Merge pull request #101 from johannag126/master

update everything

content/_index.md

@@ -3,5 +3,5 @@ title: 'Home'
 date: 2018-02-12T15:37:57+07:00
 heroHeading: 'M²LInES - Multiscale Machine Learning In Coupled Earth System Modeling'
 heroSubHeading: 'M²LInES (pronounced M-square-lines) is an international collaborative project with the goal of improving climate projections, using scientific and interpretable Machine Learning to capture unaccounted physical processes at the air-sea-ice interface.'
-heroBackground: 'images/ishan-seefromthesky-eWSh-bkaKU8-unsplash.jpg'
+heroBackground: 'images/silas-baisch-ceITO2rlDgc-unsplash.jpg'
 ---

content/blog/atmosphere.md

@@ -9,3 +9,17 @@ heroBackground: '/images/retrosupply-jLwVAUtLOAQ-unsplash.jpeg'
 ---
 
 🚧 Under Development 🚧
+
+<h3 style="text-align: center;">Atmosphere</h3>
+
+##### Parameterization of moist process in the atmosphere
+###### People involved: Paul O’Gorman, Griffin Mooers, Pierre Gentine
+Much of the uncertainty in climate-model projections for surface precipitation and winds comes from the need to parameterize subgrid processes such as moist convection. We are developing and implementing new parameterizations for these processes using machine learning trained on high-resolution simulations. We aim to develop parameterizations that are robust, stable and physically consistent. See Yuval and O’Gorman 2020 and Yuval, O’Gorman and Hill 2021 for examples of these research projects.
+<center>
+<img src="/images/research/Moistprocess.png" style="width: 50vw; padding-bottom: 30px; padding-top: 0px">
+</center>
+<p style="text-align: left;"><small><b>Figure:</b> Structure of a machine-learning parameterization of subgrid moist processes in the atmosphere. The structure is chosen so that the parameterization conserves energy and water.</small></p>
+
+##### Parameterization of the boundary layer
+###### People involved: Alexander Connolly, Pierre Gentine
+Boundary layer turbulence parameterization remains a major source of uncertainties in climate models, including for low-level clouds. We aim to develop a new approach to the boundary layer parameterization by targeting high-order closure terms in the turbulence representation, leveraging Large-Eddy Simulations and machine learning/symbolic regression. 

content/blog/climate.md

@@ -8,4 +8,10 @@ heroSubHeading: ''
 heroBackground: '/images/retrosupply-jLwVAUtLOAQ-unsplash.jpeg'
 ---
 
-🚧 Under Development 🚧
+Comprehensive climate models typically include **atmosphere, ocean, sea ice, and land components** and the **coupling** between them. These models can also incorporate **land ice, atmospheric chemistry and terrestrial and marine biogeochemistry**, enabling carbon cycle simulations. Early climate models were developed in the 1970s and have increased in complexity over the years, with more process interactions, more sophisticated parameterizations of subgridscale processes, and higher spatial resolution being incorporated over time. These models have been skillful at predicting anthropogenic climate change, and even early models accurately simulated aspects of the spatial pattern of warming. However, there is still considerable uncertainty associated with model structure, and climate models which incorporate different parameterizations can differ greatly in many characteristics of future projected change. Because of this, it is imperative that there are continued developments and improvements of these modeling systems. Bringing new approaches, such as **Machine Learning**, to this challenge has the potential to rapidly accelerate progress.
+
+Studies across M²LInES are using **scientific and interpretable Machine Learning** to gain new insight on parameterization development across the atmosphere, ocean, and sea ice systems. Development and testing of these parameterizations is underway in component model configurations and work is planned to incorporate these into a number of climate models. This includes, among others, improved parameterizations of: 
+* The simulated conductive heat fluxes through sea ice ([Zampieri et al, 2024](https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023GL106760)), 
+* Ocean mixing processes, 
+* Moist convection in the atmosphere 
+

content/blog/climateprocess.md

@@ -9,13 +9,30 @@ heroBackground: '/images/retrosupply-jLwVAUtLOAQ-unsplash.jpeg'
 ---
 
 Despite drastic improvements in climate model development, current simulations have difficulty capturing the interactions among different processes in the atmosphere, oceans, and ice and how they affect the Earth’s climate; this can hinder projections of temperature, rainfall, and sea level.
-M²LInES will be focusing on understanding these key climate processes using two types of data:
-1. [Data from high-resolution simulation and observations. ](/blog/research4)
-2. [Data resulting from model errors, also called data assimilation increments.](/blog/dataassimilation)
+
+**M²LInES will be focusing on understanding these key climate processes using two types of data:** 
+
+### High-resolution simulation and observations
+* Atmospheric convection and clouds (O’Gorman, Mooers, Yuval) (see [Atmosphere](/blog/atmosphere))
+* Atmospheric boundary layer processes at the ocean and sea-ice interface (Gentine, Connolly) (see [Atmosphere](/blog/atmosphere))
+* Ocean mesoscale buoyancy fluxes (Balwada, Everard) (see [Ocean](/blog/ocean))
+* Ocean submesoscale processes (Le Sommer, Barge) (see [Ocean](/blog/ocean))
+* Ocean mesoscale momentum, energy and air-sea interactions (Zanna, Perezhogin) (see [Ocean](/blog/ocean))
+* Vertical mixing (Adcroft, Reichl, Sane) (see [Ocean](/blog/ocean))
+* Sea-ice heterogeneity and its influence on air-sea-ice interactions (Holland, Zampieri) (see [Sea-Ice](/blog/seaice))
+
+### Data assimilation increments
+
+[Data assimilation increments](/blog/dataassimilation) or DA are data resulting from model errors. We will be working on DA from 3 distinct parts of climate models:
+
+* Atmospheric (Berner, Chapman)
+* Ocean (Adcroft, Lu, Du)
+* Sea-Ice (Adcroft, Bushuk, Gregory)
+
 
 Below is a representation of the physical processes that will be studied by the team:
 
 ![title](/images/research/physical-processes-simple.png)
 
 **Learn more:**
-Come discover Pangeo Forge and how it can help us solve complex problems in climate and weather research [here](https://vimeo.com/510830389) with Ryan Abernathey, head of big data at M²LInES.
+Come discover Pangeo Forge and how it can help us solve complex problems in climate and weather research [here](https://vimeo.com/510830389) with Ryan Abernathey.

content/blog/coupledphysics.md

@@ -8,6 +8,38 @@ heroSubHeading: ''
 heroBackground: '/images/retrosupply-jLwVAUtLOAQ-unsplash.jpeg'
 ---
 
-[Developing new physics-aware machine learning tools](/blog/research2)
+The earth system is complex with many different distinct subsystems interplaying with each other. Our approach in modeling the earth system is generally to develop separate models for each component and to include additional couplers that model their interactions. For example, Earth System Models typically include models of: 
+* Atmosphere
+* Land
+* Land Ice
+* Ocean
+* River Runoff
+* Sea Ice 
+* Wave
+
+In addition to improving each component independently, it is also important to better understand and model the coupling of components.
+
+Take the air-sea interaction as an example. The large-scale dynamics of coupled physics in air-sea interactions involve the exchange of energy, moisture, and momentum between the atmosphere and the ocean. These interactions play a crucial role in regulating climate patterns, driving ocean circulation, and influencing weather systems globally.
+
+### Heat Exchange:
+
+* The transfer of heat from the ocean to the atmosphere affects temperature gradients and drives atmospheric circulation.
+* For example, the Southern Ocean's heat uptake plays a critical role in absorbing heat from the atmosphere, influencing global climate patterns, and helping to regulate the Earth's climate system.
+
+### Momentum Transfer:
+
+* Wind stress over large oceanic areas generates surface currents and contributes to the overall momentum exchange between the ocean and atmosphere.
+* This process impacts circulation patterns and storm development.
+
+### Coupled Feedback Mechanisms:
+
+* Large-scale air-sea interactions are central to climate feedback, such as the El Niño-Southern Oscillation (ENSO).
+* ENSO represents complex feedback between the tropical Pacific Ocean and the atmosphere, which significantly influences global climate patterns.
+
+### Climate Regulation and Teleconnections:
+* These interactions contribute to teleconnections, where climate changes in one region can affect distant regions. 
+* Large-scale coupling helps balance heat and energy across latitudes, playing a critical role in Earth's climate regulation.  
+
+
+In M²LInES, there are projects using observational data to better model these air-sea fluxes, in collaboration with scientists from the NSF STC LEAP.  Several other projects also focus on the study of large-scale processes, which enhance our understanding of climate variability and change. This ultimately leads to improvements in climate model accuracy and better predictions for the earth system as a whole.
 
-🚧 Under Development 🚧

content/blog/projectgoals.md

@@ -10,19 +10,22 @@ heroBackground: '/images/retrosupply-jLwVAUtLOAQ-unsplash.jpeg'
 
 ## Project Goal
 
-_We aim to reduce biases at the air-sea-ice interface in existing global climate models for reliable seasonal to multidecadal timescale projections, focusing on fundamental ocean, atmosphere, and sea-ice processes._
+### _We aim to reduce biases at the air-sea-ice interface in existing global climate models for reliable seasonal to multidecadal timescale projections, focusing on fundamental ocean, atmosphere, and sea-ice processes._
 
 Two leading sources of errors contribute to climate model biases: missing processes and numerics. The missing or inadequate representation of multiscale ocean, sea-ice, and atmosphere processes (e.g., clouds, mixing, turbulence), are not resolved by the current generation of climate models due to computational limitations. Another error source arises from the climate models' numerics, which include spatial and temporal discretizations and numerical dissipation. These errors can accumulate or compensate for each other, making improving climate models intricate and requiring a range of approaches.
 
-To tackle these biases and reduce the potential sources of error, **M²LInES’ strategy is to leverage advances in machine learning & "interrogate” the data to**
+To tackle these biases and reduce the potential sources of error, **M²LInES’ strategy is to leverage advances in machine learning & "interrogate” the data to**: 
 
 1. Develop data-informed, interpretable & generalizable subgrid physics models (ocean, ice, atm);
 2. Produce error corrections derived from observational products for climate model components.
 
 By improving model physics, this strategy ensures a more faithful representation of feedbacks and sensitivities under different climates.
 
 ### Our vision
-1. _​​Generate new scientific knowledge_ in climate science from innovative use of data and machine learning: e.g., which physics did we overlook that might be important for scale interaction?
-2. _Accelerate end-to-end, from development to delivery, for a new generation of climate models_; this includes learning and testing parameterizations in global frameworks to tackle significant biases in climate models.
-3. _Drive a change of direction in the field by building models and tools centered around data-driven methods_ for the community to advance climate science discovery.
-4. _Enable a new generation of versatile scientists working at the interface of machine learning, climate science & numerical modeling._
+💡 **_Generate new scientific knowledge in climate science_** from innovative use of data and machine learning: e.g., which physics did we overlook that might be important for scale interaction?  
+
+💻 **_Accelerate end-to-end, from development to delivery, for a new generation of climate models_**; this includes learning and testing parameterizations in global frameworks to tackle significant biases in climate models.  
+
+⚙️ **_Drive a change of direction in the field by building models and tools centered around data-driven methods_** for the community to advance climate science discovery.  
+
+👩‍🏫 **_Enable a new generation of versatile scientists working at the interface of machine learning, climate science & numerical modeling._**

content/blog/seaice.md

@@ -7,5 +7,23 @@ heroHeading: 'Sea Ice'
 heroSubHeading: ''
 heroBackground: '/images/retrosupply-jLwVAUtLOAQ-unsplash.jpeg'
 ---
+<h3 style="text-align: center;">Sea Ice</h3>
 
-🚧 Under Development 🚧
+<center>
+<img src="/images/research/seaice.png" style="width: 50vw; padding-bottom: 30px; padding-top: 0px">
+</center>
+<p style="text-align: left;"><small><b>Figure:</b> Snapshot of Arctic sea ice thickness (Greenland, Barents and Kara Seas) from a CM4 simulation (⅛-degree ocean and sea ice). The Fram strait (between Greenland and Svalbard) exports approximately 10% of the total Northern Hemisphere sea ice cover out of the Arctic basin annually.</small></p>
+Sea ice is the thin layer of frozen ocean that exists at the high latitudes, forming when ocean temperatures are sufficiently cooled by the atmosphere. It plays a major role in the Earth’s climate and ecosystems, reflecting incoming solar radiation to keep surface temperatures cool, regulating large-scale ocean currents through the re-circulation of salt and nutrients, providing an integral platform for connecting Arctic communities, and also a natural habitat for endemic species. Sea ice loss due to anthropogenic climate change is posing a threat to the balance of these various human and natural systems. Over the past decade for example, the average summer Arctic sea ice cover was more than 60% lower than in the 1980s.
+<br/><br/>
+The importance of sea ice in the climate system was illustrated in the seminal works of Manabe & Stouffer in 1980, who showed that, due to positive ice-albedo feedbacks, the dominant climate response to quadrupling CO₂ is the seasonal loss of the Arctic sea ice cover. Even before this however there had been concerted efforts to understand and model the physical processes that control sea ice evolution – efforts which would subsequently lay the foundations for nearly all sea ice models over the next 4-5 decades! For example, all of the latest-generation (CMIP6) climate models still use the ice thickness distribution formulation devised by Thorndike in 1975, as well as some variation of the viscous-plastic sea ice rheology scheme developed by Hibler in 1979. Since the 1970s however, there have been several important sea ice physics developments, including: an energy-conserving model for sea ice thermodynamics (Bitz & Lipscombe, 1999), the incorporation of multiple scattering sea ice radiative transfer (Briegleb and Light 2007; Holland et al 2012), the formulation for sea ice surface melt ponds (Flocco et al 2012; Hunke et al 2013), and also recent work on sea ice rheology to better represent damage mechanics (Maxwell Elasto-Brittle rheology; Dansereau et al., 2016).
+<br/><br/>
+Although sea ice models are continuously improving, there is still work to do. We can see this in the figure below, which shows the sea ice extent climatology of the GFDL SPEAR (1-degree) climate model in both the Arctic and Antarctic. Here we see that SPEAR generally has a good representation of the total Arctic sea ice cover. On the other hand, SPEAR’s Antarctic biases are more severe, with too little sea ice across the summer–fall seasons, and then too much sea ice in winter.
+<center>
+<img src="/images/research/SPEAR.png" style="width: 30vw; padding-bottom: 30px; padding-top: 0px">
+</center>
+<p style="text-align: left;"><small><b>Figure:</b> Sea ice mean state (1979-2010) of the GFDL SPEAR large ensemble (historical simulation). Thin grey lines represent individual ensemble members and the black line is the ensemble mean. Shown for (top) Arctic, (bottom) Antarctic.</small></p>
+Biases in the sea ice mean state are due to a myriad of factors, and are often difficult to isolate. Biases may originate from missing physics within the sea ice model itself (e.g., absence of a sea ice ridging scheme), or from errors in sea ice model parameters (e.g., sea ice albedo or snow thermal conductivity). Furthermore sea ice is strongly coupled to the atmosphere and ocean, hence biases in either one of these components can imprint on the sea ice.  
+<br/><br/>
+M²LInES is working to improve sea ice model biases by developing new data-driven sea ice model parameterization schemes. Recent highlights include work from M²LInES members <b>Lorenzo Zampieri</b> and <b>Marika Holland</b>, who used in-situ data from the recent MOSAiC expedition ( <b>M</b>ultidisciplinary drifting <b>O</b>bservatory for the <b>S</b>tudy of <b>A</b>rctic <b>C</b>limate) to derive a new parametric correction to sea ice and snow conductive heat fluxes within the CICE sea ice model. This work showed that simulations which do not account for local-scale sea ice and snow heterogeneity can under-estimate conductive heat fluxes through sea ice by up to 10% (see this work in <a href="https://doi.org/10.1029/2023GL106760" target="_blank">GRL</a>).
+<br/><br/>
+Additional work by <b>Will Gregory</b> and <b>Mitch Bushuk</b> has shown that data assimilation can be used to extract the systematic component of sea ice model error within sub-grid model state variables, and subsequently that neural networks can learn this error very effectively (see work in <a href="https://doi.org/10.1029/2023MS003757" target="_blank">JAMES</a>). In follow-up work they showed that this neural network can be used to bias-correct sea ice conditions during online global ice-ocean simulations (see work in <a href="https://doi.org/10.1029/2023GL106776" target="_blank">GRL</a>), and outlined how this framework of data assimilation + machine learning can also be used to improve online generalization of machine learning models.

content/jobs/_index.md

@@ -5,15 +5,11 @@ heroSubHeading: ''
 heroBackground: 'images/SouthAtlantic.A2002157.1055.250m.jpg'
 ---
 
-Updated on 10/08/2024 - The links for applications will be updated as they become available.
+Updated on 11/22/2024 - The links for applications will be updated as they become available.
 
 M²LInES affirms the value of differing perspectives in Sciences. As such, we strongly encourage applications from women, racial and ethnic minorities, and other individuals who are under-represented in the profession, across color, creed, race, ethnic and national origin, physical ability, gender and sexual identity, or any other legally protected basis.
 
 ### Princeton University/GFDL
 
 Postdoctoral researcher or more senior scientist for Ocean Surface Boundary Layer Mixing Parameterizations using Machine Learning. [Apply here](https://puwebp.princeton.edu/AcadHire/apply/application.xhtml?listingId=36662)
 
-
-### Columbia University (NCAR location)
-
-Associate Research Scientist at the interface between climate modeling and machine learning. The successful candidate will be hired by Columbia University but the main work location will be at NCAR (Boulder, CO). [Apply here](https://apply.interfolio.com/140294)