1. Conclusion
The spatial decision support system for coal mine reclamation in this study was designed to provide decision-makers with supportive information for better planning practices. New technologies developed in geographic information system (GIS) and also techniques from other fields were utilized in collecting information and in building the system. The revised universal soil loss equation (RUSLE) model was introduced to calculate the erosion rate factor. Image processing was used to extract useful information for the system, and multi-criteria decision making was adopted as the model for decision making. All of the techniques were integrated into a GIS environment.

This system was used in the South Fork Patoka River watershed to assist in the abandoned coal mine reclamation activities there. This area has been severely interrupted by strip mining in the past 60 years. The environment has been degraded resulting from the extensive mining activities. Reclamation became extremely important for the improvement of the environment while a lot of abandoned mines are still causing problems in the area. This system endeavors to help the agencies identify those sites, with their priorities for reclamation, depending on the degree with which they are affecting the environment within the whole area. In this way, it provided insights into the reclamation planning.

Three physical criteria were selected for generating the priority map for an abandoned coal mine reclamation in the study area. They were erosion rate, proximity to streams, and soil acidity.

Data collection and processing were the initial and critical steps.  Data were acquired from various sources in different formats. To be compatible with each other in the same system, the data were converted into identical projection systems, coordinate systems, and file formats. The projection system for this research was that of the Universal Transverse Mercator (UTM) type, and the map unit selected was the meter. File formats were in standard ArcInfo formats such as coverage for vector data and grid for raster data. ArcInfo and ArcView software were used for the data collecting and editing. Some of the data were not directly used in the system, however they were helpful in analyzing other data.

The erosion rate factor was calculated using the Revised Universal Soil Loss Equation (RUSLE) model which was composed of five major factors. They are:

The rainfall factor was extracted from the CITY database attached to the RUSLE program.  The soil erosivity factor was obtained from the digitized soil map.  The slope factor was calculated from the Digital Elevation Model (DEM) using the ArcInfo Macro Language (AML), thus benefiting from the Open Development Environment (ODE) technology.  The cover management factor was acquired by analyzing two sets of Thematic Mapper (TM) images by means of image processing in the environment of the ERDAS Imagine. The conservation management factor was considered as a constant in the whole area.

Soil acidity was another criterion. The acidity data were directly derived from a digitized soil map.

The proximity to streams criterion was scored using ArcInfoís Ďbufferí function and conditional commands.

Since the Multi-Criteria Decision-Making (MCDM) technique was used as a decision-making model, all the criteria had to be scored, weighted and analyzed. The method of scoring a criterion was to assign values to the criterion attributes according to their quality. Since the spatial decision support system for coal mine reclamation was to generate the priority map for reclamation and a higher value represented higher priority, the attribute that was leading to more environmental pollution was assigned higher value.  The aim of weighting a criterion was to evaluate the importance of the criterion in the context of the whole set of criteria, and then assigning to it a weight value according to its importance. The Weighted Linear Combination (WLC) algorithm was chosen to perform the evaluation. Since the MCDM required that all the criteria scores had to be standardized and weights of all criteria summed to be 1, data normalization was employed. As the final output value range was set to be from 1 to 10, each criterion score was then standardized between 1 to 10 before the WLC algorithm was applied.

The user interface was built for easy access and easy operation. Delphi software was selected to establish the main frame of the system. A Multiple Document Interface (MDI) was used for viewing multiple windows simultaneously. Displaying multiple images within the system was accomplished by applying MapObjects software. The interaction of the user and the system was very convenient.  The user input his or her preferences on criterion scoring and weighting, and the system took the input and processed the requirement.  A final priority map was then created from the system by analyzing the userís preference.

Two sets of weighting matrices were imported into the system and two priority maps were generated. The results indicated that userís inputs or preferences played important roles in the final result. The final map reflected the priority of reclamation based on the userís scoring and weighting of the criteria. It provided the overall information of the whole study area, and it definitely guided the decision-makers to make more reasonable planning strategies.

Overall, the significance of the research can be concluded as providing:

    1. The integration of all processes in a single system, which can be served as a prototype for any spatial decision support system for reclamation planning.
    2. A general systematic view for study areas, which will prevent overlooking some important area that might be thr case traditional methods.
    3. A visualization and comparison method in a multiple display environment, which provides a better presentation of data.
    4. An easy and user-friendly interface, which allows users to accomplish their tasks with little difficulty.
Limitations of current version of the SDSS are:
    1. Fixed number of criteria.
    2. Since the modeling window was built with three tab forms for criteria, no more criteria could be added into the system. However, a user can change the criteria.

    3. Data inaccuracy.
    4. The soil map for the study area was digitized from older versions of soil survey maps .  This provided someXXX inaccurate information. Another factor affecting the data accuracy was that soil survey maps tend to generalize soil properties in coal mine affected areas without detailed information, which, as a result, affected the precision of the soil data.

    5. No social factors included.
In this study, because of lack of social data sources, only three physical factors were considered to determine the priority for reclamation. Even though the system provided a means to display any data available, the unavailable social data could not be used as input to the system. In general, the Spatial Decision Support System (SDSS) implemented in this research provided an easy and integrated environment for decision-makers to obtain alternatives for reclamation based on evaluating possible physical criteria. The final output map of reclamation priorities, generated according to the preferences, identified the sites that were to be reclaimed on the basis of their environmental significance. It may serve as a basic guide for environmental agencies in the future to manage mine reclamation programs. However, since the criteria of this study were not adequate to cover all possible factors relating to mine site reclamation, the results XXX needXX to be examined carefully by these agencies, with modification of the final model being done as deemed appropriate. Also, many economic, social and political variables were not easily controlled, and, in reality, could not be accurately covered in this system.

Future improvements upon this version of a spatial decision support system could be made in the following areas:

    1. More flexible user controls.
    2. It would be more efficient if the user could have more controlX over the system.  For example, the user could have control on the numbers of criteria, the end points for standardization of scores, and the range of standardized values.

    3. Adding more social, economic and political data.
    4. If social data are available, an entry for each factor should be made in the system so XXX the system could take these factors into consideration.

    5. Improving the accuracy of data.
Checking the availability and accuracy of data sources before any research starts. The main lesson learned in this research is that the data collection was overwhelmed by system building, therefore, the data accuracy and availability were not checked before hand, just taken for granted. In fact the soil data was not as accurate as expected and it caused some inaccuracy in the final calculation.