Supplementary online resource material

Each of the models is briefly described below and in Table S1, but the reader is referred to Holman and Harrison (2011) where full details of each meta model, including their development and validation, are given. In addition, Kebede et al. (In Press) describes the results of a sensitivity analysis across the climate and socio-economic input parameter space of the linked models within the IA Platform.

  • Urban: The Regional Urban Growth (RUG) metamodel simulates urban growth as a function of changes in socio-economic variables (population, GDP per capita) and societal values (strictness of planning constraints, household location preferences). The model also takes into account local geography, travel times with the existing infrastructure and city typology (e.g. mono- versus polycentric). The RUG meta-model in the IA platform consists of a look-up table of artificial surfaces per grid cell from running the original RUG model (on a 1 x 1 km grid) with all possible combinations of Platform input values and aggregating the data to the 5 x 5 km grid. The differences between the baseline simulation and the observed data from CORINE are on average around 2-3%, with most values falling below 7%.
  • Forest: MetaGOTILWA+ is an artificial neural networks (ANNs) that emulates the performance of the GOTILWA+ model. The ANN was trained on GOTILWA results for 889 cells that spanned the range of environmental conditions across Europe. For each cell, GOTILWA simulations were conducted for all combinations of a range of characteristic species, 3 different management regimes and with four different levels of effective soil volume. The predictions of the ANN were tested against GOTILWA data from cells which were not used for training and a strong 1:1 relationship found, with r2 values for metamodel indicators all greater than 0.9.
  • Flooding: The Coastal Fluvial Flood (CFFlood) meta-model consists of three coupled sub-model components: (1) Coastal flood, (2) Fluvial flood and (3) Habitat change/loss components. The 2D model identifies the area at risk of flooding based on topography, relative sea-level rise or change in peak river flow and the estimated Standard of Protection of flood defences. An estimate of the people living in the flood risk zones is calculated using population density. The flood damages for residential properties (both contents and structure) are calculated based on urban areas and people at risk of flooding, flood water depths, and Gross Domestic Product (GDP). CFFlood model also assesses possible changes in the area of coastal floodplain habitats, based on accommodation space, sediment supply and the ratio of relative sea-level rise to tidal range (Richards et al., 2008);
  • Water: The WaterGAP meta-model (WGMM) uses a look-up table to reproduce the outputs of the WaterGAP3 model run at a 5’ x 5’ resolution for about 100 spatial units (single large river basins or clusters of smaller, neighbouring river basins with similar hydro-geographic properties) larger than 10,000 km². Under a changed climate, relative deviation of average discharge simulated by WGMM from aggregated WaterGAP3 output is ±5% for most of Europe. The meta-model water use results for the manufacturing, domestic and thermal electricity production sector match the WaterGAP3 results very well (R²>=0.975).
  • Crops: The crop yield metamodels use ANNs combined with temperature thresholds to prevent crops growing in unsuitable territories. The ANNs for each of the 12 crops were trained on simulated outputs of the ROIMPEL model, with the training and validation datasets including over 150,000 data points. The training datasets were sampled to adequately cover the whole range of both soil and climate predictors and the predicted variables, e.g. sowing date or actual yield. Overall the RMSE for the yield estimates is in most cases below 0.5 t/ha and the MBE that is close to 0 indicating that there is low/no systematic bias.
  • Rural land allocation: The SFARMOD meta-model simulates the behaviour of the full SFARMOD-LP model, using SFARMOD-LP outputs from 20,000 randomly selected sets of input data that fully cover the parameter input space. The metamodel is based on a series of regression equations that estimates first the percentage of the area of each crop, then the costs of dairy cows (concentrates) then the fixed costs of labour and machinery with the profitability being the difference. Up to 10 iterations based on profitability and food demand are allowed to determine the final land allocation and food production. The SFARMOD meta-model had a <5% misclassification compared to the full LP model.
  • Biodiversity: The SPECIES model (Spatial Estimator of the Climate Impacts on the Envelope of Species) simulates the suitable climate space of over 100 species selected to interact with the agricultural, forest, coastal and water sectors and to indicate a range of ecosystem services. SPECIES is based on ensembles of ANNs, which utilise bioclimatic (climate and soil moisture) variables to characterise bioclimatic suitability envelopes. The model is trained using existing empirical data on the European and North African (north of 15oN) distributions of species to enable the full climate space of a species to be characterised. All species models show Area Under the Receiver Operating Characteristic Curve (AUC) statistics greater than 0.8, indicating good discrimination ability.

Table S1 Details of the seven meta-models included within the IA Platform (based on Holman and Harrison 2011; adapted from Harrison et al., 2013)

Meta-model / Sector / Original model / Meta-modelling approach
Meta-RUG / Urban growth / Regional Urban Growth (RUG)
Reginster I., and Rounsevell M., (2006) Scenarios of future urban land use in Europe. Environ Plan B - Plan & Des 33: 619-636 / Look-up tables
Meta-Crop yield (winter wheat and spring wheat, winter barley and spring barley, winter oil seed rape, potatoes, grain maize, sunflower, soybean, cotton, grass, olives) (Audsley et al., In Press) / Agricultural crop yields / ROIMPEL
Audsley E, Pearn KR, Harrison PA, Berry PM (2008) The impact of future socio-economic and climate changes on agricultural land use and the wider environment in East Anglia and NorthWest England using a meta-model system. Clim Chang 90:57-88
Audsley E., Trnka M., Sabaté S. and Sanchez A. (In Press) Interactive modelling of land profitability to estimate European agricultural and forest land use under future scenarios of climate, socio-economics and adaptation. Clim Change / Soil/climate clustering combined with artificial neural networks
Meta-GOTILWA+
(Audsley et al., In Press) / Forest management / GOTILWA+
Morales P, Sykes MT, Prentice IC, Smith P, Smith, B, Bugmann H, Zierl B, Friedlingstein P, Viovy N, Sabaté S, Sánchez A, Pla E, Gracia CA, Sitch S, Arneth A, Ogee J (2005) Comparing and evaluating process-based ecosystem model predictions of carbon and water fluxes in major European forest biomes. Glob Chang Biol 11-12:2211-2233 / Artificial neural networks
Meta-SFARMOD
(Audsley et al., In Press) / Rural land allocation (land profitability / land use) / SFARMOD
Holman I.P., Rounsevell M.D.A., Shackley S, Harrison P.A., Nicholls R.J., Berry P.M., Audsley E. (2005) A regional, multi-sectoral and integrated assessment of the impacts of climate and socio-economic change in the UK. Clim Chang 71(1):9-41 / Soil/climate clustering combined with artificial neural networks
WGMM (Wimmer et al., In Press) / Hydrology, water availability and consumption / Water - Global Assessment and Prognosis (WaterGAP3)
Döll P, Kaspar F, Lehner B (2003) A global hydrological model for deriving water availability indicators: model tuning and validation. J Hydrol 270:105-134
Verzano K (2009) Climate change impacts on flood related hydrological processes: Further development and application of a global scale hydrological model. PhD thesis, International Max Planck Research School on Earth System Modelling, University of Kassel
Wimmer F., Audsley E., Savin C.-M., Malsy M., Dunford R., Harrison P.A., Schaldach R. and Flörke M. (In Press) Modelling the effects of cross-sectoral water allocation schemes in Europe. Clim Change / 3-dimensional surface response diagrams
Coastal Fluvial Flood meta-model (CFFlood – Mokrech et al., In Press) / Flood impacts / RegIS2 (Mokrech et al. 2008) and DIVA (McFadden et al. 2007)
Mokrech M, Nicholls RJ, Richards JA, Henriques C, Holman IP, Shackley S (2008) Regional impact assessment of flooding under future climate and socio-economic scenarios for East Anglia and North West England. Clim Chang 90:31-55
Mokrech M., Kebede A.S., Nicholls R.J. and Wimmer F. (In Press) An integrated approach for assessing flood impacts due to cross-sectoral effects of climate and socio-economic scenarios and the scope of adaptation in Europe. Clim Chang
McFadden L, Spencer T, Nicholls RJ (2007) Broad-scale modelling of coastal wetlands: what is required? Hydrobiol 577:5-15 / Simplified process-based model
SPECIES / Bioclimatic suitability for biodiversity / SPECIES (Harrison et al. 2006)
Harrison PA, Berry PM, Butt N, New M (2006) Modelling climate change impacts on species’ distributions at the European scale: Implications for conservation policy. Environ Sci & Policy 9(2):116-128 / Artificial neural networks

Harrison PA, Holman IP, Cojocaru G, Kok K, Kontogianni A, Metzger MJ, Gramberger M (2013) Combining qualitative and quantitative understanding for exploring cross-sectoral climate change impacts, adaptation and vulnerability in Europe. Reg Environ Change 13(4): 761-780

Holman IP, Harrison PA (eds) (2011) Report describing the development and validation of the sectoral meta-models for integration into the IA platform. CLIMSAVE Deliverable 2.2. . Accessed 13 March 2014

Kebede A.S., Dunford R., Audsley E., Harrison P.A., Holman I.P., Mokrech M., Nicholls R.J., Rickebusch S., Rounsevell M.D.A., Sabaté S., Sallaba F., Sanchez A., Savin C.-M., TrnkaM.andWimmer F. (In Press). The sensitivity of cross-sectoral impacts to climate and socio-economic drivers for key European sectors. Clim Change

Richards, J., Mokrech, M., Berry, P.M. & Nicholls R.J. (2008). Regional assessment of climate change impacts on coastal and fluvial ecosystems and the scope for adaptation. Climatic Change, 90, 141-167

Table S2: Scotland-wide changes in selected socio-economic drivers under the four CLIMSAVE socio-economic scenarios for the 2050s

Tartan Spring / Mad Max / The Scottish Play / MacTopia
Well-being and lifestyle distribution / Disparate / Disparate / Equitable / Equitable
Natural Resource status / Surplus / Deficit / Deficit / Surplus
Storyline overview / A far reaching, poorly regulated privatisation changes Scotland from a prosperous country with abundant natural resources to one with an eroded soil fabric and low standard of living, culminating in a ‘Tartan Spring’ revolution / Driven by crises, a new self-centred paradigm emerges, which leads to a growing disparity in society. Survival from day-to-day prevail, while ‘clans’ are ruling Scotland again / Building on traditional Scottish values, a lack of resources is dealt with by changes in lifestyle towards reducing, re-using and recycling, leading to a poorer but greener and happier population / Initially stimulated by a resource surplus, Scotland makes a transition towards an equitable and sustainable society to eventually becomes an IT, life sciences, green technology and finance frontrunner lead by a powerful middle lass
SOCIAL DRIVERS(% change from current):
Population / +18 / -11 / +12 / +34
Water savings due to behavioural change / +29 / -44 / +8 / +45
Dietary preferences for meat / -7% / +10 / -9 / -21
TECHNOLOGICAL DRIVERS(% change from current):
Agricultural mechanisation / +26 / +15 / +8 / +26
Water savings due to technological change / +29 / -44 / +8 / +45
Agricultural yields / +26 / +58 / +26 / +58
Irrigation efficiency / +26 / +58 / 0 / +26
ECONOMIC DRIVERS(% change from current):
GDP / +172 / -39 / +25 / +259
Food imports / 0 / 0 / 0 / +30
Bioenergy production / +2 / +2 / +2 / 0
Oil price / +273 / +152 / +273 / +273

Table S2: Cross-sectoral summary of changes in the 25th and 75th percentiles of distributions for Scotland and four regions for the 2050s with baseline (2000-2010) socioeconomics and the four CLIMSAVE socio-economics scenarios

2050 Baseline / 2050 MacTopia / 2050 MadMax / 2050 Scottish Play / 2050 Tartan Spring
Scot / H&I / S / C / NE / Scot / H&I / S / C / NE / Scot / H&I / S / C / NE / Scot / H&I / S / C / NE / Scot / H&I / S / C / NE
Artificial surfaces (%) / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ↑ / ↑ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / + / + / ◦ / ◦ / ◦ / + / ↑
People flooded (%) / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦
Biodiversity VI (-) / ↓ / ↓ / - / - / ↓ / - / - / - / + / - / - / - / - / - / - / - / - / - / - / ↓ / - / - / - / - / ↓
Intensively farmed (%) / + / + / ↓ / ↑ / ↑ / ◦ / ◦ / ↓ / ↓ / ↓ / ◦ / + / ↓ / ↓ / - / ◦ / ◦ / ↓ / ↑ / ↑ / ↑ / ↑ / ↓ / ◦ / +
Extensively farmed (%) / + / + / + / ↑ / - / - / - / - / - / + / + / + / ↑ / ↑ / ↑ / + / + / ↑ / ↑ / + / + / + / ↑ / ↑ / +
Food production (%) / ↑ / ↑ / ↑ / ↑ / ↕ / ↕ / ↓ / ↓ / - / ↑ / ↑ / ↑ / ↑ / ↑ / ↑ / ↑ / ↑ / ↑ / ↕ / ↑ / ↑
Forest area (%) / - / ↓ / - / ↓ / ↓ / - / - / - / ↓ / - / - / - / - / ↓ / - / ↓ / ↓ / - / ↓ / - / - / ↓ / - / ↓ / -
Unmanaged land (%) / ◦ / ◦ / ◦ / ◦ / ◦ / ↑ / + / ↑ / ↑ / + / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦
Intensity index (-) / ◦ / ◦ / ◦ / + / + / - / ◦ / - / - / - / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / + / ◦ / ◦ / ◦ / + / + / ◦
Landuse diversity index (-) / + / ↕ / ◦ / + / + / - / - / - / - / + / ◦ / + / - / + / + / + / + / ◦ / + / + / + / + / ◦ / + / +
Water availability (%) / ◦ / + / ◦ / - / ◦ / ◦ / + / ◦ / - / ◦ / ◦ / + / ◦ / - / ◦ / ◦ / + / ◦ / - / ◦ / ◦ / + / ◦ / - / ◦
Water Exploitation Index (%) / ◦ / ↑ / ↕ / + / ↓ / ↓ / - / ↓ / - / + / ↑ / ↑ / + / ↑ / - / - / + / - / - / ↕ / - / + / - / ↕
Irrigation usage (%) / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦ / ◦
↑ / Increase >50% in either the 25th or 75th percentile with a non-negative change in the other
+ / Increase >5% in either the 25th or 75th percentile with a non-negative change in the other
◦ / Change < ±5% in both the 25th or 75th percentile
- / Increase >5% in either the 25th or 75th percentile with a non-positive change in the other
↓ / Decrease >50% in either the 25th or 75th percentile with a non-positive change in the other
Decrease > 5% in 75th percentile and Increase > 5% in 25th percentile – contracting distribution
↕ / Increase > 5% in 75th percentile and decrease > 5% in 25th percentile – widening distribution


Figure S1 Map of the four catchment-based regions of Highlands and Islands, Southern, Central and North eastern Scotland