Improving hurricane observations with data from scalable sensor networks

Sofar Ocean

Over the past year, we’ve nearly tripled our coverage of buoys, with coverage across the Pacific, Atlantic, Indian, and Southern Oceans (access data here). This extremely dense network of Spotter buoys enables us to obtain--in real-time--high fidelity observations of extreme ocean weather with unprecedented frequency. 

Current limitations with existing data collection platforms

Historically, observations of extreme events like tropical cyclones, hurricanes, or typhoons have been achieved with satellites or large, moored buoy platforms (e.g. those from the NDBC). While satellite overpasses provide invaluable data, they are of limited value in extreme conditions due to poor reliability under heavy rain conditions like those found in the most intense regions of a storm system.

Moored buoy platforms, while providing unparalleled in situ observations over a wide range of relevant variables, face extreme costs of deployment and maintenance which limits how many can be deployed--leading to fewer opportunities for extreme weather observation. In the past 37 years, there have been fewer than 100 observations made by moored, NDBC buoys with winds > 25 m/s (see Tamizi and Young 2020). 

Area of observation, Southern Pacific Ocean
Figure 1: Area of interest, Southern Pacific Ocean (source: Google Earth)
Spotter buoy locations and named tropical cyclone tracks from December 2020 to March 2021 in the southwest Pacific Ocean.
Figure 2: Sofar Spotter buoy locations (yellow) and named Tropical Cyclone tracks from Dec '20 - Mar '21

By comparison, during the most recent summer season in the Southern Hemisphere (Dec 2020 - Mar 2021), our Spotter buoy network in the South Pacific observed six cyclones, producing unique datasets on each one. As we continue to scale Spotter coverage globally, we expect to observe a growing number of extreme weather events across other basins (e.g., Northwest Pacific). The observational data--both historical and real time--of these events will go beyond anecdotal and become statistically relevant to understanding and predicting these physical phenomena. With multiple Spotters observing a single storm track from different locations, an unprecedented dataset on the temporal evolution and spatial organization of each storm will begin to emerge!

Insights gained from a high density Spotter platform

Spotter buoys transmit several relevant marine weather variables describing the waves, wind, currents and surface temperature. Even better, not only does Spotter report (in real time) the bulk wave statistics such as significant wave height and peak period, but also the full two-dimensional (2D) wave spectrum. With all of this data in hand, in real-time, we can monitor and update our understanding of extreme weather. Let’s take a look at how the Spotter platform performed during a recent cyclone.

Case Study: Tropical Cyclone Niran and Spotter-0627

On February 27, 2021, Tropical Cyclone Niran formed in the Coral Sea and ultimately grew to be a Category 5 severe tropical cyclone (maximum sustained winds of 260 km/h). Due to our high Spotter network density in the region, several Spotters were within a few radii of the cyclone eye. Here we present data from one such buoy (Spotter-0627).

A photo of a globe highlighted the area of interest regarding Tropical Cyclone Niran off the northeastern coast of Australia
Figure 3: Area of interest, Tropical Cyclone Niran (source: Google Earth)

A motion image showing Tropical Cyclone Niran's path directly over Spotter-0627 from March 4th - 5th, 2021
Figure 4: Spotter-0627 in the path of Tropical Cyclone Niran, March 4th - 5th, 2021

Significant wave height

With winds over 200 km/hr, TC Niran was expected to generate some impressive waves - and it did just that. Spotter reported significant wave heights up to 7 meters as the cyclone eye wall passed within 50 km (approximately one eye-radius).

Energy spectrum

Even more interestingly, Spotters are all reporting the full wave energy spectrum every hour. While this turns into a massive amount of data transmitted, it provides unprecedented real-time observations of the wave field as the cyclone approaches.

Why might this matter? Significant wave height, while useful, is a single statistical property of the wave field experienced by the buoy -- specifically representing the mean height of the top third highest waves. But for a wave field generated by such a violent and rapidly evolving weather event, a great deal of variability will exist in the finer details - how large are the local wind waves versus the swell (more on Sea vs. Swell here), what direction are those very different components coming from, how is the energy in the wave field (and hence the destructive power of the waves) evolving in time?

To answer these questions, we turn to the wave spectrum. And to start, we'll consider the hourly one-dimensional (1D) wave spectrum which measures the total energy at each wave frequency integrated over all directions.

Two graphs showing wave spectra data recorded by Spotter-0627
Figure 5: Wave height, energy and frequency over time recorded by Spotter-0627

The coloring of each line on the energy spectrum plot corresponds to the same color data point/dot in the significant wave height figure. Therefore, at every point in time, we have an estimate of the significant wave height (left plot), but also a full spectrum describing the energy at each frequency (right plot). Time progressing goes from dark red (warm) to dark blue (cold). So, early on (red) we can see that the peak energy corresponds to waves with a period of just over 10 seconds (or under 0.1 Hz). As time goes on (moving to green colors), we see two things happen:

The height of the peak, E(f), increases nearly an order of magnitude. Now we can see why the bulk parameter of significant wave height increased during this period - the more energy, the higher the waves!

The location of the peak on the frequency-axis moves left to lower frequencies. Ultimately, at the most intense time of the storm encounter, the peak is located at a frequency of approximately 0.06 Hz - corresponding to a wave period of 16 seconds - some long waves!

As the storm passes, we see the reversal of its approach -- peak frequencies shift higher and the overall energy decreases.

Important things to consider here

Because the wave field is ultimately a reflection of the wind forcing - there exist theoretical concepts to predict the wave field produced by a wind field modeled as a vortex. A key point of consideration is the 'sector' of the hurricane from which the observation is occurring. That is, in the reference frame of the hurricane, are you to the front left or the back right? The reason for the consistent asymmetry in the wave field owes to:

  • (1) Coriolis causes all cyclones to rotate in the same direction (clockwise, in this Southern Hemisphere case)
  • (2) the cyclone is also propagating forward in space, adding (or subtracting) to the wind strength depending on relative location
  • (3) waves require strong winds, time, and space (fetch) to develop 

For this last point, an incredibly powerful wind will only produce large waves if it blows over a long enough distance of water for enough time.

The combination of these factors manifests in a consistent structure of wave fields in cyclones. For the Southern Hemisphere, the strongest winds will be to the left of the eye of the storm (as viewed relative to its direction of travel) as the local winds add to the wind due to the forward motion of the cyclone. In terms of waves, each quadrant around the eye will also be uniquely structured. To the left, waves are expected to be more fully developed and aligned with the winds because, due to the alignment of the cyclone winds and direction of propagation of the storm, the waves will propagate forward with the storm and thus remain in the high wind regions for extended periods of time, extending the fetch (see Young, 2006). Conversely, on the right-hand-side of the eye, the wave field is expected to be highly disorganized, with contributions from the immediate local winds as well as waves propagating from the most energetic regions of the storm (i.e., storm swell).

Ahead of the cyclone

Figure 6: Spotter-0627 2D wave data measurements with Cyclone Niran approaching

In advance of the cyclone, the wave spectra (with directional convention as "going to") are well-aligned with waves propagating directly outward from the cyclone (to the southeast) and then skewed to align with the direction of the wind at higher frequencies (to the south-southeast).

In the cyclone

Figure 7: Spotter-0627 2D wave data measurements with Cyclone Niran overhead

Within the strong winds of the cyclone and closer to the storm eye, there is a clear directional split in the wave field observed (despite the one-dimensional spectra being consistently unimodal). The low frequency swell energy is largely directed to the southeast and the high frequency locally-generated wind seas are predominantly directed to the north-northwest. This quadrant of the cyclone in which the Spotter is located is limited-fetch (versus the extended-fetch region on the other side of the eye); therefore, one would expect rather complicated seas - and we can see this in the distribution of energy across a variety of directions and frequencies. Interestingly, we see the direction changes rather significantly over the frequencies. This is likely because swells at different frequencies travel at different speeds, therefore we are observing a collection of swells that originated when the cyclone was in different locations.

After the cyclone

Figure 8: Spotter-0627 2D wave data measurements after Cyclone Niran passed

Following the close pass with the cyclone eye, observations in the rear-right sector of the cyclone show a very broad spread in energy across directions - with the lower frequency energy spanning south to southwest and the higher energy spanning nearly 90 degrees from the northwest to the northeast. These complex seas are likely due to the observations occurring on the 'fetch-limited' sector of the cyclone - where a single, coherent wave field is unable to develop due to a number of factors including rapidly changing wind direction and the passage of the cyclonic wind field, resulting in a complex mix of wave ages and directions.

Looking ahead

The Spotter platform was designed to gather air-sea interaction data from the open ocean. As the network continues to scale, we are excited to continue sharing observations from some of these extremely high energy events with full-spectra data now being transmitted to help us better understand these weather phenomena.

With the learnings from observing these summer storms in the Southern Pacific, we are looking forward to seeing how Spotters perform during summer in the Northwest Pacific where these weather systems (typhoons) are both the most numerous and the strongest.

Figure 9: Full 2D data over Cyclone Niran's progression from Spotter-0627

Sofar in the News
In the News
February 26, 2024
Better, Faster, Sooner: Voyage optimization goes digital
In the News
October 23, 2023
Wayfinder delivers 8% total equivalent fuel savings on Brazil-China (BDI C14) voyages arriving in early October
In the News
October 2, 2023
What was different about the latest Advanced Naval Technology Exercise

Related Stories