Journal of Geographic Information and Decision Analysis
Journal of Geographic Information and Decision Analysis, vol. 1, no. 2, pp. 90-100, 1997


An Integrated Expert Geographical Information System for Soil Suitability and Soil Evaluation

Constantine P. Yialouris
Informatics Laboratory, Agricultural University of Athens, 75 Iera Odos, 118 55 Athens Greece
yialouris@auadec.aua.ariadne-t.gr
http://www.aua.gr

Vassiliki Kollias
Soils Laboratory, Agricultural University of Athens, 75 Iera Odos, 118 55 Athens Greece
lsos2kob@auadec.aua.ariadne-t.gr

Nikos A. Lorentzos
Informatics Laboratory, Agricultural University of Athens, 75 Iera Odos, 118 55 Athens Greece
lorentzos@auadec.aua.ariadne-t.gr

Dionisios Kalivas
Soils Laboratory, Agricultural University of Athens, 75 Iera Odos, 118 55 Athens Greece
lsos2kob@auadec.aua.ariadne-t.gr

Alexander B. Sideridis
Informatics Laboratory, Agricultural University of Athens, 75 Iera Odos, 118 55 Athens Greece
as@auadec.aua.ariadne-t.gr



ABSTRACT An integrated Expert Geographical Information System (EXGIS) is presented and applied for the evaluation of the suitability of soil and climatic conditions of in area of southern Greece, for five crops. EXGIS is an integration of an Expert System shell, designed for the manipulation of knowledge concerning soil suitability for agricultural uses, with the commercial GIS package PC ARC/INFO. The work was carried out for the purposes of a research program concerning the development of a Geographical Information System (GIS) for the management and evaluation of natural resources of the Pinios River basin, located in West Peloponnese, in South-Western Greece. Both the FAO system for soil evaluation and the local experience and knowledge of soil and climatic conditions were combined for the formulation of the rules of the knowledge base of EXGIS. The shell of the Expert System communicates with the commercial GIS PC ARC/INFO, under a common operating environment.

KEYWORDS: Expert system, geographical information system, land evaluation.

Acknowledgement The present work has been produced as part of the project 'Development of a Geographic Information System for Natural Resources Evaluation and Environmental Assessment' (No. 91ED449), funded by the Greek Secretariat of Research and Development. The authors are grateful to Professor Yassoglou  for providing the soil data of the area.



Contents

1. Introduction

Since the mid 1970's, Expert Systems (ES) have been used in a variety of application areas. One such area is the evaluation of natural resources (Loh et al. 1994; Kettal 1994). ES applications have successfully captured and focused human expertise in the evaluation of natural resources. Some systems have achieved a level of performance that compares that of human experts. The land evaluation procedures are stored in a knowledge base and are used by an ES. At the same time, a Geographic Information System (GIS) is used, to process and display spatial data (Pereira et al. 1982; Rossiter 1990; De la Rosa et al. 1992; Bouma et al. 1993).
        The present paper concerns the design considerations and the development of an Expert Geographical Information System, EXGIS, which combines the capabilities of a commercial GIS, PC ARC/INFO ver. 3.4, with those of a rule-based ES. This ES has been based on a specially developed shell which satisfies certain requirements, dictated by the Greek agricultural extension services. It has been implemented in CLIPPER because the files produced by the latter (DBASE III+ files) can subsequently be processed by ARC/INFO. EXGIS is currently running under MS DOS. It is used as a tool for the assessment of land suitability for certain agricultural uses. The motivations for the project were the following:

The new system has been applied to study an area of about 30,000 ha, located at the Pinios River basin of west Peloponesse in south-western Greece (Figure 1). This area is of major agricultural interest since the soil and climatic conditions are almost excellent for a variety of crops, although bad soil management and excessive use of fertilisers have caused problems to some soils of the area. The distribution of land uses of the study area is illustrated in Table 1.
 
 


Figure 1  The  study  area:  the Pinios River basin of west Peloponesse in the south-western Greece
 



 

Table 1  Distribution of land uses

Land Use
Percentage of the Area
1. Maize
19% 
2. Olive Trees
17%
3. Forage Crops
14%
4. Citrus
11%
5. Tomatoes
11% 
6. Fallow
10%
7. Wheat
9%
8. Vineyards
6% 
9. Natural Vegetation 
2%
10. Towns
1%

 
 

2. Knowledge Acquisition

Knowledge acquisition plays a crucial role for the development of a reliable ES and requires the close collaboration of the knowledge engineer with the domain experts. For the development of our system we followed a knowledge acquisition procedure consisting of the following four stages:
Domain Acquaintance: This stage aimed at the acquaintance of the knowledge engineer with the application domain. To this end, he studied books and papers related to soil and land evaluation issues under the guidance of the domain experts.
Meetings with the Experts: This stage aimed at the development of an understanding of the experts' way of thinking. At the same time, effort was made to elicit the experts' opinion for the ES's contribution to soil evaluation procedure. To this end, two meetings were organised with three soil scientists. After the second meeting, the experts were asked to propose:

        In response, the experts proposed five crops for cultivation: Maize, Olive Trees, Tomatoes, Wheat and Vineyards. The soil classification as well as the soil and climatic attributes which were proposed by the experts are shown in Tables 2 and 3. As is shown in Table 2, the soil and climatic requirements for Maize (also for any other cultivation) defines four suitability classes, by the maximum limitation method, namely Class S1 (Excellent), Class S2 (Good), Class S3 (Moderate) and Class N (Not Suitable). As can be seen in Table 2, more than one value can be assigned for a given attribute.
Main Knowledge Acquisition: As opposed to the previous two stages, which were preparation for the objectives of the project, this third stage aimed exactly at acquiring the experts' knowledge. Based on the input of the previous stage, we designed forms (Tables 2 and 3) which had to be filled in with the experts' knowledge. This filling was achieved in a number of structured interviews, which were organized separately with each expert. This stage ended with three distinct sets of completed forms.
Resolving of Conflicts: As had been expected, conflicts were identified between the forms which had been completed by the exclusive collaboration with each individual expert. Hence, this stage aimed at resolving exactly these conflicts. This was achieved by the organising of a series of meetings in which all the experts were present. This way, we ended up with a single set of completed forms, which had the approval of all the experts and represented the coded domain knowledge.
 
 


Table 2. The soil requirements of maize

Soil requirements of : MAIZE
FAO framework S1 S1
S2
S3
N
Restriction levels of SYS
0
1
2
3
4
Parametric evaluation of restrictions
100
95
85
60
40
Soil requirements
Slope B C, D E F
Drainage A, B,  C D E F G,  H
Texture
    0 - 25 cm
3,  4,  5 2,  5 2* 1, 2*  
    25 - 75 cm
2, 3 2,  3,  4 2, 2* 1  
    75 - 150 cm 
2, 3 2,  3,  4 1,  2*,3*  1  
Soil Depth >90 (75, 90] (50, 75] (20, 50] <=20
Erosion E0,E1 E2 E3 E4  
CaCO3 0,1 2 3    
pH (6, 7.5] (5.5, 6]
(7.5, 8.3]
(5, 5.5]
(8.3, 6]
(8.3, 8.6]
(4.5, 5]
(8.6, 8.9]
Organic Matter >1.5 (1, 1.5] (1, 0.7] <0.7  
CEC (meq/100gr) >18 (10, 18] (5, 10] (2,5]  
Base Saturation >70 (50, 70] (35, 50] <=35  
                       * gravels

Table 3 The climatic requirements of maize

Climatic requirements of: MAIZE
FAO framework S1 S1
S2
S3
N
Restriction levels of SYS
0
1
2
3
4
Parametric evaluation of restrictions
100
95
85
60
40
Climatic requirements
Average temperature during growth period (22, 25] (18, 22]
(25, 30]
(16, 18]
(30, 35]
(14, 16]
(35, 40]
>40
<=14
Average minimum temperature during growth period (16, 18] (12, 16]
(18, 24]
(9, 12]
(24, 28]
(7, 9]
(28 , 30]
<=7
>30
Average relative humidity during growth period (%) (50, 80] (42, 50]
> 80
(36, 42] (30, 36] <=30
Average relative humidity during mature period (%) (30 , 50]  (24, 30]
(50, 75]
(24, 20]
(75, 90]
<20
>90
 
Sunshine during growth period (r/p)* (0.5, 0.6] (0.4, 0.5]
(0.6,0.75]
(0.35, 0.4]
>0.75
(0.3, 0.35] <= 0.3
Sunshine during growth period (r/p)* > 0.7 (0.7, 0.5] <0.6    
3. Architecture of EXGIS

General Knowledge Based Systems and GIS have been successfully used to develop Land Information Systems (He et al. 1992; Abdelmoty et al. 1994). EXGIS was developed as a tool which makes use of all the available information, to assist in the optimal allocation of land uses. It is an integrated Expert Geographical Information System and consists of two components, ARC/INFO and the ES Shell. Its architecture is shown in Figure 2. It was developed on a PC 486 IBM compatible microcomputer operating under MS DOS. We chose to build it in a PC environment in order to keep its cost as low as possible. ARC/INFO is used for the storage and processing of spatial data. The ES shell was developed for the purposes of the project. It has been implemented in CLIPPER because the files produced by it (dBase III+ files) can subsequently be processed by ARC/INFO. Interfaces have been written in SML (Simple Macro Language), the language provided by ARC/INFO, which integrate the ES and ARC/INFO under the same operating environment. A spatial database, a conventional database and a knowledge base are used to store and process the spatial data, the tabular data and the rules, for land evaluation.


Figure 2 Components  of EXGIS

  The system uses a number of soil maps. The spatial data of these maps are stored in spatial databases created and manipulated by ARC/INFO. These databases also include basic topographic elements, e.g. roads, towns, main railways, rivers, contours and coast line. For each polygon of the map, the soil parameters used for the identification of soil cartographic units are stored in conventional dbf files. Additional parameters are also stored in these files, concerning: This file contains the meteorological data of each individual meteorological station such as: average, maximum, minimum temperature and rainfall, relative humidity, and sunshine hours. The spatial and conventional files are linked through unique polygon identifiers. Macros written in SML control the communication between the different modules of EXGIS for linking spatial and conventional databases. Although there are several methods for designing expert systems, rule-based systems have emerged as the most popular. Deriving their knowledge from easily understood rules, these systems offer satisfactory power and versatility. The Knowledge Base (KB) of EXGIS contains more than 600 rules for land evaluation. The format of each rule is
If <conditions> then <conclusion>;
i.e.,  each rule consists of one or more conditions and one conclusion. Each condition is composed of three parts, an <object>, a <property> and a <value>. The <object> specifies a parameter, the <property> is an operator and the <value> is an expression whose evaluation is related to the <object> through the <property>. For example, the components of condition drainage IS B are

<object>: drainage

<property>: IS. Other properties are <, >, =, IS NOT, etc.

<value>: B.

The conclusion of a rule represents the evaluation of a map unit for a certain land use, with respect to a particular parameter.
        The soil and climatic requirements of each crop under consideration are used for the formulation of the ES rules. For example, in the case of maize for the study area, Tables 2 and 3 contain the soil and climatic requirements. A different knowledge base is built for each crop. The modular design of the system enables its extension with additional knowledge bases, one for each new crop.

By examining the data of a region, specified by the user, and taking into account the requirements of the crops which are stored in the KB, the ES produces an output evaluation file, Eval. This file contains the result of the evaluation of each map unit accompanied by an appropriate justification.

4. System Operation

As is obvious, the system development requires two tasks:
Build up the KB  For a certain geographical area, the experts and the knowledge engineer have to represent the domain knowledge (soil and climate requirements of the crop) for each particular crop they want to include in the KB.
Data acquisition and area digitization This task includes a soil analysis for individual parts or the total area, meteorological data collection of the area and digitization in the ARC/INFO. With this procedure the necessary spatial and conventional DB are built having as result the availability of a detailed map of the study area. The whole area is divided into soil map units. Each unit has a set of properties (characteristics) concerning the soil and meteorological information and it has at least one characteristic with a different value from each neighbor soil unit.
        The system makes use of the rules of the KB and the data of the spatial and conventional databases. It can perform, according to the user's demand, the following:

The ability of the system, to evaluate each soil unit for alternative crops, helps agricultural decision makers in suggesting changes to land uses, in the case that the agricultural management practice (use of fertilizers and pesticides) has caused damage to the environment.
EXGIS can answer and provide justification for the following three types of queries:
Type A: What is the suitability of the soil units for the cultivation of a particular crop?
Type B: Which n crops, out of m different ones, fit best in each soil unit?
Type C: What is the suitability of each soil unit of an area for its current land use ?
        The procedure, to answer a question of type A, is the following:
(i) The user is prompted to supply the system with one particular crop, from maize, tomatoes, wheat, vineyards and olive oil trees.
(ii) The data of the next soil unit are selected.
(iii) The rules of the KB, related to the particular crop, are applied to the input data.
(iv) The result of the evaluation is stored into file Eval (see Figure 2).
Steps (ii)-(iv) are repeated until the data of all the soil units of the specified region have been processed. Next, Eval becomes available to the GIS, which processes and displays the soil units in the way the user has specified.
        To answer a question of type B, the procedure is the following:
(i) The user is prompted to supply a subset of m alternative crops (in the current system, m = 5, since five different types of cultivation have been considered).
(ii) The user is also prompted to supply an integer n (n <= m), which represents the n best alternative solutions, which must be considered for each soil unit.
(iii) The data of the next soil unit are selected.
(iv) The rules of the KB, related to each of the m chosen crops, are applied to the input data.
(v) The result of the n best solutions is stored into Eval.
Steps (iii)-(iv) are repeated until the data of all the specified soil units have been processed. Next, the GIS can display, for each soil unit, the crops which fit best for cultivation. Alternatively, the user can check whether other crops also fit equally well with the first cultivation proposed by the ES, by using the data in file Eval. Every distinct map represents one optional solution.
        The procedure, to answer a question of type C, is the following:
(i) The data of the next soil unit are selected.
(ii) The rules of the KB, related to the current cultivation of the soil unit, are applied to the input data.
(iii) The result of the evaluation is stored into Eval.
Steps (ii)-(iii) are repeated until the data of all the specified soil units have been processed. Next, Eval becomes available to the GIS, which displays a map with the evaluation of each soil unit.

5. Case Study

For the region under consideration, a composite map was produced from the overlay of seven available soil and land use maps, consisting of about 2,500 distinct map units. The farming units of the area are very small, if it is taken into consideration that 3,500 ha with 10 different land uses and 328 soil polygons resulted in 2500 distinct polygons with individual soil - land use combinations. For each type of land unit, additional information was also stored, concerning the use of fertilizers, agricultural management practices and crop yield.
        The seven soil maps were transformed into the scale 1:20,000 and, using manual digitisation, were stored in a spatial database created under ARC/INFO. The rectification of soil maps was achieved with the use of topographic maps (1:5,000 and 1:50,000) of the Geographic Service of the Greek Army. The basic topographic elements  (e.g., roads, towns, main railways, rivers, contours and coast line) were also digitized and stored in spatial files.
        In the following paragraphs, we describe the soil, climatic and land use characteristics of the study area, as well as the basic methodology for land exaluation.

A detailed soil survey of the area carried out by N. Yassoglou (1964) was used as the basic source of information. According to the Soil Taxonomy (Soil Survey Staff 1975), the major soil great groups in the area are Xerofluvents, Xerorthents, Haploxeralfs, Rhodoxeralfs, and Xerochrepts. Seven soil maps 1:10,000 cover the whole area (Figure 1). The soil map units can be identified according to texture (surface 0-25 cm, subsoil 25-75 cm and substratum 75-150 cm), drainage, slope, erosion of the soil surface, the presence of carbonates throughout the soil profile, and the dominant soil classification (Yassoglou et al. 1973). The Entisols occupy 62% of the area, Alfisols 35% and Inceptisols 3%. A land use map (1:20,000) was produced for one of the regions (3,500 ha), which was estimated to be representative of the whole area. The map was produced using manual interpretation of air photographs and field observations. Ten different land uses were distinguished in the area. Table 1 shows the distribution of land uses in the study area. The soil and climatic requirements of five crops, maize, tomatoes, wheat, vineyards and olive oils trees, were assessed using the existing bibliography (FAO 1976, 1983, 1984a, 1984b,1985; Sys 1985) as well as the experience and good knowledge of the local conditions of the experts. The land characteristics and crop requirements were used to define four suitability classes, by the maximum limitation method,         According to the evaluation criteria which have been used, the majority of the soils of the area have been evaluated, in the suitability scale, as S1 or S2 for the crops which are cultivated. Figure 3 illustrates the result of the evaluation of a part of the study area with respect to maize. Table 4 illustrates the evaluation of the study area in a tabular form, with respect to five cultivation types.


Figure 3  Lan suitability  for  maize

        The application of EXGIS to the area of Pinios has shown that, in their majority, the land units qualify for annual irrigated cultivation, with minor differences in maize, wheat and tomato. The excessive use of acidifying fertilizers (NH4)2SO4 has caused problems to some Alfisols of the area, the majority of which are slightly acid. In the last 30 years the pH of the affected soils has decreased by 1 to 1.5 units. Consequently, their suitability was downgraded by 1 or 2 classes. It has been estimated that approximately 15% of the Alfisol soils in which wheat and vineyards are cultivated are unsuitable for such crops, due to low pH values.

Table 4 Evaluation of the study area concerning 5 cultivation types according to FAO framework and SYS restriction levels
S1 S1
S2
S3
N
0
1
2
3
4
Maize 2.55% 61.00% 36.45% 0.00% 0.00%
Tomatoes 0.00% 5.16% 72.48% 22.36% 0.00%
Olive trees 8.53% 43.90% 32.15% 15.42% 0.00%
Wheat 0.56% 7.58% 91.86% 0.00% 0.00%
Vine yards 2.06% 60.16% 26.44% 11.34% 0.00%

6. Conclusions

EXGIS is a low cost but powerful prototype system for land evaluation. Its modular design enables its easy application to various soil conditions, climatic conditions and working environments. Since the requirements of each crop are stored in a different KB the application of EXGIS to new crops is straightforward, and no software amendments are required. The evaluation of the system has been very satisfactory, i.e., the conclusions drawn by it, match those of a human expert. Thus, the latter can be relieved from a substantial amount of work.
        Further work includes the application of EXGIS to the evaluation of other Greek regions. In addition, however, EXGIS is an efficient tool which can also be applied to other areas, such as the management of the forestry ecosystem and land registry. Optimal land use involves interrelated agricultural and economic decisions. Thus the integration of the system with procedures for economic and environmental impact analysis will produce a useful tool for land use planning.
 

References

Abdelmoty, A.I., Paton, N.W., Williamas, M.H., Fernades, A.A.A., Barja, M.L. and Dinn, A. (1994) Geographical Data Handling in a Deductive Object-Oriented Database. Proceedings of the 5th International Conference on Database and Expert Systems Applications, DEXA 94, Athens, Greece, September 7-9, 1994, pp 445-454. Berlin: Springer-Verlag.

Bouma, J., Wagenet, R.J., Hoosbeek, M.R., and Huston, J.L. (1993) Using expert systems and simulation modelling for land evaluation at farm level: a case study from New York State, Soil Use and Management, 9, 131-138.

Davidson, D.A., Theocharopoulos, S.P. and Bloksma, R.J. (1994). A land evaluation project in Greece using GIS and based on Boolean and Fuzzy set methodologies. International Journal of Geographical Information Systems 8, 369-384.

De la Rosa, Moreno, J.A., Garcia, L.V., aand. Almotza, J. (1992). MicroLEIS: A microcomputer-based Mediterranean land evaluation information system. Soil Use and Management 8/2, 89-96.

FAO (1976) A Framework for Land Evaluation. FAO, Rome: Soils Bulletin 32.

FAO (1983) Guidelines: Land Evaluation for rainfed agriculture. FAO, Rome: Soils Bulletin 52.

FAO (1984a) Guidelines: Land Evaluation for irrigated agriculture. FAO, Rome: Soils Bulletin 55.

FAO (1984b) Guidelines: Land Evaluation for Forestry. Paper 48. FAO, Rome.

FAO (1985) Guidelines: Land evaluation for rainfed agriculture. FAO, Rome: Soils Bulletin 55.

He, B., Peterson, C.L. and Mahler, R.L. (1992) An Expert System Linked with a GIS Database for spatially variable fertilizer application. ASAE International Winter meeting presentation, Tennessee, 15-18, December, 1992.

Kettal, Y. (1994) Expert Spatial Knowledge an Expert Geographical Information System in Diffusion of water. Proceedings of the 5th International Conference on Database and Expert Systems Applications, DEXA 94, Athens, Greece, September 7-9, 1994, pp 455-464. Berlin: Springer-Verlag

Loh, D.K., Hsieh Y.C., Choo H.Y.K., and Holtfrerich D.R. (1994) Integration of a rule-based expert system with GIS through a relational database management system for forest resource management, Computer and Electronics in Agriculture 11(2,3), 215-228.

Pereira, L.M., Sabatier, P., and de Oliviera, E. (1982). Orbi - an expert system for environmental resource evaluation through natural language. Report 3J82 Dept. de Informatica, Universidade Nova de Lisboa.

Rossiter, D.G. (1990) ALES: a framework for land evaluation using microcomputer. Soil Use and Management. 6(1), 7-20.

Soil Survey Staff (1975) Soil Taxonomy. A Basic System of Soil Classification for Making and Interpreting Soil Surveys. Washington: D.C. 754p. U.S.D.A. Agriculture Handbook 436, U.S. Dept. of Agriculture.

Sys, C. (1985) Land evaluation. Part I, II, III. Publication Agricoles. State University of Ghent.

Van Lanen, H.A.J., Van Diepen, C.A., Reinds, G.J., De Koning, G.H.J., Bullens, J.D. and Bregt, A.K. (1992) Physical Land evaluation methods and GIS to explore the crop growth potential and its effects within the European Communities. Agricultural Systems, 39, 307-328.

Yassoglou, N. (1964) Soil Survey of Pinios river basin, Ilia Province Greece. Athens: Greek Nuclear Research Centre, Soil Laboratory.

Yassoglou, N., Antonakopoulos, G., Tsakaleris, P., Soulis, K. and Koutalos A. (1973) Soil Survey of North Area of Pinios River, Ilia Province, Greece. Athens: Soil Institute of Greek Ministry of Agriculture. 



ContentsJGIDA vol.1, no. 2
Journal of Geographic Information and Decision AnalysisJGIDA Home