GIS Server Application for Avalanche Studies

“GIS Server Application for Avalanche Studies” by Bryan Joseph Raymer Forecasting avalanches is a complex task that the “GIS Server Application for Avalanche Studies” project prototype attempts to address. This server application provides detailed information -such as name, average activity rate, elevation, and aspect -on each known avalanche path within this study area. The prototype serves as a tool for forecasters to share their knowledge with other civilians who frequent the study area, as well as other forecasters. The primary goal of this project was to establish a tool that utilizes historical weather and avalanche activity data to determine areas where avalanches have occurred in the past and are most likely to occur again, based on forecasted weather conditions. The creation of this GIS model to process data and determine focus areas would allow forecasters to spend more time in the field by eliminating land areas of less concern. This type of model would be valuable for forecasters when they are asked to make predictions for unfamiliar regions, such as forecasting for international events like the winter Olympics or foreign countries to support military efforts. In order to accomplish this, an Arc model was created which allows individuals to enter predicted weather values such as temperature and precipitation. The tool then executes a query on historical information to find other dates with very similar conditions. Upon completion of this query, a secondary query is conducted to determine what avalanche activity occurred on those days. This information is then transferred to a shapefile with the same color scheme as the Colorado Avalanche Information Center’s (CAIC) avalanche probability Danger Rose, which is used by forecasters to communicate vii avalanche probability. Avalanche areas are symbolized as green, yellow, or red depending on whether the most similar days had zero, one, or more than one avalanche, respectively. This feature gives forecasters and other individuals access to tools to examine potential avalanche danger without accessing the areas themselves.

occur on different degree slopes, but they are much less common.
Other than slope, the next most significant factor is snow stability. Snow stability is the potential for snow to slide (unstable) or not to slide (stable) relative to a given load (Greene et al., 2004). The difficulty in gauging snow stability is that it may change frequently, sometimes on an hourly basis. Many temporal factors affect the stability, and if analyzed properly these factors can be used to predict the snow stability (Kozak, Elder, Birkeland, and Chapman, 2003).
The first factor when considering stability is temperature. Snowpack becomes more stable as snow melts and refreezes. However, rapid increases in temperature may cause snow to become unstable and increase the risk of an avalanche. Another factor that must be accounted for is new snowfall. Avalanches are far more likely to occur shortly after new snowfall (LaChapelle, 1980). Additionally, wind must be considered. Both wind speed and direction are important to the probability of an avalanche occurring, due to the transferring of snow from one area to another adding new weight to the snowpack. This process is known as wind loading (LaChapelle, 1980). If the snowpack is unstable, the weight of the new snow can cause the snowpack to fail, resulting in an avalanche.
One further fact that has to be considered is the orientation of the slope. For an avalanche to occur there must be snow accumulation. Snow typically accumulates in more shaded areas that do not have extreme amounts of direct sun light. Consequently, slopes with a northeast aspect (in the northern hemisphere), can be expected to generate avalanches on a more frequent basis.
Lastly, the amount of vegetation can give great insight to the probability of avalanches.
If there is a sudden gap in a patch of trees along a mountain side, it may be inferred that avalanches happen quite regularly in that area. Conversely, expansive areas with unbroken, mature vegetation are much less likely to slide.
Now that the conditions that lead to avalanches are known, the problems associated with their prediction can be examined.

Problem Statement
The copious amounts of data and the frequency at which the data may change make forecasting avalanches a very complex task. Scientific understanding of avalanches, as well as knowledge of the local patterns of avalanche activity (gained through experience), are crucial for avalanche forecasting (McClung 2002a). This criteria limits avalanche forecasting to a few highly trained professionals. Consequently, there are vast areas where avalanches may occur but there are very few people qualified to predict them.
Currently, there are no methods other than experience to determine focus land areas for avalanche forecasters. This means that forecasters are required to be involved in every phase of avalanche forecasting, limiting the amount of time that they may spend in the field doing research where their expertise may be most utilized. The creation of a GIS model to process data and determine focus areas would allow forecasters to spend more time in the field by eliminating land areas of less concern. This type of model would be very valuable for forecasters when they are asked to make forecasts for unfamiliar regions, such as forecasting for international events like the winter Olympics or forecasting in foreign countries to support military efforts.

Study Area
The study area for this project is located in the southern San Juan Mountains just north of the town of Silverton, Colorado. It consists of 26 avalanche paths that come in contact with Colorado Highway 550, the only direct connection to Silverton (Figure 1.2). The frequency of avalanches here has made this area a hotspot for avalanche research.
Previous research completed in this area allowed the CAIC to provide the data needed to complete this study. The focus area within this study is comprised of five of the most active avalanche paths. These five paths are associated with the Red Mountain Pass weather station, which provided all the historical weather for this study. The previous avalanche research, weather station, and high avalanche activity made this an ideal study site.

Client
The primary client for this project is Ethan Greene, the Director of the Colorado Avalanche Information Center (CAIC). The CAIC is part of the Colorado Geological Survey Department, and is responsible for forecasting avalanches in the backcountry and along highway transportation routes for the Colorado Department of Transportation (CODOT). Additionally, the CAIC educates the public about risks associated with avalanches in an effort to limit the negative effects avalanches have on the residents of Colorado (CAIC, 2007).

Client Expectations
The client expects this project to address four goals.
1. Create a map of known avalanche paths that affect Highway 550 near Silverton, Colorado.
2. Create a database of attributes for the avalanche paths. The database would include historical information on the frequency of avalanche activity for the given paths, as well as historical meteorological data.
3. Create a tool to analyze historical avalanche and metrological data.

Background and Literature Review
Most of the literature involving the use of GIS for avalanche study discusses the high level of uncertainty and corresponding factors that cause uncertainty when mapping or modeling avalanche terrain. Scale and spatial data variability are the major causes of inaccuracy when using a GIS for avalanche studies (Campbell, Jamieson, and Hageli, 2004). Over the past two decades, the majority of the research associated with avalanches has been conducted to better understand scale issues and the nature of spatial variability of snow stability.
The largest limitation of the use of a GIS for avalanche work is scale. The discrepancies between the real world and the scale of the GIS representations of avalanche terrain can be immense (McCollister & Birkeland, 2006). These differences may be attributed to the restrictions of available data that can be used within a GIS. For example, to calculate the slope and aspect of an area, one would use a Digital Elevation Model (DEM). However, the most commonly available DEMs are 30 meter DEMs which are not accurate enough for avalanche forecasting (Campbell, Jamieson, and Hageli, 2004). "Subtle changes in slope angle can mean big changes in avalanche frequency, so knowing the difference between a 28 degree slope (which almost never slides), a 33 degree slope (which sometimes slides), and a 38 degree slope (which commonly slides) is critical, and these changes are much too subtle to be picked up with a 30 meter DEM" (  While the accuracy of the DEM causes the most glaring miscalculations, the frequency at which other spatial and temporal data changes also limits the ability to use a GIS to accurately map or model avalanche terrain. Spatial variations create large amounts of uncertainty for mapping or modeling avalanche terrain. Spatial and temporal variations in snow properties present significant challenges for the research of snow processes (Deems, 2002). These variations are most apparent when trying to determine snow stability. The challenge arises from the fact that snow stability, a key component in avalanche forecasting, is very difficult to measure. It is nearly impossible to have enough observation points to collect the data needed to accurately map snow stability (Campbell, Jamieson, and Hageli, 2004). To counter this problem, forecasters must interpolate stability test results to untested areas. To accurately interpolate snow stability, avalanche forecasters routinely perform their stability tests at carefully selected sites to reveal conditions presumed to be representative of nearby avalanche terrain (McClung and Schaerer, 1993). In snow studies, this process of interpolation is termed "scaling" (Campbell, Jamieson, and Hageli, 2004). While this may help the forecasters generate predictions about an area, the use of interpolated data brings uncertainty into the forecast and places limitations on the size of forecast areas.
These size limitations are a direct result of Tobler's law: things that are near each other are more similar than things that are far apart (Maguire, 1991). Since the stability tests were performed to determine snow stability at a specific site, the further away from that site one tries to interpolate conditions, the greater the uncertainty (Logan, 2005).
As mentioned before, temporal variations can have a large impact on snow stability as well. As a snowpack ages, it is exposed to variables, including temperature, solar radiation, and wind exposure, that cause the layers to undergo a metamorphosis (Kozak, Elder, Birkeland, and Chapman, 2003). As the layers change, they may become more or less stable, depending on the variables they are exposed to. Consequently, forecasters must understand the effects of these variables (Kronholm, 2004 with geographic attributes, such as aspect and elevation, they can be mapped, analyzed, and viewed with a GIS (Stoffel et al., 1998). Mapping these data and linking the digitized data with historical slide data could prove to be very useful for forecasters.
Additionally, this would allow forecasters to analyze patterns of slides and slide data (McCollister, 2004). The resulting data gathering could help forecasters know where to direct their attention. For example, if chute "A" does not historically slide until late January, forecasters would know when it is more likely to slide. Additionally, this information could also be used with the conditions that were present when avalanches have occurred historically. For example, if chute "B" has historically slid when temperatures are between 28 and 34 degrees Fahrenheit and after four or more inches of new snow has fallen, forecasters would know when and where to focus their attention.
While this project will not be able to make a large scale prediction possible, it will help forecasters in local areas. Having this data in a GIS will also make it easier to disseminate the data to other forecasters and agencies. Hopefully, closer examination and dissemination of data will lead to a more robust large scale GIS avalanche forecasting model. Until then, geographic information systems must be used to the best of their capabilities to gain as much insight into the phenomena of avalanches as possible (McCollister & Birkeland, 2006).

Proposed Solution
The proposed solution is an ArcGIS Server application providing two separate functions.
The first function will provide an overview of the area that the Silverton forecasters are responsible for. This portion will also serve as an educational tool for individuals not familiar with the area, allowing searches by path name or most active avalanche paths.
The second function of this ArcGIS Server application will allow a forecaster to use historical weather data to determine avalanche focus areas. This prototype feature will permit individuals to enter forecasted weather data into a form, and then the application will query this information against historic data to see what effect similar weather conditions have had in the past.

Geodatabase Design
In order to create this application, a personal geodatabase was created. This personal geodatabase was selected by the client because of their use of Microsoft Access. The database is comprised of four feature classes, two data tables, and a background raster

Known Avalanche Paths Feature Class
The Known Avalanche Paths feature class is comprised of polygon shapes that represent the known avalanche areas within the San Juan Mountains that affect Colorado Highway 550. This feature class has attribute fields such as name, aspect, elevation, activity per year, and activity per year that affects specifically Colorado Highway 550. The attributes of this feature class are utilized as the queries in the ArcGIS Server application.

Lakes Feature Class
The Lakes feature class consists of polygons representing the lakes and reservoirs in the study area which have attributes such as name, shape, and area. This feature class was included for area familiarization and to show where water naturally accumulates.

Rivers Feature Class
The Rivers feature class is illustrated with vector lines representing the rivers and streams in the study area. This feature class has attributes such as name and length, and was included for familiarization and to show natural low tracts.

Streets Feature Class
The Streets feature class is represented by a vector line representing Colorado Highway 550. This feature class is included to show areas in which the avalanche paths intersect the highway. This is particularly important class because keeping the highway open is one of the top priorities of the Silverton forecasters.

Focus Area Feature Class
The Focus Area feature class is comprised of polygons that represent the avalanche paths associated with the Red Mountain Pass weather station. This feature class is used to complete the prototype portion of this application, which uses historical data to determine possible avalanche risk. This feature class is used for symbology. Importing symbology from this feature class allows the layer on the map output to follow the color symbology parameters of the Avalanche Danger Rose.

Historical Weather and Avalanche Activity Data Table
The Historical Weather and Avalanche Activity table was created by combining two separate databases, both provided by CAIC. The first database presented historical avalanche activity for five of the paths in the study area that intersect Colorado Highway 550. The second database provided weather information that was recorded by the Red Mountain Pass weather station. This database consists of average, maximum, and minimum daily temperatures, as well as daily and cumulative precipitation. The combined database will provide the data which will be queried as part of the prototype application provided by the ArcGIS Server application. When combined, the database consisted of one row for each of the 1825 days in the study period, with temperature and precipitation data for each day, as well as information on avalanche activity. A subset of that database containing only the rows for the active avalanche days appears in Figure   3.2.

Application Design
This section gives an overall description of the design of the "GIS Server Application for Avalanche Studies" intranet mapping project. Subsequent chapters will go into greater detail regarding how each portion of the project was created, functions, and satisfied the client's requirements for the project.

Interface
"GIS Server Application for Avalanche Studies" is a prototype web mapping application and associated ArcGIS tool designed for the CAIC. The purpose of this application is to design a prototype application that can display known avalanche areas, and provide detailed information on an individual avalanche path. Forecasters and researchers will be able search for avalanche paths by path name or by avalanche activity. The purpose of this site is to allow users to explore and to learn more about the study area without physically entering the avalanche territory.
This application is designed for avalanche professionals who may not be familiar with GIS. With this in mind, this application was kept as simple as possible. The map was created by using natural colors, with the avalanche paths represented in pink.
Furthermore, the query menus are designed as either drop-down menus or with the most common inputs as defaults.

Avalanche Forecasting Tool
The Avalanche forecasting tool is slightly more complex than the rest of the tools within this application, and may be used outside of the "GIS Server Application for Avalanche Studies." In order to use this tool, data layers and data fields for the tool must be added to process. The user must then enter target information, such as temperature and precipitation. To successfully utilize this application, the user would need some familiarity with GIS software. The uses of this tool and the outputs will be discussed further in the following chapters.

System Architecture
This application is powered by ArcGIS Server, and was built using ArcGIS Server  There are three main inputs for this system: 1. Historical Weather and Avalanche Activity table.
2. Vector data of the avalanche paths within the study area.
3. Imagery for Red Mountain Pass.

Data storage
The data storage device for this project is a personal geodatabase. This will provide all the capabilities needed for the geospatial analysis portion of this project.

Data dissemination
The data dissemination will be left to the client's discretion. The client has proposed making data and forecast examples available on their ftp site. However, due to liability risks the client will not be able to share the actual forecast with the general public. .

Data analysis
The geospatial analysis was accomplished within the ArcGIS 9.2 commercial, off-theshelf software. The nearest neighbor data analysis, which will be discussed in Chapter Six, was completed by using a created ArcGIS tool. This tool executes a series of queries and then displays the results via a table and a new shapefile layer.

Data and product output
At this time, there are two output products this project will provide. The main output product will be a new shapefile of the focus area, establishing the likelihood for the focus area to produce an avalanche. This will be demonstrated much like the Danger Roses that avalanche forecasters currently use to communicate forecasts. The areas will be symbolized with different colors: green, yellow, and red, with red indicating the highest probability for avalanches. The second output will be a data table of the most similar days.

Avalanche Base Map
The avalanche base map was created using the ArcGIS desktop products ArcMap and ArcCatalog. Before creating the map, it was necessary to gather geographic data. The streets and water data was obtained from ESRI's US base map data ( Figure 5.1), which is provided with the software. In order to find this data, one would navigate to the file where the software was installed, then navigate through the data folder to the maps and data folder and select USA (Data\Arc9\Maps_and_Data\usa).

Figure 5.1 ArcCatalog navigating to USAbasemap.lyr
The Avalanche Paths shapefile was provided by Douglas Scott of Avalanche Mapping Corporation.
The final layer for this map document was a high resolution image of the study area, acquired from ESRI's ArcWebServices, which served as a base image for the MXD. In order to include this image, an ArcWeb Services account had to be established. A trial ArcWeb Services account may be acquired at http://www.esri.com.
Once all of the data layers were collected, they had to be re-projected into an appropriate projection using the project tool. This projection selected was NAD_1983_UTM_Zone_13N, placing all the layers into the same projection as the avalanche paths layer.

Designing the application with ArcGIS Server Manager
The "GIS Server Application for Avalanche Studies" was designed using ArcGIS Server Manager, a web application that complements the ArcGIS Server application. This webbased application allows administrators to create, publish, and manage web mapping applications and services. The ArcGIS Server Manager is also used to manage the GIS server on which the applications are stored.

Publishing Avalanche Base Map document
When creating a web mapping application with ArcGIS Server Manager, the first step is to publish the associated MXD. In order to accomplish this, permission was granted to allow the Server Object Container (SOC) to access the MXD, then the SOC account was added to the folder containing the all the data used in the "GIS Server Application for Avalanche Studies" (Figure 5.1).

Figure 5.1 Giving the SOC permissions to access avalanche geodatabase
Once this was done, ArcGIS Server Manager was used to publish the MXD. This tool provided a wizard menu, which walks users through the process for publishing a MXD.
The first step within the wizard is to specify the MXD, by navigating to where the MXD was stored. Then a folder was created on the server to publish the resource to ( Figure   5.2).  With the publishing of the MXD completed, the next step was to create the web application with ArcGIS Server Manager and add tasks to the application. Tasks serve as tools that allow the user to analyze or gain more information from the map. There are six standard tasks included with ArcGIS Server. The six tasks are editing, find address, find place, geoprocessing, query attributes, and search attributes ( Figure 5.4).

Figure 5.4 Wizard used to set up task
The "GIS Server Application for Avalanche Studies" uses a total of four tasks. Two of the tasks are Query Attribute tasks titled "Locate Avalanche Path by Name" and "Calculate Average Road Crossings per Year Tool". The third task is a Geoprocessing task titled "Nearest Neighbor Avalanche Forecasting Tool," which will be discussed in Chapter Six. The fourth and final task is a print task titled "Print Map." The print function, not an out-of-the-box function, was created by Bryan Baker at ESRI (Baker, 2007). The print task is available as a sample task for the Web ADF/ArcGIS Server and is scheduled to be included in the ArcGIS Server 9.3 release. Once the required tasks were added, they were configured using the Task Configuration tool within ArcGIS Server Manager. The Task Configuration tool allows for the creation of titles for each of the tools, customization of the button text, and customization of task-specific settings such as default values for the queries. This tool also allows one to choose how the selected items would be displayed within the Web application. The "Locate Avalanche by Path Name" tool configuration window is depicted in Figure 5.5.

Figure 5.5 Locate Avalanche by Path Name task configuration window
In order to add or edit tasks using these wizards, the user must first select the map service for the task to be added. In this instance it was the Avalanche Base Map. Next, the MXD layer that the attribute query task queries must be selected. In this instance, it is the Colorado Highway 550 avalanche paths layer. The button text and the title of the tool were set in the dialog. The task was then configured such that users could select paths by name from a drop down list, given that many users would not be able to remember each path by name. The final step in designing this task was to choose the display color of the selected path. Yellow was selected for this application because it is bright and easily seen against the background map.
The task titled Calculate Average Road Crossings per Year Tool was designed following these same methods.

Customizing the application in Microsoft Visual Studio 2005
The final customization of the "GIS Server Application for Avalanche Studies" web application was completed using Microsoft Visual Studio 2005. The final customizations that were completed were determined by consulting the client, the advisor, and committee member Bryan Baker. The first customization was to modify the title bar. The University of Redlands logo, and links to the CAIC's and the Avalanche Mapping Organization's website, were added to the title bar of the application. Then the color of the application theme was changed. The client determined that the default blue application theme color was not visually appealing with the corresponding map. Based on this the client choose the green application theme color. The next step was deciding which typical GIS tools, such as zoom, zoom to full extent, pan, identify, measure, and magnify, to include.

Figure 5.6 GIS tools offered before customization
After consulting with the client, it was decided to include the zoom tools, the pan tool, and the identify tool. While some of these tools could have all been removed to simplify the application, the client wanted them included in the event that the prototype would be expanded.

Figure 5.7 GIS offered after customization
The final customization that occurred was to cause results to be expanded immediately upon being added to the task results panel. This change required a change in the application code for a property of the task results panel called expansion depth. The original expansion depth in the code was set at "1". At this level users would need to be familiar enough with GIS applications to intuitively know how to expand the results. To make this section of the application more intuitive for non-GIS users, the results expansion depth was reset to "3". The results of this change are illustrated in Figure 5.8. There are two data mining techniques that are commonly used in avalanche forecasting.
The first technique involves the use of regression trees to analyze the days before an avalanche event, to determine what conditions cause avalanches (McCollister, 2004). The second technique, and the one used for this research, is called the "Nearest Neighbor" technique (McCollister, 2004). This technique examines only the actual day that an avalanche occurred to determine the conditions that cause avalanches. This latter technique was used in this project, at the client's request. The Nearest Neighbor Avalanche Forecasting tool is an ArcGIS model that utilizes a Python script to query a Microsoft Access database. This model sorts variables and distinguishes possible areas of interest to avalanche forecasters. However, the model is not intended to make avalanche predictions for large areas, but will focus on relatively small, well-known areas. The "Nearest Neighbor Avalanche Forecasting" tool allows the user to employ historical weather and avalanche activity to identify areas where avalanches have taken place in the past under similar weather conditions. To use this product, forecasters will input the needed data. The forecasters will begin by selecting the path or paths to be forecasted. They will then input the predicted incoming weather data precipitation and temperature. The model will complete a number of queries and use this data to create a new shapefile to demonstrate the probability of an avalanche to occur in the given focus areas based on the predicted weather conditions.

Avalanche Forecasting Python Script
The Python script embedded within the "Nearest Neighbor Avalanche Forecasting" tool is the core of this model. The first task that this script completes is to calculate the difference between the user defined inputs and all of the corresponding historical data.
Differences are calculated for both maximum temperature and precipitation for every day within the database. Once these differences have been calculated, a second script adds each historical day's differences together. Since all avalanche areas are different, forecasters can elect to weight these parameters to make the model more effective for their forecast area. Then another query examines the different values and identifies the days with the least total difference in values.
After determining the five nearest neighbor days, another query is initiated. This query searches the avalanche activity field in the five nearest neighbor days and calculates a sum of avalanche activity, which is the basis for the final output shapefile, to be symbolized on. If the sum of the avalanche activity is less than one, the avalanche paths will be symbolized in green, signifying that no avalanches had occurred under similar weather conditions. If the sum of the avalanche activity is equal to one, the avalanche paths will be symbolized in yellow, signifying that moderate avalanche activity had occurred under similar weather conditions. If the sum of the avalanche activity is greater than one, the avalanche paths will be symbolized in red, signifying that a high level of avalanche activity occurred under similar weather conditions. Figure 6.1 represents the finished script as an ArcGIS Tool.

Developing the Avalanche Forecasting Model
The nearest neighbor Arc model was designed to make this tool user-friendly. Without this model, users would have to know what specific data types and data fields would be needed to run the analysis. The model simplifies this process by allowing the data and the data fields to be set using the Python script. The data inputs for this model are avalanche paths and historical weather data. Additionally, the model also allows the parameters of the model to be set. The parameters for this model are maximum temperature (max temp) and precipitation (prec). However these parameters can be changed in order to use minimum or average temperature. The two parameters are entered by the user and represent forecasted weather data.
The final step of the model is to designate the location where the model's output will be stored. Since this tool will be utilized via the World Wide Web, the outputs are stored in memory.

Publishing the Nearest Neighbor Avalanche Forecasting Tool
Once the "Nearest Neighbor Avalanche Forecasting" tool was completed, it was published to the "GIS Server Application for Avalanche Studies". In order to accomplish this, the Server Object Container (SOC) needed permission to access the tool ( Figure   6.2).

Figure 6.2 Granting the SOC permission to access the model
Once permission was granted, ArcGIS Server Manager was utilized to finish publishing the tool. Within ArcGIS Server Manager, the publish service wizard was used. The first step was to navigate to where the tool was stored and to name it (Figure 6.3)

Figure 6.3 Directing Server Manager to the avalanche forecasting tool
Once the tool was added to the server it had to be added to the applications itself. To do this, the application was edited using the edit application tool (Figure 6.4).

Figure 6.4 Edit application tool
Once inside the editing session, the add task link was clicked, launching a GUI that allowed the geoprocessing service to be added to the application ( Figure 6.5).

Figure 6.5 Adding a geoprocessing services to the GIS Server Application
After adding the services, the associated supporting services were established ( Figure   6.6).

Figure 6.6 Adding a supporting service to run the geoprocessing services
Once the associated service and the geoprocessing tool were configured, the defaults on the remaining wizard windows were accepted until the tool was successfully added to the application.

Using the "GIS Server Application for Avalanche Studies" Tools
The "GIS Server Application for Avalanche Studies" application was designed to allow users to learn more about the study area without actually having to go out into the field.
This application supplies four tasks for users to gather and share information about the study area. The four tasks provided are: "Select Avalanche Path by Name," "Calculate Average Road Crossings per Year," "Forecasting Tool," and "Print Map."

Locate Avalanche Path by Name
The "Locate Avalanche Path by Name" tool allows users to quickly and correctly find the locations of a given avalanche path. The tool was designed particularly for users who might not be familiar with the study area. To make this tool user-friendly it was designed with a drop down menu. To use this tool the user must select "Path Name" and then choose an avalanche path from the drop down menu (Figure 7-1).

Figure 7.1 Locate Avalanche Path by Name tool and associated drop down menu
Once the user selects one of the path names the results will be displayed in the results window In order to display this result on the associated map, the user must check the box next to the associated path. Checking this box will display the corresponding path on the map in yellow (Figure 7.2).

Calculate Average Road Crossings per Year Tool
The "Calculate Average Road Crossings per Year Tool" tool allows the users to sort and identify avalanche paths by number of avalanches they produce that affect Colorado Highway 550. Avalanche forecasters in this area concentrate their efforts on the paths that have the most potential to affect Colorado Highway 550. This tool allows forecasters to query the Historical Avalanche Activity database to determine which avalanche paths have had the largest effect on Highway 550. To use this tool, users first need to select the task. A simple fill-in-the-blank menu is then displayed (Figure 7.3).

Figure 7.3 Average Avalanches on Road per Year GUI
This fill-in-the-blank Graphical User Interface (GUI) is designed with a default value of "1". However, the user may choose any numerical value to input into the GUI and select evaluate. The tool will then query all the paths and only select those that produce this number or greater number of avalanches which have affected Colorado Highway 550.
The results of this query will be displayed in the results window (Figure 7.4).

Figure 7.4 Results from Calculate Average Road Crossings per Year Tool
To display these results on the associated map, the user must select which paths they want to display. This function takes slightly more effort and GIS familiarity, but gives users the ability to display the paths one at a time (Figure 7.5) or all at once (Figure 7.6).

Using the Nearest Neighbor Avalanche Forecasting Tool
The Nearest Neighbor Avalanche Forecasting tool allows users to quickly query the associated database to search for historical avalanche activity that has occurred on historically similar weather days. This tool was designed to be user friendly and for individuals who are not familiar with GIS. To launch this tool, the user simply selects the tool in the task menu. Once the tool is launched, a simple fill-in-the-blank GUI appears (Figure 7.7). The GUI is designed so that users may input forecasted maximum temperature and precipitation, and the tool will search the database for historically similar days.
Once the user enters the values and selects submit, the tool runs. The values, a tool status message, and the results of this tool will display in the results window. To view the results, the box next to the output must be selected.

Using the Print Function
The Print Map function allows users to print the map being viewed within the web browser. To use this function, users need to zoom in or pan to the area of interest. Once this area is clearly in view, the user can simply select the print map task and a print map dialog box is launched. This dialog box allows users to change certain settings such as title, size, resolution, and scale (Figure 7.9).

Figure 7.9 The print task GUI
Once this GUI is completed, the user selects the "create print page" to execute the print function.

Conclusion
The "GIS Server Application for Avalanche Studies" is a prototype tool that allows avalanche forecasters and avalanche researchers to retrieve information about this avalanche study area without venturing into avalanche territory. The user friendly application provides the following tools for forecasters and avalanche researchers.
Tools included in "GIS Server Application for Avalanche Studies": 1. Quickly and correctly identifies avalanche path by name.

Quickly and correctly identifies avalanche paths that have historically affected
Colorado Highway 550.
3. Provides a prototype avalanche forecasting tool.
4. Allows forecasters to utilize the wealth of historical data that have been compiled.

Future Work
The "GIS Server Application for Avalanche Studies" provides several opportunities for future research endeavors. These endeavors include but are not limited to testing and expanding the study area.

Initial Testing
The next logical step for this application is further testing of the prototype "Nearest Neighbor Avalanche Forecasting Model." The first test that should be completed would be a test to see if the avalanches actually occurred when the model said that avalanches would be most probable. One option for further testing could be completed by obtaining more data. For example, the historical weather data from known paths that have generated avalanches could be input to the model to determine if the program recognizes these paths as likely candidates to produce slides. This testing could provide evidence for or against the continuing use of the maximum temperature and precipitation parameters for avalanche forecasting.

Parameter Testing
A parameter test could also be done by using the same data that was used in creating the prototype tool. For example, the parameters currently used for the tool are maximum temperate and precipitation. The temperature parameter could be changed to minimum temperature or average temperature, while the precipitation parameter could be changed to year-to-date precipitation. The model could be executed with any or all combinations of these parameters. The output results when using these different parameters could then be compared. If each of these parameters produced different results, the results could be tested in a similar way as the first test discussed. Testing such as this could help improve the overall accuracy of the model.

Expanding the Study Area
Expanding this study area to include surrounding avalanche areas is another possibility for future work. There are two possible ways that this application could be expanded. The first technique would be to create separate applications for each separate avalanche area for which the CAIC is responsible. For example, an organization such as the CAIC could build a separate application for each of their focus regions, allowing individuals to gather and spread information about each avalanche area.

Compare Results from Different Study Areas
In some avalanche areas it is believed that parameters such as wind have a higher correlation to avalanche activity than temperature and precipitation. Both wind speed and direction are important to the probability of an avalanche occurring, due to the transferring of snow from one area to another adding new weight to the snowpack , a process known as wind loading (LaChapelle, 1980). One way to illustrate this is to compare model results from different study areas. For example, if a model proves accurate in one study area and highly inaccurate for another study area, it could illustrate the need to use different or more parameters, such as wind, in future models.

Future Works Conclusion
Due to the complex nature of avalanches, it is currently impossible for one model to predict avalanches for all areas. Tests such as these are, however, very important in furthering avalanche research. Only through research and tests such as these can researchers begin to build accurate and useful avalanche models. Hopefully, future research will lead to a robust large scale GIS avalanche forecasting model. But until then, GIS may be used to gain insight into the phenomena of avalanches as possible (McCollister & Birkeland, 2006). Only by gaining information can researchers and policy makers counter the negative effects that avalanches produce. gp.tabletotable_conversion(WeatherTable,"in_memory",TableName + "_temp") tempTable = "in_memory" + "\\" + TableName + "_temp"