Trajectory prediction for the 2006 Home Reef in the South Pacific Ocean for August-November 2006 using QuikSCAT/SeaWinds, Jason-2/Poseidon-3 and Moderate Resolution Imaging Spectroradiometer (MODIS) observations and imagery.
Team Location: Langley Research Center, Hampton, Virginia
Michael Bender (Pennsylvania State University)
Joshua Kelly (University of Rhode Island)
Maureen Kelly (University of Maryland College Park)
Corey Walters (Saint Louis University)
Kenton Ross, Ph.D. (NASA, DEVELOP National Science Advisor)
Beth Brumbaugh (DEVELOP Langley Center Lead)
Lauren Makely (DEVELOP Langley Assistant Center Lead)
Pumice rafts are expansive masses of pumice clasts floating on the oceansurface produced by silicic shallow submarine and subaerial explosive volcanic eruptions. The goal of this project was to enhance knowledge of pumice rafts and develop accessible and practical methodologies for predicting the movement of pumice rafts in the South Pacific region. Two volcanoes in this region have recently erupted and formed pumice rafts: Home Reef volcano (Tonga) in 2006 and Havre Seamount (Kermadec Islands, New Zealand) in 2012. These raft events were used as examples to test the trajectory prediction model since they occurred during times at which high spatial and temporal resolution true color imagery were being collected and they have been frequently described in peer reviewed literature, both of which were crucial in providing validation for our models.
Project partners included Dr. Bradley Scott from GNS Science New Zealand and Dr. Greg Vaughan from the U.S. Geological Survey. They are particularly interested in learning how to predict the movement of pumice rafts for enhanced navigational advisement to maritime authorities. Remote sensing data acquired from NASA’s Earth Observing System (EOS) satellites Aqua and Terra were used to image and track the pumice raft produced from the 2012 Havre Seamount eruption. Additional data acquired from NASA’s EOS satellites Jason-2 and QuikSCAT were used to predict the trajectory of the pumice raft using the General NOAA Operational Modeling Environment (GNOME). GNOME is a modeling tool used to predict the possible trajectory a pollutant might follow on a body of water using wind and ocean current satellite data.
Learning more about the processes and transport mechanisms of pumice rafts is significant for a number of ecological and economic reasons. Pumice rafts pose a hazard to marine transportation as individual clasts can block seawater intake valves of large ships and cause hull damage to smaller vessels. Rafts can also be detrimental to fisheries, a large kill of deep-sea fish followed the arrival of pumice rafts during the 1984 Home Reef eruption. Additionally, rafts have the potential to introduce harmful invasive species to pristine areas as they drastically increase dispersal distances for otherwise benthic or relatively sedentary organisms. This novel and easily adaptable methodology can be used by island and coastal nations and fishery managers to forecast when and where a pumice raft will be, drastically enhancing maritime navigational warnings and response times to eventual pumice landfall.
Great job writing this up and on the video as well. Are there any ideas about the purpose pumice rafts might serve in nature since the eruption of volcanoes and resulting pumice rafts are natural phenomena and not at all anthropogenic?
P.S. I realize the very practical emphasis of this project was on predicting an areas potential for developing pumice rafts and the movement of them, but I also noted some of the negative aspects of pumice rafts and wondered if there was anything good about them.
@anjward7 That's a great question and something we haven't really thought about considering we were concerned with the pumice rafts' impacts on humans. There's a great article in USA Today from earlier this year about how pumice rafts could have been the "Cradle of Life" billions of years ago. A variety of marine organisms have been observed to attach themselves to pumice rocks as they provide a hard substrate. These organisms such as macroalgae, mollusks, and barnacles are essentially hitching a free ride thousands of kilometers across the globe to areas where they may not have been introduced before. This process could be one of the most efficient processes for organism dispersal in both modern and ancient Earth environments. Here's the article describing in more detail - http://www.usatoday.com/story/tech/columnist/vergano/2013/02/08/life-pumice-origin/1902231/
Hi - I would be interested in comparing the results of the oil spill drift model to results from the models that have been developed for the "bathfub toe" and "Nike shoe" spills from container ships in the North Pacific Ocean. (I expect that similar models are being applied to the tsunami debris which is drifting toward the U.S. and Canadian West Coast.) As I recall, these models take surface winds more into account than oil spills as the higher profiles of floating debris make them mini-sails, more susceptible to wind effects than oil spills, which will move mostly in response to wind-driven currents. I would think that pumice floats so well that wind direction would have an effect on their movement.
@got_giovanniHi got_giovanni, thanks for the question! So far as I know, GNOME (the model we used) is the primary model that NOAA is using operationally to track the marine debris. This website talks about the very issue you brought up (i.e., windage affecting how quickly objects travel), under the context of the tsunami debris they are tracking right now. http://marinedebris.noaa.gov/tsunamidebris/debris_model.html
For the windage, we used 1-4%, which is the default in GNOME. We used this because we expect that the windage is actually quite similar to oil. Rafts stick very slightly above water and go only a few inches deep. The more notable parameters we changed were weathering (to non-weathering) and persistence (to infinite).
And you are correct! The pumice certainly was more influenced by wind than currents, and the model's algorithms automatically take that into account based on the windage we assigned. Had it been more influenced by the currents, it likely would not have reached Papau New Guinea, as our model suggested (and in situ observations compiled by the Smithsonian Institute suggests).
@got_giovanni got_giovanni, I've seen you posting around the different blogs, and I want you to know I'm a huge fan! I must say, you are out of character here, but I'm interested in knowing what "bathfub toe" and "Nike shoe" are.
1) The pumice rafts can and commonly do congregate in harbors and narrow waterways under the influence of tides and strong nearshore winds. This is when humans are most affected by pumice rafts as it makes marine transportation and fishing incredibly difficult. Here's a report from the USGS about a boat sinking in the harbor of Rabaul in Papua New Guinea as a result of pumice raft inundation.
2) The individual pumice clasts are typically a 5-10 centimeters in diameter. The pumice raft itself is fairly wide upon formation (>50 km) but they rapidly thin and disperse as a result of wind shearing to less than 1 km.
3) Out of all the pumice rafts that I have studied, only the 2012 Havre Seamount raft has drifted back towards its source volcano. I'm sure it has happened in the past, although we are limited in our abilities to observe and track pumice raft movement.
1) A clast is just another term to describe an individual rock fragment. So the pumice raft is made up of thousands of individual pumice clasts.
2) Pumice rafts erode on the sea surface at an extremely low rate, mostly due to wind erosion. The typical lifespan of a pumice raft is around 1.5 years and assuming they drift at a rate between 10 to 20 km/day, they typically make landfall before absorbing enough water to sink.
3) Pumice rafts are mostly detrimental to marine life. Pumice rafts block sunlight and oxygen from penetrating the ocean's surface, both of which are essential for photosynthesis and fish respiration. Additionally, pumice rafts can transport invasive species for thousands of kilometers across the sea surface to delicate coastal ecosystems. On the other hand, it's been observed that fish gather around pumice rafts as they act as a natural Fish Aggregating Device. Some hypotheses as to why they do this is that the raft hides smaller fish from predators, provide a source of food (algae), and they can use the pumice as a hard substrate to lay eggs.
1) What is the next step in the application of this data? Will there be outreach and easy access to coastal nation governments, shipping companies etc. so that they might best take advantage of your findings?
2) Can this modeling methodology be adapted to apply to the trajectory of other, similar substances such as oil spills or red algae colonies?
1) This is the most important step of the project. We wrote a highly detailed user manual for how to use GNOME to predict the trajectory of a pumice raft. It has been written in a manner so non-scientific users from threatened island and coastal nations, merchant mariners, and fishery managers are able to use the model without a problem. Additionally, all of the data and programs necessary to run this model are completely free of charge.
2) Funny you ask, the GNOME model was developed by NOAA in 1999 and was originally intended for modeling the trajectory of oil spills. In 2011, NOAA adapted it for tracking debris from the 2011 Tohoku tsunami in Japan. GNOME is very adaptable to these different types of materials by changing various parameters such as windage, persistence, and pollutant type (weathering or non-weathering).
@HeikkiVesanto Heikki, thanks for the questions! I do just want to clarify, the model is not ours but NOAA's Office of Response and Restoration's model. The model was adapted for pumice by a) changing the pollutant type to non-weathering, as pumice does not undergo the processes that oil would; and, b) setting persistence to infinite, as pumice rafts by definition do not sink and resurface (because of near-neutral buoyancy) like oil does.
Otherwise, the parameters were quite similar.
That is a very fair question to ask. One caveat we offer to any potential end-user is that we used this methodology to hindcast historic events, and it has not been used yet operationally. Thus, the data we were using was observed data, unlike the data (forecast data) end users would be using in real-time. For that reason, and because even we experienced slight error in our model as time progressed, we expect the model would only be skillful in predicting accurately for the immediate term (say, 1-3 days out), and giving a more general outlook on the period of a month or so.
We would point to the successes by NOAA in using this model for oil and other marine debris, as well as publications which indicate the model can accurately predict oil's location within 1 km 48 hours out. Only time will tell of course, as pumice rafts occur so infrequently.
@AlexandraPerillo Thank you! We had trouble classifying pumice rafts in imagery due to the fact that they do not have a unique spectral signature. After doing supervised and unsupervised classifications in ENVI we found that the rafts signature was too similar to that of the edge of clouds and were not able to eliminate the clouds without taking most of the rafts. If we had more time we would have explored further possibilities for identifying the pumice rafts in remote sensing imagery.
Great job guys! Can your project methodologies be utilized to track movement of other marine debris? Could this work be useful for the U.S. Navy as they transition a large part of their fleet to the Pacific in future years?
@LaurenCGGreat question! A methodology for general marine debris would rely on similar datasets as our project, and in fact using GNOME for pumice was actually inspired by its current use for marine debris. Here is some good information on how NOAA used the model for the 2011 Tohoku Japan tsunami. https://ams.confex.com/ams/93Annual/webprogram/Paper219319.html
As for the U.S. Navy, any ship that relies on sea-water intake cooling systems is susceptible to pumice rafts blocking them and causing them to overheat. The model can be used (and has been for tsunami debris) to track pumice rafts either for long term outlooks or for short term warnings.