fm

ª Œª Œ 27ƒ 6D Á 2007 11œ pp. 809~817 ˆ «~ ª m y d w» A Modified Technique for Prediction of Future Land Use Change Á½ Lee, Yong JunÁKim, Seong Joon Abstract The purpose of this study is to suggest a prediction method of land use change by modifying technique. In the modified method, a logarithmic function was reflected for the trend of past land use change of each item. Data of water quality protection area and green belt area were considered to include systematic factor. In addition, the minimal preserved probability that is the percent of upper limit of land use change between land use classes in the process of prediction was applied to prevent unrealistic prediction of future land use. The prediction results of original and modified were evaluated by comparing indices (α: the ratio of matched cell number of the predicted to the total cell number of the known, β: the ratio of matched cell number of the predicted to the total cell number as sum of sets of the known and the predicted, γ: the ratio of cell number of the predicted to the cell number of the known) which compare the spatial fit between the known and the predicted. For Anseongcheon watershed (371.1 km 2 ), the 2000 land uses were predicted using the 3 past land use data (1985, 1990, 1995). The values of α, β, and γ for urban area were 0.69, 0.63, 0.80 for the modified and 0.52, 0.45, 0.68 for the original respectively. Keywords : land use change, markov chain, cellular automata, prediction, m d»» w wš m w. w m w m ƒ w LOGy» w, y ù p y Á z w. w, x m d w» w ƒ w w y w y (Minimal preserved Probability) w. w α, β, γ (α: m ew», β: m ew» ù d γ : d» ) œ e ùkü, (371.1 km ) 2 1985, 1990, 1995 m 2000 w, CA- Markov 0.69, 0.63, 0.80 š»» 0.52, 0.45, 0.68 œ e ƒ y w. w : m y, Markov Chain, Cellular Automata, d, 1. ù w ù, w y,, w m y, w w xk y ã w yƒ. m y y ql dw w, p y v * z Á w w zy lœw ** z Á Á w y w w zy lœw (E-mail : kimsj@konkuk.ac.kr) 27ƒ 6D 2007 11œ 809 m dw y w ù ù z ƒ ( m, 2000). m y ƒ ƒ, w w ùkú m yƒ w mw k w»» (, 1994). ù, m y p w x d w

w w. m d š s ƒ l v w, y yƒ e w dw p y w z w w z w. x m k wš xk w yƒ w» w w. w w m ƒ y k» (Remote Sensing Technique) w w w. š e w œw. w k» w wš x y w d w m y d z w. m y d w w ½ y (1994) Markov w x q w w w m dw, ½ z(2002) GIS Logistic z w «m y w e w ü w, y(2002) CA(Cellular Automata) x w m y d w. (2003) Landsat l k Markov Chain» w m y dw. ½ (2005) mw m y r. m (2006) m w SD(System Dynamics)» GIS» œ» ww m x m y w. ½ (2007) ³ œ w» w. m y w û m p w œ Markov w Wood x. Turner(1987) œ w w m w yw œ w š (Spatial Influence Algorithm) w. Clake(1996) x w» w CA w. w Clake(1997) MAIA (The Mid-Atlantic Integrated Assessment) CA kp (Deltatron) ww y w. Mundia(2006) fÿ w y w Landsat w w. Thomas(2006) M-CA» w v g p- m d w.» w mw w m w p w z œ w (Transition Matrix) w m d ƒ w» g k m d wš w. 2.»» Markov Chain CA(Cellular Automata) ww». Markov Chain e» GIS l l, x m y w w ƒ š w y (transition probability) w wš e w» ƒ w, m v y w ùkü». œ w w w ƒ y š w w ys ƒ œ. CA» w w, CA x x œ y w, w w œm k k y k. y k œ w w p xk w. 2.1 Markov» Markov Chain ƒ š p w y dw» w w». Markov l k w x l kƒ w š k w x w Markov Chain y ƒ wš. l y w y y l p» w k k w y w. sww w w y ƒ w ƒ y (X t ) y (stochastic process) w. y X t (t = 1, 2,...) t k(state) w(s 1, S 2,..., S k ) wù x š w, k S i S j ƒ (step) š, S i S j y, y P ijƒ k S i š k ƒ y Markov Process w. { } P P ij = P PX n = S j X n = 1 S i { } = P PX n = S j X n = 1 S j y P ij w P w, P = P 11 P 12 P 1k P 21 P 2k P k1 P k2 P kk w ùkù», w. k P ij = 1 j = 0 (1) (2) (0ßP ijß1) 810 ª Œª Œ

2.2 CA(Cellular Automata)» Cellular Automata p w ù w» w 1963 John von Neuman Stanislaw Ulam w cellular space. z CA w, yw, w, w, ful œw sww w œw w x w. GIS CA GIS w d y š. CA œ (discrete time and space)» w x y l (Gutowitz, 1991). Cellular œ CA» w j» w ww j» œ w, ƒx, ƒx, ƒx xk š. k» ƒ k ùkü ƒ ƒ. (focus cell) wš w w m, w y ƒ w. y³e ƒƒ» yw ³ w» eƒ. y³e w œm³e p š CA ƒ w. k w ù w w k w w. œm š y³e w k yw š, k» k. CA w y w ƒ š. 3. w œ d w, 371.1km 2, w 26.8km, s³ s 14.0km, s³tš 113.6m, s³ 10.6%. w, ƒ ƒƒ 90.2%, 1.0%, 8.7% w ù 20 œ, k, v, ü m yƒ ã. ü š Ÿ swwš ( 1). 4. 1. 4.1»» x m w m k d w. w ù 70-85 w y m yƒ j ù 85-x ( 2)¾ m y w w yw š. m dw» w x» m š w» m š w m dw w q. w (Maximum Likelihood Classification) mw m w, y r ƒ j (, 2003). w mw m training site w ƒ ƒ w y wš y ƒ j w w q w j w w ƒ w w» y w ù ù y w ƒ w ( 3). 2. ü m y (1985~2000) 27ƒ 6D 2007 11œ 811

IV. m» 75% w. V. y ù p y œ ë. 3. (Image Processing and Interpretation -Morro Bay, Califonia, 2006) m y w w.(burnham, 1973) w» w m»» zù š w. ww w ù y ë w m yƒ ù»» w š w. w w w» w ƒ wš w. I. m y LOG z x y w. II. y w. III. ù ù y m y w. w w» w 4 ƒ m v w LOG z y v mw m v w w. x ƒ w w m» m yƒ x w e š ƒ w» x. w LOG z y ƒ m w m y w m y z ƒ w ƒ š» m v w m w w. w m v» ƒ» w vl m v» w, t 1 ƒ m w 75% m w yw š m w. ƒ m w 75% y w (t 1, 5). zù w» w II~III ƒ ³e w, ƒ m w y ù p y m yƒ w w w ( 6). š w 7. m 4. 1985~2000 LOG z y ù t 1. 1985~2000 1985 2000 ( % ) water 6.45 6.13 105.36 urban area 12.51 42.11 29.70 bare field 4.31 15.43 27.97 grassland 9.42 29.09 32.40 forest 269.78 251.72 107.17 paddy field 236.17 189.45 124.67 upland crop 42.65 47.37 90.04 average 74.90 812 ª Œª Œ

4.2 y LOG z y ƒ yw» w w mw ƒ w ƒ w m kƒ y y w. z ƒƒ m w w. ³e(Transition Rules) w yƒ q w. z w m w w ù m y ƒ m w z LOG z y ƒ š w m y j. y 8. 5. w w ³e(Transition Rules) y (Minimal preserved Probability) w z yy w w w. z m w w (Moore Neighbourhood Cell) w m y w z m y w LOG z y ƒ w y w. 4.3 m m m w w» xy m k 1985 l 2000 ¾ 5 w. m Landsat ETM + Landsat TM (Maximum Likelihood Supervised Classification) w 7 (1., 2., 3.ù, 4., 5., 6., 7. ) w ( 9). 6. y p 7.» š 27ƒ 6D 2007 11œ 813

그림 8. 최적화의 흐름도 그림 9. 최적화의 흐름도 5. 결과 및 고찰 5.1 개선된 검증 본 연구에서는 개선된 기법이 실세계를 얼마나 반영하는지 아래와 같은 방법을 통하여 검증 하였다 (주용진, 2003). α는 각 항목별 추정된 셀과 기준이 되는 토지이용과의 일치율을 나타내는 지수이고 β(a shape index, 1997). 예측 모형은 과거의 통계적 성장 패턴과 일치 시킬 수 있 어야 하며 추정치가 현실을 반영 할 수 있어야 한다(Clarke, a measurement of spatial fit between the model s growth 814 大韓土木學會論文集

and the known. urban extent for the control years) ƒ w m» m œ ù ew ƒ ùkü (Lee and Sallee, 1970), γ d m Á w. α, β, γ 1 ƒ¾ m œ ew š. m ew α = ----------------------------------------------------------------------------------------------------- (3)» m ew β = ----------------------------------------------------------------------------------------------------------------- (4)» m m γ = -------------------------------------------------------------------------- (5)»» w y w 1990, 1995 m w IDIRISI v ü w 2000 m dwš,» d 2000 m w.» 2000 w» xy m w ( 10~11). 10 (a)~(b)» CA- Markov» ù, p œ e ƒ, d m Á w 11» e ùkü. m w 10. 2000 w» xy m» α, β Modified t 2. Modified,, (0~100%) (0~100%) (0.00~2.00) 11. γ (1 ƒ¾ e) Modified Modified water 84.8 69.1 78.7 68.3 0.93 0.70 urban 69.9 51.8 63.4 44.7 0.80 0.68 bare 61.7 22.0 57.1 20.6 0.70 0.29 grass 19.8 20.2 10.1 10.5 1.16 1.12 forest 80.5 81.1 67.2 67.0 1.00 1.02 paddy 66.5 65.0 47.8 45.0 1.06 1.09 crop 16.0 18.3 7.9 8.1 1.17 1.44 27ƒ 6D 2007 11œ 815

12.» d m œ e ƒ û r m d, m ƒ» w y ƒ û ù (t 2). 5.2 1985, 1990, 1995, 2000 w» xy m m» w m dw» m w m y r.( 12). ù ã w ƒw w ùkþš, ù, w w ùkü. 6. m dw x» ù w w w w r š ƒ ƒ mw x w w z m y d w. w, ù m y w q w m w m» w ³e w. w ³ y p sww w.,» w mw w m w p w z œ w (Transition Matrix) mw m dw.,» œ s mw w y w. m w w r. d w x ¾ w ƒ š q w. 2000 ¾ ƒw d ƒ w w w. 90 z l v û, v l 816 Á z ql w». w»» x g m y w w w. œ e w α, β, γ ƒƒ 69.1%, 63.4%, 0.80» CA- Markov (51.8%, 44.7%, 0.68) œ e ƒ y w. w m y, e y ƒ» w m» ƒ ƒ š w w x w w sww š. ywš w m z ƒ, e Á z š w y m dw». 2007 ( w» ) w w w (No. R01-2006-000-10343-0). š x m (2000) z y (IV),»z y z w sƒ:23-30. m (2006) œ mw m x (II). ½, (2007) ³ œ w ƒ» w m d e w, w wz, w wz, 10«2y, pp. 58-70. ½, y,, ½ (2005) k y w ƒy y w, w wz, w wz, 8«2y, pp. 23-30. ½ y(1994) œ m v d w x w, w GISwz, Vol. 2, No. 1, pp. 47-51. ½ z(2002) GIS p z w m y, m, 33«, pp. 153-164. (1994) w, :1-23., w (2001) 1985-2000 w, m z, Vol. 37, No. 4, pp. 41-54., y,, ½ ¼(1999) LANDSAT TM JERS- 1 OPS w m y, w wz, w wz, 2«1y, pp. 73-84. y(2002) m k x w m y d, m z, 37«4y, pp. 125-132., ½ (2003) šw w l, wz, 9«3y, pp. 301-301., y(2003) w m y d y x x, wz, w wz, 37«4 y, pp. 373-385. (2003) w m y d y x x, w, w w. w,,, (2002) z ³ JERS-1 m v y sƒ, w wz, w wz, 5«1y, pp. 27-38. http://rst.gsfc.nasa.gov/sect1/sect1_19.html Burnham, B.O. (1973) Markov intertemporal land use simulation model. Journal of Hydrology, Southern Journal of Agricul- ª Œª Œ

tural Economics, pp. 253-258. Clark, S., Starr, J., Foresman, T. W., Prince, W., and Acevedo, W. (1997) Development of the temporal transportation database for the analysis of urban development in the Baltimore-Washington region. Processing of ASPRS/ACSM Annual Convention and Exhibition, Baltimore. MD, April 22-24. Vol. III, pp. 101-110. Clarke, K.C., Hoppen, S.S and GaydosS L. (1996) Methods and techniques for rigorous calibration of a cellular automaton model of urban growth. Third International Conference/Workshop on Integrating GIS and Environmental Modeling. Santa Fe, New Mexico, January 21-25. Santa Barbara: National Center for Geographic Information and Analysis. Mundia, C.N. and Aniya, M. (2006) Dynamics of Landuse/cover changes and degradation df nairobi citiy kenya, Land Degrad. Develop, Vol. 17, pp. 97-108. Gutowitz, H. (1991) Cellular Automata : Theory and Experiment, MIT Press, Cambridge. Lee, D., Thomas, T., and Sallee, G. (1970) A method of measuring shape. Geographical Review, Vol. 60, pp. 555-563. Turner, M. G. (1987) Spatial simulation of landscape change in Georgia, A comparison of three transition models. landscape Ecology, Vol. 1, pp. 29-36. Houet, T. and Laurence, Hubert-Moy (2006) Modelling and projecting landuse and landcover changes with a cellular automaton in considering landscape trajectories: An improvement for simulation of plausible future states, EARSe, Vol. 5, pp. 63-76. ( : 2007.7.12/ : 2007.10.4/ : 2007.10.4) 27ƒ 6D 2007 11œ 817