Tag page additions and fixes

content/blog/dataassimilation.md

@@ -0,0 +1,35 @@
+---
+title: 'Data Assimilation'
+draft: false
+featured: true
+weight: 1
+heroHeading: 'Data Assimilation'
+heroSubHeading: ''
+heroBackground: '/images/retrosupply-jLwVAUtLOAQ-unsplash.jpeg'
+---
+
+<h3 style="text-align: center;">Climate models & observations</h3>
+
+Climate models are powerful tools to simulate and predict the Earth’s climate system, but they aren't without flaws. Due to missing physics, imperfect parameterizations, and inaccuracies in the underlying numerics, climate models often show structural errors. On the other hand we have direct observations of the climate system. However, observations come with their own set of limitations. They have limited space and time coverage and also contain errors due to measurement noise.
+
+
+<center>
+<img src="/images/research/DAillustration-logo.png" style="width: 20vw; padding-bottom: 30px; padding-top: 0px">
+</center>
+<p style="text-align: left;"><small><b>Figure:</b> Hoteit, I., et al., 2018: Data assimilation in oceanography: Current status and new directions. In "New Frontiers in Operational Oceanography"</small></p>
+
+<h3 style="text-align: center;">Data assimilation</h3>
+
+To overcome  the challenge of uncertainty inherent in both sources of information–models and observations– we employ a method known as data assimilation. Data assimilation merges the information obtained in models and observational data, and produces a “best guess” of the climate system. The schematic above outlines the core process of sequential data assimilation. The model state variables (blue) are continuously adjusted based on the most recent observations (red). This adjustment process (dashed red arrow) is often referred to as the analysis step or analysis increment.
+
+<h3 style="text-align: center;">Learning from analysis increments</h3>
+
+The analysis increments provide useful information on model errors. For example, if the analysis increments always correct the model to a warmer state (i.e., all dashed arrows point upward), this may reflect that the model has a systematic cold bias. Several members of M<sup>2</sup>LInES employ machine learning techniques to learn such model errors from analysis increments.
+
+<h3 style="text-align: center;">Learning model error</h3>
+
+Machine learning algorithms can also be used to understand and correct biases that are due to the combined error of physics and numerics. One way to learn this combined error is to use analysis increments as a training dataset. Analysis increments represent the adjustments made to a model to bring it closer to observations during the data assimilation process. The information contained in analysis increments allows therefore for the development of correction schemes that improve the reliability and accuracy of model predictions.
+
+<h3 style="text-align: center;">Emulation of the full model dynamics</h3>
+
+Another important application of machine learning in climate modeling is the development of emulators. Climate model emulators are surrogate models that mimic the behavior of complex climate models. By capturing the essential features and relationships within the original models, emulators provide a computationally efficient alternative for exploring climate model outputs.

content/blog/machinelearning.md

@@ -7,16 +7,18 @@ heroHeading: 'Machine Learning'
 heroSubHeading: ''
 heroBackground: '/images/retrosupply-jLwVAUtLOAQ-unsplash.jpeg'
 ---
-<center>
-<img src="/images/research/Colored_neural_network.svg.png" style="width: 20vw; padding-bottom: 30px; padding-top: 0px">
-</center>
+
 Machine learning methodologies, particularly neural networks, are computer algorithms inspired by the structure and function of the human brain. These algorithms are capable of learning from data, identifying patterns, and making predictions or decisions without being explicitly programmed. Neural networks consist of interconnected nodes organized into layers, including input, hidden, and output layers. During training, they adjust the strength of connections between nodes, called weights, based on examples from a dataset, allowing them to learn complex patterns and relationships in the data. Neural networks are widely used in various applications, including image recognition, natural language processing, and climate modeling.
 
 
+<center>
+<img src="/images/research/Colored_neural_network.svg.png" style="width: 20vw; padding-bottom: 30px; padding-top: 0px">
+</center>
+<p style="text-align: center;"><small>Source: wikipedia</small></p>
 
 <h3 style="text-align: center;">Machine learning for climate modeling</h3>
 
-Climate models often face challenges in representing complex processes of large-scale simulations. Machine learning presents innovative approaches to confront these obstacles. It provides new methodologies to learn missing model physics and model errors directly from data. Moreover, machine learning holds the potential to act as a feasible substitute for emulating the complete dynamics of models, thus offering an alternative to traditional climate modeling approaches. The next three paragraphs describe three applications of machine learning for climate modeling that are pursued within the M2LInES project.
+Climate models often face challenges in representing complex processes of large-scale simulations. Machine learning presents innovative approaches to confront these obstacles. It provides new methodologies to learn missing model physics and model errors directly from data. Moreover, machine learning holds the potential to act as a feasible substitute for emulating the complete dynamics of models, thus offering an alternative to traditional climate modeling approaches. The next three paragraphs describe three applications of machine learning for climate modeling that are pursued within the M<sup>2</sup>LInES project.
 
 
 <h3 style="text-align: center;"> Learning missing physics (“parameterization learning”)</h3>

content/blog/ocean.md

@@ -0,0 +1,58 @@
+---
+title: 'Ocean'
+draft: false
+featured: true
+weight: 1
+heroHeading: 'Ocean'
+heroSubHeading: ''
+heroBackground: '/images/retrosupply-jLwVAUtLOAQ-unsplash.jpeg'
+---
+
+<h3 style="text-align: center;">Ocean modeling and turbulence</h3>
+
+<center>
+<img src="/images/research/ocean1.png" style="width: 50vw; padding-bottom: 30px; padding-top: 0px">
+</center>
+<p style="text-align: left;"><small><b>Figure:</b> The ocean circulation is composed of many coherent features distributed over a wide range of scales. Snapshot of surface relative vorticity (swirliness) from a global 1/10 degree simulation with MOM6, showing the presence of ocean eddies, sharp boundary currents, tropical instability waves, shelf circulations, etc.</small></p>
+
+The ocean is a turbulent fluid, in which motions on a wide range of scales – from centimeters to thousands of kilometers – interact and exchange energy. Ocean turbulence plays an important role in the climate system because it can affect large-scale ocean currents as well as ocean heat uptake. Among the key features of ocean turbulence are mesoscale eddies, submesoscale eddies, and vertical mixing processes.
+
+<ul>
+    <li><b>Mesoscale eddies</b> are large swirling vortices with horizontal scales typically ranging from tens to hundreds of kilometers. These eddies play an important role in the ocean energy cycle (see below) and are often associated with the meandering of ocean currents such as the Gulf Stream. Mesoscale eddies play a crucial role in transporting heat, salt, and nutrients across ocean basins.
+    </li>
+    <li><b>Submesoscale eddies</b> have smaller horizontal scales, typically ranging from hundreds of meters to tens of kilometers, and occur in the form of fronts, filaments, and spirals. Submesoscale eddies exhibit relatively large vertical velocities, which promote the exchange of properties between the mixed layer and ocean interior and control restratification of the upper ocean boundary layer.
+    </li>
+    <li><b>Vertical mixing</b> processes, such as turbulent diffusion and shear-induced mixing, are fundamental for redistributing heat, nutrients, and dissolved gasses throughout the water column. Mixing occurs across a range of spatial and temporal scales, from turbulent overturns driven by wind and surface waves to internal waves generated by tidal currents and density gradients.
+    </li>
+</ul>
+
+<h3 style="text-align: center;">Ocean energy cycle</h3>
+
+The ocean energy cycle plays an important role in transporting heat, carbon, and nutrients throughout the world’s oceans, and therefore needs to be accurately represented in climate models. However, much of the ocean energy cycle is not resolved in current models because many of the important energy pathways and exchanges are driven by mesoscale eddies.
+
+<center>
+<img src="/images/research/ocean2.png" style="width: 20vw; padding-bottom: 30px; padding-top: 0px">
+</center>
+<p style="text-align: center;"><small><b>Figure:</b> Zanna, Bachman, Jansen et al. (2020)</small></h>
+
+The schematic above shows a simplified schematic of how mesoscale eddies contribute to the the ocean energy cycle in three stages:
+<ol>
+<li><b>Forward Cascade of APE:</b> Available potential energy (APE) arises from horizontal density (or buoyancy) gradients associated with variations in temperature and salinity. As mesoscale eddies form and grow, they tap into this reservoir of APE, and transfer large-scale APE to smaller scales in a forward cascade.
+<li><b>Conversion of APE to eddy KE:</b> Near the deformation scale, where mesoscale eddies reach their maximum size and strength, a significant portion of the APE is converted into eddy kinetic energy (KE) by a process called baroclinic instability.
+<li><b>Inverse Cascade of KE:</b> As eddies merge and interact, they transfer their kinetic energy to larger-scale structures, thus triggering a kinetic energy inverse cascade.
+</ol>
+
+<h3 style="text-align: center;">Parameterization in Ocean Models</h3>
+
+Due to computational cost, the grid cells in most ocean models are not small enough to simulate ocean turbulence phenomena such as mesoscale & submesoscale eddies and vertical mixing. To still account for the effects of ocean turbulence on the resolved scales and the ocean energy cycle, one can use simplified representations of these processes: parameterizations. In M<sup>2</sup>LInES, we aim to use machine learning approaches to develop such ocean turbulence parameterizations. Specifically, we target
+
+<ul>
+    <li>Mesoscale eddy buoyancy parameterizations, which parameterize the forward cascade of APE.
+    </li>
+    <li>Mesoscale eddy momentum parameterizations, which parameterize the KE inverse cascade.
+    </li>
+    <li>Combined mesoscale eddy momentum + buoyancy parameterizations, which jointly parameterize the forward cascade of APE, the conversion of APE to eddy KE, and the KE inverse cascade.
+    </li>
+    <li>Submesoscale parameterizations</li>
+    <li>Vertical mixing parameterizations</li>
+</ul>

content/team/JiarongWu.md

@@ -6,10 +6,8 @@ jobtitle: "Affiliate, Postdoc"
 promoted: true
 weight: 52
 Website: "https://jiarong-wu.github.io/"
-
+tags: [Ocean, Coupled Physics]
 ---
 
 
 NYU
-
-*Ocean, Coupled Physics, Ocean Surface Waves*

data/tag_links.yml

@@ -1,6 +1,6 @@
-Ocean:
+Ocean: /blog/ocean
 Climate:
 Machine Learning: /blog/machinelearning
-Data Assimilation:
+Data Assimilation: /blog/dataassimilation
 Sea Ice:
 Atmosphere:

themes/hugo-hero-theme/assets/scss/pages/services/_page-services-single.scss

@@ -1,18 +0,0 @@
-.page-services-single {
-  .content {
-    > p {
-      &:first-of-type {
-        font-size: 1.2rem;
-        line-height: 1.5;
-        margin-bottom: 40px;
-        font-weight: 300;
-        @include media-breakpoint-up(md) {
-          font-size: 1.4rem;
-          line-height: 1.4;
-          margin-bottom: 40px;
-          font-weight: 300;
-        }
-      }
-    }
-  }
-}

themes/hugo-hero-theme/layouts/blog/single.html

@@ -3,11 +3,12 @@
 {{ define "header_classes" }}{{ end }}
 
 {{ define "main" }}
-{{ partial "hero-image-setheight.html" (dict "background" .Params.heroBackground "heading" .Params.heroHeading "subheading" .Params.heroSubHeading "section" .Section "content" .)}}
+{{ partial "hero-image-setheight.html" (dict "background" .Params.heroBackground "heading" .Params.heroHeading "subheading" .Params.heroSubHeading  "content" .)}}
 <div class="container pt-4 pt-md-10 pb-4 pb-md-10">
   <div class="row justify-content-center">
     <div class="col-12 col-md-8">
       <div class="service service-single">
+
         <div class="content">{{.Content}}</div>
       </div>
     </div>