Tag Archives | Visualization

The Circle of OOI

By Sage Lichtenwalner on September 6, 2013 in Visualizations

Let’s be honest. This image won’t be winning any design awards. In fact, it’s not much more than a rehash of an example script for Reingold-Tilford Trees, though I did have to delve into some intricacies of d3 and SVG to figure out how to implement line-wraps of all those long names.

But the crude design of this image belies the incredible complexity behind its subject mater, the Ocean Observatories Initiative or OOI.

The OOI is currently constructing a large network of ocean sensors that will be deployed at 6 sites around the Western hemisphere, with “771 instruments on 38 moorings, 12 seafloor instrument packages, and 27 mobile assets.” (Check out the OOI Instrument Tables for all of the details.)

When you take into consideration all of the data products that can be created by each instrument – for example, wherever you measure temperature, salinity and pressure, you can also calculate density and sound velocity – there will soon be several thousand streams of data spewing forth, as construction is completed and arrays are deployed over the next 18 months. In fact, Station Papa was just deployed last month in the Gulf of Alaska, though it will be a while yet before its data is made available to researchers.

The OOI promises to provide an amazing wealth of data for the up-and-coming generation of oceanographers to explore. And it is sure to provide a ripe source of cool new science for the next 25+ years. We better start preparing for the onslaught.

So why did I create this (admittedly, not very pretty) image?

For the past two years, I’ve been working with of a small team of developers on the OOI project to build a suite of tools for undergraduate educators. Developing web applications and data visualization tools is not an easy task on a normal day. But perhaps our biggest challenge on this project has been to wrap our heads around the giant scope of the OOI, both in terms of the complex structure of its network of sensors, and its grand goals to investigate a full swath of science themes, from carbon cycling and ocean acidification to turbulent mixing and sea-floor volcanic processes.

The image above depicts the top four layers of the current design of the system, comprising 7 arrays and their associated sites and sub-sites at each location. The arrays include 2 coastal sites (the Endurance Array off of Oregon and the Pioneer Array in the Mid Atlantic), 4 global high-latitude sites (off the coasts of Argentina, Greenland, Alaska and Chile), and a regional cabled observatory (also off the coast of Oregon).

I could have included more levels of information in this diagram (remember, there are over 750 instruments planned), but with so much data to display and process, sometimes simple is better. In fact, oftentimes that is the case. After all, any good physicist will tell you to think of the cow as a sphere.

Our team is tasked with designing interfaces that will allow teachers and students to easily visualize and analyze OOI data, and creating this image was just one exercise we used to develop our understanding of the full-scale design of the OOI.

You can expect to hear more about the OOI over the next year, and particularly the educational tools we’re working on. And we’ll definitely be looking for your help as we try to make the data and information from the OOI easy to use in educational environments.

Which is why I thought I’d share this image now, to invite you into the circle.

The End of Upwelling

By Sage Lichtenwalner on July 19, 2013 in Visualizations

What a difference a week makes.

Late last week, the waters off New Jersey were between 5-15 degrees below normal thanks a persistent pattern of coastal welling in which warmer surface waters were pushed offshore and replaced by colder waters from below.

This year’s upwelling, which typically occurs this time of year, was longer than usual due to a strong Bermuda High. The High also stalled weather fronts along the eastern seaboard, carrying a lot of moisture up the East Coast from the Gulf of Mexico. But it was the upwelling that caused a lot of consternation among beachgoers in New Jersey, particularly on Long Beach Isltand where the upwelling was strongest. (Warning, the comments on that last link are a sad example of the urgent need for scientists to become more involved in communicating science.)

However, this week, the Bermuda High shifted west, causing record high temperatures across much of the East coast and odd rainstorms out west. It also reversed the coastal upwelling pattern enough so that surface waters could return to their seasonal norms.

The image above shows the change in sea surface temperature over the last week, between July 19 and July 12. (Technically, this map shows the difference between two 7-day composites, one ending on the 19th and the other on the 12th.) Almost the entire region warmed up a few degrees thanks to the strong sun and cloudless days we’ve had. But it is the coastal waters off NJ that increased the most, from 5-12 degrees Fahrenheit. The shift in winds allowed warmer waters from offshore to head back towards the coast, while at the coast, previously upwelled cooler waters were subject to downwelling.

For the ocean, and all the fish in it, this is a huge change to cope with in just a few short days. But if you’re at the beach, it means a more pleasant swim is in your future.

Image Note: I created a new colormap for this image that includes 4 color points instead of the traditional 2 for a divergent color scale (e.g. blue to white to red). Let me know what you think!

The Ocean in Red, White and Blue

By Sage Lichtenwalner on July 4, 2013 in Visualizations

To celebrate Independence Day, I thought it would be fun to dress up the ocean in a little red, white and blue.

If you’re curious, the image above represents the gradient of sea surface temperature (SST) at each point, and is based off of today’s 7-day composite of SST collected by the AVHRR instrument on NOAA’s polar orbiting satellites.

For every point, if the temperature change from its left (or below) neighbor to its right (or above) neighbor increases, then the gradient is considered positive and is colored red. Similarly, if the temperature decreases as you go from left to right or bottom to top, then the gradient is negative and is colored in blue. If the temperature doesn’t change much, white is used. The darkest colors (red or blue) represent a temperature change of around 2-3 degrees Fahrenheit over 2km of distance.

The map above is actually the sum of the horizontal and vertical gradients (i.e. dT/dx + dT/dy), so areas that are blue, indicate areas where colder temperatures can be found towards the Northeast.

In the image, a few patterns stand out, particularly the north wall of the Gulf Stream which shows up as blue streaks, indicating colder temperatures to the North. The waters off New England show up as a dark mess of blue and red, due to the large number of clouds in the area over the past few days, not to mention all the hot weather warming up the cold waters, both of which resulted in uneven measured temperatures, and therefore chaotic calculated gradient values.

Scientists often use a gradient calculations to identify large-scale features in the ocean, like fronts and eddies, which can be seen in the image. However, to do so they generally would calculate the gradient over larger areas than the 1km pixels used above or use a combination of filtering or averaging to smooth out the features and make them stand out more.

Using raw SST data to calculate gradients results in an image that is very noisy. But then, fireworks are noisy too… and they are also quite beautiful to behold.

Happy 4th!

Tropical Storm Andrea clouds up the ocean

By Sage Lichtenwalner on June 7, 2013 in Visualizations

This week, was the start of the 2013 Hurricane Season, and already forecasters have declared the first storm of the season. So with one week down, I’d say we’re on track to meet NOAA’s prediction of an active to extremely active season.

Tropical Storm Andrea started as a small storm system in the in the Gulf of Mexico earlier this week, and by Wednesday evening she had grown (barely) into a low level Tropical storm. Tropical Storm Andrea then made landfall in the Big Bend region of Florida, causing some minor coastal flooding and wind damage in the Tampa area, before heading up the eastern seaboard. For the most part, Andrea was primarily a major rain event, dampening the spirits of many who are anxious for summer to finally arrive.

Large storm systems are also a nuisance to satellite oceanographers, who generally need a clear view of the ocean to measure physical variables like sea surface temperature, or the amount of chlorophyll and sediment in the water.

The image above was generated using data collected by the AVHRR instrument on NOAA-18 as it flew over the area at 4:36pm. AVHRR does not collect data in the visible light range, so this false-color representation was created by converting data from the red and infrared channels on the satellite into an image resembling a true-color photo.

What is clear from this image, is that the sky is not so clear over the ocean. In fact, the only clear areas over water are off the coast of South Carolina and over the Great Lakes, which show up as dark blue. To an oceanographer then, this image doesn’t offer much to look at, but for a meteorologist, it’s a different story.

For more on the aftermath of Tropical Storm Andrea’s deluge, I encourage you to check out the New Jersey CoCoRaHS site tomorrow to see how much rain fell on the state.

Streamflow and Conductance on the Delaware

By Sage Lichtenwalner on April 22, 2013 in Visualizations

Rivers play an important role in our ecosystem. They provide water for drinking and irrigation of crops, a habitat for fish and other organisms, and routes to easily transport goods. For these reasons and more, it is important to monitor the quality of river water, including its physical, chemical, and biological characteristics.

One often measured parameter is specific conductance. Conductance is a measurement of a substance’s ability to conduct electricity and is related to the amount of ions, like salt, that are dissolved in the water.

Where rivers meet the ocean, the salt typically comes from seawater flowing upstream into the river. How far upriver the saltwater can reach (often called the salt front or salt line) depends greatly on an estuary’s type and the current streamflow.

Further upstream, where the ocean doesn’t have as much influence, the amount of dissolved salts in river water is generally related to how much precipitation there has been. When rainfall is light, more water on land can evaporate before it reaches the river, which concentrates the amount of dissolved salts in the water that remains. When rainfall is heavy, water tends to flow more quickly into rivers and streams, with smaller concentrations of dissolved salts.

The image above shows the relationship between river flow (discharge) and conductance over a 3+ year period on the Delaware River in Trenton, NJ. In general, the conductance is quite low, and well below accepted salt front cutoffs of ~400-1,000 micro-Siemens per centimeter, which correspond to chloride concentrations of 100-250 milligrams/liter. However, there is clearly an inverse relationship between conductance and discharge. When discharge is strong, water conductance is low, though it never gets below ~100 µS/cm. Likewise, when discharge is light, conductance is 2-3 times higher. While the salt concentration on the Delaware River near Trenton is generally low, it does depend on the streamflow.

Knowing the location of the salt front is important, especially on rivers where water is drawn for human consumption or irrigation, and for protecting riverine infrastructure like ships that corrode more easily in salty water. The Delaware River Basin Commission regularly monitors the salt front location in order to control its location by storing and releasing water in reservoirs upstream.

Streamflow on the Delaware

By Sage Lichtenwalner on April 12, 2013 in Visualizations

Scientists who study river streamflow do not have an easy job. Unlike many weather measurements like temperature, pressure or humidity that change with more predictable variation throughout the course of a year, streamflow is more closely correlated with major rain and snow events. These events occur sporadically throughout the year, often in large doses.

The graph above is a good example of this complexity. The data shown was measured from a stream gauge on the Delaware River in Trenton, New Jersey. The daily mean streamflow (also referred to as “discharge”) from the most recent year is plotted in red, ending April 11, 2013. The long-term average streamflow statistics are calculated from a 99-year data set collected between 1913 and 2011.

The mean daily streamflow includes many large peaks over the course of the year. These peaks correspond to major precipitation events that occurred in the upper Delaware river basin. On the other hand, the lines showing the long-term average streamflow change more gradually throughout the year, with the highest streamflows observed in the spring (especially this week), and lower streamflows in the summer and fall.

Looking at this data, several interesting observations can be made.

In October 2012, Hurricane Sandy hit New Jersey causing extensive damage, however the observed streamflow during this time was not as high as some of the other events. As it turned out, most of the damage was caused by coastal flooding and high winds and not from flooding rivers.
Over the course of the winter, several large events were observed. These correspond to the major nor’easter’s that passed through the area, including the (strangely-named) winter storms Nemo and Saturn.
Last month was unseasonably cold and, while there was a nor’easter early in March, the month also ended up unseasonably dry. However, following today’s rainstorm, it is likely that the streamflow will rise to the more typical values expected during this time of year.

Ultimately, when studying rivers it’s important to remember that individual events will not always neatly line up with long-term averages, but over time, the trends should match.

Riding the Waves of the Seasonal Roller-Coaster

By Sage Lichtenwalner on March 29, 2013 in Visualizations

It is often said in Pennsylvania that March comes in like a lion and goes out like a lamb. And while this March felt more like a ride on an Arctic roller-coaster that wouldn’t end, the good news is that, based on past years, we should soon be on the downward slope towards more calmer weather.

In the Mid-Atlantic, the winter months usually bring with them strong storms and high winds, like the nor’easter we saw earlier this month. In the ocean, strong winds lead to larger significant wave heights, as can be seen in the graph above that depicts the average monthly wave heights off the coast of New Jersey over the course of a year.

This graph was created using 8 years of significant wave height data from NDBC Buoy 44025, which is a little more than 40 miles from the New Jersey coast. Each line represents a particular percentile level, indicating the percentage of measurements that fall below the indicated level. The 50% percentile level is commonly called the median average value. For each month, half of the measured data over the course of the 8 years fell above the median value while the other half fell below.

This graph shows a distinct difference between the seasons. The median wave height in December, January and February is around 4.5 feet, while in the summer months of June, July and August, the average is closer to 3 feet. While the median value is higher in winter months than summer ones, the change is even larger for the 80th percentile line. For that, we can thank those large winter storm events that turn the ocean into one rough ride.

Thankfully, spring will soon give way to summer, and if the past averages hold true, the summer months of this roller-coaster should include calmer waters.

A Rough and Significant Winter

By Sage Lichtenwalner on March 22, 2013 in Visualizations

It has been a rough winter in New Jersey, especially on the coast. First, Post Tropical Storm Sandy struck Atlantic County on October 29th 2012, becoming the costliest natural disaster in New Jersey’s history. Over the next 5 months, several additional strong storms made their way across the state, bringing with them heavy winds, coastal and inland flooding and significant snowfalls.

Strong storms, many of which are called Nor’easters, are common occurrences in the Mid-Atlantic during winter months. Their strong winds also lead to high waves in the ocean. But this past winter was rather exceptional.

The above graph shows an analysis of wave heights measured by NDBC Station 44025. Each bar depicts the maximum wave height reached for each of the 12 largest events (each lasting 2-3 days) recorded over the past eight years (from January 2005 to today). Of the 12 events with the highest waves, 6 of them have been in the last 6 months. The largest recorded wave event was, of course, due to Sandy. The 12th largest occurred during the nor’easter that struck earlier this month.

It’s important to note that this does not (yet) represent a significant trend. Taking the top 20 events into account, only the same 6 events occurred this winter. Extreme events can often occur in spurts, triggered by prevailing climatic conditions, so the likelihood of many major events coinciding is not uncommon. In addition, an 8-year dataset is far too limited to make any assumptions about long-term climate changes.

However, it should be quite clear from this evidence that New Jersey residents are certainly ready for Spring, and more importantly, calmer weather.

Significant Waves

By Sage Lichtenwalner on March 15, 2013 in Visualizations

Last week a major snowstorm travelled across the continental United states, becoming a strong nor’easter over the Mid-Atlantic. While snowfall amounts in New Jersey were far less than some had predicted, the wind and waves that battered the coast were still quite severe. Dunes in Mantoloking, NJ that were heavily damaged last fall by Hurricane Sandy were again breached, causing flooding and further hindering repairs.

Wave heights at NOAA Station 44025, just 43 miles off the coast of New Jersey, reached 18.4 feet on the night of March 6th. The blue line above shows the significant wave heights measured by the NOAA buoy over the course of the last week.

The red horizontal lines signify the percentage of hourly wave measurements recorded between 2005 and 2012 that were less than the indicated height. The top line, at 31.6 feet, represents the maximum wave height reached during the 8-year record, which occurred as Hurricane Sandy made landfall.

The maximum wave height during last week’s storm reached the 99.9th percentile. Only 1 hourly measurement in 1000 hours of measurements (the equivalent of 42 days) ever reach this level. After the peak, wave heights remained between the 90 and 99.9th percentile for 3 days, which indicates the significance of this storm.

Matlab tip: If you’re interested in calculating significant wave heights at various percentile levels at other stations or for other parameters, here’s some code to play with.

% Load in the concatenated NDBC Datafile
fid=fopen('data/44025_8yr.txt','r');
data = textscan(fid,'%4f %2f %2f %2f %2f %f %f %f %f %f %f %f %f %f %f %f %f %f %*[^\n]','HeaderLines',2,'CommentStyle','#');
fclose(fid);
 
% Calculate time and remove bad datapoints
dtime = datenum(data{1},data{2},data{3},data{4},data{5},0);
wvht = data{9};
wvht(find(wvht==99)) = NaN;
 
% Calculate percentile levels and convert meters to feet
wvd = sort(wvht(find(~isnan(wvht))));
disp([.9 wvd(round(length(wvd)*.9))*3.28084]);
disp([.99 wvd(round(length(wvd)*.99))*3.28084]);
disp([.999 wvd(round(length(wvd)*.999))*3.28084]);
disp([1 wvd(round(length(wvd)*1))*3.28084]);

A Colorful Winter Storm

By Sage Lichtenwalner on March 8, 2013 in Visualizations

A major winter storm made its way across the continental United States this week, dropping snow across the Dakotas, then the Midwest and the Mid-Atlantic before finally heading out to sea over the Northeast. While snowfall from the storm was difficult to forecast precisely, nonetheless it still caused major damage across many states.

The image above shows what the storm looked like at 9am EST on March 8, 2013, through the false-color eyes of the AVHRR instrument on board satellite NOAA-16. Unfortunately, AVHRR was not designed to measure visible light as many more modern satellites do. It’s primarily used for measuring the surface temperature of land and the ocean. The colors in the above image were approximated with a computer algorithm that converted AVHRR’s red, near-infrared and infrared channels into red, green and blue, creating this non-traditional colorful image of the Mid-Atlantic.

While the colors in this image can not be regarded as real, they are still useful. The white clouds are colder and generally higher in the atmosphere, while yellow clouds are slightly warmer and lower. Most of the clouds connected with the storm system are yellow. Storm bands are also visible as semi-circles pushing in towards Massachusetts and on down into New Jersey. This counterclockwise rotation is a common feature of a Nor’easter.

In advance of the storm, scientists at Rutgers deployed an underwater glider to measure how the storm will mix sediment in coastal waters. The glider certainly saw a lot of action, as wave heights reached 14 feet at the New York Harbor entrance, 24 feet off the coast of Virginia Beach, and 30 feet at the Hudson Canyon as the storm’s center passed by late on March 6th.

Special thanks to Steve Miller at NRL-Monterey for the code used to create this image.

1 2 Next →