CMOS Congress Outstanding Presentations

On June 3-6, 2024, the 58th annual Canadian Meteorological and Oceanographic Society Congress was held virtually with help from the Winnipeg Centre and the BC Interior-Yukon Centre. The theme of the Congress was “Extreme Events in a Changing Climate,” wherein session topics spanned the latest science and solutions on the climate emergency to high impact weather and the global water crisis, among others. More information on the Congress can be found here.

To highlight the diverse and impactful research presented, we asked session chairs to choose one talk or poster to share with the Bulletin. We then reached out to these individuals and to ask what drew them to their research and what they found interesting. Here, we are sharing answers from presenters and testimonials from session chairs on why they nominated the presentations from their sessions. Abstracts for each presentation can be found at the bottom of the article.

Canada’s Three Oceans multi-decade ocean – Sea-ice hindcast: A Hudson’s Bay sea-ice study

– By Sarah MacDermid – 

1. What drew you to this area of research?

I was first introduced to sea-ice modelling during my Master’s Research at McGill University very much by accident. I wanted to do field work to study extreme weather, but all the professors working within that field at the time were taken. Looking back, I have no regrets as I get car sick easily and I got to work with Bruno Tremblay and Anne De Vernal, who were both very supportive and inspiring throughout my studies. I was also lucky and spend 6 weeks in the Arctic aboard the CCGS Amundsen Icebreaker operating the Rosette. I have continued to work with sea-ice and ocean models on and off ever since; from regional coupled and uncoupled systems to larger regional climate-scale set-ups.

2. What is an interesting or unexpected finding you had?

While studying a Cold Air Outbreak that affected much of Atlantic Canada in February 2024, my team at DFO and I noticed a region of warm coastal water shown by the Coastal Ice Ocean Prediction System near Anticosti Island. While we haven’t yet confirmed whether this event occurred in reality, the model clearly shows a short lived warm water upwelling event. Looking at the situation it does makes sense, because in the winter the subsurface waters in the Gulf of St Lawrence are warmer and the event was accompanied by strong winds driving the upwelling. It was just surprising to see such a warm pool of water appear during such a frigid event.

3. What are you looking forward to most next?

There has been a clear desire within the community, and my department, to set up a coupled Arctic Air-Sea-Ice forecasting system at Environment and Climate Change Canada and I am looking forward to working within a team of highly motivated, multi-disciplinary scientists to get it up and running.

Leveraging Data-Driven Weather Emulators to Guide Physics-Based NWP Models: A Fusion of Forecasting Paradigms

– By Syed Husain – 

” The recent rise of AI-based weather emulators is challenging the traditional approach to meteorological forecasting, which mainly depends on physics-based numerical weather prediction (NWP) models. These AI models have limitations, such as coarser resolution and a limited range of predicted variables. This study proposes a hybrid forecasting system that combines accurate large-scale information from a data-driven model with NWP models to enhance predictions. The improvements in predictability gains are significant, equivalent to several years of research, offering the best of both worlds.” – Miguel Tremblay, Advances and Applications of Artificial Intelligence in Meteorology Chair

Severe and Extreme Convective Storms: Update on the Meteorological Service of Canada’s Convective Alert Modernization project

– By Bradley Power – 

1. What drew you to this area of research?

Since I graduated from University, I have felt there were many opportunities to improve the convective warnings in Canada.  After I became a forecaster with the Meteorological Service of Canada (MSC), I created a small tool that queried post processed Radar data to help forecasters quickly create the text that is added to our warnings. Though not widely adopted outside of my home office, it formed the nascent idea of how the convective warning process could be streamlined. About a decade later, the MSC was ready to begin modernizing its alerting systems and I was asked to lead the development effort for improving our summer time convective warnings. I have been truly blessed to be part of such a talented and devoted team who is making significant strides in helping potentially save Canadian lives.

2. What is an interesting or unexpected finding you had?

Though it seems like a straightforward process, the system to get reliable and understandable warning information quickly into the hands of the Canadian public is more complex and interconnected than you would expect. Though the MSC is responsible for producing the warning messages, we must rely heavily on our partners in Canada’s weather and telecommunication industries to amplify our warnings. Without our partners, the public would not have warning information delivered to them in less than 30 seconds after the forecaster sounds the alarm.

3. What are you looking forward to most next?

I am really looking forward to some of the changes coming to the MSCs alerting program. The change that I am leading will introduce more precise severe thunderstorm and tornado warnings to reduce the amount of areal over-alerting present with our current system. Another change in development will see the introduction of new colour-coded alerts with escalating levels for high-glance information on potential impacts of severe weather events. Though these first steps will be seemingly small, it is hoped they will have a large impact in helping Canadians make safer, more informed decisions to mitigate their exposure to weather related hazards. These changes will be stepping stones opening up new opportunities to be built upon for years to come. Knowing that my team and I played a small part in potentially saving peoples lives will be quite rewarding.

Ocean acidification is not a slow burn in some coastal regions: rapid modulation of corrosive conditions in the northern Strait of Georgia

– By Wiley Evans – 

“The talk uses the specific example of the Strait of Georgia (due to its exceptional data record) to challenge the idea that changes in ocean acidification are gradual. Unexpected rapid change is likely to occur in many coastal regions leading to short lived extremes which can adversely affect organisms. The Arctic and many coastal areas with limited buffering capacity are particularly susceptible to rapid shifts in CO2 levels. These areas are important for global biodiversity and understanding these rapid changes may help us prepare for future impacts. ” – Amber Holdsworth,  Advancing Research on Marine Extremes Chair

Atmosphere, Ocean, and Climate Dynamics: Climate Change and the Changing Dynamics of the Beaufort Gyre

– By Rosalie Cormier – 

1. What drew you to this area of research?

The ocean has always awed me with not only its capacity to sustain life, but also the way its intricate, multiscale dynamics give the impression that the ocean has a life of its own. I am curious to understand the many interrelationships between the physical fields that comprise the ocean, because this allows us to gain a deeper understanding of how the ocean shapes our climate. I am passionate about both learning and sharing my knowledge of the physics of the ocean, particularly in the context of mitigating further climate change.

2. What is an interesting or unexpected finding you had?

It has been interesting to get to look at observational and hindcast data illustrating the changes to the large-scale dynamics of the Beaufort Gyre throughout the past several decades. It is well known that the Beaufort Gyre is one of the Arctic regions undergoing the most rapid decline in sea-ice concentration, and it has been interesting to observe the simultaneous changes in other properties of the gyre, including the migration of its centre and strengthening of its surface circulation. 

3. What are you looking forward to most next?

I am looking forward to continuing to use numerical simulations of an idealized model to better understand the dynamics of baroclinic eddies in the Beaufort Gyre. We are working on understanding the relationship between surface forcing and the manifestation of baroclinic instability in the gyre, and it will be interesting to compare our results to theoretical predictions and to observations.

Urban and Non-Urban Environmental Components and their impacts on Vector-Borne and Human Infectious Disease risks via effects on local Climate in Canada

– By Sukanya Ghosh – 

“Sukanya Ghosh has a good story to tell, in a field that is clearly set to grow and develop as we become more familiar with high resolution satellite and model data, and are looking to make challenging, nonlinear associations with new climate-related risks for Canadians.  Her CMOS presentation seemed only to touch on work that she might be publishing now, but I suspect that the Bulletin may also reach at least as many people.” – Rick Danielson, Multidisciplinary – Theoretical to Applied Science Chair

Earth System Modelling in Canada: A New Estimate of the Climate Sensitivity of CMIP Earth System Models

– By Prof. Ivy Tan – 

1. What drew you to this area of research?

Climate change impacts everyone.  Clouds are a primary concern in the context of climate change due to the uncertainties associated with how they form, evolve, and dissipate.  Narrowing the uncertainty in climate projections will largely hinge on how well we understand cloud processes.

2. What is an interesting or unexpected finding you had?

A number of climate models have apparently overcompensated for their previous underestimates in the proportion of liquid in cold clouds.  Adjusting for this overcompensation in the proportion of liquid in cold clouds implies less projected warming.

3. What are you looking forward to most next?

I am looking forward to seeing the evolution of climate science in the next decade as new tools and methods are being developed.  I’m particularly excited to see new developments in Canada’s CanESM model, which collaborative efforts will play a pivotal role in, and also Canada’s innovative development of satellite instruments to observe clouds and aerosols as a part of the High-altitude Aerosols, Water vapour and Clouds (HAWC) mission.  Together, these efforts are expected to enable us to better understand cloud processes and provide a better consensus on future climate change.

Calibration of Parameters of Distributed Land Surface Models Using a Deeping Learning Technique

– By Qingyun Duan – 

“This work is worth recommending to a broad audience because it introduces an innovative deep learning approach, GAN-PO, that significantly enhances the accuracy and spatial consistency of land surface model (LSM) simulations. The method’s application to the Variable Infiltration Capacity (VIC) model demonstrates substantial improvements in simulating evapotranspiration, a critical process for understanding water cycles. Additionally, the use of advanced neural networks to address calibration challenges represents a significant advancement in Earth system modeling, making it highly relevant for researchers and practitioners in climate science and hydrology. ” – Yanping Li, Leveraging Artificial Intelligence for Enhanced High-Resolution Regional Climate Modeling of Extreme Events under Climate Change Chair

Canada’s Three Oceans multi-decade ocean – sea-ice hindcast: A Hudson’s Bay sea-ice study.

By Sarah MacDermid, Youyu Lu, Li Zhai, Xianmin Hu, David Brickman 

We will briefly report on the progress of a collaboration project entitled “Assessing on-going ocean climate change: A high-resolution climate simulation for Canada’s Three Oceans from 1958-to-Present day” supported by the Competitive Science Research Fund of Fisheries and Oceans Canada.

The “Three Oceans” model is based on version 3.6 of Nucleus for European Modelling of the Ocean (NEMO) and version 3 of the Louvain-la-Neuve Sea Ice Model (LIM3). The model domain covers north of 45°N in the North Pacific, the Arctic, and north of 7°N in the North Atlantic Oceans. The model includes tides, and is driven by hourly ERA5 atmospheric forcing and monthly lateral boundary conditions, provided by ORAS5. Monthly varying river runoff and Greenland ice sheet meltwater is also included.

While the project will be completed using a grid with a nominal horizontal resolution of 1/12° in longitude/latitude and 75 z-levels in the vertical, we make use of a coarser model resolution of ¼° to make testing and parameter tuning quicker. We will be presenting results from a 64 year simulation of this coarser model, concentrating on possible trends and historical changes in sea-ice phenology in the Hudson’s Bay.

Leveraging Data-Driven Weather Emulators to Guide Physics-Based NWP Models: A Fusion of Forecasting Paradigms 

By Syed Zahid Husain,  Leo Separovic,  Jing Yang,  Christopher Subich,  Rabah Aider

Numerical weather prediction (NWP) models that rely on a physics-driven approach to simulate atmospheric processes have long been the gold standard for meteorological forecasting. However, the advent of data-driven models inspired by artificial intelligence (AI) has recently started to seriously challenge this well-established paradigm. These AI models are generally based on some form of deep neural network architecture. A number of these models and their trained weights have recently been made open-source, e.g., GraphCast by Google’s DeepMind, Pangu-Weather by Huawei, and FourCastNest by NVIDIA.

By training on the ERA5 reanalysis dataset from ECMWF (European Centre for Medium-Range Weather Forecasts), the weights of these models are calibrated to make predictions that emulate ERA5. As a result, models like GraphCast and Pangu-Weather can even surpass the accuracy of ECMWF’s Integrated Forecasting System (IFS) in certain metrics. More importantly, they can make predictions with computational efficiency that is orders of magnitude higher than any NWP model.

Despite their advantages, AI models can suffer from excessive smoothing of fine-scale features that may progressively worsen over longer lead times, affecting resolutions up to 1000-1500 km. To address this limitation, efforts are underway at Environment and Climate Change Canada (ECCC) to combine the AI and NWP modelling efforts through well-designed spectral nudging of NWP forecasts towards the large-scale states predicted by an AI model. Such an approach may help to improve NWP guidance while eliminating fine-scale smoothing and give operational meteorologist access to all the prognostic and diagnostic variables they are used to. A pertinent study at ECCC in this regard is aiming to identify the strengths and limitations of both AI and NWP models, with preliminary findings indicating higher spectral coherence in AI models over a wide range of scales. The detailed results from this ongoing comparative study will be shared at the conference.

Update on the Meteorological Service of Canada’s Convective Alert Modernization project

By Bradley Power

The Convective Alert Modernization (CAM) project, initiated in early 2022, aims to modernize the production and delivery of convective warnings in Canada. The current practice of creating convective alerts for predefined zones results in areas being alerted where there is no actual threat expected. The project aims to reduce areal over-alerting of tornado and severe thunderstorm hazards by introducing forecaster defined free-form polygons. This change will more precisely represent the boundaries of the predicted convective threat areas. Internal evaluations have demonstrated potential service improvements and highlighted challenges with the introduction of this new threat boundary paradigm in Canada. This presentation will discuss the approaches being considered to address the challenges and will also provide a project update.

Ocean acidification is not a slow burn in some coastal regions: rapid modulation of corrosive conditions in the northern Strait of Georgia

By Wiley Evans, Justin Del Bel Belluz, Katie Campbell, Carrie Weekes, Jessy Barrette, Eva Drew Jordison, Kimberly Bedard, Jonathan Bergshoeff, Ian Giesbrecht, Alex Hare, Colleen Kellogg, Jennifer Jackson

There is a perception that ocean acidification (OA) is a gradually intensifying phenomenon; however, recent studies have illustrated large rates of change in weakly-buffered seawater within the ocean interior and in the Arctic. Rapid changes in marine CO2 chemistry are also likely to occur in many coastal regions that exhibit a weak capacity to buffer natural and anthropogenic CO2 additions. Rapid abatement in adverse conditions may also occur, leading to short-lived extremes that manifest on time scales dictated by the nature of physical and biogeochemical forcings. The Strait of Georgia, on the northeast Pacific coast, is one such region that has exhibited short-lived extremes in marine CO2 chemistry. Here, we evaluate inter-annual physical and biogeochemical variability using an 8-year record of bi-weekly measurements from an oceanographic station to show how the seasonal manifestation of extremely corrosive, low-pH, and hypercapnic conditions in the northern terminus of this region is related to wintertime wind and summertime productivity season intensities.

Climate Change and the Changing Dynamics of the Beaufort Gyre

By Rosalie Cormier, Francis Poulin  

The Beaufort Gyre (BG) is one of the two major currents in the Arctic Ocean and is driven anti-cyclonically by surface winds. The transfer of momentum from the winds to the gyre is partially damped by the seasonally-fluctuating sea-ice cover, which has, on average, dramatically declined over recent decades. Wind forcing mixes the uppermost layer of the BG and skews the gyre’s isopycnal surfaces to create a strong vertical shear below the mixed layer. The baroclinic instability associated with this shear generates baroclinic eddies, which stir the water column and lift warm water from depth. Field measurements of the BG and data-assimilation models reveal that an increased upward heat flux correlates with recent sea-ice retreat in the BG region. This talk will explore the changing dynamics of the BG through two complementary approaches. First, we will chronicle the dynamics of the BG over the past several decades via an analysis of ECCO (Estimating the Circulation and Climate of the Ocean) model data. We will focus on how the temperature, density, and velocity profiles of the BG have changed, and how these changes correlate with the loss of sea ice. Second, we will present an idealized model of the BG, informed by our ECCO-data analysis, that can be simulated numerically using the Julia-language library Oceananigans.jl. We discuss how this model is used to more precisely parameterize baroclinic phenomena in the BG.

Urban and Non-Urban Environmental Components and their impacts on Vector-Borne and Human Infectious Disease risks via effects on local Climate in Canada

By Sukanya Ghosh, Philippe Gachon, Nicholas H. Ogden

Climate change driving global warming is due to human-induced activities that include modification of Land-Use and Land Cover (LULC) alongside the continuous emission of greenhouse gases into the atmosphere. As a result, agricultural and forestry lands are partially converted to increased impermeable surfaces and non-vegetated space, which exacerbates the urban heat island (UHI) and its effects on local temperature. The effects of warming and other climate change indicators, such as altered precipitation and humidity patterns, may have a significant impact on vectors (mosquitoes, fleas, and ticks), as well as our capacity to prevent and control vector-borne diseases. Many vector-borne illnesses are currently a threat to North America, including Canada. Variability and changes in the daily, seasonal, or annual climate can lead to the adaptability of vectors and pathogens, as well as changes in their geographic locations. Thus, this study aimed to find the association between climate-change, the impact of LULC change, and potential changes in UHI on the transmission of vector-borne diseases in North America. Furthermore, it is important to comprehend and manage the historical and existing LULC characteristics to evaluate the association between the climate and vector-borne diseases. The study concentrated on the major Urban (and non-Urban) Environmental Components (UEC) that include LULC, vegetation and build-up indices, land surface temperature (LST), UHI along with the on-going climate change (i.e. effects on temperatures and precipitation) affecting the spatial and seasonal distribution of vector-borne populations and diseases in North America. Satellite Images are used to identify the spatial correlation between UEC and climate fields that influence the distribution of vector-borne populations. Landsat-8 series and MODIS available datasets covering a period from 2003-2023 are used and investigated using geospatial techniques. The UHI, derived from LST obtained from satellite images, is strongly associated with changes in LULC over the past 20 years. This evaluation or relationship has been made possible by the application of cutting-edge analysis techniques, improvements in computing power, and the freely available five-decade archival remotely sensed datasets. Those include different study area characteristics, LULC classification, normalized difference vegetation index (NDVI) and normalized difference built-up index (NDBI) with the association of spatial and temporal changes of LST and climate modifications of UHI. The results obtained from satellite data for climate modifications are validated using the reanalysis ERA-5 land datasets. The following steps will be to develop a model-based infectious and vector-borne diseases over North America using these geospatial information and regional climate model simulations to evaluate the changes in the risks of infectious and emerging diseases in Canada under future climate conditions.

A New Estimate of the Climate Sensitivity of CMIP Earth System Models

By Ivy Tan, Chen Zhou,  Aubert Lamy, Catherine Stauffer  

The projected change in Earth’s global mean surface air temperature in response to a doubling of atmospheric carbon dioxide concentrations — known as climate sensitivity — remains highly uncertain.  Attempts to narrow Earth’s climate sensitivity using Earth System Models (ESMs) have remained elusive in large part due to clouds.  Previous studies have shown that exaggerating the proportion of ice in cold clouds in ESMs participating in CMIP5 were linked to underestimated climate sensitivity values.  A number of ESMs in CMIP6 have since exaggerated the proportion of liquid in cold clouds that were potentially linked to high climate sensitivity values.  Here, we analyze the CMIP5 and CMIP6 ESMs and find a linear relationship emerge between the change in cloud opacity in response to global warming and the proportion of liquid in cold clouds on the global scale, thus offering a potentially promising means of constraining climate sensitivity to more realistic values.  We use this relationship, with underpinnings rooted in the physics of mixed-phase clouds, alongside theory and global satellite observations to derive a new estimate of the climate sensitivity of the CMIP5 and CMIP6 ESMs.

Calibration of Parameters of Distributed Land Surface Models Using a Deeping Learning Technique

By Qingyun Duan,  Ruochen Sun

Land surface models (LSMs) are an important component of the Earth system models (ESMs), playing the role of simulating various land surface processes, including water, energy and carbon cycles over land. Calibration of LSM parameters presents an enormous challenge due to the compression of information inherent in model outputs and observations into a single-value objective function, which leads to uneven spatiotemporal performance of the model. We propose the use of a deep learning technique to fully utilize spatiotemporal information from the model outputs, observations as well as from land surface characteristics. Here, we presents a generative adversarial network-based Parameter Optimization (GAN-PO) method, which leverages a deep neural network to discern model spatial biases to produce spatially consistent parameter fields and minimizes the differences between simulations and observations. We applied GAN-PO to the Variable Infiltration Capacity (VIC) model to simulate evapotranspiration (ET) over China’s Huaihe basin. We will show that GAN-PO can diminish errors in simulated ET across nearly all grid cells within the study region. Notably, due to the discriminator’s explicit identification of model spatial biases, GAN-PO excels in maintaining spatial consistency.