Land use-cover conversion from forest to

another land was contributing for total GHG

emission, especially form forestry sector. Land
Change modeler from IDRISI Selva is one of
many tools which concern for environmental
spatial analyst. We can predict future land
cover based on actual and previous land cover,
also included some driver variable such as,
road, river, and settlement (distance),
elevation, slope, etc.

Land Change
Modeler
IDRISI Selva

Prepared by Arif Prasetyo

1
INTRODUCTION
The processing workflow of this tool can be see as below:

There are 6 tab of this panel tool, but I was used to 3 or 4 tab to predicting the future
land cover and estimating emission of GHG. There are Change Analysis, Transition
potential, Change prediction, and REDD Project. Implication and Planing tab are focused
for habitat and biodiversity analyst.
Prepare / make Project Directory and Project File.

2
Purpose of this step is to make directory as database for all processing file and project as
long as LCM tool going on.

In this process, I was using some spatial datas as below:

a. Land Cover (actual and previous)
b. Elevation and Slope
c. Road and Distance from road
d. Distance from River
e. Distance from settlement and distance from city (point / center of urban area)

1. CHANGE ANALYSIS TAB

a. LCM Project Parameter
This panel allows us to specify the essential files associated with the land cover change
analysis of a specific study area, as well as a project name.
For the change and prediction analyses, a minimum requirement is the specification of
two land cover maps that can be used as the basis of understanding the nature of
change in the study region and the means of establishing samples of transitions that
should be modeled.

Land Change Modeler has the following requirements for the input land cover maps that
can be resolved with the Harmonize tab:
a) That legends in both maps are the same.
b) That the categories in both maps are the same and sequential.
c) That the backgrounds in both maps are the same and have a value of zero.

3
 d) That the spatial dimensions including resolution and coordinates are the
same.
If these requirements are not met, the Harmonize tab will launch to resolve these issues.
There is the option to specify REDD project parameters for REDD modeling. Selecting
this option will allow for the specification of the project start and ending dates, including
the reporting interval. Setting these parameters will dictate the settings for the Change
Prediction process and the future scenarios for the REDD modeling tab.
b. Change Analysis
The Change Analysis panel provides three graphs of land cover change between the two
land cover maps specified in the Project Parameters panel, we can view these graphs in
a variety of units (cells, hectares, square kilometers, acres, square miles, % of change
and % of area). Note that changing the units on this panel causes the units on the
Change Maps panel to also change, and vice versa

c. Change Maps
This panel provides the ability to create a variety of change maps, including maps of
persistence, gains and losses, transitions and exchanges.

In this case, I choose transition from Forestland to All, it’s mean deforestation.

4
c. Spatial Trend of Change
In landscapes dominated by human intervention, patterns of change can be complex,
and thus very difficult to decipher. This panel provides the ability to map trends with a
best fit polynomial trend surface to the pattern of change

a) The intention of this module is to provide a means of generalizing about the

pattern of change. The numeric values do not have any special significance. The

5
 surface is created by coding areas of change with 1 and areas of no change with
0 and treating them as if they were quantitative values.
b) Trends up to the 9th order can be calculated. However, the time needed to
calculate the surface increases substantially as the order is increased.

2. TRANSITION POTENTIALS
a. Transition Sub- Models
The tables on this panel lists all transitions that exist between the two land cover maps
(except those filtered out in the Change Analysis panel) and requires the user to specify
which transitions are to be used for calculating the transition potentials. We can have as
many sub-models as we have transitions, or we can group transitions into named sub-
models, assuming these grouped transitions share the same underlying driving
determinants for prediction.
We must have specified a project in the Project Parameters panel before accessing this
these tools.

In this case, I used transition from forestland to non-forestland (deforestation).

b. Variable Transformation Utility
The Variable Transformation Utility panel is an optional panel for transition potential
modeling that provides a selection of commonly used transformations. The
transformations available include: natural log, exponential, log it, square root, power,
and evidence likelihood.
These are particularly critical if the Logistic Regression modeling option is chosen since it
requires that the variables be linearly related to the potential for transition.

6
The Multi-Layer Perceptron (MLP) and SimWeight options do not require the variables
to be linearly related, but transformation can sometimes make the task easier for it to
solve in cases of strong non-linearity’s, thus yielding a higher accuracy.  Two require
special mention.
- The natural log transformation is commonly effective in linearizing distance decay
variables (e.g., proximity to roads).
- The evidence likelihood transformation is a very effective means of incorporating
categorical variables into the analysis. Categorical variables must either be converted
into a set of Boolean (dummy) variables, or transformed using the Evidence Likelihood
transformation option (highly recommended).
We must have specified a project in the Project Parameters panel to access these tools

c. Transition Sub-Model Structure

The Transition Sub-Model Structure panel provides a table for specifying and describing
the driver variables for a specific sub-model as specified in the Transition Sub-Models:
Status panel. If we have more than one sub-model we will need to identify the driver
variables for one sub-model and create transition potential maps for each sub-model in
turn.
 Variables can be added to the model either as static or dynamic components.

7
 a) Static variables express aspects of basic suitability for the transition under
consideration, and are unchanging over time (may be possible for elevation,
river, and slope).
b) Dynamic variables are time-dependent drivers such as proximity to existing
development or infrastructure (roads) and are recalculated over time during the
course of a prediction (may be possible too for distance from deforestation and
settlement).
d. Run Transition Sub-Model
The MLP option can run multiple transitions, up to 9, per sub-model. But SimWeight and
Logistic Regression can only run one transition per sub-model.
a) Multi-Layer Perceptron (MLP) Neural Network
 The Transition Sub-Models: Status panel and the Transition Sub-Model
Structure panel must be filled out prior to running the model, i.e., we must
specify a sub-model and its associated environment variables.
 Open the Run Transition Sub-Model panel in the Transition Potentials tab.
 Select the MLP neural network option to calculate the transition potentials.
 Note that MLP can model multiple transitions that are grouped together in a
common sub-model from the Transition Sub-Model: Status panel. Initially
the dialog for the MLP neural network may seem daunting, but most of the
parameters presented do not need to be modified to make productive use
of this very powerful technique.
 Choose a sample size.
 The sample size must be smaller than the smallest of the cells that
transitioned and persisted between the two dates of the land cover maps.
Generally for MLP, use the largest sample size possible. Note that MLP will
use half of the sample for training and half for testing.
 Click the Run Sub-Model button.
 When the Run Sub-Model button is clicked, samples are extracted from the
two landcover maps of areas that underwent the transitions being modeled
as well as the areas that were eligible to change but did not.
 When it is finished, click the Create transition potential button to produce
the transition potential model(s)

8
b) SimWeight
 The Transition Sub-Models: Status panel and the Transition Sub-Model
Structure panel must be filled out prior to running the model.
 Open the Run Transition Sub-Model panel in the Transition Potentials tab.
 Select the SimWeight option to calculate the transition potentials.
o Note that the SimWeight option cannot model multiple
transitions that are grouped together. They must be modeled
separately.
 Choose a sample size.
o The sample size must be smaller than the smallest of the cells
that transitioned and persisted between the two dates of the
land cover maps. With SimWeight, the number of samples will
strongly affect the time it takes the model to run. Therefore it is
advisable to select the smallest sample size that is
representative.
 Select the K parameter. K determines the number of most similar instances
that are used to calculate the transition potential.
o K must be less than the total sample size. A larger K makes the
system more general while a smaller K makes it more specific. If
artifacts are evident in the result, try using a larger K – the
artifacts are the result of over-training.
 Click on the Calculate Relevance Weights button. The system will calculate
and graph the weights to be assigned to each variable. To ignore variables
with too low a weight, set the minimum weight threshold – doing so will
increase the speed of calculation.
 Click the Run Sub-Model button.

9
 Based on the validation data (half of the overall sample), the system will
report the hit rate, false alarm rate and the Peirce Skill Score

c) Logistic Regression
 The Transition Sub-Models: Status panel and the Transition Sub-Model
Structure panel must be filled out prior to running the model.
 Open the Run Transition Sub-Model panel in the Transition Potentials tab.
 Select the Logistic Regression option to calculate the transition potentials.
o Note that the Logistic Regression option cannot model multiple
transitions that are grouped together. They must be modeled
separately.
 Indicate whether to apply a stratified random or systematic sampling and
the sampling rate.
o Specifying a sampling scheme can significantly decrease the
processing time, as well as reduce the negative impact of spatial
interdependence.
 Click the OK button to begin the process.
The resulting transition potential map will be displayed when the process is
complete, as will a .txt file

10
All of Transition Sub-Model type process, will produce Potential Transition Map like
below :

11
The higher value mean that there is more probable / potential to change, in this case,
change from Forestland to Cropland.
3. CHANGE PREDICTION TAB
a. Change Demand Modelling
This panel allows us to determine the amount of change that will occur to some point in
the future using the Markov Chain prediction process or a user-specified model

The default procedure, Markov Chain, determines the amount of change using the
earlier and later landcover maps along with the date specified. The procedure
determines exactly how much land would be expected to transition from the later date
to the prediction date based on a projection of the transition potentials into the future
and creates a transition probabilities file. The transition probabilities file is a matrix that
records the probability that each landcover category will change to every other
category.
b. Change Allocation
The Change Allocation panel predicts future scenarios.
Recalculate stages is interval of years that will be producted land cover map.
In this case, land cover map prediction from 2010 to 2040 with stage maps in 2015,
2020, 2025, 2030, and 2035 (stage = 5).

12
13
4. REDD PROJECT TAB
a. REDD Project File
The REDD Project File panel allows us to set our REDD project file environment to save
and retrieve previously run scenarios. We can create a new project or open an existing
project. We can save different REDD scenarios with the Save project option.

b. REDD Project Spesification

The REDD Project Specifications panel allows us to set the project and leakage area
input files that will be used in the calculation of GHG emissions. The project start and
ending dates along with the reporting interval are specified in the LCM Project
Parameters panel under the Change Analysis tab. These parameters determine the
calculation and reporting of GHG emissions in the subsequent panels.

14
c. Calculate CO2 Emission
This panel allows for the identification of the types of carbon pools to be included in the
analysis along with their respective carbon densities in terms of tons of carbon per
hectare (tC ha-1). Six carbon pools are available for inclusion in the calculations: above-
ground (AB), below-ground (BB), dead wood (DW), harvested wood products (HWP),
litter (L), and soil organic carbon (SOC).

d. Calculate Non-CO2 Emission

The Calculate Non-CO2 panel allows us to specify and calculate non-CO2 GHG emissions
due to biomass burning. Specifically, methane (CH4) and nitrous oxide (N2O) emissions
can be estimated per carbon pool, excluding harvested wood products and soil organic
carbon, when deforestation is due to fire.

15
e. Calculate GHG Emission
This panel estimates the net GHG emissions based on a REDD project's estimated
effectiveness. For each stage of the REDD project, the effectiveness rate is calculated
based on estimated leakage rate due to displacement of agricultural activities within the
project area and the overall success rate of the project. These estimates will alter the
overall total baseline GHG emissions. Once this panel is completed, a Microsoft Excel
spreadsheet will be generated detailing the GHG emissions

16
Source:
IDRISI Selva Help System
17