Alleviating Spatial Constraints

ESRI collects more cash with Spatial Analyst and its integrated ModelBuilder, which make ESRI ArcView GIS capabilities accessible to more users

Michael L. Gonzales     

The power of spatial analysis has long been correlated with a high level of difficulty. Although spatial analysis can reveal important insights veiled by spreadsheets, making it desirable to business users, the spatial community widely accepts that the intricacy of the data and subsequent analysis leaves the timid or casual user unable to grasp it. As a result, spatial analysis often breeds exclusive communities of analysts with stovepipe applications.

Spatial architects endure several challenges when blending traditional data with spatial objects and transforming this collective data into cohesive, meaningful, and timely information. Of the many challenges, one of the most pressing is to build an infrastructure that can mold complex data and withstand intricate analysis from an easy-to-use, readily accessible environment. There are, of course, those custom applications that developers have painstakingly created with the goal of concealing spatial complexity in a user-friendly package. The GPS navigator system now available for cars is an example of these custom efforts. But, generally speaking, “user-friendly spatial analysis” is oxymoronic.

ArcView Spatial Analyst (SA) 2.0 represents substantial progress in resolving this long-standing contradiction. An extension to ArcView GIS and an upgrade from SA 1.0, this release of SA integrates raster-based spatial analysis with vector-based mapping. The ability to work with both raster and vector data types is an achievement in itself. Raster data is the oldest form of modern mapping data, vector the newest. In SA 2.0 the difficulty of dealing with these data types, especially raster, is transparent to the analyst.

Moreover, SA 2.0 introduces an integrated toolset called ModelBuilder. Through the use of wizards and drag-and-drop features, this toolset lets users create process flow diagrams that define and sequence complex raster operations. You can store these diagrams (or models) and therefore rerun them as often as necessary. When users rerun these predefined spatial models, they can change data sets and function parameters. As a result, users can calibrate the models or examine how the models perform under the influence of different data sets or constraint values. (See Figure 1.)

FIGURE 1 Users can refine the models after observing their performance with different data sets or constraint values.

What does all this mean in practical terms? Imagine a fire is raging in the La Canada Mountains. Firefighters need to know how the fire will spread and in what direction. Will it go southeast to Altadena or south toward Eagle Rock? And how quickly will it move? To make these predictions, the analysts use a process model that architects have already built in ModelBuilder for this purpose. Analysts simply feed parameters into the model, such as present wind speed, wind direction, elevation, and temperature. It outputs a prediction of which areas the fire will affect. The analysts can continue tuning this prediction or can research alternative scenarios as needed during the course of fighting the blaze. Thus, you can readily design and build a complex data and analysis requirement into a single model. You can, in turn, store the model in a library with other predefined processes for on-demand use.

The benefits of spatial modeling have led to widespread application acceptance from a broad range of user communities, including:

• The Earth Data Analysis Center at the University of New Mexico used SA to help fight the raging forest fires that recently ravaged northern New Mexico.

• The Las Vegas Police Department uses SA to identify, analyze, and track crime statistics in an effort to identify trends, patterns, and serial crimes.

• Lava Cap Winery in northern California uses SA to produce isothermal and iso-degree-day contour maps as part of their ongoing temperature monitoring network.

• Planning South Australia used SA to model the accessibility of health services in the region.

ModelBuilder is unique among competitors in spatial analysis. The notion of creating reusable process flows resonates throughout other computer application areas — data warehousing, for example. And it represents a critical feature for efficiently creating complex models that you can exploit repeatedly.

Additionally, this type of functionality lets you delineate spatial analysis tasks according to effort and skill:. Skilled developers and architects can build the model and deal with the intricacies of the complex data and prerequisite formulas. Analysts and casual users can focus on applying the model to different data sets and parameters as needed.

Aside from ModelBuilder, you have to measure the balance of SA 2.0 against the established functionality of raster-based GIS and vector-based mapping, both pylons of the spatial industry. These technologies support most of the work that spatial applications conduct. The imagery and cell-level granularity of raster data lets engineers, scientists, and developers probe for soil erosion, water runoff, and land-use, among other factors. Moreover, unlike image processing applications, a raster GIS provides robust analysis in imageless processing operations, such as view shed, cost-surface models, and surface interpolation.

A strong spatial analysis product will support traditional map algebra as well as provide a means for critical operations (see Geographic Information Systems and Cartographic Modeling by C. Dana Tomlin, Prentice-Hall, 1990) such as:

• Local: operations that let you merge raster cells of multiple maps into a single map

• Focal: filtering operations that examine surrounding neighborhoods

• Zonal: operations that target all areas of a single class, such as “desert”

• Incremental: operations that examine different aspects, such as slope.

For brevity, I provided Table 1 to rate SA 2.0 according to characteristics fundamental to all GIS products, including: map algebra, imagery, and the industry-defined operations I just outlined. You can use the table’s eight characteristics to help determine the product’s applicability to you.

Suggested Improvements

I can appreciate the tenuous nature of segmenting products and tiering prices in order to maximize your profit potential. Nevertheless, 3D viewing and extensive statistical analysis are components you would naturally expect of a product like Spatial Analyst. ESRI’s supplemental products, Splus and 3D Analyst, should be standard parts of SA, embuing SA with equal depth in virtually every category listed in Table 1. Coupling these strengths with tools like ModelBuilder creates an unparalleled, formidable product. With all those capabilities, Spatial Analyst would be the clear industry leader and benchmark for any future competitors. To offer anything short of that is to play Russian roulette, considering how fickle the buyer community is.

ArcView Spatial Analyst 2.0 offers a robust environment for blending raster GIS and vector-based mapping with strengths in most critical areas and extensions to supplement others, such as 3D viewing and extended statistics. Furthermore, inclusion of the ModelBuilder toolset ensures SA 2.0 a spot in the top tier of product functionality. The notion of building process models is critical for any area of information systems involved in data integration and transformation. Taken further, this ability to create predefined process models allows the development of process libraries analysts can use indefinitely. With such use, SA 2.0 essentially evolves from a robust model builder into a graphical, easy-to-use interface for complex data and analysis. Now even military intelligence seems within reach.