As　an　extension　of　traditional　data　mining,　distributed　data　mining　(DDM)　is　a　well-known　issue　that　has　been　extensively　studied　to　extract　some　hidden,　useful　and　interesting　information　from　data　by　using　various　kinds　of　resources　distributed　in　different　geography　positions.　In　the　literature,　there　are　many　methods　and　technologies　proposed　to　implement　large-scale　distributed　data　mining,　but　owing　to　lack　of　autonomy,　intelligentization　and　the　limitation　of　expansibility,　most　of　the　works　fail　to　reach　a　practical　application　in　industry　and　commerce　fields.　With　the　rapid　development　of　distributed　sharing　infrastructures　and　the　emergence　and　mature　of　grid　technology,　providing　knowledge　services　in　grid　environment　is　becoming　the　trend　of　knowledge　discovery,　and　it　will　certainly　make　the　technology　of　data　processing　into　a　broader　way.　In　this　paper,　we　elaborate　on　a　geographical　knowledge　services　grid　test-bed,　which　is　developed　to　support　distributed　&　parallel　geological　problems　solving,　as　an　example,　a　distributed　spatial　outliers　mining　algorithm　based　on　Delaunay　triangulation(D-TIN)　is　introduced,　and　the　perform　scheduling　of　Delaunay　triangulation　construction　and　spatial　outliers　mining　is　presented　too.