Report on the Morning Session
The morning session included two papers of general interest to snow studies and one of direct interest to the avalanche community. Dr. Richard Armstrong spoke about the World Data Center at the University of Colorado which is part of the international system of repositories for snow and ice data sets, maps and other information sets. Data is both saved and made available for general use, demand for which is increasing because of widespread interest in global climate. As a result, the scale on which such data is taken and archived has changed from the point scale of snow surveyors to the regional or global scale which allows, for example, the resolution of continental features. The availability of such a data base allows the examination of features of the earth's climate heretofore unavailable, such as the relationship between mean temperatures over a continent and its aerial snow cover. It also affects current trends in avalanche forecasting, since current thinking in that field has shifted from physically-based to data driven models like the nearest-neighbor models which aid forecasting for current conditions by providing information about similar scenarios in the record.
The scale problem is evident when trying to gather information on the mesoscale while using a technology which relies on knowledge of processes on the scale of snow grains. To use remote sensing to gather information about snow quickly and efficiently, knowledge of the processes and properties of the snow cover being evaluated are usually required. However, there are various conditions under which useful information can be gathered as Dr. Jeff Dozier discussed. Microwave remote sensing is especially difficult in mountainous regions because of its large footprint and, because of the sensors that are available, visible and near-infrared radiation is more readily measured. There are problems with these measurements and both the physics and the instrumentation are very different in different parts of the spectrum. However, not only can information be obtained about important snow properties like grain size, but the energy input can be determined and this controls the development of many snow features. An example of how remote sensing can provide otherwise unobtainable information on a large scale is the use of three wavelengths which allow location of snow and clouds, the differentiation between snow and clouds, and finally, the grain size of just the snow. In each case a wavelength is chosen to obtain specific information.
It is still difficult to transfer information from flat to mountainous regions but much more sophisticated instruments are being flown at this time with the promise of impressive developments to come. The development and use of these instruments must take place with increased knowledge of the physical properties since snow parameters like grain size are very important when the e-folding distance is similar to the grain size. Stereology, the science of interpreting three-dimensional materials from their two-dimensional sections, can provide objective, transferable analysis of various types of snow. Parameter such as the ratio of grain volume to surface area are useful measures of grain size since this ratio seems to simulate the equivalent sphere for optical purposes and other definitions are equally useful for other purposes.
Increased use of the remote sensing suggests the availability of very large quantities of data and that will require increased use of computers and development of computer skills among a wide variety of snow scientists and practitioners. Even the amount of information which is now available to an avalanche forecaster can be overwhelming as described by Dr. Sue Ferguson. While 95% of avalanches can be forecasted, the remaining 5% plague the forecasters. These often involve multiple shear weaknesses which greatly complicate the forecasters problem. Dr. Ferguson felt that process models are still valuable because they can be updated as the season progresses and used to extrapolate over a wide area. To improve these models there are many specific research projects which need to be accomplished. Unfortunately, in the past many of the projects have produced results from a limited range of conditions so it is difficult to extrapolate these results for a particular forecast. Information is needed about the failure of weak layers of faceted crystals or depth hoar, about pinning points on slopes, about the general variability of snow on a slope and about the variability of snow and avalanche types. Questions were raised about whether explosions help stabilize a slope, a problem whose answer probably lies in the size of the explosive and particular conditions on the slope.
It is difficult and expensive to close large sections of mountains and thus it is necessary to advance avalanche forecasting in at least three areas. First, Better data collection, transmission and analysis are needed for both research and forecasting purposes. Second, more research into the properties of snow on a slope and the conditions which lead to instability. Third, it is necessary to educate people to respond properly to the dangers involved on mountainous slops, an endeavor in which physical scientists have not been traditionally involved.
While the morning session did not provide the immediate answers to some very important and old questions about snow and avalanches, it did provide some insight into what are the best questions to ask and what are the best ways to address those questions. As Dr. Ferguson said, the science has not kept pace with the needs of the forecasters, so it is important to focus on the research or changes in practice which are most likely to lead to useful results. This session was useful because it provided some background on several important aspects of the current state of the art. Issues of data collection, manipulation and storage were addressed. The needed research into the physical properties of snow and snow-covered slopes were identified. Methods developed for large-scale remote sensing were discussed and some of these methods may be applicable to both large-scale and small-scale problems in the future.