IEA (2021), The Role of Critical Minerals in Clean Energy Transitions, IEA, Paris https://www.iea.org/reports/the-role-of-critical-minerals-in-clean-energy-transitions, Licence: CC BY 4.0
Mineral requirements for clean energy transitions
Introduction
Clean energy technologies require a wide range of minerals and metals
Clean energy technologies – from wind turbines and solar panels, to electric vehicles and battery storage – require a wide range of minerals1 and metals. The type and volume of mineral needs vary widely across the spectrum of clean energy technologies, and even within a certain technology (e.g. EV battery chemistries).
Critical mineral needs for clean energy technologies
Copper | Cobalt | Nickel | Lithium | REEs | Chrominum | Zinc | PGMs | Aluminium* | |
---|---|---|---|---|---|---|---|---|---|
Solar PV | |||||||||
Wind | |||||||||
Hydro | |||||||||
CSP | |||||||||
Bioenergy | |||||||||
Geothermal | |||||||||
Nuclear | |||||||||
Electricity networks | |||||||||
EVs and battery storage | |||||||||
Hydrogen |
Importance | High | Moderate | Low |
This chapter assesses the aggregate mineral demand from a wide range of clean energy technologies under the IEA’s Stated Policies Scenario (STEPS) and the Sustainable Development Scenario (SDS), including:
- Low-carbon power generation: solar PV, wind, other renewables and nuclear;
- Electricity networks;
- Electric vehicles and battery storage;
- Hydrogen (electrolysers and fuel cells).
For each of the clean energy technologies, we estimate overall mineral demand using four main variables: clean energy deployment trends under the STEPS and SDS; sub-technology shares within each technology area; mineral intensity of each sub-technology; and mineral intensity improvements.2
Projected mineral demand is highly dependent on the stringency of climate policies (reflected in the difference between the STEPS and SDS) as well as potential technology development pathways such as different solar PV module types or EV battery chemistries. We explore the impacts of varying technology evolution trends through 11 alternative cases under both the STEPS and SDS, in addition to our base case.
Overview
Rising deployment of clean energy technologies is set to supercharge demand for critical minerals
The global clean energy transitions will have far-reaching consequences for mineral demand over the next 20 years. By 2040, total mineral demand from clean energy technologies double in the STEPS and quadruple in the SDS.
In both scenarios, EVs and battery storage account for about half of the mineral demand growth from clean energy technologies over the next two decades, spurred by surging demand for battery materials. Mineral demand from EVs and battery storage grows tenfold in the STEPS and over 30 times in the SDS over the period to 2040. By weight, mineral demand in 2040 is dominated by graphite, copper and nickel. Lithium sees the fastest growth rate, with demand growing by over 40 times in the SDS. The shift towards lower cobalt chemistries for batteries helps to limit growth in cobalt, displaced by growth in nickel.
Total mineral demand for clean energy technologies by scenario, 2010-2040
OpenElectricity networks are another major driving force. They account for 70% of today’s mineral demand from the energy technologies considered in this study, although their share continues to fall as other technologies – most notably EVs and storage – register rapid growth.
Mineral demand from low-carbon power generation grows rapidly, doubling in the STEPS and nearly tripling in the SDS over the period to 2040. Wind power plays a leading role in driving demand growth due to a combination of large-scale capacity additions and higher mineral intensity (especially with growing contributions from mineral-intensive offshore wind). Solar PV follows closely, with its unmatched scale of capacity additions among the low-carbon power generation technologies. Hydropower, biomass and nuclear make only minor contributions given their comparatively low mineral requirements and modest capacity additions.
The rapid growth of hydrogen use in the SDS underpins major growth in demand for nickel and zirconium for use in electrolysers, and for copper and platinum-group metals for use in fuel cell electric vehicles (FCEVs). Despite the rapid rise in FCEVs and the decline in catalytic converters in gasoline and diesel cars, demand for platinum-group metals in internal combustion engine cars remains higher than in FCEVs in the SDS in 2040.
Demand for rare earth elements (REEs) – primarily for EV motors and wind turbines – grows threefold in the STEPS and more than sevenfold in the SDS by 2040.
Clean energy technologies are set to emerge as a major force in driving demand growth for critical minerals
For most minerals, the share of clean energy technologies in total demand was minuscule until the mid-2010s, but the picture is rapidly changing. Energy transitions are already the major driving force for total demand growth for some minerals. Since 2015, EVs and battery storage have surpassed consumer electronics to become the largest consumers of lithium, together accounting for 30% of total current demand.
As countries step up their climate ambitions, clean energy technologies are set to become the fastest-growing segment of demand for most minerals. Their share of total demand edges up to over 40% for copper and REEs, 60-70% for nickel and cobalt and almost 90% for lithium by 2040 in the SDS.
A wide range of futures are possible, mainly related to level of climate ambition and action, as well as technology uncertainties
Demand projections are subject to considerable uncertainty, with different levels of climate ambition and various technology development pathways resulting in a wide range of mineral demand.
For example, lithium demand in 2040 may be 13 times higher (if vanadium redox flow batteries rapidly penetrate the market in the STEPS) or 51 times higher (if all-solid-state batteries commercialise faster than expected in the SDS) than today’s levels. Cobalt and graphite may see 6- to 30-times higher demand than today depending on the direction of battery chemistry evolution.
Among non-battery materials, demand for REEs grows by seven times in the SDS, but growth may be as low as three times today’s levels should wind companies tilt more towards turbines that do not use permanent magnets in the STEPS context.
Demand for renewables- and networks-related minerals from clean energy technologies in 2040 relative to 2020 under different scenarios and technology evolution trends
OpenDemand for battery-related minerals from clean energy technologies in 2040 relative to 2020 under different scenarios and technology evolution trends
OpenThese large uncertainties around possible futures may act as a factor that hampers mining and processing companies’ investment decisions, which could in turn cause supply-demand imbalances in the years ahead. Despite the promise of massive demand growth, mining and processing companies may be reluctant to commit large-scale investment given the wide range of possible demand trajectories.
The biggest source of demand variance comes from the uncertainty surrounding announced and expected climate ambitions – in other words, whether clean energy deployment and resulting mineral demand follows STEPS or SDS trajectories. Governments have a key role to play in reducing uncertainty by sending strong and consistent signals about their climate ambitions and implementing specific policies to fulfil these long-term goals.
These efforts also need to be accompanied by a range of measures to dampen the rapid growth in primary supply requirements such as promoting technology innovation for material efficiency or substitution, scaling up recycling and extending the lifetime of existing assets through better maintenance (see Reliable supply of minerals).
Low-carbon power generation
Rapid deployment of solar PV in the SDS underpins more than doubling of mineral demand for solar PV by 2040 despite continued intensity reductions
Worldwide solar PV capacity has increased by almost 20 times over the past decade, spurred by declining costs and strong policy support in key regions. In both the STEPS and SDS, solar sets new records for deployment each year after 2022, representing 45% of total power capacity additions by 2040.
In the SDS, capacity additions in 2040 are triple those of 2020, resulting in a near tripling of copper demand from solar PV. However, potential material intensity reductions could significantly dampen demand growth for both silver and silicon, with 2040 levels only 18% and 45% higher than in 2020.
Solar PV capacity additions in 2040 in the STEPS are 25% lower than in the SDS. However, slower assumed improvements in material intensity for silver and silicon offset the lower capacity additions, resulting in similar demand for silver and silicon in the two scenarios.
Mineral demand from solar PV by scenario, 2020-2040
OpenWhile crystalline silicon modules are expected to continue to dominate the solar PV market, further progress on alternative technologies could see these technologies achieving growing market shares by 2040, which we explore in three alternative cases: high cadmium telluride, high perovskite, and high gallium arsenide.
In the High cadmium telluride case, demand for cadmium and tellurium grows sevenfold by 2040 in the SDS to 1 300 tonnes and 1 400 tonnes respectively. This rapid growth would put pressure on supply capacities, as current production levels are around 23 000 tonnes for cadmium and 500 tonnes for tellurium.
In the High pervoskite case, silicon demand is 10% lower in 2040 compared to the base case in the SDS, while lead demand is 45% higher.
In the High gallium arsenide case, demand for arsenic, gallium and indium in 2040 are around twice as high compared to the base case in the SDS. The additional demand for arsenic represents around 25% of global production today, while additional demand for gallium is about 10 times more than current production of high-purity refined gallium.
The growing market for wind turbines with permanent magnets – particularly for offshore projects – could dramatically increase demand for rare earths over the coming decades
Global installed capacity of wind power has nearly quadrupled over the past decade, spurred by falling costs and policy support in more than 130 countries. Over the next two decades, wind power is set for strong growth, with the offshore wind industry maturing and adding to developments in onshore wind on the back of technology improvements and low-cost financing.
Wind turbines require concrete, steel, iron, fibreglass, polymers, aluminium, copper, zinc and REEs. Mineral intensities not only depend on the turbine size, but also on the turbine type. For example, turbines based on permanent-magnet synchronous generators – which dominate the offshore market due to their lighter and more efficient attributes as well as lower maintenance costs – require REEs.
In the SDS, demand for REEs in wind – neodymium and praseodymium in particular – is set to more than triple by 2040, driven by the doubling of annual capacity additions and a shift towards turbines with permanent magnets. Copper demand reaches 600 kt per year in 2040, propelled by offshore wind requiring greater cabling. Offshore wind accounts for nearly 40% of copper demand from wind despite accounting for only 20% of total wind capacity additions.
In the Constrained REE supply case, manufacturers are assumed to gradually switch to non-magnet technologies and project developers adopting hybrid configurations with a gearbox and a smaller magnet. As a result, neodymium demand in the SDS is contained at around 8 000 tonnes in 2030 and 40% lower in 2040 compared to the base case. Demand for praseodymium and dysprosium are 15% and 32% lower respectively compared to the base case in 2040.
Mineral demand from other renewables varies significantly
The expansion of concentrated solar power increases demand for chromium, copper, manganese and nickel. Between 2020 and 2040 in the SDS, chromium demand from CSP grows by 75 times (to 91 kt), copper demand grows by 68 times (to 42 kt), manganese demand grows 92-fold (to 105 kt), and nickel demand grows 89-fold (to 35 kt).
Mineral demand from geothermal more than quadruples between 2020 and 2040 in the SDS. Despite accounting for less than 1% of all low-carbon power capacity additions in 2040, geothermal power is a major source of demand for nickel, chromium, molybdenum and titanium from the power sector. Of the total mineral demand from all low-carbon power sources in 2040, geothermal accounts for 80% of nickel demand, nearly half of the total chromium and molybdenum demand, and 40% of titanium demand.
In contrast, hydropower and bioenergy have relatively low mineral intensity compared to other renewable power sources. Hydropower and bioenergy each account for only about 2% of the total demand for copper from all low-carbon power capacity additions in 2040.
Modest growth in mineral demand from nuclear power
Nuclear power is the second-largest source of low-carbon power behind hydropower, accounting for about 10% of global electricity generation in 2020. Global installed capacity of nuclear power grows modestly to 2040 (by 15% in the STEPS and 45% in the SDS compared to 2020), as capacity declines in North America and Europe are offset by growth in emerging economies.
Along with hydropower and bioenergy, nuclear has relatively low critical mineral intensity. In the SDS, total mineral demand from nuclear power – mostly chromium, copper and nickel – grows by around 35% compared to 2020 levels, reaching almost 70 kt by 2040. However, demand for these minerals from nuclear accounts for less than 6% of overall demand from low-carbon power.
Electricity networks
Electricity networks are the backbone of secure and reliable power systems, and have a vital role in integrating clean energy technologies
Many of the features that characterise a clean energy system – the growing role of electricity in final consumption, rising contributions from renewables in electricity supply and the greater need for flexibility – all necessitate significant expansion of electricity grids.
The projected requirement for new transmission and distribution lines worldwide in the STEPS is 80% greater over the next decade than the expansion seen in the last ten years. In the SDS, the annual pace of grid expansion needs to more than double in the period to 2040. Around 50% of the increase in transmission lines and 35% of the increase in distribution network lines are attributable to the increase in renewables.
In addition to additional lines, there is scope to refurbish grids to strengthen the resiliency of electricity systems to climate change and extreme weather events. Refurbishment of electricity grids is also strongly linked to digitalisation, given the rising need for smart and flexible grids.
Growing need for grid expansion underpins a doubling of annual demand for copper and aluminium by 2040 in the SDS
The huge expansion of electricity grids requires a large amount of minerals and metals. Copper and aluminium are the two main materials in wires and cables, with some also being used in transformers. Copper has long been the preferred choice for electricity grids due to its high electrical and thermal conductivity. However, copper is over three times heavier by weight than aluminium and is more costly.
Copper is widely used for underground and subsea cables where weight is not a major concern and superior technical properties (e.g. corrosion resistance, tensile strength) are required. By contrast, aluminium is commonly used for overhead lines given its weight advantage. In some instances, aluminium is also used for underground and subsea cables.
Annual copper demand for electricity grids grows from 5 Mt in 2020 to 7.5 Mt by 2040 in the STEPS and to nearly 10 Mt in the SDS. Aluminium demand increases at a similar annual pace, from 9 Mt in 2020 to 12.8 Mt in the STEPS and 16 Mt in the SDS by 2040.
Minerals account for a considerable share in total investment costs for grids. Using average prices over the past 10 years, copper and aluminium costs are estimated to represent around 14% and 6% of total grid investment respectively.
One option to reduce raw material costs is to switch from copper to more affordable aluminium. If aluminium takes a higher share in underground and subsea cables, copper demand could be reduced by 3.6 Mt (down by a third) in 2040 while raising aluminium demand by 5.8 Mt (up by over a third).
Another option is to adopt HVDC systems more widely, which uses one-third less metal compared to AC systems and are capable of transporting more electricity. A wider uptake of HVDC systems could reduce combined demand for copper and aluminium in 2040 by 4 Mt (or 15%) in the SDS.
Electric vehicles and battery storage
The adoption of EVs and battery storage is set to accelerate rapidly over the coming decades
Electric car sales worldwide climbed 40% in 2020 to around 3 million, reaching a market share of over 4%. As a result, more than 10 million electric cars are now on the road globally. In the SDS, electric car sales exceed 70 million in 2040, alongside the rapid electrification of light commercial vehicles, buses and freight trucks.
As of the end of 2020, around 15.5 GW of battery storage capacity were connected to electricity networks. After annual installations of battery storage technologies fell for the first time in nearly a decade in 2019, they rebounded by over 60% in 2020. In the SDS, global installation of utility-scale battery storage is set for a 25-fold increase between 2020 and 2040, with annual deployment reaching 105 GW by 2040. The largest markets for battery deployment in 2040 are India, the United States and China.
The evolution of cathode and anode chemistries could drive mineral use for batteries in varying directions
Lithium-ion batteries are often categorised by the chemistry of their cathodes, such as lithium iron phosphate (LFP), lithium nickel cobalt aluminium oxide (NCA) and lithium nickel manganese cobalt oxide (NMC). The different combination of minerals gives rise to significantly different battery characteristics.
As it has become evident that reducing cobalt content in the cathode and striving for higher energy density are key concerns for many manufacturers and countries, the base case scenario sees a shift away from cobalt-rich chemistries. While most heavy trucks are reliant on LFP batteries in the medium term, our base case also sees modest growth in the market share of LFP for cars due to its increasing use in China and entry-level models.
Significant improvements in energy density and further declines in battery prices will likely require technologies beyond liquid electrolyte-based lithium-ion batteries. Such a breakthrough is expected from the advent of lithium metal anode all solid-state batteries (ASSBs). The base case sees ASSB becoming commercially available by around 2030 and requiring another five years for manufacturing capacity to build up.
Overall mineral demand from EVs in the SDS grows by nearly 30 times between 2020 and 2040, with demand for lithium and nickel growing by around 40 times
In the SDS, battery demand from EVs grows by nearly 40 times between 2020 (160 GWh) and 2040 (6 200 GWh). Overall demand for minerals under the base case assumptions grows by 30 times between 2020 and 2040, from 400 kt to 11 800 kt. In the STEPS, battery demand from EVs grows just 11 times to nearly 1 800 GWh in 2040, with demand for minerals growing ninefold to around 3 500 kt in 2040.
In the SDS, nickel demand grows by 41 times to 3 300 kt, while cobalt increases by only 21 times, as cathode chemistries shift away from NMC 111 towards lower-cobalt chemistries (NMC 622 and NMC 811). Lithium demand grows by 43 times, while copper demand grows by 28 times.
Graphite demand grows 25 times from 140 kt in 2020 to over 3 500 kt in 2040. Silicon registers the largest relative growth, up over 460 times, as graphite anodes doped with silicon grow from a 1% share in 2020 to 15% in 2040. Demand for REEs grows 15 times to 35 kt in 2040.
The alternative cases demonstrate the considerable sensitivity and uncertainty of mineral demand to the future mix of EV battery chemistries
We explore three alternative scenarios to assess how the demand outlook for various minerals could change under varying technology evolution trends.
A delayed shift to nickel-rich chemistries (and away from cobalt-rich chemistries) results in nearly 50% higher demand for cobalt and manganese in 2040 compared to the base case. Nickel demand is 5% lower in 2040 compared to the base case.
The faster uptake of lithium metal anodes and ASSB results in 22% higher lithium demand in 2040 compared to the base case, but also much lower demand for graphite (down 44%) and silicon (down 33%).
Moving rapidly towards a silicon-rich anode results in nearly three times as much silicon demand in 2030 compared to the base case, and a slight decrease in graphite demand (down 6%). By 2040 silicon demand is only 70% higher, owing to a higher adoption of silicon-rich anodes even in the base case.
Mineral demand for storage in the SDS grows by over 30 times between 2020 and 2040, with demand for nickel and cobalt growing by 140 times and 70 times respectively
Safe and cheaper LFP batteries for utility-scale storage are expected to dominate the overall battery storage market. The remaining demand is covered by the more expensive, but energy-dense, NMC 111 and NMC 532 used predominantly for home energy storage. The NMC variants transition towards NMC 622 and NMC 811 in a similar way to the market for EV batteries, albeit with a delay owing to the time needed for transfer of technology and sufficient reduction in prices. Vanadium flow batteries (VFBs) first become commercially suitable in 2030 with a small share, growing modestly to capture a wider market for storage applications in large renewables projects.
In the SDS, battery storage grows by 11 times between 2020 (37 GWh) and 2040 (420 GWh). Overall demand for minerals in the base case grows by 33 times between 2020 and 2040, from 26 kt to nearly 850 kt. Overall mineral demand outpaces battery demand growth, as the market share for LFP batteries is displaced by more mineral-intensive NMC chemistries. The largest relative growth is seen in nickel, which grows more than 140 times from 0.4 kt in 2020 to 57 kt in 2040. Cobalt demand increases by 70 times while manganese demand increases by 58 times.
For battery storage, we explore two alternative scenarios. A more rapid adoption of wall-mounted home energy storage would make size and thus energy density a prime concern, thereby pushing up the market share of NMC batteries. The rapid adoption of home energy storage with NMC chemistries results in 75% higher demand for nickel, manganese and cobalt in 2040 compared to the base case. A faster uptake of silicon-rich anodes also results in 20% greater demand for silicon compared to the base case in 2040.
If flow batteries achieve widespread commercialisation earlier than expected, then utility-scale storage technology could shift away from LFP batteries towards vanadium flow batteries. The early commercialisation of vanadium flow batteries results in 2.5 times more demand for vanadium compared to the base case in 2030 and 50% more demand in 2040. As a result of lower market shares for NMC chemistries, demand for nickel, cobalt and manganese are about 20% lower in 2040 compared to the base case.
Could mineral prices be an obstacle for further battery cost declines?
The average cost of lithium-ion batteries has fallen dramatically over the past decade, reaching USD 137/kWh in 2020. Further cost reductions are necessary for EVs to achieve the adoption rates observed in the SDS. However, with major technological improvements achieved over the past decade, raw materials now account for the majority of total battery costs (50–70%), up from around 40–50% five years ago. Cathode (25–30%) and anode materials (8–12%) account for the largest shares.
Given the importance of material costs in total battery costs, higher mineral prices could have a significant effect on achieving industry cost targets. For example, a doubling of lithium or nickel prices would induce a 6% increase in battery costs. If these events happen at the same time, the cost increase would eat up the anticipated learning effects associated with a doubling of capacity. It is therefore of paramount importance for governments and industry to work to ensure adequate supply of battery metals to mitigate any price increases, and the resulting challenges for clean electrification.
High prices for rare earth elements could see a shift away from permanent-magnet motors towards induction motors, increasing demand for copper or aluminium
Over 90% of the EVs marketed today use permanent-magnet synchronous motors due to their high efficiency, compact size and high power density. However, their use of REEs such as neodymium, praseodymium, dysprosium and terbium – upwards of 1 kg per motor – raises concerns given the geographical concentration of raw material and processing in China, the lack of recycling pathways and high price fluctuations.
There are several pathways to reducing REE use in EV motors: (i) improving material efficiency in magnet production to obtain NdFeB magnets with less REE content but with similar performance; (ii) reducing the amount of NdFeB magnets in permanent-magnet synchronous motors; (iii) substituting permanent-magnet motors with REE-free motors.
Hydrogen
Electrolysers and fuel cells could drive up demand for nickel, platinum and other minerals, but the market effects will depend on the shares of the different electrolyser types
Electrolyser capacity for low-carbon hydrogen production rises to around 1 400 GW in 2050 (in electricity input terms) in the SDS. Compared with today’s level of electrolyser manufacture – for which factory capacity is under 5 GW worldwide – this is a large increase and will require a corresponding scale up of mineral inputs.
Estimating mineral inputs for electrolysers is complicated by the different mineral intensities of the competing electrolyser designs. There is uncertainty about which of the three main types of electrolyser might dominate the market.
Alkaline electrolysers are currently the most widely used. They have low capital costs, partly because of their avoidance of precious metals, but current designs do require nickel in quantities of more than one tonne per MW. Reductions in nickel demand for alkaline electrolysers are expected, but nickel is not expected to be eliminated from future designs. However, even if alkaline electrolysers dominate the market, then nickel demand for electrolysers would remain much lower than that for batteries in the SDS. Proton exchange membrane (PEM) electrolsyers are more expensive today, but are already being deployed in large facilities as they are smaller and more flexible. While PEM uses more precious metals than alkaline – for example, around 0.3 kg of platinum per MW today – it is not expected to become a dominant source of platinum and iridium demand, even at the deployment levels in the SDS. Solid oxide electrolysers are at an earlier stage of development, though their higher efficiencies and reversibility make them a potentially attractive option. Like alkaline electrolysers, they currently use nickel, as well as rare earth elements like lanthanum, yttrium and zirconium.
Despite having a marginal impact on total energy demand for critical minerals in the SDS, the mineral requirements of electrolysers are a significant cost component and this could impact their competitiveness if mineral prices rise in response to demand, for example from batteries, fuel cells and other clean energy technologies.
In the SDS, platinum demand for vehicles in 2040 remains dominated by catalytic converters and not fuel cells
While the automotive sector is set to become a dominant source of global demand for lithium, nickel and cobalt for EV batteries, it already leads demand for platinum and palladium for use in catalytic converters. For these so-called platinum group metals, a key issue is whether new demand from fuel cells will offset declining demand from internal combustion engine vehicles.
While fuel cells for converting hydrogen to electricity have been in production for many years, the introduction of commercial passenger FCEVs has spurred innovation to reduce the use of platinum to limit costs. For example, Toyota’s second-generation Mirai, released in 2020, uses over 80% less platinum per kW of output than the 2008 prototype and roughly a third less than the first-generation from 2014. If targets to reduce platinum loading per kW are met, demand for FCEVs in the SDS would grow platinum demand to just over 100 tonnes by 2040.
Catalytic converters represent around 40% of global platinum demand today, and are also the major source of demand for two other platinum group metals: rhodium and palladium. In the SDS, an increase in the coverage of emissions regulation to include all new cars by 2030, coupled with continued sales of internal combustion engine, especially hybrids, keeps demand for platinum group metals for use in catalytic converters above that for fuel cells by 2040.
References
This report considers a wide range of minerals and metals used in clean energy technologies, including chromium, copper, major battery metals (lithium, nickel, cobalt, manganese and graphite), molybdenum, platinum group metals, zinc, rare earth elements and others (see Annex A for the complete list). Steel and aluminium are not included in the scope for demand assessment, but aluminium use in electricity networks is exceptionally assessed given that the outlook for copper is closely linked with aluminium use in grid lines (see Introduction).
See Annex for methodologies and data sources.
This report considers a wide range of minerals and metals used in clean energy technologies, including chromium, copper, major battery metals (lithium, nickel, cobalt, manganese and graphite), molybdenum, platinum group metals, zinc, rare earth elements and others (see Annex A for the complete list). Steel and aluminium are not included in the scope for demand assessment, but aluminium use in electricity networks is exceptionally assessed given that the outlook for copper is closely linked with aluminium use in grid lines (see Introduction).
See Annex for methodologies and data sources.