Transcript Predicting land use changes in the Lake Balaton catchment (Hungary)

Predicting land use changes in the Lake
Balaton catchment (Hungary)
Van Dessel Wim1, Poelmans Lien1, Gyozo Jordan2, Szilassi Peter3,
Csillag Gabor2 , Van Rompaey Anton1
1
Physical an Regional Geography Research Group, K.U.Leuven, Belgium
2
Geological Institute of Hungary, Hungarian Geological Survey, Hungary
3
Szeged University, Deparment of Geography, Hungary
International Workshop: European Union Expansion: Land Use Change and Environmental Effects in Rural Areas
Introduction
•
Land use changes: caused by socio-economic evolution (often at a
macroscale): political decisions, econmic development, changing lifestyle
•
Land users determine the spatial pattern of these land use changes
•
e.g. Due to economic or political pressure a farmer can be forced to take land
out or in production
•
Wich parcels will be chosen depends on a lot of criteria (ex. soil parameters);
personal experience and motivation of the farmer
•
Evalutation of the quality/characteristics of the parcels
•
Can we model the behaviour of the farmer and simulate the spatial pattern of
his decisions?
Situation Study Area
Pécsely Basin: 24 km²
Objectives of the study
• Which land use changes have recently occured and where?
• Determine the landscape characteristics influencing the
spatial pattern of the land use transitions
• Can we use information from past changes to predict
patterns future land use change?
• Investigate the impact of recent and future land use
changes on soil erosion and sediment yield
Method
• Satellite images (spatial pattern of the changes; resolution 30m)
• 1992: Landsat 4 Thematic Mapper
• 2003: Aster
• Aerial photographs
(parcel size, …)
Digitizing test sites
Supervised classification
77% accuracy
• Physiographic characteristics
Topography
Pécsely
Vászoly
Land Use
Pécsely Basin
2003
2003
Arable land
Pasture
Vineyard
Forest
Build up area
Land use
Arable land
Pasture
Vineyard
Forest
Buid up area
Total
3%
ha
492.12
632.35
271.51
961.63
75.11
2432.72
20%
%
20.23
25.99
11.16
39.53
3.09
100
arable land
pasture
40%
vineyard
• Based on Aster satellite image
26%
11%
forest
build up area
Historical land use changes
Land use around Pécsely in 1955 (a) and 1971 (b)
Source: Museum of Military History, Budapest
Historical land use changes
•
•
•
•
•
•
•
•
•
1949:
1952:
1955:
1956:
1957:
1961:
1968:
1989:
1994:
Start collectivisation
Opposition against collectivisation
Collectivisation
Revolution against collectivisation
Flexibilization
“Complete” collectivisation (90%)
New economic mechanism (more independent farms)
Republic
Privatization
Farmers can claim their land back
Recent land use changes
• Comparison of satellite images and aerial
photographs
• Construction of land use transition maps
• Analysis of the characteristics of the
transition zones
• Calculation of conditional transition
probabilities
Recent land use changes
3%
1992
20%
arable land
pasture
40%
3%
2003
pasture
vineyard
forest
forest
26%
build up area
2%
11%
arable land
40%
vineyard
26%
20%
build up area
3%
11%
20%
20%
30%
40%
22%
26%
26%
Based on landsat satellite image
11%
Based on Aster satellite image
Arable land: equal area; smaller parcels
Heterogeneous pattern
Evolution in land use (1992 –2003)
No change
Changed to arable land
Changed to pasture
Changed to vineyard
Changed to forest
Build up area
Land Use Changes (in Ha)
1992
arable land
pasture
vineyard
forest
build up area
total
arable land
176
108
164
39
5
492
pasture
166
168
220
74
4
632
2003
vineyard
46
56
144
24
2
272
forest
101
181
102
578
1
963
build up area
6
13
6
0
49
74
total
495
526
636
715
61
2433
5 changes in red represents 72% of all changes
Unchanged: 1066 ha (44%)
Actual land use changes
• Arable land
Pasture
166 ha
• Pasture
Arable land
Forest
108 ha
181 ha
• Vineyard
Arable land
Pasture
Forest
164 ha
220 ha
102 ha
Statistical analysis
• Parameters
•
•
•
•
•
Hight
Slope
Soil texture
Distance to road
Distance to village
Statistical analysis
• The problem: which factors control land use changes ?
• Relative importance of different factors ?
• Prediction of future land use changes ?
Which variables contribute significantly to the land use change pattern
Chi-square analysis
Logistic regression
Conclusion Statistics
• Which physical and “infrastructure” parameters
determine the spatial pattern?
Small differences observed between both methods
because the first one handles with categorical
variables and the second one with continuous
variables. Chi-square analysis handles each factor
separately.
Transition Probabilities
• Based on the logistic regression analysis
• Transition probability map for each type of
land use conversion
Probability map arable Probability
land to pasture
map pasture to arable
land
Probability
map pasture to forest
Simulation of Land Use Changes
Arable land
Pasture
Vineyards
Forest
Stochastic allocation procedure was used to generate land use
pattern for different scenarios
Predictions for 2015 when the actual trend persists???
Consequences of Land Use
Changes
WATEM/SEDEM is a spatially distributed
erosion and sediment delivery model (Van
Rompaey et al., 2001, Van Oost et al., 2000, Verstraeten et al., 2002)
Hillslope Sediment routing
CALCULATION OF DISTRIBUTED
PATTERN OF MEAN ANNUAL SOIL
EROSION RATES (E)
CALCULATION OF DISTRIBUTED
PATTERN OF MEAN ANNUAL
TRANSPORT CAPACITY (TC)
(RUSLE-based)
ROUTING OF SEDIMENT
VIA FLOWPATHS TO
THE RIVER CHANNELS
TC > E +
SED_INPUT
TC < E +
SED_INPUT
RIVER
CHANNEL
SEDIMENT
TRANSFER
SEDIMENT
TRANSFER +
SEDIMENTATION
SEDIMENT
DELIVERY
Results (erosion reduction)
• Pécsely SY: 0.030 ton/ha year
• Kali Basin SY: 0.018 ton/ha year
(1975 – 1994)
(1981 – 1989)
Very low SDR-values as a consequence of relatively flat centre of the
basin
Predictions for 2015???
Conclusions
• 1949 – 1989:
• 1989 – 2004:
Collectivization
Privatization
Fragmentation
Increase of non-cultivated areas
Forest: 715 to 963 ha
Pasture: 526 tot 633 ha
Vineyards: 636 to 272 ha
Arable land: constant
• Driving Forces: (Chi² and Logistic Regression)
Transition Probability Maps
Scenario Development
GEOMOPRHOLOGICAL IMPACT
Thank You !!!