The crystal globe: A GIS-based operational area production model

The Crystal Globe : A GIS-Based Operational Area Production Model

Yousif Ali Bussin Johan Bode Alfred De Gier
The Internation Institute £or Aerospace Survey
and Earth Sciences (ITC)
7500 AA Enschede, The Netherlands
Fax: (31) (53)874-399
E-mail: [email protected], [email protected]

Abstract

The Area Production Model (APM) is a simulation model of land use changes in response to growth population, gross domestic product and agriculture productivity. Given these factors, the model can predict the amount of land that will be transferred from forest to agriculture land use. APM was developed by order of the Food and Agriculture Organization (FAO) of the United Nation. This model was first written in fortran computer language and later implemented within LOTUS 123 spreadsheet software. APM is a numerical model, e.g. the data input in form of number such as tables and the output in forms of numbers such as tables or graphs. Earlier, the authors have resolved the model spatially and r apply it to an area in East java, Indonesia. This implementation include the land transfer part of the model. The above mentioned researchers used five different software packages to achieve the spatial implementation of APM. The objective of this research was to create an automated operational spatial APM in ILWIS (Integrated Land and Water Information System) image processing and GIS software. In most cases in countries like Indonesia and Thailand it is not possible to establish an agriculture field on a cleared forested land. Therefore, forest land will be change to scrub land. This new implementation of the model can be used for predicting forest degradation.

Introduction

As in ancient Biblical times, e.g. the history of King Saul and the female prophet of Endor (I Samuel 28 vss. 1-25), people always wanted to know what would happen in the future. One of the later developed methods to get knowledge of the future, people thought, was to look into a crystal globe and see clearly what would happen. Nowadays, people have not changed in the curiosity for the I future. They still want to know what will happen in the future. But what have r changed is the materials and methods. In our time we would not use crystal globes to look in the future but we develop computer models which are able to predict the future based on regression curves. There are plenty of subjects useful for prediction. One of these subjects is the issue that the world , scientists worry about and would like to know what will happen in the future, is tropical forest degradation. Forest degradation is defined as all biological, chemical and physical processes that result in loss of the productive potential of natural resources in areas that remain classified as forest (World Bank, 1991) .The reasons of forest degradation are logging, human activities, animal grazing and natural factor (e.g. storms, lightening) (Palo , et al., 1986) .One of the computer programs which can predict the future land , use and therefore predict forest degradation, is the Area Production Model (APM).

The development of the Area Production Model (APM) has been an important. step in predicting land use changes. The APM is intended for simulation of long term land use changes and the prediction of primary and secondary yields from agricultural and forest lands. Among other things, the model simulates the future need for agricultural land. The model’s demand and supply scenarios for agricultural products and land are generated primarily by the growth rates of population and GDP, and by changes in land productivity. The model is comprehensive, but does not have excessive data requirements

The model uses three different agricultural classes: land for subsistence crops (used mainly for home consumption), land for market crops (produced mainly for the local markets), and land for cash crops (destined primarily for markets outside the area) .

An important aspect of the APM is the concept of land allocation. If demand for land in a particular class increases, transfers of land from another class to that one are generated by the model. When the donor class concerned is exhausted, land from another class is transferred. The sequence of the donor land use transfer classes is specified by the APM user. This sequence usually goes from currently under-utilized agricultural land, through the conversion of one category of agricultural land into another one, and finally to forest land.

The APM has been tested by FAO in East Java, Indonesia (FAO, 1986a and 1986b) and by FAO and USAID in two parts of Peru. In both cases tropical rainforest areas were included. The conclusions were that the model was adequate and sufficiently adaptable for extension as well as modification. Williams (1987) converted the original FORTRAN version to LOTUS 123. APM assumes that the forests are degraded mainly by human activities (expansion of the agriculture) . Indicators which are used for predicting the future land use are the population growth and the growth of the gross domestic product (GDP) (FAO, 1986a) .The APM requires only numerical data input. Its output is in tabular and graphic form. Spatial data are neither required nor produced. It is therefore insensitive to the location at which the need for more land exists and to the location of the donor land use class from which land is to be transferred. This is considered a major limitation, because it is not possible to predict where the land for the specific classes will be needed, and where the land of the relevant donor class is located. But if one would not only like to predict how much forest will be degraded but also where it will degraded, one have to take into account more factors than the numerical model takes (e.g. slope and distance) .Furthermore, a spatial model would allow to grade accessibility to land, by using friction coefficients, and it could simulate the locational effects of decisions that influence the need for land transfer.

De Gier and Hussin (1993) described how the spatial component was developed and linked to the model for the Kali Konto area in east Java, Indonesia. All spatial data used were in digital form. The results of this spatial implementation show the suitability of GIS for combining the spatial component and the numerical output from the APM. The results also indicated that the spatial component of the APM significantly improved the model’s behavior and interpretation capabilities. Later, Hussin et al., 1994 verified the results of spatial version by the numerical one and established the idea of predicting forest degradation using APM. While this paper can be considered as an step in the ongoing research of creating a computer model which will be able to predict spatially the forest degradation. It will gives ideas on how the model was transferred to GIS-ILWIS.

The objective of this research was to create an ILWIS-GIS spatial model which makes it possible to predict forest degradation using the principle ideas of APM in the numerical and spatial forms. The model will consist of two parts: Automation part, with the help of batch and text files for GIS-ILWIS, the process of forest degradation can be done more than once. This enables the user to quickly see the different results depending on different values of input factors. Second part is the forest degradation. The result of the model will be presented as a map. Depending on the idea of how forest degradation will occur, the contents of this map could/will differ. Four maps can be distinguished depending on the assumption about the idea of forest degradation made.

General Approach

In order to reach the research’s aim a method was developed. The first step was to do manually the process of forest degradation in GIS-ILWIS. This derived information on which commands successively have to be given to create the desired map. Secondly these commands were written into a batch file in order to automate the process of forest degradation. The third step was to verify numerically and spatially the model on the reality.

In the first instance, spatial APM must be numerically verified. In the numerical APM the predicted surface of agricultural land depends on both the population growth for subsistence crop and the GDP for the cash crops. But in spatial APM, the prediction of agricultural land only depends on the average r population growth, in order to simplify the calculation. However, it must be verified in how far the prediction of agricultural land by using the numerical APM which approximates the prediction of agricultural land. Spatial verification means that we have to find the most correct data and make the best decisions in the spatial APM resulting in a spatial prediction of forest degradation which approximates as best the reality. The right data have to be found for the tables POPULATION, PRIORITY and YEAR.

Making the best decisions means that there has to be chosen whether the plantations are degradable or not and that there has to be chosen from where the forest degradation must start (the agricultural area, the existing scrub land : or from both) .

After the numerical and spatial verification, the spatial APM can be seen .as most optimized for the situation on Kali Konto. Now it has been made possible to solve the research question: describing the impact of the input factors Slope, Distance, Population Growth, Population Pressure and Priority on the process bf forest degradation in Kali Konto and classify the most important one(s)

The Area Production Model in LOTUS (Numerical APM)

The main idea behind the process of land use change is that land use classes (mainly forest) will transfer after each other at the benefit of theexpanding agricultural land use class. The model assumes that the increasing demand for agricultural land is the most important factor for the land use , changes. The agricultural development is assumed to be controlled by two parameters: First, the demand for new agricultural land differs according the , types of crop (subsistence crops, market food crops and cash crops) .Second, the production depends on the area cultivated and the productivity.

The next formulas are used to calculate for each crop type the demand for new land:

Demand for new land for subsistence crops = Present Area * P(pop) P(prod)
Demand for new land for market food crops = Present Area * P(GDP) / (prod)
Demand for new land for cash crops = Present Area * P(GDP) / P(prod)

where:
P(pop) = Growth rate of population
P(prod) = Growth rate of the Gross Domestic Product
P(GDP) = Growth rate of the agricultural productivity

The numerical APM which run in LOTUS is working according to the following steps:

STEP 1. Choosing the option DATA to fill in the data for the specific region ;this involved in:

The region: Starting year, total population, rural population, annual GDP/capita
and the total land area.

The land use: The surfaces (ha) of the several land use classes.

Transfer: Choosing which land use classes will increase and which land use classes will decrease.

The priority: Making a ranking among the land use classes which are supposed to transfer (=decrease) Priority I means that this land use class will firstly be transferred into for example agriculture. The growth factors: Fill in for the next 10 years the growth factors of the rural and total population (Ppop)’ the Gross Domestic Product (PGDP) and the agricultural productivity (Pprod) for the ” tree possible crops (subsistence, market and Cash).

STEP 2. Choosing the option CALCULATE. This causes a several results. The .factors governing change in land use (Ppop, PGDP and Pprod) are extrapolated for the next 50 years. The results are show in a table called table I and graph 1. With the help of these factors the totals in population, GDP and GDP/capita, are calculated and printed in a second table called table 2. With the help of tbe three formulas mentioned above, the known surfaces of the land uses ( as mentioned in step 1) and the extrapolated values of the Ppop, PGDP and Pprod ,it is possible to calculate the future area which has to be transferred which is = .(subsistence area + market food area + cash crops area) .The results are
printed in table 3 and graph 2. Now the total transfer land for every future year is known, it will be divided over the land use classes which are proper for transfer into agriculture. While the proper land use classes will transfer into agriculture one by one using the known priority for change by the division. The result is printed in table 4 and graph 3.

STEP 3. The final step is choosing the option results which makes it possible to print out the tables and the graphs. In this case the most interesting tables are table 1-4 and the most interesting graphs are the graphs 1-3 (FAO, 1986) .

The Crystal Globe : A GIS-Based Operational Area Production Model

Results and Discussiohs

The results of this research is a computer ILWIS batch file which includes all steps of the spatial APM. This model is based on the numerical model which was described earlier. This model was developed in GIS-ILWIS. In spite of the fact that the model is not difficult to use (e.g. it is fully automated), it is recommendable to get firstly knowledge of the assumptions of the model and how it is working.

Assumptiona

Every model will be an approach of the reality. It cannot take into account every variable factor of life which does influence the process of the model e.g. forest degradation. Therefore, it is important to know the assumptions of a model. In this model the following assumptions were made:

1. The most important assumption is that in this model the forest classes do not degrade after each other as in numerical APM. It is assumed that the forest classes can be degraded at the same time. Every pixel gets a value depending on: slope, distance from agricultural/village land, priority for forest classes of the rural people, population density and population growth. None of these is extra weighted nor does have a primary key. These factors were chosen their data were available in raster maps.

2. It is assumed that each of these factors has the following influence: The steeper the slope the slower the forest degradation and the higher the friction value; The greater the distance the slower the forest degradation and the higher the friction value; The higher the priority value the slower the forest degradation and the higher the friction value. Note that the highest priority means the lowest priority value, the higher the population growth the faster the forest degradation and the higher the starter value. The higher the population density the faster the forest degradation and the higher the starter value.
3. It is not clear whether it has to be assumed or not that it is possible for plantation forest to degrade into scrub. It was argued that plantation forest could in practice only be thinned or harvested. While this is still not clear, it has been made possible for the user to decide whether he/she want to have the plantation forest as degradable forest or not. If it is decided that the plantations are not degradable, it means that the user has to make an absolute barrier of these land use classes which could easily be done by giving the plantations a negative value in the priority table.
4. In the first instance, preference was set on using the real values of the input factors: slope, distance, priority, population growth and population pressure. This would guarantee a proper relation among the values within an input factor. But taken into account all these, two arguments plead for reclassification of the values of every input factor between 1 and 10. First of all, using the real values of an input factor give a weighting among the input factors. For example, using the real values would give a high weighting to the map slope distance (values 1-18825) and a low weighting to the population growth (values 1-5) .The second reason is that the real values of the input factors would, if you want to calculate the distances, result in values which exceed the maximum pixel value or the capacity of the computer.
5. Theoretically, forest degradation seems to start at the fringes of the scrub land. It is not clear whether the model is taking into account this theory. Therefore, it has been made possible for the user to chose whether forest degradation must start from the agricultural land or the existing scrub land or from both.
6. The process of forest degradation is simulated by creating a time-of-change-map. Firstly, this is done by calculating distances in a friction map in which each pixel has a weighting factor. This weighting factor includes the values of the DISTANCE, the SLOPE and the PRIORITY for change. This gives a direction to the forest degradation. Secondly, computing this map with the output of “population density (e.g. pressure) * population growth, creates the time-of-change-map.
7. It is assumed that the area of forest which will be degraded into scrub can be calculated in two steps. First, the future agricultural land is calculated with the help of the following formula:

   Na=Ba*(p/100)n

   where:
   
   Na = Total agricultural area in the future year (ha),
   Ba = Total begin agricultural area (ha)
   
   p = Population growth (%),
   n = future years (years)
   
   Second, the degraded area is calculated with the helpof the following formula:

   Na = X1 * (Na)1/2 -X2 * Na -C
   
   where:
   Na = Total scrub area in the future year (ha)
   Na = Total agricultural area in the -future (ha)
   X1, X2, C = The coefficient of X1, X2, and the constant of the regression model between APM-predicted agricultural land and the real developed scrub land.

How is the new mode1 working in ILWIS-GXS

The new spatial APM working according to the following steps:

STEP 1. The model uses four input maps: General, Slope distance, Forest and
Villages; and three tables: Priority, Population, and Year; Two batch files: Program.bat and Program1.bat; Twelve text files: FORMO.txt up to FORM9.txt, The DOS command file: ASK. corn.

STEP 2. Start ILWIS, then exit to Dos, change the directory to where the files of step 1 are loaded. The model will start by typing PROGRAM and then [ENTER]

STEP 3. The Model comes up with the tables population and Priority which can be altered by the user.

The table population contains information about the number of people, the surface of agricultural land and the population growth for each village. In FORMO.txt the user (s) alternations, the pressure on land, the Vilfact (e.g. village population) for each village and the maximum Vilfact are automatically stored and/or calculated in table POPNEW.

The table Priority contains information about the people’s priority for transfer of a forest class. In FORM1.txt the user’s alternations and the maximum priority are automatically stored and/or calculated. It must be noticed that a forest class with a negative value for its priority will be considered as an absolute barrier and will therefore not degrade into scrub.

STEP 4. In ILWIS-MCalc Form2.txt the map STARTER is realized. Each village owns apart of the agricultural land. This village land is given an integer value between 1 and 10 based on the Vilfact values of table population. With the help of the Distances Module in ILWIS a Thiessen map Thies is created. In Thies, this Vilfact value of each village land is extrapolated into the
non-agricultural land use classes.

STEP 5. The user have to chose whether the process of forest degradation will start from the agricultural land or the scrub land or from both the agricultural and scrub land using the files FORM4_1.txt, FORM4_2.txt and FORM4_3.txt

STEP 6. In ILWIS-MCalc Form4 x.txt the map FRICTION is realized. Each pixel of a non-agricultural land use class has an integer value between 1 and 10. This value is based on the slope, distance from the village land and the priority. With the help of the ILWIS Distances Module the time of change into scrub land has been made visible in the friction map. The time of change depends on the roughness (e.g. friction value) of the area.

STEP 7 .In MAPOUT the time of change into scrub land has been weighted by the “Vilfact of each village. ;

STEP 8. The Histogram (Mapo) of the pixel values is derived from mapout in Histogram Form6 .txt .

STEP 9. The program comes up with table Year. The user has the possibility to change the begin area of agriculture, the percentage population growth and the future year for prediction. Based on these three values the total area of agricultural land for the future year is calculated (in Tabcalc Form7.txt), then transformed into a total needed area of scrub land and finally printed in table New year.

STEP 10. In ILWIS Tabcalc, the file Form8.txt, the total area of scrub land in the future year is translated into total needed pixels. If the cumulative number of pixels of table MAPO is lower than these needed pixels, then it will get a value of 23, otherwise it will get a value of 0. The result is printed in column (value) of table MAPON. The pixel groups with the value 23 will change into scrub land in the future. The other groups will not yet change.

STEP 11. Finally Map4 is constructed as follows: First, the pixels outside the area, the lake, the existed scrub land and the agricultural land use class will get the same value as the map Forest. Second, the pixel groups of table MAPO which has a value of 23 must be adjusted. These are the new developed scrub lands and will get a value of 23. Finally the rest pixels will also get the value of the map Forest.

References

• De Gier A. and Hussin Y.A., (1993) Spatially Resolved Area Production Model in Kali Konto, Indonesia, GIS/LIS ’93, Annual Conference and exposition, 31 Oct. -4 Nov., 1993, Minn, USA, Vol 1. 157-169 pp.

• Hussin Y.A., A. de Gier and Hargyono (1994) Forest Cover Change Detection Analysis Using Remote Sensing: A test for the Spatially Resolved Area Production Model, Fifth European Conference and Exhibition on Geographic Information System, EGIS/MARI ’94 Proceedings, Paris, France/Maroh, 29-April 1, 1994, Vol 111825-1834 pp.
• FAO, (1986) Manual for using the area production model (APM), Case Studies, Asia-Pacific Region. GCP/RAS/106/JPN. Field Document 12:2, May, 1986. 99 pp.
• FAO, (1986) Users guide to area production model (APM), Asia-Pacific Region. GCP/RAS/106/JPN. Field Document 12:1, May, 1986. 65 pp.
• Hargyono, (1993) Occurrence and prediction of forest degradation, a case study
  of Upper Konto Watershed East Java Indonesia ,lTC, Enschede.

• Palo M., M. KANNINEN, G.MERY and A. SELBY. 1986. Forest-based socio-economic development and deforestation in developing countries: A feasibility study for a major research project. Proceedings of IUFRO World Congress, Ljubljana, Yugoslavia.
• World Bank, 1991, A world Bank Policy Paper, The Forest Sector, The World Bank,
  Washington DC.

• Williams, D. H., (1987) LOTUS APM, version 1: A spreadsheet version of the area
  production model 1. Asian seminar on forest planning, Kuantaun, Malaysia.
  November 5-7, 1987. 12 pp.