Archive

Archive for July, 2010

Mid-week update.

July 28th, 2010 1 comment

Figure 1: IOOS Overview. iRobot Seaglider is at the drill site, Navy Seaglider 137 is heading south, away from the drill site and towards the Loop Current Eddy. Navy Seaglider 135 and SIO/WHOI Spray 40 are at the northeast edge of the Loop Current Eddy. USF and Mote Slocum Gliders are on the Florida shelf. Strong currents from the drill site up the Canyon to the northeast are observed in the HF Radar current field (red arrows).  Several drifters are southwest of the drill site, all with complicated tracks at this scale.

Figure 2: Zooming into the drifters just southwest of the drill site, we see the deepwater drifters all have the scallop pattern in their tracks characteristic of inertial waves.  We will probably see several days of strong, slowly decaying, inertial currents spinning in the wake of  Tropical Storm Bonnie.

Figure 3: We check in on the HyCOM model.  Color is sea surface height, red is high, blue is low. HyCOM  also has strong currents heading northwest up the canyon from the drill site. The Spray 40 and Seaglider 135 are in the shear zone between the Loop Current Eddy flow to the southeast and the currents that turn back to the northwest around the local low (blue) in sea surface height.  USF Slocum Sam is heading alongshore to the south through the shelf side of this Loop Current Eddy interaction with the Florida shelf.

Tags:

July 26, 2010

July 26th, 2010 No comments

Storm and Gulf activities have impacted Gulf operation, but  gliders are doing well.  We we have 7 gliders in the water.  Two up in the spill vicinity. 2 offshore in the cyclonic eddy.  3 along the west cosat of Florida.

So lets start with the 2 in the spill vicninity.  There we have the Navy glider 127 and the Robot/APL Seaglider. The NAVO shows a continued picture of a stratified system, a subsurface chlorophyll maximum, enhanced CDOM at depth.  The iRobot/APL glider shows the slow continued decline in the midwater CDOM maxima. Backscatter continues to show dramatic spatial complexity.  The system is stratified, and generally the oxygen values show slight increases at depth.

For the two offshore gliders.  The Navy glider 135 shows declines in the optical backscatter as the glider heads offshore.  It has not reached the eddy as the temperature, salinity, and chloropyll do not exhibit any of the changes such as observed by the Scripps/WHOI glider as it headed into the eddy.  The offwaters remain stratified.

The gliders offshore Florida show that waters have told a consistet story for over several months now.  The three gliders include the USF SAM glider, Mote’s Waldo, and the U Del Ble Hen.  The U Del glider will soo be recovered this week.  The Florida waters are stratified.  Offshore the chlorophyll shows a subsurface maximum.  The optical bcackscatter is enhanced at depth.  On the shelf the bottom waters show enhanced concentrations in particles, chlorophyll, and CDOM.  There is in shallower waters enhanced concentrations of particles, CDOM and chlorophyll.  These features show local variability.

Tags:

Bonnie Dissipates over Louisiana

July 25th, 2010 No comments

Figure 1: Cut and Paste from the NOAA National Hurricane Center early morning update on Bonnie:
THE REMNANT LOW OF BONNIE…CENTERED ABOUT 25 MILES EAST-SOUTHEAST
OF GRAND ISLE LOUISIANA…CONTINUES TO PRODUCE DISORGANIZED SHOWERS AND A FEW THUNDERSTORMS. RE-DEVELOPMENT OF THE SYSTEM IS NOT EXPECTED BEFORE THE LOW REACHES THE SOUTH CENTRAL LOUISIANA COAST EARLY THIS MORNING. THERE IS A LOW CHANCE…NEAR 0 PERCENT…OF THIS SYSTEM BECOMING A TROPICAL CYCLONE DURING THE NEXT 48 HOURS.

Figure 2: Oceanweather’s compilation of ship and offshore platform weather report’s has strong winds towards the coast of Mississippi and Alabama on the back side of Bonnie’s Low.

Figure 3: Wave forecast from Oceanweather has the largest waves on the right side of the Bonnie’s track, just off the coast of Mississippi & Alabama.  Just like the textbook says.

Figure 4: IOOS Overview showing the official NOAA track (in red) for Bonnie and this morning’s location for the Low (the red L) just at the edge of the Louisiana coast. The track passed very close the drill site, within 5 nautical miles by my google earth measurement. The current cloud coverage shows how Bonnie has dissipated. Probably the most striking observation are the strong currents in the HF radar (normally green arrows all red) off the coast of Mississippi and Alabama where we saw the strong winds and waves in Figures 2 & 3. To better compare these currents with the SABGOM forecast later in this report, we switch to the hourly currents in this image.  This is much easier to do when the wind driven currents are strong and are much larger than the tides.  Under normal conditions when the wind driven currents are similar in size to the tides, we almost always have to look at the 25 hour average currents to remove the tides for comparisons to the model.

Figure 5:  First we turn on the HyCOM model to look at the overall view of the Gulf, the Loop Current and the Loop Current Eddy.  Here we plot the sea surface height (high is red, low is blue) and the surface currents (white arrows).    The red colors indicate the areas where the warm surface layer of the upper ocean is thickest.  The ocean’s thick warm layers of the fuel for the tropical storms.  Bonnie’s track is northeast of the thick layer of warm water associated with the Loop Current Eddy.

Figure 6: First we zoom into the Florida shelf and switch to the nested high resolution SABGOM model.   We see a collection of assets, the Spray 40, Seaglider 135, a Horizon Marine drifter and  Navy drifter heading from deep water straight at the Florida shelf.  The USF Slocum has turned south along the outer shelf towards this area.

Figure 7: Moving north to the coast of Louisiana, Mississippi & Alabama, and again switching to the high-resolution SABGOM model to compare the forecast coastal currents with the hourly HF Radar observed currents.  North of the drill site and on the continental shelf, we see the strongest HF radar currents in red.  Both the observed HF Radar currents and the SABGOM forecast currents are heading north across the entire shelf towards the coast of Mississippi and Alabama.  In the deeper water of the canyon to the northeast of the drill site, both the HF Radar and the forecast have the currents heading to the northeast, up the canyon. On the shelf to the east of the drill site, and east of the canyon, we see a clockwise circulating eddy on the outer shelf, with the HF radar seeing the northern half of the clockwise circulation.  Moving nearshore, to the far northeast corner of the HF Radar coverage, we see a current reversal, with currents heading due east, opposite most of the other currents in the field.  An initial thought may be bad data.  But when you look at the forecast model, we see a strong coastal current running north and east along the Florida coast, and the far western edge of this coastal current is just inside the HF Radar coverage.  This is an interesting point, since as an oceanographer, when you get a current observation, you first try to explain it in terms of the forcing to determine if it is believable.  If we only look at the windfield, the first guess would be that the currents cannot possibly be in the direction opposite to the wind.  Perhaps they are wrong.  But then you look at the SABGOM model, and there are more forcing functions than just the winds.  SABGOM says a coastal current has developed in this region with flow to the east, we see flow to the east in the observations, and suddenly we are very happy with our sensors.

I know I’ve said this many times in this blog, but once again, I am amazed by our community’s growing ability to forecast the ocean.  Perhaps it comes from being in the forecasting business for 30 years and seeing the progression.  It may also come from the perspective of the ensemble forecast.  These are common in weather forecasting,  like the many tracks of Tropical Storm Bonnie in yesterday’s forecast, but much less common in ocean forecasting. Often we get one ocean forecast, we compare it with one current meter at a fixed location, we see they don’t agree but don’t know why, and we conclude the models don’t work.  But one of the most important things to forecast in the ocean is where is the water going.  We do this with the U.S. Coast Guard to compare various methods of forecasting the dispersive cloud of an ensemble of drifters.  The drifter cloud dispersion does not depend on a single forecast current value being exactly correct.  It integrates the effects over many current values along a drift path.  It tells you whether the general circulation pattern is correct or not, and if something that is in the water, be it oil or people, are where you expect them to be.

Tags:

Bonnie heads towards Louisiana

July 24th, 2010 No comments

Figure 1: Official track forecast for Tropical Storm Bonnie from the NOAA National Hurricane Center.  Official discussion indicates the track guidance has Bonnie heading down the center of the envelope.

http://www.nhc.noaa.gov/text/refresh/MIATCDAT3+shtml/240847.shtml?

Figure 2: Collection of the offshore wind reports from Oceanweather. The counterclockwise wind pattern around Bonnie’s broad low pressure field is observed.

Figure 3: Highest waves, running about 8-10 ft significant, are on the northeast corner of Bonnie on the Florida shelf.  Just like the textbooks.  The largest waves are on the right side of the hurricane track.

Figure 4: IOOS Overview with the most recent spaghetti plots from this morning’s early cycle (6:00 Z)  guidance.  The red L marks the center of Bonnie at the time of the forecasts. The many tracks heading northwest from the Low are the ensemble of forecast model runs.  The NOAA National Hurricane Center uses this ensemble and others to produce the forecast envelope in figure 1.  The current cloud image shown in this google earth image has Bonnie moving northwest along these tracks.

Figure 5: Looking at the Florida shelf behind Bonnie to see where the storm has been, it has passed over 5 gliders deployed for the spill. Two in deepwater at the edge of the Loop Current Eddy, two in shallow water to the left of the track, and one in shallow water to the right of the track. I can’t remember a Tropical Storm with so many gliders deployed since Ernesto in 2006.  Behind the storm where the waves are still big, the HF Radar coverage (green arrows) is huge.  The 25-hour average nearshore flow is offshore to the southwest.

Figure 6: Now we turn off the clouds so we can look at whats in the water in front of the storm.  By the end of the day, Bonnie will have passed over the two gliders currently operating over the drill site, breaking Ernesto’s record for the number of gliders hit by a single tropical storm.  We’ll soon have a great dataset for future ocean forecast sensitivity studies in the archive.  In front of the storm where the waves are smaller, the HF radar coverage area is reduced.  We’ll see this coverage area increase over the course of the day and tomorrow as the storm passes.

Figure 7: Impressed by the weather models and their ability to forecast Bonnie, and the validation data we are are collecting for the ocean models, how are we doing with the ocean models?  Here we overlay the SABGOM forecast of sea surface height in color and surface currents as the black arrows. On the shallow side,  we see general agreement with the HF Radar behind the storm, where the 25 hour average currents and the storm response in the model are both offshore.  On the deep side, we see that the Spray, a Navy Seaglider and a Horizon Marine drifter are all on the eastern side of  the Loop Current Eddy where it is undergoing a strong interaction with the Florida shelf break, sending water onshore and east. Tropical Storm Bonnie is passing right over the convergence zone where the offshore flowing waters are meeting the onshore flowing waters on the Florida shelf.  Up where the storm is passing now, many small scale features are observed in the model and the HF Radar, they are evolving rapidly, and these comparisons will require a more careful alignment of the observed currents and the modeled currents.  We also have many Horizon Marine and Navy drifters in this region.  All these will help sort out the impact of this tropical storm on the complicated flows off Louisiana.

Tags:

Bonnie approaches Florida

July 23rd, 2010 No comments

Figure 1:  The official NOAA track forecast for Tropical Storm Bonnie.  The track crosses over the west Florida Shelf late friday night into early saturday morning.  By early Sunday morning it is passing the drill site.

Figure 2: Here are the marine weather reports compiled by Oceanweather. Today’s winds are from the east over much of the Gulf.  Something we will see change as Bonnie tracks across.

Figure 3: Here are the clouds from Tropical Storm Bonnie overlaid on the usual IOOS Overview.   The two Slocum gliders on the southern side of the west Florida Shelf, Delaware’s UD134 and Mote’s Waldo will be the first to see Bonnie’s impact later today.  Next in line will be the Navy Seaglider 135 and the SIO/WHOI Spray 40, both of which are located just east of the Loop Current Eddy.  Later in the day on Saturday will be Navy Seaglider 137 and iRobot’s Seaglider 515′s turn as the Tropical Storm crosses the drill site.

Figure 4:  Since the strengthening of tropical storms and hurricanes is related to the temperature and thickness of that warm surface layer of the ocean, here is the HyCOM forecast of the sea surface height.  Bonnie is already north of the thick warm waters (red) of the Loop Current.  Its path will take it alongside the thick warm waters (red) of the Loop Current Eddy.  The official NOAA forecast indicates the environment in the Gulf is not favorable for significant strengthening.

Tags:

Glider data for July 22, 2010

July 22nd, 2010 1 comment

We have seven gliders flying in the Gulf.  All looks well and they are all making good progress in their respective surveys.  We are keeping an eye on the tropical depression which is likely to enter the Gulf on the weekend.The straight yellow line is some artifact and rest assured does not mean one of the gliders has been hijacked.

Looking at the NAVO gliders, we see one glider (sg135) has headed offshore and is swinging east towards Florida, and if lucky we might be able to rendevous with the Spray/WHOI glider.  The data from the offshore glider shows a relatively constant stable pyconcline and a decrease in optical backscatter.  There is a subsurface chlorophyll maximum with CDOM enhanced at depth.  The NAVO glider that is closer to shore, shows similar features.

The iRobot/APL Seaglider continues it high resolution survey near the oil spill site.  The glider shows a continue decrease in the mid-water CDOM.  Subsurface deep optical backscatter continues to show impressive complexity.  Note that the enhanced backscatter is much deeper the mid-water CDOM maximum, but is coincident with a deep water CDOM maxima. There is a noteable exception to this in the last day with a mid-water peak in optical backscatter at ~275m. Lowest oxygen values are not associated with the CDOM and optical backscatter maxima.

Some beautiful images collected by the Spray/WHOI glider.  The glider entered into the eddy and has since then headed east and broken through the eddy wall heading towards Florida.  The depth of the eddy is evident to be around 250m.  The strong zonal current velocities entering and exiting the eddy are easy to see.  The backscatter signals, assuming they are a proxy for critters, show that the biomass concentrations were lower in the deep eddy.

The three gliders on the West coast of Florida show a consistent story thanks to the great efforts of USF, Mote Marine, and U. Delaware.  The system remains stratified with enhanced optical backscatter, chlorophyll fluorescence and CDOM associated with bottom waters.  The features span the entire water column nearshore where  stratification weaker.  There is variability in the subsurface maxima in chlorophyll fluorescence, optical backscatter and CDOM.  Also the U Del Blue Hen identified a beautiful deepening of the thermocline at ~360 km of its transect which had a dramatic impact on the bio-optical properties of the water column.

Tags:

Mid-Week Spatial Update

July 22nd, 2010 No comments

Figure 1: IOOS Overview of the Gulf.  NOAA Slick forecast remains steady.  For the gliders, starting in deepwater, Navy Seaglider 135 is approaching the SIO/WHOI Spray 40. Navy Seaglider 137 and iRobot Seaglider 515 remain near the drill site.  Moving east, the USF Slocum Sam is on the Florida shelf near the north side. The Mote Slocum glider Waldo has joined the UDel Slocum UD134 on the southern shelf of Florida.  Both HF Radar networks are operating.

Figure 2: Zooming into the U-shaped slick, surface currents in the HF Radar network are still heading north from the top of the U to shore.

Figure 3: Lets expand the current coverage to the SABGOM model.  SABGOM sea surface height is shown as color fill, and the surface currents as white arrows.   SABGOM currents are lined up with the western side of the U-shaped slick, all heading northeast.

Figure 4: CSTARS has 3 excellent images of the spill over the last few days.  We overlay all three images here. The darker areas are the slick.

Figure 5: Moving east to the Florida shelf, we see that Slocum UD134 is heading in and is being replaced by Slocum Waldo. Currents over much of the inner to middle shelf are generally offshore, to the west or to the southwest.

Figure 6: Zooming out to the HyCOM model to take a look t the Loop Current Eddy, the Navy Seaglider 135 is riding the strong currents of the northeast out edge as the Spray 40 heads east from the center along the major axis of the elliptical eddy.

Figure 7: And one more image for the east coast.  The Horizon Marine Drifter is on the southern side of a warm feature in the Sea surface temperature and is circulating in a large clockwise pattern in the Sargasso Sea.

Tags:

A quick look for the next glider deployment

July 19th, 2010 No comments

Figure 1: Monday morning shot of the IOOS Overview.  Navy Seaglider 137 is getting close to the iRobot Seaglider 515 near the drill site.  Navy Seaglder 135 is heading south towards the Loop Current Eddy and the SIO/WHOI Spray 40 is heading north away from the center of the Loop Current Eddy.  USF and U.Delaware Slocum gliders are outbound on the Florida Shelf.  Plans are being made for the next Mote Slocum deployment.

Figure 2: Here we zoom into the NOAA slick forecast and see there is no change from yesterday.  Checking CSTARS, there are no satellite passes from yesterday, July 18, over the slick.  The main change we see in this image from yesterday are the evolving current patterns in the HF Radar coverage. The counterclockwise circulation cell in the western side of the coverage is still present, and even more well defined.  The pathway to Dauphin Island is still visible.

Figure 3: And now a quick look at the SABGOM forecast for the Florida shelf in preparation for the Mote glider deployment. The hand drawn red line still shows the location of the mid-shelf jet flowing to the southeast.  In between the USF Slocum on the north and the UDel Slocum on the south, the flow along the outer shelf is to the northwest.  Nearshore currents where the glider will be deployed are weaker.  The HF Radar field and the model both agree that the nearshore currents are south or southwest, heading towards the mid-shelf jet.

Tags:

Glider data for July 18 2010

July 18th, 2010 No comments

The gliders have been surveying throughout the Gulf this weekend. The data streams from the NAVO gliders are back online. Unfortunately our hand shaking with the Scripps/WHOI glider has dropped off, but we will get this rectified at the beginning of the week. The U Del glider will be turned inshore and it begin flying north for recovery maybe at the beginning of next week. Tomorrow, WALDO rejoins the party and a Rutgers glider should join towards the end of the week.

The two NAVO gliders have split apart,  The glider that is heading offshore shows a deepening pycnocline with a general decline in optical backscatter and chlorophyll fluorescence. The nearshore glider shows similar features and in general the optical backscatter data has been showing a general decline over the last week.  The iRobot/APL glider is also showing some changes.  The interesting mid-water CDOM is dissipating and the deepwater CDOM also is showing general declines.  The optical backscatter shows a  decline.

The two gliders offshore Florida (USF SAM & UDel Blue Hen) show a consistent picture which we have seen the last few weeks.  The general features show enhanced CDOM, optical backscatter, and chlorophyll fluorescence at depth.  The UDel shows interesting the features with the chlorophyll fluorescence showing variability not seen in the optical backscater.  The CDOM values appear to be dropping in the deeper waters offshore.

Tags:

2010-2020: The Ocean Forecasting Decade

July 18th, 2010 No comments

Figure 1: IOOS Overview.  I see a few new icons are being phased in.  The Navy profilers now look real – like the vertical yellow cylinders with an antenna on the top. The Navy drifters are starting to look like the red drogues.  The SIO/WHOI Spray 40 glider is heading into the center of Loop Current Eddy Franklin.  it is surrounded by 2 Navy and 1 Horizon Marine drifter.  Heading north, we encounter Navy Seaglider SG135 between the Loop Current Eddy and the slick. In the slick we see two Seagliders, the Navy’s SG137 and iRobot’s SG515. On the Florida shelf we see the Slocum gliders from USF and U.Delaware.

Figure 2: As usual, we’ll start today’s analysis in deepwater, looking at the general configuration of the Loop Current, the Loop Current Eddy, and its interactions with the shelf to the north and east.  Then we’ll zoom into the shelf regions.  Here we look at the Navy global HyCOM sea surface height (color fill) and the surface currents (white arrows). The meander crest in the Loop Current continues to develop in the HyCOM forecast, heading straight north from Cuba and making a sharp right hand turn to the east, flowing along the shelf break and out of the Gulf.  The elliptical Loop Current Eddy (Franklin) has rotated so that its major axis is oriented east-west.  The western side of the major axis has that lingering trail of an interaction with the Loop Current, and the eastern side has the evolving interaction with the Florida shelf. Interpretation of the link between the western side of Franklin and the Loop Current is aided by the Horizon Marine Drifter to the south of Franklin. The Slocums on the Florida shelf will help with the eastern side. In this orientation, the strong clockwise currents of the Loop Current Eddy are much farther south of the oil spill site than they were when the eddy orientation was north-south.

Figure 3: Here we switch to the nested higher-resolution SABGOM forecast from the ROMS model run at North Carolina State. Sea surface height is in color, and the surface currents vectors are white arrows so we can better see the heights. The main deepwater features are similar, and we commented yesterday how the small clockwise spinning eddy in the SABGOM forecast south of Franklin looks to be in even better agreement with the Horizon Marine drifter.  In this image, we also see a band of light blue running north-south on the middle to outer Florida shelf. With the distance between the Loop Current Eddy Franklin and the spill site increasing, this ridge of slightly higher water on the Florida shelf is an important feature we will return to further down in today’s analysis.

Figure 4:But first lets continue moving north and zoom into the slick. First we turn off the SABGOM sea surface height so we can see the slick, but we leave the SABGOM surface current vectors turned on with a color change to black arrows.  First we see that the  NOAA oil slick forecast (light, medium & dark blue) has the same U-shape  as yesterday with a similar area of uncertainty (thin black line).  The base of the U-shaped slick is a bit wider than yesterday, and the tops of the U have been bent over a bit to the east.  As we suspected from yesterday’s SABGOM forecast, the area of uncertainty along the southern coast of Louisiana has extended a bit farther to the west along the coast. Yesterday’s westward flowing coastal jet is much less prominent in today’s SABGOM forecast, so we expect less change in this portion of the area of uncertainty tomorrow.  Looking north of the slick, we have the HF Radar surface currents as the patch of mostly green arrows.   On the western side of the HF Radar coverage, we see a counterclockwise circulation pattern, with a strong jet towards the coast in the region of Dauphin Island. On the eastern side of the coverage we see the currents flowing to the northeast and turning to the east in the alongshore direction.  Where they overlap, the black arrows of the SABGOM surface current forecast are in very good agreement with the HF Radar currents.

So lets use our combined observation and modeling database to look at transport pathways. I have drawn three paths as red lines on the image.  We start with the most western red line that heads from the oil slick in towards Dauphin Island.  Recall from yesterday that the CSTARS satellite imagery clearly identifies the oil slick at the base of this red arrow.  From that location, the SABGOM currents flow north, turn northeast, than turn north in a jet that snakes its way  in towards Dauphin Island. Along that transport pathway, we see a Horizon Marine Drifter, a small piece of the oil slick that has broken off, and the surface currents from the HF Radar array confirming the direction and strength of this shoreward transport pathway.  As an old ocean forecaster, it is amazing for me to see how far our ocean forecast models have come.  In oceanography, we witnessed the growth of Ocean Observatories over the last decade.  This next decade we will see the explosion in the use of ocean forecast models.

Figure 5: So lets take that same SABGOM forecast model which just amazed us with its capabilities in Figure 4, and look for other transport pathways to the southeast. The most prominent feature is the alongshelf transport pathway that runs midshelf along Florida to the southeast.  The tops of the U’shaped oil slick both join this midshelf jet near 86W. With the Loop Current Eddy pulling away from the spill site, this may be a transport pathway of growing importance for Mote’s upcoming glider deployments.

Figure 6: Turning the SABGOM sea surface height back on, we see that this mid shelf jet is associated with the eastern side of the ridge of slightly higher (light blue) water on the middle to outer Florida shelf. The USF and U.Delaware Slocums gliders are in excellent positions to investigate the northern and southern sections of this jet.  Both are heading offshore into this feature. The Mote glider deployment planned for this coming week will be deployed in between the two Slocums, giving us an excellent assimilation and validation dataset for this critical feature.

Figure 7: Here we take Figure 6 with the SABGOM forecast, and add the HyCOM surface currents as the white arrows.  Comparing the black arrows of the SABGOM forecast and the white arrows of HyCOM, we see many areas of agreement. This is important.  A lot of time, effort and money is spent on the glider deployments, and we want as much knowledge of the environment as possible before we deploy.  We can see the midshelf jet is present in both.  We can also look at the outer shelf and see that many of the smaller scale interaction zones between deepwater and the shelf are also in good agreement.  For example, the interaction between the eastern side of the Loop Current Eddy and the Florida shelf is observed in both models.  The details vary, just as with an ensemble of weather forecast models, but the major features are reproduced in both.

Figure 8:  So lets zoom in and produce one more image for the mission planners at Mote. The USF Slocum is heading southwest across the midshelf jet towards a small clockwise eddy that is interacting with the shelf break with significant cross-shelf break flows.  The U.Delaware Slocum is heading northwest across the midshelf jet and towards another clockwise eddy interacting with the shelf break and causing the strong cross-shelf break flows.  The Mote Slocum deployments are being planned for somewhere in the middle, where the flow at midshelf is alongshelf to the southeast in both models and in the HF Radar current fields (green arrows). In this central region, the  flow along the outer shelf also is alongshelf to the northeast.  The spacing of the three gliders is about as good as you can do with three.

Tags: