Inside this Issue

• Global Tsunami Model
• Alaska Weather Modeling
• Geochemical Interfaces
• VRcussion
• Clusters at ARSC
• Intern Summer Research Program
• Visualizing Frost Boils
• Permafrost Visualization

Dr. Vladimir Romanovsky, UAF Geophysical Institute
Dr. Ronald Daanen, UAF Department of Mining and Geological Engineering/ Geophyical Institute
Patrick Webb, ARSC

Story by Lorien Nettleton

To see the effects of warming climate trends on permafrost, much of what researchers must do is watch and wait. Whereas climate changes are visible year-to-year, changes in air temperature generally precede changes in permafrost temperature, and studies show that permafrost temperatures reflect changes in air temperature over periods as long as decades and even centuries. Long-term study is necessary in order to collect information about the depth of the active layer above permafrost, see the patterns of climate change and draw conclusions about their effects on permafrost temperature. The UAF Geophysical Institute’s Permafrost Laboratory began an intensive comparative investigation of permafrost dynamics in different regions of Siberia and Alaska in 1998. Headed by Dr. Vladimir Romanovsky, UAF Associate Professor of Geophysics, the Permafrost Observation, Analysis and Interpretation project has recorded and archived daily temperature data for the active layer, the portion of ground closest to the surface that annually thaws and freezes, and for shallow permafrost. This work is the continuation and further development of a comprehensive network of Permafrost Observations initiated and implemented in the 1970s and 1980s by UAF Professor Emeritus Thomas E. Osterkamp.

Permafrost temperature data is available for discontinuous periods in the mid-1950s and the 1970s, and continuous data is available from 1983 forward. As detailed permafrost temperature data is only available for intermittent periods, in order to see averages or trends, researchers have developed a model to simulate the active layer temperatures based on climate data, which has been continuously recorded over most of the last century.

The GIPL2.0 model simulates the active layer and mean annual ground temperature based on climate data. The model uses a numerical calculation of all the temperature dynamics down to a depth of one kilometer, simulating permafrost and active layer features to create a dataset spanning the years 1900 to 2100. Presently, the model runs at 1/2 degree resolution over the whole state of Alaska, with over 1000 grid points representing the active layer and permafrost. The model includes vegetation and snow cover, as well as air temperature as primary drivers. GIPL2.0 is calibrated by comparing climate data from the National Weather Service (NWS) station closest to the area where non-continuous permafrost data is available for the same time period. Once calibrated, the model can be used to determine past permafrost data based on NWS meteorological records, and can be used to predict permafrost dynamics when future climate change scenarios are used as input.

Visualizing the Patterns

Numerical permafrost data has traditionally been displayed in two-dimensional graphs, a format that restricts such representations to a single point over time, or many points for a single time. In order to represent multiple points over a range of dates, a new format for presenting information was developed by undergraduate interns Patrick Webb and Jordanna Chord, who spent the summer of 2004 working with mentor Glenn Chappell, Assistant Professor of Computer Science at UAF, to develop a three-dimensional visualization of the permafrost cycle across a transect of Alaska. The model utilizes the stereoscopic visualization resources available at ARSC’s Discovery Lab. The product of the internship became FrostByte, a program that graphically represents seasonal permafrost cycles.

FrostByte is visualization software that plots the warming and cooling patterns of the active layer and permafrost along a transect of Alaska that roughly parallels the Alyeska Pipeline stretching approximately 800 miles from Prudhoe Bay to Valdez. The visualization application converts GIPL2.0’s output into image data, allowing researchers to see thawing and freezing patterns through the 200-year dataset. Using the numerical temperature and thaw depth output of GIPL2.0, FrostByte transfers the location data into a visual display on a cutaway three-dimensional map of Alaska.

The cutaway is a wall that transects the state, and temperature differentials are indexed by color—temperatures below 0 degrees Celsius are presented in shades of blue, while temperatures above 0 degrees Celsius are presented in shades of red. A shaded green band indicates areas where the active layer does not re-freeze down to the level of the permanently frozen ground after the thaw cycle. The portions that do not re-freeze are indicative of permafrost degradation.

The FrostByte visualization includes topographical details and dynamic seasonal textures that indicate seasonal changes such as vegetation growth and decline, as well as the snow cover that plays a role in the freeze-up of the active layer by insulating the ground from colder air temperatures.

One beneficial aspect of modeling soil thawing and freezing patterns is that it allows researchers like Romanovsky and Ron Daanen, Post- Doctoral Fellow at the Geophysical Institute’s Permafrost Laboratory, to look at persistent trends and identify anomalous details. As a tool for evaluating the GIPL 2.0 output, FrostByte shows how in some cases the modeled data just doesn’t look right.

“We immediately saw strange behaviors in some places,” says Romanovsky. “On the north slope of the Brooks Range, the snow jumped up at the start of winter and tapered down until spring. These are the kinds of things you can see by looking at the data if you are practiced at it. But this visualization can be useful for sharing those observations with other people who don’t know what to look for. It is also a useful tool for scientists who don’t have a background in permafrost to become familiar with the issues.”

Recent Developments

The concept and appearance of FrostByte remain largely the same from its creation by Webb and Chord. Over the past year, Webb has steadily worked to implement a number of tweaks and adjustments to the dataset, including modifications that streamline the program and how it is accessed.

The initial program was based on a seven-point grid, covering 400 miles from Prudhoe Bay to north of the Alaska Range. The current incarnation is fed by a 21-point dataset, running the length of the state. Webb added complete visual seasonal changes and relief details of the landscape topography, as well as dynamic textures to the map of Alaska which provide visual cues to the time of year, with environmental representations such as snow cover and the greening and browning of spring and autumn. The new incarnation also features a series of contour lines that allows the user to scale perspective of the active and deeper permafrost layers in order to emphasize or de-emphasize the cyclical changes. Now a user can scale the depth from 10 to 90 meters, effectively condensing the deeper permafrost into the lower 20 percent of the display, making the upper five-meter layer the dominant feature of the virtual representation. This scaling feature can be performed on the fly, allowing the virtual driver to expand and reduce the representation of the most dynamic zone in relation to the other permafrost and topsoil features as the program runs.

Webb improved the way the datasets are preprocessed to run in one uniform format. Due to the number of researchers contributing datasets for visualization, many files were provided in differing formats – and rarely in the ASCII or binary formats which FrostByte reads. To resolve the time-intensive challenges of manually converting the datasets to a binary form, Webb created a conversion tool that automates the process, freeing the programmer to focus on other aspects of the program.

Perhaps the most exciting new feature of FrostByte, called directed visualization, allows scientists to manipulate input data as the program is imaging it. This allows them to evaluate the variables visually and change them dynamically. Parameters such as fire, snow and heavy rains can be adjusted, with instant change in the imaging display, giving feedback about climate conditions. For example, the researchers could be viewing the representation and decide to see what it would look like with air temperatures that are 10 degrees higher. They could input the command and watch as the model adjusts its display to reflect that change.

For Webb, the most rewarding part of being the primary developer is dedicating the time to develop the program beyond its original edition.

“In some ways it’s completely different from the way it started out,” Webb says. “The pieces are more elegant and smooth, they run more efficiently together.”

For the last year, Webb was the sole developer of the program and refined a number of its features. Webb, now a graduate student of computer science at UAF, is leaving the development of FrostByte in the hands of a new round of students while he moves on to new challenges.