Application of Classification and Regression Trees:
Selection of Avalanche Activity Indices at Mammoth Mountain

Robert E. Davis*

Kelly Elder**

ABSTRACT

We focus on the the importance of the parameters describing the degree of avalanche activity. Classification and regression trees were trained on weather, snow plot and avalanche occurrence observations from the Sierra Nevada, California, a maritime influenced region. Avalanche activity was characterized by the total number of releases, the sum of the sizes and the maximum size class. The number of cases for the data set was 482. We show that the accuracy of overall classification depended on which activity parameter was selected, while the ranking of the critical input variables remained identical. The probability of correct classification was the highest for the maximum size class, followed by the sum of the sizes, and the total number of releases.

INTRODUCTION

Works in the literature report that the importance of various meteorological and snow plot variables associated with avalanche hazard changes with geographic region. This observation appears consistent despite the use of various parameters to describe the level of avalanche hazard or the degree of avalanche activity. Thus its seem plausible that for a given region, a list of critical variables can also state the order of importance to avalanche hazard or activity. Such a list should, ideally, be constructed from variables dictated by experience and ranked by quantitative analyses. Once a list can be made, a further question arises about the suitability of using it for distinguishing among different intensities of avalanche activity.

Describing avalanche activity, the dependent variable, by a meaningful metric that is physically justified and statistically unique represents one of the fundamental problems of statistical or deterministic studies. Many attempts throughout the world have partitioned avalanche response into a variety of genetic and morphologic classifications. Approaches using different dependent variables run the risk of confounding comparison of forecasting methods. Definitions of avalanche activity or response range from individual path observations and descriptions (Judson and King, 198:S) to hazard levels based on frequency of events (Elder and Armstrong, 1986) to binary outcome of avalanche-day versus non-avalanche-day, (Bois et. al., 1974) to the sum of avalanche sizes on a given day (McClung and Tweedy, 1993).

Correct identification and quantification of independent variables leading to avalanche release potentially present a more difficult problem. Unfortunately, data availability often represents the most severe constraint and scientists are forced to make do with data that has already been collected. While these data are necessary because long-term databases are critical to all non deterministic forecasting techniques, they have usually been collected for another purpose (weather forecasting for cities, agriculture, etc.). Collection sites are often located in valley bottoms, urban areas, and at low elevations, which make it difficult to extrapolate to conditions in avalanche starting zones.

BACKGROUND

There is general agreement that it will be sometime, if ever, before quantitative methods can substitute for the human element in avalanche forecasting. However, forecasters seem increasingly dependent on quantitative analysis, from the daily weather predictions to more sophisticated techniques relating to slab release. Great improvements have been made in physically-based deterministic modeling, but we believe it is likely that future avalanche forecasting aids will arise from hybridization between empirical and physically based modeling because of the intensive data requirements of physical models. Two long-standing challenges have been 1) to identify and rank variables that can be measured-efficiently and reliably, according to their importance as indicators of avalanche activity and 2) to characterize the avalanche activity by some useful parameter. Much effort has gone toward assessing the former, while little attention has been paid to the latter.

Perla ( 197()) revisited Atwater's ( 1954) ten contributory factors for avalanche hazard evaluation and found precipitation and wind direction to be the most important parameters. Fohn et al., ( 1977) compared conventional forecasting techniques with four statistical methods ranging from principal components analysis (PCA) and discriminate analysis of local and regional data to cluster analysis of local data. They found that all the methods produce about the same results at 70 to 80 percent accuracy, with some slightly better than others. Each method had distinct advantages and disadvantages.

The nearest-neighbor method has been applied in a number of climates with a variety of input variables. Buser ( 1989) gave results from a nearest-neighbor forecasting program introduced by Buser (1983) and used operationally in Switzerland by the ski patrol in the Parsenn area. The program identified the ten days in the record with the most similar conditions to the day in question. Similarity is based on the proximity of weighted meteorological and snowpack variables in data space. The program also creates "elaborated" variables, for example the time trend in a particular meteorological measurement.

Buser et al. (1985) reviewed a broad range of avalanche forecasting methods for short and long time scales and over local and regional spatial scales. Input data collected by conventional field methods and by instruments designed and built for specialized tasks, such as FMCW radar, were discussed for different applications. Forecasting methods from conventional induction to complex statistical models were reviewed.

Although not directly addressing forecasting, Jaccard (199()) used fuzzy factorial analysis to identify important interactions of avalanches related to snowpack, meteorology, terrain and vegetation parameters based on expert opinion. Slope angle and aspect, overall weather conditions and precipitation were found to be the most important factors related to avalanches. Avalanche hazard forecasting has been addressed from a number of different angles and approaches from nearly all of the affected regions of the world. Tables I and II summarize some of the key research on the subject. The lists in Tables I and II are not exhaustive and represent only a portion of the research published in the English language.

Table 1. Key articles in literature on quantitative analysis methodologies.



TREE-BASED MODELS

Decision trees are an efficient form for representing decision processes for classifying patterns in data or piecewise constant functions in nonlinear regression. A tree functions in an hierarchical arrangement; data flowing "down" a tree encounters one decision at a time until a terminal node is reached. A particular variable enters the calculation only when it is required at a particular decision node, and only one variable is used at each decision node. Moreover, a variable that is unimportant except for a special case can be used in that special case without impacting" the decisions in other cases. These characteristics contrast with multivariate analyses and clustering, which use all critical variables for each case. In a study comparing the performance of neural networks and classification  trees, Atlas et al. ( 1990) showed that in most cases the two methodologies exhibited comparable accuracy for prediction. In a related study, Sethi (199()) showed that decision trees can be trained on a data set then used to construct artificial neural networks by mapping the hierarchical layers in the tree to layers in the net.

There are two types of simple binary decision trees; regression and classification. Regression trees are appropriate where the dependent variable is a ratio scale data type. In other words, if the dependent variable can assume any value over the range of observations, and if the differences are quantitative and consistent, then we want a model that can predict these values and one that is not constrained to particular members. An example is number of avalanches per day. A regression model will predict somewhere between zero and u reasonable maximum number of avalanches for a given day based on the independent variables.

A classification tree is appropriate where the independent variable itself belongs to the data types nominal (named) or ordinal (ordered). Nominal data includes such variables as slope aspect: east, west, etc. Ordinal data exhibits relative, rather than quantitative differences: for example, magnitude 1 through 5 avalanche events. Avalanche magnitudes, like earthquake magnitudes, are expressed on a log scale of magnitude. The difference is that earthquake magnitudes are objectively measured, while avalanche magnitudes are estimated by an observer. Thus a magnitude 4 event is larger than a 2, but not necessarily 10² as large. A regression tree would not make sense in this case because it would predict unsuitable results such as a magnitude 2.76 or 4.89 event.

Table II. Advantages and disadvantages of using selected quantitative methods.

The type of model chosen, regression or classification. depends in part on the dependent variable type. You cannot apply a regression tree model to classification data. However, you can apply a classification tree model to ratio scale data by generalizing the data into classes. Days with any avalanche activity could be called "avalanche days" and days without activity called "non avalanche days" (as has been done in many previous studies). Then a classification tree model could be used on number of events observed, where the observations have been re-expressed into nominal data, avalanche versus non-avalanche days.

Advantages of tree-based regression and classification models over alternative methods (such as those listed in Table 1) include:

* The models are not affected by monotone re-expressions of data, so results are independent of data form and magnitude. In other words, linear mathematical operations such as addition or multiplication can be performed on the data without affecting the results. This characteristic also means that one variable may be expressed in millimeters and another in kilometers without affecting the model.

* Gaussian assumptions are not violated by the distribution of one or more independent variables, (tree-based methods are nonparametric or "distribution free"). Trees are valid even using mixed data sets containing multiple distributions. It is not necessary that data be normally distributed or that non-normal data be transformed before analysis.

* Model results are less dependent on missing values in the independent variables (methodology finds "surrogate" values for each decision node). Many statistical models cannot use data sets where one or more attributes for a given observation are missing. Binary trees can use the existing data to statistically predict what the missing elements should be, or to use only the elements that do exist.

* Tree-based models allow complex interactions between the independent variables, which must be specified a priori in standard linear models. For example, snow accumulation may increase up to a critical elevation, then decrease with increasing elevation above that critical point. Standard linear models can only take advantage of that t:act if a mathematical expression for the relationship is formulated and expressed before model implementation.

*Interpretations of complex interactions are clear and often more easily understood than other model constructions. A tree is t:ar more easily interpreted by most people than mathematical expressions or nonlinear equations.

Binary decision trees or predictive tree classifiers of the type used in this study take a vector of measurements x, (x_m,m = 1,2,...) of variables from the measurement space X of a result y and calculate the probabilities (P₁,P₂ ) that y is in each of the possible classes. The tree is constructed by repeated partitioning of subsets of X into two descendent subsets or nodes, where X itself is the root node and the partitions end in a set of terminal nodes. The terminal nodes are assigned a value based on the probabilities that they belong to a given class y. The partition or split at each node is made on the values in y conditionally on values in the sample vector x, based on a single variable in x. For ordinal or ratio scale data, splitting decisions are posed in the form: is x_m < c? where c is within the domain of x_m . For categorical variables, the decisions may be expressed as: is x_m E 5 ?, where S includes all possible combinations of subsets of the categories defined in x_m.

In the present study these decisions take the form: is new snow depth < l0 inches or is the snow surface temperature <-4.0 degrees C ? The categorical analog would be similar to: does the azimuth of the starting zone of path x_m belong to the subset north ? A portion of the finished binary classification tree may look like the following:

if (SST_i < 65degrees C) and (MAXWS_i < 21 .5 mph )

then avalanche activity AA_i is most likely to be in final decision class AA₂,
and a final decision set for a node in a binary regression tree may look like the following:

if (SST_i < 6.5 degrees C) and (MAXWS_i < 21.5 mph )

then avalanche activity AA_i is most likely to produce 22 releases under current conditions,

where SST, MAXWS, AZ are the independent variables of snow surface temperature, maximum wind speed, and slope azimuth, respectively; i is the co-registered datum of the variables.

A collection of such decision rules is obtained through a technique referred to as recursivepartitioning. Three elements must be defined before the sample data may be recursively partitioned into a binary decision tree:

1) method for determining the best split at each node,
2) basis for deciding when to continue or stop splitting a node,
3) method for assigning class probabilities for each terminal node.

The details of these decisions are beyond the scope of this paper but are explained at length in the standard reference on classification and regression trees (Breiman et al., 1984).

We have used both the tree-based model implementation in CART (Breiman et al., 1984) and in the S-PLUS mathematical language, which follows closely the development in Breiman et al. (1984). Both software packages have unique advantages and the user should explore both implementations. Details of the S-PLUS software are explained in Chambers and Hastie (1992). Two applications of tree based models in the natural sciences can be found in Michaelsen et al. (1987 and in press). The output of the software packages includes a ranking of the independent variables in order of importance as primary decision makers, or as surrogates for other independent variables, as well as the decision tree.  This is the focus of our discussion.

METHODOLOGY

In this study, decision tree methods were applied to observations from Mammoth Mountain, California. Mammoth Mountain ranges in elevation from 2,59() to 3,371 m in the eastern Sierra Nevada and is the major site of the Mammoth/June Ski Resort, who provided the avalanche and weather observations. Data from Mammoth Mountah1 spanned the winters 1989- 1990, 1990-1991 and 1992-1993, which consisted of snow plot measurements, weather variables and avalanche release observations, 482 individual cases with no days with missing data. The weather and snow data were collected from a snow study plot by the Main Lodge at Mammoth Mountain, at an elevation of 2,743 m on the northern base of the mountain. Avalanche observations were from the entire in-bounds ski area and the vast majority of avalanches resulted from artificial release by explosives. Variables used for this analysis were those recorded at the study plot or nearby, as listed in Table III.

Table III. Input data used in CART analysis from daily data record at Mammoth Mountain.
__________________________
1) Total snow depth
2) Storm total snow depth
3) New snow depth
4) Snow water equivalent - storm
5) Fractional density, new snow
6) Average wind speed
7) Maximum 24-h air temperature
8) Minimum 24-h air temperature
9) Current air temperature
10) Snow surface temperature

Control activities and avalanche observations were recorded at Mammoth Mountain in a format consistent with the standard U.S. Forest Service avalanche control and occurrence chart. This protocol consists of codes for the date, time, path, patroller identification, control type, control number, control surface, avalanche class type (hard slab, soft slab, etc.), avalanche trigger mechanism, avalanche size, and so forth (Perla and Martinelli, 1978). It should be noted that the avalanche size class is somewhat subjective when comparing the data from different areas, but consistent within this study area.

Avalanche observations were aggregated h~to three response variables, the total number of avalanche releases on a given day, and the maximum size class. Our premise for specifying these avalanche activity characteristics was that the number of releases may provide an indication of how widespread the avalanche hazard (i.e. spatial dispersion), the sum of the sizes may indicate the overall intensity of the activity, and that the maximum size may provide an index of the local h~tensity of the hazard. Therefore, a regression tree method was used to evaluate the data with the total number of releases and the sum of the sizes as the response variable; and a classification tree method was used to evaluate the data with the maximum size class on a given day as the response variable.

RESULTS

Both the regression tree and the classification tree analyses produced the same ranking of weather and snow plot variables, also the same as Davis et al., 1992. This shows the robustness of the method.

The overall probability of a case falling into the correct terminal node for the regression trees depended on the avalanche activity variable; the total number of releases (range 0 - 41 ) was 0.68, and the sum of the sizes (range 0 - 69) was 0.71. The overall probability of correct classification for the classification tree (maximum size class with a range 0 - 5) was 0.90. The classification matrices showed some details in how various values of result were predicted. In Table V the entire classification matrix is shown for the outcome of the maximum size class.

Table V. Classification matrix: classification tree constructed using maximum size for avalanche response variable.


The results in Table V show the overall classification accuracy of 0.90 - the sum of the diagonal of correct classifications divided by the total number of cases. The accuracy of classification of a non-avalanche day was 0.98. while the accuracy of classification of individual avalanche sizes was substantially lower or undefined. The accuracy of individual values of the total number of releases and the sum of the sizes was also lower than for the individual size classes.

DISCUSSION

The goal of this work was to explore avalanche response variables other than avalanche day or non-avalanche day. We were able to show for the data in this study that the probability of success in obtaining reasonable results seems highest for the maximum size class. Overall it can be said that the rankinig; of input variables in terms of sensitivity to avalanche activity appears insensitive to the avalanche response variable for the example here and in Davis et al. ( 1992.)

Decision tree analysis may not be able to accurately predict details of avalanche activity in terms of numbers or size of releases with only inputs of observations from the current day. This is clearly the experience of the Swiss, reported in many classic works. Much more effort is needed to condition the data sets and specify elaborated variables (e.g. Buser, 1989). Other t:actors also may come into play because we are dealing primarily with artificially released avalanches.

* Cases in these data involve avalanche paths that are repeatedly shot during an avalanche cycle. Therefore the probability of deep slab release is likely to decrease over time.

* There were situations where conditions were ripe for release, but control operations were delayed until the weather improved. This may explain the cases where the prediction was for no avalanche, but releases were observed (top row in Table V).

In order to test this technique effectively and objectively, we need to study other data sets from areas with longer records, which will allow model construction and validation either through unique elements or cross validation. We would also like to test the method in different snow climates to assess model performance and objectively confirm the existence of different snow climates and avalanche response. Both studies are in progress at this time. However, it will be tricky to compare avalanche records where releases from one area are natural or skier triggered, and releases in another area are explosively triggered.

SUMMARY

Various statistical techniques have been tried to rank critical variables in terms of their sensitivity to avalanche activity. Classification and regression tree methodology shows promise because of distinct advantages over standard statistical techniques. While the ranking is robust with respect to the choice of dependent variables describing avalanche activity, the accuracy of classification and regression trees shows sensitivity to the choice of the dependent variables. In the example used for discussion in this work, the methodology was better at correctly classifying the maximum size of releases than it was at predicting the sum of the sizes or total number of releases.

ACKNOWLEDGEMENTS

This work was funded under Army Corps of Engineers Project 4A762784AT42 with other generous support from Mammoth/June Ski Resort and the University of California. Mammoth/June Ski Resort staff, including Eddy Bouzglou, Mary Waller, Stacey Tracy and Gary Reitman provided much help. The idea to use CART to select critical variables came after detailed discussions with Othma Buser of the Swiss Federal Institute of Snow and Avalanche Research.

_______________________________________________________________________________________________
* U.S. Army Cold Regions Research and Engineering Laboratory, 72 Lyme Road, Hanover, New Hampshire 03755- 1290

** Center for Remote Sensing and Environmental Optics, University of California, Santa Barbara CA 93106

REFERENCES

Atlas, L., R. Cole, Y. Muthusamy, A. Lippman, J. Connor, D. Park, M. El-Sharkawi, and R. J. Marks II, 1990, "A performance comparison of trained multilayer perceptrons and trained classifications trees," Proceedings of the IEEE, Vol. 78(10), pp. 1614- 1618.

Atwater, M.M., 1954, "Snow avalanches," Scientific American, vol. 190(1), pp. 26-31.

Bois, P., C. Obled and W. Good, 1974, "Multivariate data analysis as a tool for day-by-day avalanche forecast", Proceedings of the Snow Mechanics Symposium, IAHS Pub. No. 114, Grindelwald. pp. 391-403.

Bovis, M.J., 1977, "Statistical forecasting of snow avalanches, San Juan Mountains, southern Colorado, U.S.A.," Journal of Glaciology, Vol. 18(78), pp. 87-99.

Breiman, L., J. H. Friedman, R. A. Olshen, and C. J. Stone, 1984, Classification and Regression Trees, Wadsworth & Brooks/Cole Advanced Books & Software, Pacific Grove, CA.

Buser, O., 1983, "Avalanche forecast with the method of nearest neighbors: An interactive approach." Cold Regions Science and Technology, Vol. 8(2), pp. 155- 163.

Buser, O., P. Foehn, W. Good, H. Gubler and B. Salm, 1985, "Different methods for the assessment of avalanche danger", Cold Regions Science and Technology, vol. 8, no. 2, pp. 155-163.

Buser, O., 1989, "Two years experience of operational avalanche forecasting using the nearest neighbors method," Annals of Glaciology: Vol. 13, pp. 31 -34.

Chambers, J. and T. Hastie (eds.), 1992, Statistical Models in S, Wadsworth and Brooks, Pacific Grove, CA.

Davis, R.E., K. Elder and E. Bouzglou, 1992, Applications of classification tree methodology to avalanche data management and forecasting, Proceedings of the ISSW '92, pp. 126- 133.

Elder, K. and R. Armstrong, 1986, "A quantitative approach for verifying avalanche hazard ratings", Avalanche Formation, Movement and Effects, IAHS pub. No. 162, B. Salm and H. Gubler eds., International Association of Hydrological Sciences, Wallingford, UK, pp. 593-601.

Fohn, P.W. Good, P. Bois, and C. Obled, 1977, "Evaluation and comparison of statistical and conventional methods of forecasting avalanche hazard," Journal of Glaciology, Vol. 19(81), pp 375-387.

Jaccard, C., 199(), "Fuzzy tactorial analysis of snow avalanches," Natural Hazards, Vol. 3, pp 329-3401.

Judson, A., and B. J. Erickson, 1973, "Predicting avalanche intensity from weather data: a statistical analysis," Research Paper RM- 112, U. S. Forest Service, Fort Collins, CO.

Judson, A., 1983, "On the potential use of index paths for avalanche assessment," Journal of Glaciology, Vol. 29(101), pp. 178- 184.

Judson, A., and R. King, 1985, "An index of regional snowpack stability based on natural slab avalanches,'' Journal of Glaciology, vol. 31, pp. 67-73.

Kennedy, J. L., 1984, "Avalanche litigation: Technology and liability," Proceedings of the ISSW '84, pp. 99- 101.

LaChapelle, E., 1965, "Avalanche forecasting - A modern synthesis," International Symposium on Scientific Aspects of Snow and Ice Avalanches, IAHS Pub. No. 69, pp. 350-356.

McClung, D. M., and J. Tweedy, 1993, "Characteristics of avalanching: Kootenay Pass, British Columbia," Journal of Glaciology, vol. 39, no. 132, pp 316-322.

Michaelsen, J., F. Davis, M. Borchert, 1987 "A non-parametric method for analyzing hierarchical relationships in ecological data", Coenoses, vol. 2, no. 1, pp. 39-48.

Michaelsen, J., D. Schimel, M. Friedl, F. Davis and R. Dubayah, in press, "Regression tree analysis of satellite and terrain data to guide vegetation sampling and surveys", Journal of Vegetation Science.

Obled, C. and W. Good, 198(). "Recent developments of avalanche forecasting by discriminant analysis techniques: A methodological review and some applications to the Parsenn area (Davos, Switzerland)," Journal of Glaciology, Vol. 25(92), pp. 313-346.

Penniman, D., 1986, "The Alpine Meadows avalanche trial: Conflicting viewpoints of the expert witnesses," Avalanche Formation, Movement and Effects, IAHS pub. No. 162, B. Sahn and H. Gubler eds., International Association of Hydrological Sciences, Walibirtord, UK, pp. 665-677.

Perla, R., 197(), "On contributory factors in avalanche hazard evaluation", Canadian Geotechnical Journal, vol. 7, no.14, pp 414-419.

Perla, R. and M. Martinelli, 1978, "Avalanche Handbook" Agriculture Handbook 489, rev. ed., USDA Forest Service, Washington, D.C..

Sethi, I., 1990, "Entropy nets: from decision trees to neural networks,"Proceedings of the IEEE, Vol. 78(10), pp. 1605-1613.