMA?), suggestions? • Webinars (self organized or through associations): several – most recent had 500+ participants. • Canadian Smoke Newsletter • National Smoke Forum: Halifax 2014 • Develop and carry out survey of target audience re: awareness and needs and feedback on system • Produce tailored messaging for primary target audience to use in their activities – FAQs (how does this tool help you?), health link BC file on wildfire smoke, etc
Develop a Brand (Logo): Example developed through crowdsourcing
How? • Produce a BlueSky web widget that can be integrated on partner websites and key message delivers’ sites • Website links currently: BCMoE, BC Forests (Wildfire), Alberta ESRD, NRC, US AirNow, US AirFire, Others? • Include links to Kerry’s recorded webinar • Wildfire Smoke Website: Tools, Data, Information (one stop shopping) • Smoke forecasts promoted on the BC wildfire Facebook page • Develop a BlueSky and Playground app for accessibility on all devices