successive, the medium rules always use fuzzy 
methods to classify them. Here is a sample 
example, from grassland to forest, tree 
gradualy substitute grass, we can not find a 
absolute boundary to delimite grass and 
forset, so in order to classify them, fuzzy 
mathematics are used to build fuzzy 
classification model. 
(3) The advanced level is a series of 
integrative geographical planning rules, 
always is a set of experience of geographer. 
MCGES expresses geographical knowledge with 
the form of production-rule. The BNF 
defination of MCGES knowledge is as 
following: 
(1) The basic level 
<factor>::=<factor name><factor grading> 
“factor grading>::=<factor grade» 
«minimum of grade» 
Xmaximum of grade»? 
“factor grade>::=0 | 1 | 2 | ... 
(2) The medium and advanced level 
<rule>::=IF «evidence? THEN «conclusion» 
<effectiveness> 
<effectiveness>::= 
<conclusion prior probability> 
{<evidence probability 
if conclusion exist> 
{evidence probability 
if conclusion not exist>} 
<evidence> : :=<condition condition> AND 
«condition | condition» 
€condition»^::-«factor name»«factor grade»? 
«conclusion?::-«temporary conclusion»? | 
«final conclusion»? 
<temporary conclusion>::=<geographical type> 
<geographical value> 
<final conclusion”: :=<geographical process> 
<geographical type>::=geographical typel | 
geographical type2 | ... 
<geographical value>::=geographical valuel | 
geographical value2 | ... 
{geographical process>::= 
geographical processl 
geographical process? ee 
Here is an example of MCGES basic knowledge, 
#factor-1: /* from MCGES for Kouhe Soil&Water 
Conservation Expert System */ 
{ 
#name: slope; /* slope grading */ 
1: 0,2 ; /* grade 1 is 0 - 2 */ 
2: 2,5. ; /* grade 2 is 2 - 5 */ 
3: 5,8 ; /* grade 3 is 5 - 8 */ 
4: 8,15 ; /* grade 4 is 8 -15 */ 
5: 15,25; /* grade 5 is 15-25 */ 
6: 25,- ; /* grade 6,if >25 */ 
} 
266 
The effectiveness of rules points out the 
corelation between evidence and conclusion. 
P(C), P (EIC) and P(E!^C) are used to express 
conclusion prior probability, evidence 
probability if the conclusion exists and 
evidence probability if the conclusion does 
not exist. 
Examples of rule are shown as following, 
$rule-17: /* from MCGES for Kouhe Soil&Water 
Conservation Expert System */ 
{ 
#if: slope == 3, 
soil depth -- 3, 
erosion type zz water erosion, 
erosion density -- 2, 
landuse == 1; 
/* landuse type is cultivated land */ 
#then: terrace ; 
/* Using terrace to conserve S&W */ 
#effect: 0.22, 
/* terrace prior probability */ 
0.40,0.45, 
/* slope : P(E 1 C), p(E 1 "C) */ 
0.65,0.70, 
/* soil depth : P(E ! C), p(E ! "C) #*/ 
0.43,0.67, 
/* erosion type : P(E ! C), p(E ! "C) */ 
0.21,0.74, 
/* erosion density:P(E ! C),p(E ! "C) */ 
1.00,0.85; 
/* landuse : P(E ! C), p(E ! "C) */ 
frelative: rule-29,rule-36,rule-81, 
rule-90,rule-103; 
/* rules connected with #rule-17 */ 
} 
#rule-81: /* from MCGES for Kouhe Soil&Water 
Conservation Expert System */ 
{ 
#if: slope == 3, 
soil depth == 3, 
erosion type -- water erosion, 
erosion density -- 2, 
landuse -- 8; 
/* uncultivated land */ 
#then: tree planting; 
/* planting tree to conserve S&W */ 
feffect: 0.18, 
/* tree planting prior probability */ 
0.15,0.35;, 
/* slope : P(E;C), p(E}"C) */ 
0.45,0.27, 
/* soil depth : P(E!C), p(E!^C) */ 
0.43,0.43, 
/* erosion type : P(E!C), p(E!^C) */ 
0.15,0. 74, 
/* erosion density : P(EIC), p(E!"C) */ 
0.00,0.90; 
/* landuse : P(E!C), p(E!"C) */ 
  
The 
otl] 
col 
Sp: 
the 
knc 
MCC 
in 
3.1 
We 
to 
gra 
inf 
geo 
3.2 
The 
cal 
geo 
the 
the 
the 
Now 
dat. 
dat: 
geo; 
less 
real 
mat] 
succ 
usec 
mode 
we | 
same 
func 
dist 
band 
ob 
of M 
fu 
wh 
ju 
Ai 
whe 
the