Friday, October 1, 2010

Anatomy of TTUHRT Landfall Operations

Well as we head into October the tropics have quieted down a bit. TS Matthew dissipated over central America while short-lived TS Nicole and its remnants are helping to produce prolific rainfall totals across the east coast of the US.

I thought I would take a blog and discuss just how a TTUHRT deployment for a landfalling hurricane actually works. Since Lubbock is a long way from the coast, logistical concerns are pretty significant. How do we get 24 StickNets and 2 mobile Doppler radars to the coastline, amidst evacuation traffic, get probes deployed and collecting data, find suitable radar sites that are unobstructed, and then retrieve all of our instruments.

We will start first with just deciding if or when to depart... We monitor the tropics pretty closely especially during times when active cyclones are present and even provide forecast discussion emails within the university, to our private funding sources, and our government and university partners. These provide information regarding the current large-scale weather pattern that may influence the hurricane, track guidance from computer models, and the forecast of what the hurricane and the environment may look like in several days. Over the past several years we have developed a loose set of criteria to help us determine if we should conduct landfall operations. These are based in part on science objectives as well as our current level of funding and logistical concerns. An example would be, a weak Saffir-Simpson category 1 hurricane along the east coast of Florida. In this situation, a trip to this area of Florida is quite expensive given the cost of fuel, hotel. and per diem expenses for the crews. For the sake of this example, we will say we have no specific science objectives other than monitoring and documenting the hurricane wind field. Well for this situation, we already have a large database from lower end hurricanes as they are the most common. We would likely decide not to deploy for this storm. However, if there was a specific science objective, we would deploy. An example would be using the Ka-band radar systems to examine the interface of the eye/eyewall. For a very strong hurricane, this experiment would likely be too dangerous to employ, but for a weaker storm that we do not have wind load concerns on the radars, we could execute this objective and obtain a quality dataset. Also funding, as is the case with most scientific field studies, controls how many deployments we can make. Unfortunately, we do not have an unlimited budget and must balance the cost/scientific reward for each deployment.

Now lets assume we have an established Saffir-Simpson category 2 (peak 1-minute average winds of 96-110 mph) "Hurricane Ian" entering the southern Gulf of Mexico. The forecast environmental conditions are expected to be favorable for strengthening with low wind shear and the storm is likely to pass over a swath of very deep and warm water associated with the loop current. All signs point to an intensifying hurricane with the forecast tracks in good agreement on a landfall along the northern Gulf coast, like those shown to the left from Hurricane Gustav (2008). In our hypothetical situation, landfall of our system would be near 180 hours or 5 days out. It takes us approximately 12-15 hours to reach most of the northern Gulf coast. For StickNet operations we usually like to use a whole day to scout deployment sites along the coast. This also gives us a chance to investigate how close we can get probes to the immediate shoreline. We also require approximately 12 hours to complete a full StickNet deployment. For safety concerns, we try to complete the deployment prior to the onset of 50 mph wind speeds. In these sustained wind conditions, damage typically begins to occur especially to tree limbs. This allows us to begin our deployments approximately in teh 24-36 hour window prior to landfall. The timeline requires a departure from Lubbock often several days in advance of the hurricane. Once again for safety, we try very hard to avoid having to drive through the night to reach our target. However, in some instances such as Hurricane Charley (2004) which intensified very quickly, it requires driving straight from Lubbock to the target area. For Hurricane Charley, this meant driving 23 hours straight to Sarasota, FL.

On to the actual deployment... We have our target area defined for "Hurricane Ian" of the upper-Texas coast and we feel that this will meet deployment criteria, so it is time to hit the road. The StickNet crews are broken into two teams of two people. Each team is responsible for deploying 12 probes which are housed in a covered trailer. We use Dodge 4x4 diesel trucks to pull the trailers. The trucks are outfitted with cell Internet routers to allow us to monitor the storm while in transit and to use Google Earth aerial imagery to locate possible deployment sites. For each of the two Ka-band Doppler radars, we also have two people manning each. The radars are also equipped with mobile Internet. As we arrive in the area, navigating through evacuation traffic can be difficult. We often try to use roadways not specifically assigned as evacuation routes. Assuming there is time, both the StickNet and Ka radar teams will scout for potential deployment sites. For StickNets, we are looking for open exposure areas which are free from obstacles in most directions. The more obstacles means the larger frictional force to slow the wind speeds and afterall we are trying to measure the strongest winds. In the past with the larger WEMITE towers, airports have provided good deployment sites. With the versatility of StickNet, these sites have been used less. If we do use an airport or park site, the field coordinator (which has been me since 2005, although Rich Krupar will take over in 2011) will obtain the necessary permissions to allow for the probe to be deployed. The Ka-radar teams are also searching for sites which are unobstructed, as buildings and trees can block the radar beam. After a day of scouting and marking deployment sites the teams try to get a good night's rest before the action gets going the following day.
Showtime...Now we are getting to the business of making a full deployment as "Hurricane Ian" is now only 36 hours from landfall along the upper-Texas coast. Usually by this time the spread in computer model tracks has become much smaller and we can really get down to figuring out exactly where the hurricane is going to come ashore. We also evaluate the structure of the wind field as often we are targeting the area of strongest winds. Most of us know that the strongest winds in a hurricane are within the eyewall in the right-front quadrant of the storm. This is the location where the wind is aligned with the direction of motion and the flow of air is on-shore. Water surfaces are much smoother than land, thus the wind speeds are not reduced as much due to surface friction, unlike in the off-shore side where air traverses across the land. We often use a tool called H*Wind to evaluate the general wind structure of the hurricane. H*Wind was developed by Dr. Mark Powell at NOAA's Hurricane Research Division. It is a wind field model that uses observations from all types of instruments (aircraft, dropsondes, satellite wind estimates, ASOS stations, Ship reports etc... etc...) to synthesize a near real-time picture of the hurricane's wind field. An example analysis is shown on the image to the left from Hurricane Ike (2008). This allows us to target the region where we expect the maximum winds to come ashore. In the case of Hurricane Ike, we decided to concentrate most of our probes within the wind maximum in an attempt to make fine-scale measurement across this region (shown in the image to the right, H*wind maximum wind swath and TTUHRT probe locations). This allowed us to evaluate the local variability in the wind field as probes were spaced approximately 8 miles apart. As we have found, each deployment has differences and never goes exactly according to plan. We must account for storm surge issues and road networks. A good example is from Hurricane Gustav where the Atchafalaya swamp kept us from having a continuous array of probes across the landfall region.

As for the Ka-band radar deployments we have yet to have an opportunity to deploy both systems. Ka-1 made its first venture into a tropical cyclone during Tropical Storm Ida last year. The primary deployment strategy we would like to use is a dual-Doppler one. The use of two radars allows for the complete 3-dimensional wind field to be derived through some higher order math. One Doppler system can only sense 1 component of the wind (toward or away from the radar). With the proper orientation, 2 radars can get the total wind speed and direction fields. Given the wavelength of our radars, this requires the two radars to be located fairly close to each other (maybe only a couple miles apart at the most). Possibly even at the opposite ends of the same airfield. Another strategy involves single Doppler sampling of the eye/eyewall interface to examine mesovorticies within the eyewall. I'll let Pat Skinner and Scott Gunter go into some more detail regarding radar sampling strategies in some later blogs. In case you were wondering, our radar systems were designed to withstand sustained windspeeds near 120 mph. This is the main reason we have a complete radome over the antenna in order to reduce the wind loads on the antenna itself. This allows the system to be deployed in higher windspeed environments. All contingent on the lack of flying debris which is one of the most dangerous aspects of a landfalling hurricane.

After the deployment we find a safe hotel, typically not in the direct path of the eyewall. However, we do try to remain close to the array as it makes retrieval much easier. Often law-enforcement closes the impacted area to outside traffic, so remaining close by allows us to avoid this issue. This often means staying in a hotel with no power. We try to stock up on snack food and water prior to landfall and fill up the bathtub with water as anyone should do who is planning on riding out a hurricane. This sometimes means a diet of poweraid and potato chips for a couple days. After all the StickNet probes have been deployed, the teams retreat to a hotel to ride out the storm. In the case of the radar crews, their job is just beginning. A radar operator is required to be in each radar through the event to ensure the system is operating correctly with no problems. The ideal situation allows the StickNet crews to be nearby to shuttle radar personnel back and forth between the radar and the team's hotel. Unfortunately if conditions become too hazardous for relief crews to be shuttled in, the radar operators may be required to work for significant lengths of time. Once the storm has passed or data collection has been called off, the radar teams can un-deploy and depart the area. The StickNet teams however must retrieve their respective probes. This can go smoothly as in the case of Hurricane Gustav or can be quite difficult such as following Hurricanes Dolly and Ike. Downed power lines can cutoff areas as well as storm surge and freshwater flooding. In fact, Probe 110A stayed at Ft. Travis on the Bolivar peninsula for 3 weeks after Hurricane Ike (image to the left), as the area was only accessible by boat. It was eventually retrieved by personnel from the Galveston County Emergency Operations Center, and we are indebted to them for their help.

Data collection and dissemination...Following the retrieval of all of the StickNet probes, each data acquisition box is plugged into an Ethernet hub in each trailer. Data is downloaded and summary statistics are computed through a computer program. This allows TTUHRT to disseminate data summaries to pertinent agencies as well as our private funding institutions. We typically are able to distribute deployment information which includes deployment maps, site summaries, and data statistics within 12 hours of retrieval. The data collected from the Doppler radar systems takes much longer to analyze and initial plots will likely take several days following the team's return to Lubbock and Texas Tech. Once the team returns to Lubbock, products are created for display via our website. These include the summary statistics, deployment maps, and data plots for each probe.

As for the science objectives, the collected data eventually makes its way into M.S. Theses and Ph.D. dissertations, scientific journal articles, and technical reports. This process takes a significant amount of time and often results take years to produce. In later blogs, TTUHRT scientists will talk a little about their individual research topics.

No comments:

Post a Comment