Environmental Engineering and Management Journal February 2010, Vol. 9, No.2, 205-213

http://omicron.ch.tuiasi.ro/EEMJ/

______________________________________________________________________________________________

SBC-MEDIU: A MULTI-EXPERT SYSTEM FOR

ENVIRONMENTAL DIAGNOSIS

Mihaela Oprea, Daniel Dunea

University Petroleum-Gas of Ploiesti, Department of Informatics, 39, Bucuresti Blvd., 100680 Ploiesti, Romania

Abstract An efficient environmental management system needs some air, water and soil pollution analysis software tools. The paper presents a multi-expert system, SBC-MEDIU, developed for environmental diagnosis. The system has three modules: SBC-AIR, the module for air pollution analysis and dispersion assessments, SBC-WATER, the module for surface water pollution analysis, and SBC-SOIL, the module for soil erosion risk assessments. The experimental results obtained so far are also discussed. Key words: air pollution, artificial intelligence, expert system, soil pollution, water pollution

Author to whom all correspondence should be addressed: email: m_oprea@yahoo.com, Phone: 40(244)575059, Fax: 40(244)575847

1. Introduction

In the last decade, more applications of artificial intelligence (AI) were developed in the area of environmental protection. The final purpose of such applications is to improve the quality of the environment (air, soil and water), and therefore, the quality of life on the Earth. Till now several artificial intelligence techniques such as knowledge based systems (Cortés et al., 2002; Oprea and Dunea, 2008; Vouros et al., 2000), case-based reasoning (Rodríguez-Roda et al., 2001; Rubio et al., 2004), machine learning (Gibert et al., 2005; Sànchez-Marrè et al., 2002, Vellido et al., 2004), and artificial neural networks (Wieland et al., 2002) were successfully experimented in the area of environmental protection. Artificial intelligence can provide efficient software instruments for the monitoring, diagnosis and control tasks of the environmental quality at different levels of analysis: local, regional, national, or international (Kim and Platt, 2007).

Modelling the environmental knowledge is a key issue in a successful implementation of an artificial intelligence-based approach. An efficient environmental management system has to include software tools for air, water and soil pollution diagnosis. Recent years showed progress in the development of several complex systems that are

based on AI techniques (Bellamy and Jones, 2007; Oprea and Sànchez-Marré, 2004; Platt, 2008). These systems are usually specialized either to air, water or soil analysis. Most of the systems are dedicated to air and water analysis, and few of them to soil analysis. Our research work involved the integration of all three main components of the environmental analysis (air, water and soil) into an environmental diagnosis multi-expert system. Such an integrated system can provide a general overview of the environmental pollution status in a given region for a certain period of time by taking into account the interdependencies that can appear between air, water and soil pollution. In some situations, these interdependencies can be very critical, and a global analysis might provide the support for the environmental protection decision-making factors.

A multi-expert system was developed, called SBC-MEDIU, which has three modules, SBC-AIR, SBC-SOIL, and SBC-WATER, respectively for air, soil and water pollution analysis. Fig. 1 presents the modular architecture of the multi-expert system. Each module of the system has a knowledge base (KB) with specific expert knowledge represented under the form of production rules (IF-THEN rules), and all modules are using the inference engine of VP-Expert, which is an expert system generator (Friedrich and Gargano, 1989). Due to the high complexity of such

“Gheorghe Asachi” Technical University of Iasi, Romania

Oprea and Dunea/Environmental Engineering and Management Journal 9 (2010), 2, 205-213

206

system, the efforts were focused toward the inclusion of the expert knowledge for some types of environmental diagnosis. SBC-AIR provides air pollution analysis and dispersion assessment, SBC-SOIL accomplishes soil erosion diagnosis and SBC-WATER performs surface water pollution analysis.

SBC-MEDIU

KB Air

User Interface

KB Soil

KB Water

VP-EXPERT INFERENCE ENGINE

Environmental Knowledge Base (ENV-KB)

SBC-AIR SBC-WATER SBC-SOIL

Fig. 1. The modular architecture of the multi-expert system

SBC-MEDIU Fig. 2 shows the basic components of each

module of the SBC-MEDIU system: a knowledge base, the inference engine (common to all modules), and the databases with standards and regulations (for a clean air, water and soil), and time series of specific measurements (meteorological, concentrations of air pollutants, water pollutants, and soil pollutants etc). The knowledge bases of the multi-expert SBC-MEDIU system were implemented in VP-Expert.

DataBases

Knowledge Base

Inference Engine

Fig. 2. The basic components of each module of SBC-MEDIU

In the next sections, some case studies of using

the three modules of SBC-MEDIU, and the experimental results obtained so far are presented. 2. Case studies

SBC-AIR

The analysis made by the SBC-AIR module has three main parts: (1) the air quality index assessment, (2) the air pollution analysis, and (3) the air pollution dispersion assessment. The knowledge base KB-Air has rules related to these parts, written in VP-Expert, and used when running the SBC-MEDIU multi-expert system.

The first part of the analysis establishes the air quality indicator using recorded imissions pollutant concentrations (15 minutes - sulfur dioxide, hourly -

ozone, nitrogen dioxide and 24 hours values - PM10 fraction) introduced by the user providing warning capabilities related to the potential impact of air pollution on sensible individuals:

(1) Green Code – levels 1, 2 and 3 - Effects are not noticed by individuals sensitive to air pollutants;

(2) Yellow Code - levels 4, 5 and 6 - Medium effects, may be noticed amongst sensitive individuals, do not require intervention;

(3) Red Code - levels 7, 8 and 9 - Significant effects may be noticed by sensitive individuals and action to avoid or reduce these effects may be needed;

(4) Maximum Code - level 10 - The effects on sensitive individuals described for “High” levels of pollution are worsening - Maximum Alert.

The second part of the analysis carries out a more complex approach to the air pollution by taking into account apart from the air pollutant concentrations, the meteorological factors (the observed factors, the measured factors and the predicted factors), the period of the year when the analysis is done, and some factors that are specific to the analyzed urban region. The knowledge base used in the air pollution analysis includes rules from the knowledge base of the expert system DIAGNOZA_MEDIU described in (Oprea, 2005). The result of the analysis is the degree of air pollution risk, and the decision and control measures that should be taken if the risk is high.

In this section we shall focus on the third part of the air analysis, which is the air pollution dispersion assessment. SBC-AIR required the integration of a simple tool assessing what happens to pollutants in the atmosphere after they are discharged from stationary emission sources. For the present, the stochastically based Gaussian type model is the most useful in modeling for regulatory control of pollutants (Schnelle and Brown, 2002; Vogmahadlek and Satayopas, 2007).

The Gaussian plume model provided rough estimates of pollutant ground level concentrations (imissions) in the absence of monitored data, allowing air quality index calculations. Therefore, the final objective of the calculations using SBC-AIR capabilities was to determine if an emission will result in ground ambient concentrations which exceed air quality standards that have been set by reference to air quality criteria. Knowing the location of the source relative to the receptor and its characteristics would allow calculating the concentration at a particular downwind receptor using a dispersion model. The model is applicable to continuous sources of gases and particulates less than 10 m in diameter estimating the plume concentrations over horizontal distances of 102 to 104 m (Nicolescu et al., 2008).

The Gaussian plume model (Eq. 1) and related additional algorithms were used to allow SBC-AIR to compute ground level concentrations at any receptor point (Xo, Yo) in a polluted region resulting from each of the isolated sources in the emission inventory.

SBC-MEDIU: A Multi-Expert System for Environmental Diagnosis

207

2

2

2

2

2

2

2exp

2exp

2exp

2),,,(

zzyzy

HzHzy

U

QHzyxC

(1) where: C(x, y ,z, H) - pollutant concentration at any point in the plume (g m-3);

Q – emission rate of pollution from the source (g s-1);

H – the effective height of the pollution source, function of the chimney height, its diameter, speed and temperature of gas exhaustion, and air layering;

y, z – horizontal and vertical standard deviations of the pollutant concentration distributions in the y and z directions;

U – the wind speed measured at the height of the source (m s-1).

Despite its limitations in building, a complex three-dimensional image of the bulk transport of pollutants and their chemical transformations before being deposited on the earth's surface, and the building downwash and surrounding topographical features effects, the system provided preliminary information of the main air pollution dispersion factors status and trends based on real data from different emission sources in the Dâmboviţa County. The experimental results are discussed in section 3.

SBC-WATER The SBC-WATER module carries out an

analysis of river pollution with effluents from wastewater treatment plants and other industrial sources. The domain knowledge considered in the development of this module has been encoded mainly on the basis of literature review such as books (Mendes and Santos Oliveira, 2004), scientific papers, technical manuals, water quality standards, norms and regulations, and interviews with domain experts.

A large number of indicators representing the discharging limits with pollutants on the release in natural receivers, extracted from the current regulations were introduced in the knowledge base (e.g. NTPA-001/2002 technical norm contains 40 physico-chemical indicators). The knowledge base is composed by a set of rules related to standards of water quality, a set of heuristic rules given by the experts in the domain (based on their experience), and a set of rules related to some methods of water pollution analysis extracted from the literature. The uncertainty is modelled through the certainty factors (CNF), which are percentage values in the interval [0,100], that are associated to facts and rules from the knowledge base implemented in VP-Expert.

The heuristic rules were generated from some decision tables through the induction mechanism that is included in VP-Expert. The decision tables model knowledge from past cases that were solved by the human experts. The certainty factors were given by the human experts, together with explanations of the

rules. Fig. 3 shows the decision tree used by SBC-WATER.

Deoxygenation

Water_flow Effluent_flow

BOD

Rainfall Surface_runoff

Type_of_surface Slope

Water_temp

WATER_QUALITY

Type_water

Fig. 3. The decision tree of the module SBC-WATER (water quality analysis)

We have considered that water quality depends

on the type of water, on the evolution of the biochemical oxygen demand (BOD), and of the water temperature. Some examples of rules used by the module SBC-WATER are given bellow.

RULE A_1 IF Type_water = wastewater_industrial AND

Water_temp = increase AND BOD = increase

THEN WATER_QUALITY = decrease Advice = check_BOD_COD_ratio CNF 90;

RULE H_SR_12 IF Surface_runoff = low AND

Rainfall = above_normal THEN Water_flow = increase

DISPLAY “Raise stream level”; RULE R3-DT5 IF Water_flow = increase AND

Effluent_flow = decrease THEN Deoxygenation = decrease //strong decrease of deoxygenation;

The experimental results are presented in section 3. SBC-SOIL

SBC-SOIL is a simplified assessment tool for soil erosion risk that assists novice application users in evaluating environmental problems for various local weather and geo-morphological conditions. It was developed by using the elements of universal erosion formula: pluvial intensity (aggression), soil erosion capacity, vegetation and cropping system influence, versants characteristics

Oprea and Dunea/Environmental Engineering and Management Journal 9 (2010), 2, 205-213

208

and adding surface runoff effect (soil permeability and surface slope). Validation and verification steps were done according a special attention to the input and control variables definition, interface conditions definition, rule base structure, inference rules design and their transformation to actions.

Fig. 4 highlights the SBC-SOIL inner structure, consequently the decision tree that provides the evaluation of global erosion risk from a specific location based on the risk analysis of the individual components. The development of SBC-SOIL has relied on the elements that constitute the erosion universal formula, adding surface runoff effects on erosion. The decision table used by SBC-SOIL for the global evaluation of erosion risk is given in Table 1.

The rules from the knowledge base of module SBC-SOIL were generated from several decision tables, as well as from the existing speciality literature (Gavrilescu and Olteanu, 2003; Jetten et al., 2003; Muresan et al., 1992). Production rules were conceived for the following components: soil erosion capacity (9 rules), pluvial aggression (10 rules), vegetation and cropping system influence (13 rules) and versant characteristics (10 rules), adding surface runoff risk estimation (9 rules). The final rule set that evaluates the global erosion risk contains 5 rules.

3. Results and discussions

SBC-AIR

We have selected as modeling exemplification the emission source of Doiceşti power plant stacks (volumetric flows of 1074.6 and 392.5 m3 s-1). In the Doiceşti area, the power plant is emitting pollutants from crude oil and coal combustion through both tall stacks. The power plant has a discontinuous functioning regime (e.g. 384 hours in 2008). De-dusting electro-filters malfunction affects the suspended dusts concentration within the area. In 2008, real monitoring data (200 observations) for suspended dusts showed the exceedence of the 24 hours alert level (105 g m-3) with 22.5% frequency. The maximum value recorded in 24 hours was 166 g m-3 (maximum admissible concentration MAC for 24 h – 150 g m-3) and the minimum value was 11 g m-

3. Some factors such as chimney height, air currents and local topography insure a good dispersion of SO2 and NOx pollutant despite the significant discharged volumes.

The first step of the program is to compute lateral and vertical dispersion coefficients for each atmospheric stability category. The user is asked to input the distance from the stationary source (km) of the point required to assess the pollutant concentration. Plume shapes are comprehensive indicators to estimate and select the atmospheric stability class. Six stability classes for atmospheric condition were considered: Very Unstable, Moderately Unstable, Slight unstable, Neutral, Somewhat stable, and Stable. The next steps are

inputting the emission source characteristics such as: gas exit velocity (m/s), stack diameter (m), gas exit temperature and ambient temperature in Kelvin degrees. Several more parameters are required to be introduced: wind velocity measured at 10 meter height (m/s), the stack height of the emission stationary source (m), and the source emission rate (g/s). Consequently, SBC-AIR computes the ground level concentrations at any receptor point from the emission source on plume centerline in the selected region resulting from each of the isolated sources in the emission inventory. If the composition of the emission from the analyzed stack is known, SBC-AIR provides the pollutant exceeding estimate by comparing the result of the estimated concentration of ground-level pollution on plume centerline at selected distance from source with the standard limit values.

We have selected for simulation two contrasting ambient temperatures (40C and -15C). Ground concentrations were computed at 0, 0.5, 0.8, 1.5, 3, 5, 10, 20, 35, 50, and 75 km distance from the emission source. SBC-AIR provided pollution estimates at each considered wind speed (1, 5, 10, 15, 20 and 25 m s-1) for each stack and pollutant type. The air pollution dispersion analysis system is designed to test different scenarios allowing control of particular industries emissions when air quality monitoring data is not available. We present a case study of emission diffusion assessment for two emission stacks of the Power Plant of Doiceşti, which are the most significant sources in the area of Targoviste city. Table 2 presents the main characteristics of the emission sources, which were used as inputs in the SBC-AIR.

With moderately unstable condition, pollution is transported downward toward ground level. In the looping pattern situation, the sinusoidal path may bring the plume content to ground level close to the emitting source. Consequently, we showed the SBC-AIR results for diffusion of point-source pollution in unstable to neutral conditions, since the stable atmospheric conditions resulted in very low ground concentration within 75 km of source area.

Table 3 presents the maximum pollutant ground concentration (µg m-3) determined for various scenarios showing the distance and the wind speed regime where this concentration might occur. Results obtained through SBC-AIR calculations showed that the highest pollutant concentrations are possible to appear in unstable atmospheric conditions at distances between 1.5 and 3 km for stack 1 (10-15 m/s wind speed), and 0.5 and 0.8 km for stack 2 (20-25 m/s wind speed). Aggregation of these results for the pre-established distances permits the graphical visualization of the pollutant dispersion.

Fig. 5 highlights estimated concentration of ground-level suspended dust (g/m3) on plume central axis at selected distances (km) from Doiceşti power plant Stack 1 for various wind speed at 40C ambient temperatures.

SBC-MEDIU: A Multi-Expert System for Environmental Diagnosis

209

Fig. 4. The decision tree for heuristic assessment of global erosion risk of a specific site

Table 1. Decision table for the global evaluation of erosion risk

Rule 6_1 6_2 6_3 6_4 6_5 Operator OR OR AND AND OR

Soil erosion capacity erod excessive vhigh high reduced/vreduced high Pluvial aggression stand excessive vhigh high reduced/vreduced high

Vegetation veg excessive vhigh high reduced/vreduced high Versant characteristics versant excessive vhigh high reduced/vreduced high

Surface runoff risc excessive vhigh high reduced/vreduced high goal maxim excessive high reduced high

Yellow level 2 /ELSE

Global risk assessment code Display

Red Orange Yellow level 1

IDEAL CONDITIONS

moderate Green code

Table 2. Main parameters of the stack emission sources from Doiceşti Power Plant used in SBC-AIR

No. Stack Source name Gas exit velocity (m/s)

Gas Temperature

(°C)

Stack Diameter

(m)

Stack Height

(m)

Discharged pollutant

Mass Flow (g/s)

Suspended Dusts

107.47

NOx 644.8 1

Power Plant

Doicesti

Power plant generator unit

(470 MW) 25 150 7.4 200

SO2 558.83 Suspended

Dusts 39.25

NOx 196.25 SO2 785

2 Power Plant

Doicesti

Power plant generator

(< 50 MW) 20 165 5 80

CO 98.13

Stack 2 is a significant source of SO2 emissions showing in very unstable atmospheric conditions high ground concentrations at 0.5-1.5 km from the power plant stack. This could be a dangerous situation for the inner-locality residents suffering from asthma or other respiratory illnesses.

According to the national air quality standard, the MAC for 1 hour is 233.34 g m-3 for NOx and for SO2 is 350 g m-3. SBC-AIR can offer support using specific inferences establishing the most possible conditions when MAC could be exceeded:

a. NOx – Stack 2, very unstable conditions, emission rate 451 g s-1 at 40C ambient temperature

( 472 g s-1 at 30C; 493 g s-1 at 20C), 20 m s-1 wind speed.

b. SO2 – Stack 2, very unstable conditions, emission rate 778 g s-1 at 40C ambient temperature ( 813 g s-1 at 30C; 846 g s-1 at 20C), 20 m s-1 wind speed.

More factors are required to be considered in the system development for plume asymmetric propagation such as local topography and building downwash effects. Furthermore, the pollution dispersion algorithm implemented in SBC-AIR needs further calibration with real monitored data. Hourly pollutant records were not available at the time of

Oprea and Dunea/Environmental Engineering and Management Journal 9 (2010), 2, 205-213

210

system development. We will plan a monitoring campaign using GPS and a portable digital analyzer for suspended dusts allowing predicted/recorded values comparisons at selected distances.

SBC-WATER

The SBC-WATER module has been built for use by staff of the local environmental protection authority or of the wastewater treatment plant (WWTP), providing recommendations for complying with the standards and regulations demands concerning water qualitative and quantitative aspects of effluents discharging. The module should advice the analysts for each set of laboratory results either

physical, physico-chemical, chemical, heavy metals, oil residues and pesticides, or colimetry. The experiments were carried out in Targoviste South WWTP, which discharge effluents in the Ialomita River, being situated at 400 m from the riverbed. The monitoring of the effluents was accomplished by discrete sampling followed by chemical analysis in the laboratory.

In this section we present the analysis of the effluent effect on river deoxygenation or reaeration processes. Originating from industrial by-products, runoff, and maltreated waste, poor water quality can lead to health problems for the public and stressful living conditions for aquatic life.

Table 3. SBC-AIR assessment of maximum pollutant ground concentration (µg m-3) for different atmospheric conditions near

Doiceşti power plant stacks – in the brackets: distance from source and wind speed

Ambient Temperature

40°C

Atmospheric stability class

Total Suspended Particles

NOx SO2

Very Unstable 13.59 (1.5 km, 10 m/s) 81.54 (1.5 km, 10 m/s) 74.46 (1.5 km, 10 m/s) Moderately Unstable 11.25 (3 km, 10 m/s) 67.48 (3 km, 10 m/s) 61.62 (3 km, 10 m/s)

Slight Unstable 8.71 (5 km, 10 m/s) 52.28 (5 km, 10 m/s) 47.74 (5 km, 10 m/s)

STACK 1 H = 200 m Φ = 7.4 m

Neutral 3.07 (20 km, 15 m/s) 18.39 (20 km, 15 m/s) 16.81 (20 km, 15 m/s) Very Unstable 17.70 (0.5 km, 20 m/s) 88.50 (0.5 km, 20 m/s) 353. 99 (0.5 km, 20 m/s)

Moderately Unstable 14.72 (0.8 km, 25 m/s) 73.60 (0.8 km, 25 m/s) 294.42 (0.8 km, 25 m/s) Slight Unstable 13.17 (1.5 km, 20 m/s) 65.86 (1.5 km, 20 m/s) 263.45 (1.5 km, 20 m/s)

STACK 2 H = 80 m Φ = 5 m

Neutral 6.04 (5 km, 20 m/s) 30.20 (5 km, 20 m/s) 120.80 (5 km, 20 m/s)

Ambient Temperature

-15°C

Atmospheric stability class

Total Suspended Particles

NOx SO2

Very Unstable 10.55 (1.5 km, 15 m/s) 63.33 (1.5 km, 15 m/s) 57.83 (1.5 km, 15 m/s) Moderately Unstable 9.05 (3 km, 10 m/s) 54.27 (3 km, 10 m/s) 49.56 (3 km, 10 m/s)

Slight Unstable 6.90 (5 km, 15 m/s) 41.38 (5 km, 15 m/s) 37.79 (5 km, 15 m/s)

STACK 1 H = 200 m Φ = 7.4 m

Neutral 2.41 (20 km, 20 m/s) 14.48 (20 km, 20 m/s) 13.23 (20 km, 20 m/s) Very Unstable 14.28 (0.5 km, 25 m/s) 71.42 (0.5 km, 25 m/s) 284.61 (0.5 km, 25 m/s)

Moderately Unstable 11.73 (0.8 km, 25 m/s) 58.64 (0.8 km, 25 m/s) 233.61 (0.8 km, 25 m/s) Slight Unstable 10.62 (1.5 km, 25 m/s) 52.93 (1.5 km, 25 m/s) 211.71 (1.5 km, 25 m/s)

STACK 2 H = 80 m Φ = 5 m

Neutral 4.87 (5 km, 25 m/s) 24.27 (5 km, 25 m/s) 97.10 (5 km, 25 m/s)

0

2

4

6

8

10

12

14

16

0 0.5 0.8 1.5 3 5 10 20 35 50 75

Distance from source on plume centerline (km)

Gro

un

dle

vel

con

cen

trat

ion

(ug

/m3)

1 5 10 15 20 25

0

1

2

3

4

5

6

7

8

9

10

0 0.5 0.8 1.5 3 5 10 20 35 50 75

Distance from source on plume centerline (km)

Gro

un

d l

evel

co

nce

ntr

atio

n (

ug

/m3)

1 5 10 15 20 25

Fig. 5. SBC-AIR estimated concentration of ground-level suspended dust (g/m3) on plume central axis at selected distances (km) from Stack 1 of Doiceşti power plant for various wind speed at 40C

ambient temperatures - very unstable (a), slight unstable (b) atmospheric conditions

SBC-MEDIU: A Multi-Expert System for Environmental Diagnosis

211

Wastewaters contain a large variety of

contaminants (organic, inorganic, microorganisms, and suspended solids), coming from various sources and presenting important fluctuations, both in flow-rates and composition. SBC-WATER allows the analysis of industrial, urban, or sewage network effluents, which can be selected by the user.

The preliminary steps in evaluating the effluent effect on river deoxygenation or reaeration processes are several questions addressed to establish the need for more detailed analysis. User is asked to input the month, and the trends of water temperature, water flow, BOD or DO – dissolved oxygen (increase/stable/decrease) in different types of combinations extracted from the expert knowledge.

The result displays the variation of the water quality: “decrease”, “increase” or “stable conditions” together with a short advice (e.g. “Check BOD/COD ratio”; “Check DO”) and the allocated confidence level CNF. The effluent analysis continues with the estimation of the biochemical oxidation rate of the organic pollutant substances in the river, based on the water temperature and the type of the organic substances (slowly degradable substances or easily degradable substances) allocated by the operator. Each type of organic substance detains a variable deoxygenating rate (k) depending on temperature and time t. Based on these variables, SBC-WATER estimates the efficiency of pollutant oxidation on different time intervals (1-20 days). Fig. 6 presents an example of SBC-WATER module run with the allocated production rule and different analysis results for the biochemical oxidation rate in a polluted river (e.g. 20.6% day-1; 36.9% day-1).

By using this type of knowledge based approach, even if in the process of simplification some information are lost or ignored, the result will retain the essence of the process, making inferences and drawing conclusions from imperfect or incomplete data.

SBC-SOIL

We show a step-by-step example of the SBC-SOIL module run, as follows:

1) Soil erosion capacity is influenced by factors such as the structural soil aggregates dimension and water stability, granular structure, soil volume weight etc. Two factors were considered to empirically estimate this parameter: particles cohesion (low, mild, and high) and soil structure (thin, average, and good).

2) Pluvial aggression is assessed using three variables according to end-user estimation, as follows: rainfall intensity (intense, moderate, and reduced), rainfall duration (long, average, and short) and precipitation quantity (numerical value in mm of rainfall).

3) The vegetation and cropping system or vegetation arrangements influence considered the

grouping of species by erosion sensibility, and cropping system by erosion protection degree.

4) Versant characteristics effect on erosion risk was estimated based on versant length, slope, profile form and exposition.

5) The Surface Runoff Class site characteristic determined from the relationship of the soil permeability class and field slope was adapted from the Soil Survey Manual (1993) and was included in SBC-SOIL in order to increase the system complexity. The erosion risk is established based on runoff classification from soil permeability category that characterize the infiltration process and surface slope. Fig. 7 shows a screenshot of surface runoff analysis.

6) SBC-SOIL provides the final analysis giving comprehensive answers concerning the global erosion risk and brief descriptions of the negative consequences which might occur. Several color codes were considered to facilitate the understanding of final results:

(1) RED CODE – Maximum Risk; (2) ORANGE CODE – Critical Risk; (3) YELLOW CODE - High Risk (with two levels, 1

and 2); (4) GREEN CODE – Moderate or reduced Risk; (5) IDEAL CONDITIONS – Absence of Risk or

minimal Risk. Fig. 8 shows a screenshot with the final analysis result given by SBC-SOIL.

The module SBC-SOIL facilitates the understanding of the complex relationships among the main factors that are responsible for the apparition, development and acceleration of various surfaces erosion process. It has proved to be a versatile diagnosis tool for the global erosion risk, providing the user with the ability to perform a posteriori, detailed analysis of the main erosion factors that presents high risks, using SBC-SOIL resources from its knowledge base. 4. Conclusions

Environmental components (air-water-soil) are

characterized by the high complexity of the involved processes, which are difficult to be translated into deterministic models. The SBC-MEDIU multi-expert system provides an integrated decision support tool for environmental management, solving problems such as surface water quality analysis, air pollution analysis, and soil erosion analysis.

The system can be used as an educational tool for the students that study the environmental protection domain.

The preliminary use of this system might also facilitate the environmental monitoring-network design, control strategy evaluation or control-technology evaluation.

Oprea and Dunea/Environmental Engineering and Management Journal 9 (2010), 2, 205-213

212

Fig. 6. Estimation of the biochemical oxidation rate of the organic pollutant substances in a river based on the water temperature and the type of the organic substances

Fig. 7. Surface runoff analysis – screenshot of system run

Fig. 8. Example of screenshot with the final analysis result given by SBC-SOIL.

At the level of local Environmental Protection Agencies, SBC-MEDIU can be integrated in a Local Monitoring Plan to provide useful information for decisional support and reliable answers to:

standards/regulations demands concerning qualitative and quantitative aspects;

trends of environmental quality modification due to various factors;

impact of deterioration on ecosystems; and the efficiency of strategies and

management action for pollution control. As a future work we shall verify the

performance of the knowledge bases on more scenarios, with interdependencies between air, water and soil pollution, and we shall increase the system complexity in order to become an alert system serving to signal when potential pollution is high for water, air or soil, requiring interaction between control agencies and emitters.

The system will serve to locate areas of expected high concentration for correlation with health effects, or to identify environmental pollution issues.

Acknowledgements The research reported in this paper was partially funded by the Romanian Education and Research Ministry and the National Council of Academic Research (UEFISCSU-CNCSIS) under the postdoctoral research programme CEEX19-1533/2006. References Bellamy P.H., Jones R.J.A., (2007), Identifying Changes in

Soil Quality: Contamination and Organic Matter Decline, In: Air, Water and Soil Quality Modelling for Risk and Impact Assessment, Ebel A. and Davitashvili T. (Eds.), Springer, 271-279.

Cortés U., Rodríguez-Roda I., Sànchez-Marrè M., Comas J., Cortés C., Poch M., (2002), DAI-DEPUR: An Environmental Decision Support System for control and supervision of Municipal Waste Water Treatment Plants. Proc. of Eureopean Conference on Artificial Intelligence, IOS Press, Lyon, France.

Friedrich S., Gargano M., (1989), Expert system design and development using VP-expert, Wiley, London.

Gavrilescu E., Olteanu I., (2003), Environmental Quality. Analysis methods (soil), (in Romanian), Universitaria Publishing House, Craiova.

Gibert K., Sànchez-Marrè M., Flores X., (2005), Cluster discovery in environmental databases using

SBC-MEDIU: A Multi-Expert System for Environmental Diagnosis

213

GESCONDA: The added value of comparisons, AiCommunications, 18, 319-331.

Jetten V., Govers G., Hessel R., (2003), Erosion models: quality of spatial predictions, Hydrological Processes 17, 887-900.

Kim Y. J., Platt U., (Eds.), (2007), Advanced Environmental Monitoring. Springer.

Mendes B., Oliveira J.S., (2004), Water quality for human consumption, (in Portuguese), Lidel Edições técnicas, Lisboa – Porto – Coimbra.

Mureşan D., Plesa I., Onu N., Savu P., Nagy Z., Jinga I., Teodoriu Al., Paltineanu P., Toma I., Vasilescu I., (1992), Irrigations, Draining, and Soil Erosion Control, (in Romanian), Didactical and Pedagogical Publishing House, Bucharest, Romania.

Nicolescu C., Gorghiu G., Dunea D., Buruleanu L., Moise V., (2008), Mapping Air Quality: An Assessment of the Pollutants Dispersion in Inhabited Areas to Predict and Manage Environmental Risks, WSEAS Transactions on Environment and Development, 4, 1078-1088.

Oprea M., (2005), A case study of knowledge modelling in an air pollution control decision support system, AiCommunications, 18, 293-303.

Oprea M., Sànchez-Marrè M. (Eds.), (2004), Binding Environmental Sciences and Artificial Intelligence, Proceedings of the International Workshop Binding Environmental Sciences and Artificial Intelligence, 4th Edition, Valencia, Spain.

Oprea M., Dunea D., (2008), Modelling a Surface Water Pollution Analysis System with a Knowledge-based Approach. Proc. of European Meetings on Cybernetics and Systems Research, Vienna, vol. 2, 585-590.

Platt U., (2008), Air Pollution Monitoring Systems – Past - Present - Future, In: Advanced Environmental Monitoring, Kim Y.J. and Platt U. (Eds.), Springer.

Rodríguez-Roda I., Sànchez-Marrè M., Cortés U., Comas J., and Poch M., (2001), Development of a case-based system for the supervision of an activated sludge process, Environmental Technology, 22, 477-486.

Rubio M., Colomer J., Ruiz M., Colprim J., Melendez J., (2004), Qualitative Trends for Situation Assessment in SBR Wastewater Treatment Process, Proc. of the 4th Int. Workshop BESAI, 51-59.

Sànchez-Marrè M., Gibert K., Rodríguez-Roda I., and et al., (2002), Development of an intelligent data analysis for knowledge management in environmental data bases. Proc. of the 1st Int. Conf. of Int Emergency Medical Services Society, Lugano, Suiza, 420-425.

Schnelle K.B., Brown C. A., (2002), Air Pollution Control Technology Handbook, CRC Press, Boca Raton.

Soil Survey Division Staff, (1993), Soil survey manual, Soil Conservation Service, U.S. Department of Agriculture Handbook, 18.

Vellido A., Olier I. Marti E., Comas J., Rodríguez-Roda I., (2004), STREAMS project: Exploration of the ecological status of Mediterranean rivers using Generative Topographic Mapping. Proc. of the 4th Int. Workshop BESAI, 151-152.

Vogmahadlek Ch., Satayopas B., (2007), Applicability of RAMS for a Simulation to Provide Inputs to an Air Quality Model: Modeling Evaluation and Sensitivity Test, WSEAS Transactions on Environment and Development, 8, 129-138.

Vouros G. A., Pantelakis I.S., Lekkas T.D., (2000), Knowledge representation in an activated sludge plant diagnosis system, Expert Systems, 17, 226-240.

Wieland D., Wotawa F., Wotawa G., (2002), From neural networks to qualitative models in environmental engineering, Computer-Aided Civil and Infrastructure Engineering, 17, 104-118.