Predicting land cover changes using a CA Markov model under different shared socioeconomic pathways in Greece

Figures & data

Figure 1. Aster GDEM (left side) and land cover types distribution in Greece in 2001, 2011 and 2018

Figure 2. Spatial distribution of dominant land cover changes in Greece from 2001 to 2011 and from 2011 to 2018, based on MODIS land cover product and the International Geosphere-Biosphere Program classification scheme. Coordinates are in MODIS sinusoidal projection

Figure 3. Gains and losses of various land cover types in Greece from 2001 to 2011 and from 2011 to 2018

Table 1. Areas of various land cover types in Greece, in 2001, 2011, and 2018, based on MODIS land cover product

Land Cover type	2001		2011		2018
	Area (km^2)	Area (%)	Area (km^2)	Area (%)	Area (km^2)	Area (%)
Evergreen Needleleaf Forests	5396.7	4.08	8302.1	6.27	8938.8	6.75
Evergreen Broadleaf Forests	937.6	0.71	1084.0	0.82	1195.4	0.90
Deciduous Needleleaf Forests	0.5	0.00	0.0	0.00	1.7	0.00
Deciduous Broadleaf Forests	3194.8	2.41	4379.0	3.31	4572.9	3.45
Mixed Forests	2001.9	1.51	2514.5	1.90	3084.0	2.33
Closed Shrublands	30.5	0.02	26.2	0.02	13.7	0.01
Open Shrublands	299.0	0.23	188.7	0.14	211.0	0.16
Woody Savannas	15,889.9	12.00	17,053.1	12.88	19,330.2	14.60
Savannas	34,632.4	26.15	32,556.0	24.58	32,074.9	24.22
Grasslands	32,830.8	24.79	30,574.0	23.09	28,829.7	21.77
Permanent Wetlands	2033.9	1.54	2265.9	1.71	2452.7	1.85
Croplands	31,112.0	23.49	28,580.1	21.58	26,461.6	19.98
Urban and Built-up Lands	2044.6	1.54	2055.6	1.55	2066.3	1.56
Cropland/Natural Vegetation Mosaics	808.0	0.61	1650.7	1.25	1985.6	1.50
Permanent Snow and Ice	0.9	0.00	0.5	0.00	0.5	0.00
Barren	319.6	0.24	244.3	0.18	266.6	0.20
Water Bodies	901.6	0.68	959.7	0.72	948.8	0.72
Total	132,434.6	100	132,434.6	100.0	132,434.6	100.0

Figure 4. Historical climate data from 2000 to 2018 at 2.5 minutes grids (Harris et al. 2014): a) Mean Annual Precipitation (mm) and b) Mean Annual Minimum Temperature (^oC)

Table 2. Multilayer Perceptron LCM parameters

Independent Variable 1	Evidence Likelihood of Land Cover
Independent Variable 2	Altitude (ASTER GDEM)
Independent Variable 3	Mean Annual Precipitation
Independent Variable 4	Mean Annual Minimum Temperature
Independent Variable 5	Distance from previous changes
Number of pixels for training	5000
Number of pixels for testing	5000
Input layer neurons	5
Hidden layer neurons	7
Output layer neurons	7
Final learning rate	0.001
Momentum factor	0.5
Sigmoid constant	1
Acceptable RMS	0.01
Iterations	9343
Training RMS	0.0078
Testing RMS	0.0082
Accuracy rate	97.32%
Skill measure	0.8632

Table 3. LCM skill breakdown by transition and persistence

Class	Skill measure
Transition: Savannas to Woody Savannas	0.9579
Transition: Savannas to Grasslands	0.9588
Transition: Grasslands to Savannas	0.9691
Transition: Croplands to Grasslands	0.9508
Persistence: Savannas	0.7485
Persistence: Grasslands	0.7367
Persistence: Croplands	0.7212

Figure 5. LCM sensitivity analysis, a) forcing a single variable constant, b) forcing all variables except one to be constant, c) using backward stepwise constant forcing

Table 4. Classification accuracy of predicted versus observed 2018 land cover

	User’s Accuracy %	Producer’s Accuracy (%)	Overall Accuracy (%)	Kappa Coefficient
Evergreen Needleleaf Forests	81.7	94.8	96.7 (p < 0.01) 95% confidence interval: 96.51–96.72	0.959
Evergreen Broadleaf Forests	79.6	92.9
Deciduous Needleleaf Forests	76.2	99.6
Deciduous Broadleaf Forests	85.9	98.1
Mixed Forests	77.8	94.8
Closed Shrublands	27.8	26.5
Open Shrublands	66.3	97.3
Woody Savannas	99.8	92.1
Savannas	99.1	97.3
Grasslands	99.7	98.7
Permanent Wetlands	83.5	97.1
Croplands	99.8	99.1
Urban and Built-up Lands	99.5	99.8
Cropland/Natural Vegetation Mosaics	58.7	99.7
Permanent Snow and Ice	100	100
Barren	83.6	94.8
Water Bodies	98.9	97.8

Figure 6. Future climate data (2021–2040) from CMIP6 at 2.5 minutes grids (Harris et al. 2014; L. Feng et al. 2020) provided as ensembles of 4 GCMs, i.e. BCC-CSM2-MR, CNRM-ESM2-1, PSL-CM6A-LR and MIROC-ES2L: a) Projected mean annual precipitation and b) mean annual minimum temperature based on the SSP126

Figure 7. Future climate data (2021–2040) from CMIP6 at 2.5 minutes grids (Harris et al. 2014; L. Feng et al. 2020) provided as ensembles of 4 GCMs, i.e. BCC-CSM2-MR, CNRM-ESM2-1, PSL-CM6A-LR and MIROC-ES2L: a) Projected mean annual precipitation and b) mean annual minimum temperature based on the SSP585

Figure 8. Comparison of observed land cover for 2001, 2011 and 2018, and predicted land cover for 2018 and 2030 in Greece under SSP126 and SSP585

Figure 9. Spatial distribution of land cover transitions for the two examined scenarios, i.e. SSP126 and SSP585, for 2030 in Greece with reference the observed 2018 land cover

Table 5. Basic statistics of the examined climate scenarios over the study area

	Historical 2001–2018	SSP126 (2021–2040)*	SSP585 (2021–2040)*
Mean annual precipitation (mm/year)
Mean	681.2	653.6	643.3
Standard deviation	171.3	166.6	162.4
Minimum	392.3	391	381.7
Maximum	1389.1	1238	1211.9
Mean annual minimum temperature (°C)
Mean	9.8	10.6	10.7
Standard deviation	2.9	2.8	2.7
Minimum	−0.7	3.4	3.6
Maximum	17.5	17.7	17.8

*Averages of four GCMs: BCC-CSM2-MR, CNRM-ESM2-1, PSL-CM6A-LR, and MIROC-ES2L

Table 6. Predicted versus observed land cover areas in various land cover types for 2018 and predicted land cover for 2030 under SSP126 and SSP585

Land Cover type	2018 Observed		2018 Predicted		2030 Predicted SSP126			2030 Predicted SSP585
	Area (km^2)	Area (%)	Area (km^2)	Area (%)	Area (km^2)	Area (%)	% change from observed 2018	Area (km^2)	Area (%)	% change from observed 2018
Evergreen Needleleaf Forests	8938.8	6.75	8635.2	6.52	8978.6	6.78	0.03	9028.1	6.82	0.07
Evergreen Broadleaf Forests	1195.4	0.90	1117.2	0.84	1221.9	0.92	0.02	1285.6	0.97	0.07
Deciduous Needleleaf Forests	1.7	0.00	2.0	0.00	0.0	0.00	0.00	0.0	0.00	0.00
Deciduous Broadleaf Forests	4572.9	3.45	4213.9	3.18	4587.5	3.46	0.01	4597.3	3.47	0.02
Mixed Forests	3084.0	2.33	2793.7	2.11	3121.1	2.36	0.03	3133.2	2.37	0.04
Closed Shrublands	13.7	0.01	15.4	0.01	16.5	0.01	0.00	16.9	0.01	0.00
Open Shrublands	211.0	0.16	162.5	0.12	197.8	0.15	−0.01	198.1	0.15	−0.01
Woody Savannas	19,330.2	14.60	21,014.5	15.87	24,282.8	18.34	3.74	22,997.4	17.37	2.77
Savannas	32,074.9	24.22	31,565.8	23.83	30,731.3	23.20	−1.01	30,986.0	23.40	−0.82
Grasslands	28,829.7	21.77	29,132.6	22.00	26,783.8	20.22	−1.54	27,357.2	20.66	−1.11
Permanent Wetlands	2452.7	1.85	2311.6	1.75	2436.8	1.84	−0.01	2438.0	1.84	−0.01
Croplands	26,461.6	19.98	26,991.2	20.38	24,853.0	18.77	−1.21	25,184.0	19.02	−0.96
Urban and Built-up Lands	2066.3	1.56	2057.2	1.55	2068.4	1.56	0.00	2068.3	1.56	0.00
Cropland/Natural Vegetation Mosaics	1985.6	1.50	1235.6	0.93	1916.7	1.45	−0.05	1925.2	1.45	−0.05
Permanent Snow and Ice	0.5	0.00	0.4	0.00	0.5	0.00	0.00	0.4	0.00	0.00
Barren	266.6	0.20	233.0	0.18	274.6	0.21	0.01	255.9	0.19	−0.01
Water Bodies	948.8	0.72	952.7	0.72	963.4	0.73	0.01	963.0	0.73	0.01
Total	132,434.6	100.0	132,434.6	100.0	132,434.6	100.0	0.0	132,434.6	100.0	0.0

Harris, I., P. D. Jones, T. J. Osborn, and D. H. Lister. 2014. “Updated High-Resolution Grids of Monthly Climatic Observations - the CRU TS3.10 Dataset.” International Journal of Climatology 34 (3): 623–642. doi:10.1002/joc.3711.

 Web of Science ®Google Scholar

Feng, L., S. J. Smith, C. Braun, M. Crippa, M. J. Gidden, R. Hoesly, Z. Klimont, M. van Marle, M. van den Berg, and G. R. van der Werf. 2020. “The Generation of Gridded Emissions Data for CMIP6.” Geoscientific Model Development 13: 461–482. doi:10.5194/gmd-13-461-2020.

 Web of Science ®Google Scholar