| 09 Dec 2025
Operational chemical weather forecasting with the ECCC online Regional Air Quality Deterministic Prediction System version 023 (RAQDPS023) – Part 1: System description
Abstract. The online version of the Regional Air Quality Deterministic Prediction System (RAQDPS) is a chemical weather forecast system that has been employed operationally by Environment and Climate Change Canada (ECCC) since 2009. It is run twice per day to produce 72-hour forecasts of hourly 10 km abundance fields of three key predictands, NO2, O3, and PM2.5 total mass, as well as other gas-phase chemical species, PM2.5 chemical components, and dry and wet deposition for Canada, the contiguous U.S., and northern Mexico. The forecasts of NO2, O3, and PM2.5 are needed to calculate the Air Quality Health Index (AQHI), which is used to communicate current and forecasted pollutant levels to the Canadian public. Version 023 of the RAQDPS (RAQDPS023) went into service at ECCC in December 2021 and was replaced by the RAQDPS025 in June 2024. This paper provides the first full description of any version of the online RAQDPS. After giving a brief history of the ECCC operational air quality forecasting program, we provide a comprehensive description of the RAQDPS023 forecast system as well as shorter descriptions of several upstream and downstream forecast and analysis systems. The latter include two upstream operational meteorological forecast systems that were based on version 5.1.0 of the ECCC Global Environmental Multiscale (GEM) numerical weather prediction model, one which used a global configuration, the Global Deterministic Prediction System (GDPS 8.0.0), and the other which used a regional configuration, the Regional Deterministic Prediction System (RDPS 8.0.0). An emissions processing system, an Updateable Model Output Statistics-based system for bias-corrected station-specific pollutant concentration forecasts (UMOS-AQ), and a regional objective analysis system for surface pollutant concentration fields, the Regional Deterministic Air Quality Analysis system (RDAQA 2.0.0), are also described.
The RAQDPS023 itself consisted of version 3.1.0.0 of the GEM-Modelling Air quality and CHemistry (GEM-MACH) chemistry module, which was embedded with one-way coupling within GEM 5.1.0, its meteorological host model. The meteorological configuration of the RAQDPS023 closely followed that of the RDPS 8.0.0. Details covered in this paper include a summary of the dynamical representations and physical parameterizations used in the three GEM-based forecast systems, which are closely harmonized, the chemical parameterizations used in the MACH chemistry module, numerical solvers, system inputs, including both anthropogenic and natural emissions of chemical species, system outputs, and run configuration, strategies, and timings. One simplification employed to reduce RAQDPS023 execution time for operational deployment was to represent the particulate matter (PM) size distribution with only two aerosol particle size bins, one corresponding to particle diameters in the 0–2.5 µm range (“fine particles” or PM2.5) and the other to the 2.5–10 µm range (“coarse fraction” or PMcf). A second simplification was to represent the chemical composition of PM2.5 with only nine chemical components, and a third simplification was to use a longer time step (900 s) for the time integration of atmospheric chemistry than the time step used for time integration of atmospheric dynamics and physics (300 s). Even so, activating the MACH module increased RAQDPS023 run time by a factor of 4.4 on average compared to meteorology only, partly due to the cost of the integration of chemistry but partly to the increased cost of integration of the GEM dynamical core due to the advection with imposed shape preservation and mass conservation of 57 additional chemical tracers. The role of the RAQDPS-FW023, a second chemical weather forecast system that was identical to the RAQDPS023 except for the addition of near-real-time biomass burning emissions, is also described. Biomass burning emissions for Canada and the U.S. estimated from satellite measurements were first calculated by the Canadian Forest Fire Emissions Prediction System (CFFEPS) version 4.1 before each RAQDPS-FW023 run was launched. Outputs from the two RAQDPS versions were then used to produce forecasts of wildfire smoke transport and diffusion. The paper closes by summarizing the key upgrades made to the RAQDPS025, the current version of the ECCC operational chemical weather forecast system, and then describing some possible future improvements and updates. A companion paper by Moran et al. (2025) presents the results of a comprehensive, five-year performance evaluation of prospective and retrospective annual air quality simulations made with the RAQDPS023.
The manuscript provides a very comprehensive description of the RAQDPS023 modeling system used by Environment and Climate Change Canada to provide air quality forecasts to the public. As such, it is a rare example of a one-stop technical and scientific documentation that allows the reader to gain an understanding of the many different aspects that go into building such a system. It is also extremely well written and structured so that despite its length it is easy to follow. I commend the authors for the care they took in compiling the references that underly the scientific formulations of RAQDPS023. This extensive list of references in conjunction with the detailed descriptions of all RAQDPS023 science processes creates a rare repository of knowledge not only about RAQDPS023 but also about the tremendous amount of effort it takes to design, implement, and operationalize air quality modeling systems more generally. My specific comments listed below are minor and/or editorial in nature.
Specific Comments:
Page 3: “Schiermeir, 1978” should be “Schiermeier, 1978”. While the doi of the scanned copy available through the ACS legacy archives (https://pubs.acs.org/doi/abs/10.1021/es60142a608) indeed mis-spelled the author’s last name, the actual scanned copy available under that doi correctly shows the last name as “Schiermeier”
Page 7, line 218 and Table 1: Consider using “horizontal domain size” instead of “horizontal grid size” since the latter could potentially be misinterpreted to refer to the size (spacing) of individual grid cells which is identical between RAQDPS023 and RDPS 8.0.0.
Page 8, line 255: Maybe change “included a small number of meteorological tracers” to “included two meteorological tracers (water vapour and cloud water)”
Page 8, line 266: Insert “and” between “meteorological” and “chemical”
Page 14, lines 454 - 456: Please clarify relative to which starting point (e.g., 10 bin GEM-MACH configuration, RAQDPS022) this modification of the numerical solution was implemented.
Page 15, line 476: Consider changing “where” to “whereas”
Page 15, lines 495 – 496: This question reveals my lack of understanding of aerosol schemes, but I was still curious what “partially activated” refers to. If there is a critical particle radius above which all aerosol particles are activated as stated on line 489, and if the two bins each represent aerosols of a discrete size, how does partial activation occur? Does each discrete size bin assume an internal distribution of particle radii, allowing a determination of the fraction of particles in a bin that exceed the critical radius above which aerosols are activated?
Page 16, line 521: Would H2O (water vapour) profiles not be available from GEM?
Page 16, lines 537 – 539: Maybe comment on the implications for SOA formation when omitting emissions of organic acids and approaching the decision of which emissions to retain solely from a reactivity point of view.
Page 26, line 877: typo, change “haf already” to “had already”
Page 27, line 898: Could you elaborate on this additional information and how it was used to adjust the inventories?
Page 28, line 942: Should this be “the usual hourly anthropogenic and biogenic emissions” instead of just “the usual hourly anthropogenic emissions”?
Page 30, lines 995 – 1001: Was there any dependency of soil NO emissions on precipitation to represent the pulsing effect described in the Yienger-Levy framework?
Page 33, line 1111: Please check if “for 900 s” is needed here. With the way the end of the sentence is written (“for 900 s for three shorter 300 s time steps”), I am not sure if the only point of the sentence is to say that a single MACH chemistry execution every 900 s still took four times longer than the combined time it took to execute three GEM 300 s physics time steps, or if there is some additional information being conveyed here.
Page 36, lines 1200 – 1211: It might be good to provide a brief summary of which emission inputs (anthropogenic and natural) were used in these MOZART-4 simulations. Did the anthropogenic emissions represent 2009 conditions? Were aircraft, lightning NO, and soil NO emissions considered? Adding such information could set the stage for the updates to LBC in RAQDPS025 and future versions discussed in later sections of the manuscript.
Pages 45, lines 1493 – 1503: Given the episodic nature of processes affecting large-scale distributions of O3, CO, and PM2.5, why did the LBC updates in RAQDPS025 still adopt an approach based on climatology? Were there any differences in the types of anthropogenic and natural emissions considered in the MOZART-4 vs. CAM-chem simulations?
Page 46, lines 1520 – 1522: Does this update eliminate any dependence on the five broad phenological seasons described in the last paragraph of Section 3.9?