With a straightforward graphical environment—the Macro Modeler—to
develop analytical and modeling routines, along with a selection of
almost 250 geographic information system (GIS) analysis and image
processing functions, IDRISI software has served as the happy medium
between involved code writing and “black box” GIS analysis for years.
Users have enjoyed the ability to customize their analysis procedures
without programming expertise.
However, Clark Labs departs from tradition in its newest incarnation,
IDRISI Andes (Version 15), by adding a vertical application that is
dedicated to specific goals. The new release also offers revised
versions of some old modules, new graphics options and enhanced data
access. Although it’s primarily known as a raster image processing
package, IDRISI Andes also offers vector GIS functions capable of being
manipulated using SQL on a linked database.
New Features
Data access and organization have been improved for Andes with the
addition of a new Explorer bar that integrates the IDRISI Explorer, the
metadata viewer and project file path selector of older versions. These
all appear in a tabbed window at the left of the screen, thereby cutting
down on the amount of browsing necessary when naming new files. File
selection differs a little from the Windows Explorer-style setup from
earlier versions, but this feature comes in handy for quickly
referencing existing file names and layered metadata. When not in use,
the window can be minimized to make room for images.
IDRISI continues to build on its unparalleled selection of image
classification modules. One broad improvement for the Andes release
is that its image processing modules now accept any data type,
expanding their utility. Neural networks and machine learning have
been a large part of recent IDRISI product development, as evident
in the revision of the Multi-Layer Perceptron (MLP) neural network
classifier to support an automatic training mode. Two new neural
network classifiers include the Self-Organizing Map (SOM), developed
by Teuvo Kohonen in the early 1980s, and Fuzzy ARTMAP. Both SOM and
Fuzzy ARTMAP support supervised and unsupervised routines. Other new
machine learning modules offered in Andes include a fully automated
Classification Tree Analysis (CTA) module and a K-means unsupervised
classifier.
Image processing and analysis modules also have been revised and
expanded for the Andes release. For example, a new Multinomial
Logistic Regression module has been added that supports a multi-categorial
dependent variable. Additionally, the Principal Components Analysis
(PCA) module now supports an automated inverse PCA for noise
reduction, and the TREND module has been expanded to calculate
polynomial surfaces up to a ninth order. Analytical improvements
include an expanded RUNOFF module that can incorporate rainfall
duration and initial absorption variables, a SEDIMENTATION module
that builds on the program’s Revised Universal Soil Loss Equation (RUSLE)
module to calculate net soil movement within a given area, and a
more versatile CROSSTAB module allows the cross-tabulation of a
third layer and can generate results for fuzzy membership images.
The most celebrated new element in Andes is the Land Change Modeler
(LCM) for Ecological Sustainability, which was developed for
Conservation International’s Andes Center for Biodiversity
Conservation—thus the source of the version’s name. The LCM
application is designed to address the loss of habitat and
biodiversity through land cover change. LCM is an integrated
modeling environment for the analysis and predictive modeling of
land conversion and includes tools for analyzing land cover change,
modeling the potential for future change, predicting future change,
assessing the effects of land cover change on biodiversity and
integrating planning regimes into predictions.
After using the tutorial, experienced GIS
users should find the LCM easy to use and informative. Although
experienced users may be able to deduce exactly what each component
of the LCM is doing, less-experienced users may not find the answers
they are looking for in the Andes Help menu or manual. For
background on the processes in LCM, users will find the most in the
tutorial (with references), not in the Help menu as for other
modules in earlier versions of IDRISI.
A benefit of the LCM is that it provides ways to assess data through
stages of processing. For example, LCM allows users to examine the
data through bar graphs, as in the change analysis module, and
report results in .txt format. Unfortunately, these slick graphs
aren’t available for export other than screen capture, hindering
their use in reports without exporting the data and duplicating the
graph elsewhere.
IDRISI’s potential for quantitative analysis never has been matched
by its graphic output capabilities, forcing some users to export
their images and data to other software systems for graphic display.
Some attempt was made to change this with the addition of some new
graphic display options, including new north arrow designs, the
option to import a user’s own north arrow, and the ability to
include and activate outside images as insets for a primary image
using the new Photo Layer function.
Clark
Labs also touched up the histogram function for Andes. Now
users can manipulate the range and class widths of graphs in
the open HISTO function window after initial graph creation.
Thankfully, the folks at Clark Labs have done away with the
unruly histogram gridlines, which couldn’t be manipulated by
the user in earlier versions. Raster group files can be used
in the revised HISTO module, allowing users to generate
multiple graphs simultaneously. However, the multiple colors
option in the view settings menu, which allows users to apply
a color palette to the data in a histogram, is confusing in
that it doesn’t stretch color values to match the range of the
data represented in the graph, creating a “rainbow” effect
through the data. And as before, the histograms generally
disappoint as graphics when copied and pasted to the
clipboard.
Overall Assessment
IDRISI Andes’ competitive pricing, versatility and
accessibility keep it among the top choices for image
processing. Users get an improved product with expanded
capability and a respectable amount of technical support with
the Help menu, user manual, and tutorial. Experienced IDRISI
users should find Andes to be familiar and perhaps more
comfortable than previous versions. The new Explorer bar and
the interactive histogram editor are changes that are most
likely to affect the workflow of regular users. The Explorer
bar is an improvement, providing a heads-up reference that
enhances file organization on the fly. The image histogram
allows users to quickly assess their data by modifying the
display parameters without starting the HISTO module from
scratch. However, as with earlier versions, publishing
histograms or images is probably better left to other software
packages.
Clark Labs has strengthened IDRISI’s status as a solid option
for image analysis by refining some modules from previous
versions and adding the LCM for Ecological Sustainability.
Although LCM is a package designed for particular analysis
goals, the independent nature of many of the processes it
offers stays true to the utilitarian, adaptable nature of the
IDRISI line.
