Mapping global development potential for renewable energy, fossil fuels, mining and agriculture sectors

Mapping suitable land for development is essential to land use planning efforts that aim to model, anticipate, and manage trade-offs between economic development and the environment. Previous land suitability assessments have generally focused on a few development sectors or lack consistent methodologies, thereby limiting our ability to plan for cumulative development pressures across geographic regions. Here, we generated 1-km spatially-explicit global land suitability maps, referred to as “development potential indices” (DPIs), for 13 sectors related to renewable energy (concentrated solar power, photovoltaic solar, wind, hydropower), fossil fuels (coal, conventional and unconventional oil and gas), mining (metallic, non-metallic), and agriculture (crop, biofuels expansion). To do so, we applied spatial multi-criteria decision analysis techniques that accounted for both resource potential and development feasibility. For each DPI, we examined both uncertainty and sensitivity, and spatially validated the map using locations of planned development. We illustrate how these DPIs can be used to elucidate potential individual sector expansion and cumulative development patterns.


Concentrated Solar Power (CSP)
Notes on comparison values: Similar other solar suitability assessments 1,2 , we ranked resource yield (RY) as the most important criterion, followed closely in importance by distance to the electrical grid (DEG), and then distance to major roads (DMR). Following ref 2 , we ranked landcover as equal in influence with DMR. Distance to urban areas (DUA) was ranked as the lowest importance of all other criteria expect distance to railways and ports (DRP) based on ref 1 . Given that ref 3 identified distance to railroads as influential for siting of solar power and because both rail cars and ships provide an inexpensive option for transporting building materials for any renewable energy project, we included the DRP criteria but ranked it least important. Notes on comparison values: Similar to CSP, we followed other solar suitability assessments 1,2 and ranked resource yield (RY) as the most important criterion, followed closely in importance by distance to the electrical grid (DEG), and then distance to major roads (DMR). Following ref 2 , we ranked landcover as equal in influence with DMR. Distance to urban areas (DUA) was ranked as the lowest importance of all other criteria except distance to railways and ports (DRP) based on ref 1 ; however, we increased the importance value when comparing to CSP due to PV installations being more commonly located in or near developed areas 4 . Given that ref 3 identified distance to railroads as influential for siting of solar power and because both rail cars and ships provide an inexpensive option for transporting building materials for any renewable energy project, we included the DRP criteria but ranked it least important. This criterion was given an even lower comparison values than CSP, because less materials are required for PV development 5 . Notes on comparison values: Similar to other wind suitability studies 2,6,7 , we ranked resource yield (RY) as the most important criterion, followed closely in importance by distance to the electrical grid (DEG). Given the need to transport oversized wind development materials (i.e., hubs and blades) and a likelihood of these sizes continuing to increase 8 , we assigned the distance to railways and ports (DRP) and the distance to major roads (DRM) criteria as equally important and more important than landcover, inverse population density (IPD), and distance to urban areas (DUA). We ranked the landcover criterion as more important than both IDP and DUA because of a greater frequency of use in other wind suitability studies 2,6,7 . Finally, because of a trend towards larger power producing wind farms and longer transmission distances 9,10 , we graded the IPD criterion as more important than DUA. Notes on comparison values: We ranked resource yield (RY) as more important than all other criteria, because of the substantial cost benefits associated with building a hydropower dam in regions with high resource potential 11 . Given that studies identify population displacement as a substantial hurdle when establishing a reservoir 12 , we assigned the inverse population density (IPD) criterion to be slightly less important than RY but more important than all other criteria. When comparing the distance to the electrical grid (DEG) criteria with either distance to major roads (DMR) or distance to railways or ports (DRP), we ranked DEG more important than these two criteria based on ref 13 , which ranked grid connection as a higher risk on investment for building a hydropower plant than any transportation criteria (i.e., DMR or DRP). We assigned equal importance when comparing DMR and DRP criteria because materials for development can occur from either. Similar to the wind MCDA, we weighted distance to urban areas (DUA) as least important, because large transmission distances are typical for large hydropower dams (i.e., > 30 MW; ref 14 ), which is a size most common for future hydropower projects 15 . Notes on comparison values: We assigned resource yield (RY) and active coal mine density (ACMD) criteria to be equally important and higher than all other criteria, because existing coal mines occur within a coal basin where it is most profitable to operate due to factors such as low overburden, large coal seams, and/or highquality coal 16 ; and future development will seek to expand and exhaust currently mined deposits before moving to unproven deposits 17 . After RY and ACMD, we considered distance to railway and ports (DRP) and distance to major roads (DMR) criteria as the next most important given that coal transportation is a critical element for coal field development 17 . We ranked DRP slightly higher than DMR given that 70% of U.S. coal is transported by either rail or ships 18 , however, in developing countries there is still a strong dependence on trucking coal 19 . After transportation, we considered access to electricity (AE) more important than the distance from coal-fired power plants (DCPP) because of the large benefit electricity provides for all types of mining 20 and the large variation in distances existing power plants are from their coal source 17 .  21,22 . After yield, we ranked distance to producing oil/gas fields (DPOGF) as the second most important criteria given the higher likelihood of infilling and expansion of existing oil and gas fields relative to the development of new basins that require much higher capital 23 . Because oil development relies on electricity to pump and collect resources 24 , we ranked access to electricity (EA) as more important than either distance to railways or ports (DRP) or distance to major roads (DMR). Finally given the high dependency on rail and ship transportation for oil resources 24 , we ranked DRP more important than DMR but only slightly due to heavy road use in the all phases of oil field development 21,25 .

Conventional Gas
Classified  21,22 . After yield, we ranked distance to producing oil/gas fields (DPOGF) as the second most important criteria given the higher likelihood of infilling and expansion of existing oil and gas fields relative to the development of new basins that require much higher capital 23 . We assigned DPOGF as slightly higher importance value than applied in the conventional oil MCDA due to higher infrastructure needs being required to fully capture gas 24 . Because gas development relies on electricity to pump and collect resources 24 , we ranked access to electricity (AE) as more important than either distance to railways or ports (DRP) or distance to major roads (DMR). We ranked DMR more important than DRP because of heavy road use in the all phases of gas field development 21,25 and the transportation of natural gas via ships and rail limited to extremely costly liquid natural gas (LNG) 26 . We did not however remove the DRP criterion due to an increase in LNG production and exports for large gas producing countries like the US 27 .  21,22 . After yield, we ranked distance to producing oil/gas fields (DPOGF) as the second most important criteria given the higher likelihood of infilling and expansion of existing oil and gas fields relative to the development of new basins that require much higher capital 23 . Because fracking liquids used in unconventional oil practices are almost exclusively transported by trucks 28 , we considered distance to major roads (DMR) as more important than both electricity accessibility (EA) or distance to railways or ports (DRP). EA was given a slightly higher ranking than DRP because oil development relies on electricity to pump and collect resources 24 . Finally given the high dependency on rail and ship transportation for oil resources 24 , we included DRP but ranked it least important.  21,22 . After yield, we ranked distance to producing oil/gas fields (DPOGF) as the second most important criteria given the higher likelihood of infilling and expansion of existing oil and gas fields relative to the development of new basins that require much higher capital 23 . We assigned DPOGF as slightly higher importance value than applied in the unconventional oil MCDA due to higher infrastructure needs being required to fully capture gas 24 . Because fracking liquids used in unconventional gas practices are almost exclusively transported by trucks 28 , we considered distance to major roads (DMR) as more important than both electricity accessibility (EA) or distance to railways or ports (DRP). EA was given a slightly higher ranking than DRP because gas development relies on electricity to pump and collect resources 24 . Finally given the limited current role of rail and ship transportation for gas resources 26 , we ranked DRP least important but still included this criterion due to this transport method having a potential to increase in the future 27 .  29 , we ranked resource yield (RY) as the most important criterion. With proximity to transportation infrastructure being the next prerequisite for mining to occur 29,30 , we considered distance to major roads (DMR) and distance to railways and ports (DRP) only moderately less important than resource yield. In comparing these two criteria to each other, we ranked them equally important due to transportation preference often being related to the value and size of mineral being mined 31 and, thus, hard to distinguish within our MCDA. Finally, we also included electricity accessibility (EA) as a feasibility criterion due to the vast amount of energy commonly used to mine and process metallic minerals at the mine site 32 , but ranked EA as the least important criteria due to on-site energy production often occurring in remote mining locations 31,33 .  29 , we ranked resource yield (RY) as the most important criterion. With non-metallic mining being heavily dependent on proximity to populated areas that use the materials mined 34-36 , we assigned distance to aggregate demand centers (DADC) as more important than either distance to major roads (DMR) or distance to railways or ports (DRP). Finally, we ranked DMR and DRP equally since all forms of transportation (i.e., truck, train, and ship) are commonly used and transportation selection is based on whichever type is closest to the mine site and can move materials to the demand center 37 .

Croplands Expansion
Classified Notes on comparison values: In addition to resource yield (RY) criteria, we included two feasibility criteria within our cropland expansion MCDA: market accessibility (MA) as proxied by travel time to cities 38 and land supply elasticity (LSE) map as proxied by nighttime lights 39 . Next to RY, market access is a widely recognized contributor to cropland expansion in both developing and developed countries [40][41][42] . At the same time, lands close to major urban areas compete with more profitable land uses such as residential and built-up areas and thus exhibit a lower LSE 43,44 and have a higher likelihood of crop intensification than crop expansion 45 . Therefore, in our AHP comparisons, we ranked RY only slightly more important than both MA and LSE; and ranked MA slightly more important than LSE, because MA is cited more often in the literature as a factor affecting cropland expansion 41 .

Biofuels Expansion
Classified Notes on comparison values: Similar to cropland expansion, we included with our biofuels resource yield (RY) criterion, two feasibility criteria: market accessibility (MA) as proxied by travel time to cities 38 and land supply elasticity (LSE) map as proxied by nighttime lights 39 . We followed same logic for criteria comparisons as cropland expansion ranking RY only slightly more important than both MA and LSE; and ranked MS slightly more important than LSE.

Metallic mining development potential index (DPI) example: Methods applied in generating criteria weight ranges for DPI uncertainty analysis.
We provide a detailed example for Metallic Mining (MM) DPI of how we define criteria weights ranges. Ranges were iteratively sampled with Monte Carlo procedures described in main text Uncertainty section. This process was sector-and criterion-specific and was applied for all 13 DPIs and their criteria (see Online-only Table 5 for all criteria and sector weight range results).
For all criteria related to the MM DPI (Supplementary Table S2), we modified the importance value (i.e., 1-9 and reciprocals) of the original AHP judgement matrix (e.g., Supplementary Table S3). We increased scores two points up (Supplementary Tables S4, S6, S8, and S10) or down (Supplementary Tables S5, S7, S9, and S11), and for each score modification, we adjusted the criterion's judgement matrix column and row (highlighted in grey) to maintain consistency ratios within the < 0.1 guideline 46,47 . This process required two additional AHP matrices for each sector and criterion combination, which produced 8 new matrices for MM DPI given it had 4 criteria, and overall produced 130 new AHP matrices across all sectors and criteria.  Fig. S1. Example sensitivity results for minimum and maximum criteria weight changes applied to the active coal mining density (ACDM) criteria within the coal development potential index (DPI) analysis. Mapped examples of sensitivity analysis for the ACDM criterion within the Coal DPI comparing the minimum (i.e., -20%) and maximum (i.e., +20%) criterion weight change with the original coal DPI (DPI Orig ). Row (a) compares binned DPI maps and percentage of cells falling within each bin. Row (b) highlights the spatial difference in the continuous coal DPI maps with row (c) showing cell-by-cell continuous value histograms and the binned breakpoints associated with row (a). Finally, row (d) displays the final change maps for the minimum and maximum criterion weight change and highlights those cells within the greatest change from the DPI Orig . This example is one of three criteria within DPI sectors (i.e., resource yield -wind, resource yield -hydropower, ACDM -coal) which had a more detailed sensitivity analysis ran to further examine absolute cell value changes as reported in Table 3.

Supplementary
Sensitivity Run for -20% ACDM Criterion Weight DPI Orig Sensitivity Run for +20% ACDM Criterion