An effective way to mitigate climate change is to electrify most of our energy demand and supply the necessary electricity from renewable wind and solar power plants. Spatio-temporal predictions of electric load become increasingly important for planning this transition, while deep learning prediction models provide increasingly accurate predictions for it. The data that are used for training deep learning models, however, are usually collected at random using a passive learning approach. This naturally results in a large demand for data and associated costs for sensors such as smart meters, posing a large barrier for electric utilities when decarbonizing their grids. Here we investigate whether electric utilities can use active learning to collect a more informative subset of data by leveraging additional computation for better distributing smart meters. We predict ground-truth electric load profiles for single buildings using only remotely sensed data from aerial imagery of these buildings and meteorological conditions in the area of these buildings at different times. We find that active learning can enable 26–81% more accurate predictions using 29–46% less data at the price of 4–11 times more computation compared with passive learning.@en

In the version of this Analysis originally published, the total learning rate for parabolic trough stations with 6–8 hours of thermal storage (Fig. 2b) was calculated to be 25%. After publication, the authors found a code error that caused the learning curve fit function to believe that the first station in the dataset was marked as 1 GW and not 0 GW. As a result, the estimated learning rates for the complete timespan were too high. The correct learning rates should be 2.7% for Fig. 2a and 6.8% for Fig. 2b (instead of 5.2% and 25.2%, respectively). These learning rate fit curves have been updated and the captions have been corrected. For consistency with Fig. 2a, a fit for 2011–2017 has been added to Fig. 2b, showing a learning rate of 17.5% (R2 = 0.337). The text has been modified in the abstract and the sections ‘Observed investment cost development and learning rates’, ‘Policy regime impacts on cost development’ and ‘Conclusions’ to reflect the quantitative changes to the learning rates. Supplementary Notes 1, 3 and 4 and Supplementary Figs. 2, 6 and 7 and their captions have also been updated to reflect the new learning rates. In the caption of Supplementary Fig. 2b, “(2008–2017 learning rate=0.21, R2=0.468)” has been changed to “(2008–2017 learning rate=0.06, R2=0.513; 2011–2017 learning rate=0.079, R2=0.072)”. In the caption of Supplementary Fig. 6b, “(2008–2017 learning rate=0.289, R2=0.715)” has been changed to “(2008–2017 learning rate=0.077, R2=0.631; 2011–2017 learning rate=0.225, R2=0.498)”. In the caption of Supplementary Fig. 7, “(a) parabolic trough (PT) stations with <1 hour thermal storage (2011–2014 learning rate=0.297, R2=0.972 (US$) and learning rate=0.27, R2=0.909 (€)); and (b) PT stations with 6–8 hours of thermal storage (2008–2017 learning rate=0.252, R2=0.621 (US$) and learning rate=0.138, R2=0.502 (€))” has been changed to “(a) parabolic trough (PT) stations with <1 hour thermal storage (2011–2014 learning rate=0.297, R2=0.972 (US$) and 2011–2014 learning rate=0.27, R2=0.909 (€)); and (b) PT stations with 6-8 hours of thermal storage (2011–2017 learning rate=0.175, R2=0.337 (US$) and 2011–2017 learning rate=0.072, R2=0.149 (€))”. The underlying data were correct as originally published and remain unchanged.@en

There is a large literature suggesting that improvements in energy efficiency support efforts at climate mitigation. Addressing a conceptual gap in that literature, however, we evaluate whether there are any conditions under which policies to promote improvements in energy efficiency could be counterproductive to efforts to limit climate change to 1.5 °C global warming from pre-industrial times. We identify three conditions under which this could be the case. The first condition is if policies for energy efficiency have a political opportunity cost, in terms of crowding out or delaying policies aimed at decarbonizing energy supply. There is an extensive literature in the fields of political science and policy studies to suggest that this is possible, but there have been no studies examining whether it has actually happened or is likely to happen in the future. The second condition is if investments in energy efficiency improvements come at a higher cost, per unit of fossil energy avoided, than do investments in new renewable energy supply. Current cost estimates suggest that there are some energy efficiency investments for which this is the case, but it is difficult to predict whether this will remain the case in the future. The third condition is if policies for energy efficiency, or specific investments in energy efficiency, were to delay the complete decarbonization of energy supply by more than some critical value. We show that critical delay is quite short—measured in weeks to months—in the case of a 1.5 °C temperature target, assuming constrained availability of negative emission technologies. It is impossible to say whether any of these conditions is likely, but in theory, each of them would appear to be possible.@en

Concentrating solar power (CSP) capacity has expanded slower than other renewable technologies and its costs are still high. Until now, there have been too few CSP projects to derive robust conclusions about its cost development. Here we present an empirical study of the cost development of all operating CSP stations and those under construction, examining the roles of capacity growth, industry continuity, and policy support design. We identify distinct CSP expansion phases, each characterized by different cost pressure in the policy regime and different industry continuity. In 2008-2011, with low cost pressure and following industry discontinuity, costs increased. In the current phase, with high cost pressure and continuous industry development, costs decreased rapidly, with learning rates exceeding 20%. Data for projects under construction suggest that this trend is continuing and accelerating. If support policies and industrial structure are sustained, we see no near-term factors that would hinder further cost decreases.@en

In a recent article, Trainer argues that electricity from concentrating solar power (CSP) in winter would be unreliable and prohibitively expensive, even if generated in premium desert sites. However, he does not carry out a detailed analysis of the reliability or the cost, but bases his conclusion on five arguments, each of which is either irrelevant or erroneous. In particular, his research question, concerning the cost of a CSP kilowatt-hour during winter is irrelevant, and the answer misleading, because the power station will deliver electricity in summer too. A more relevant and not misleading question would be about the performance - the yearly levelised costs of electricity and the reliability - of a CSP fleet. We argue based on a detailed analysis of the performance of CSP in four deserts worldwide, that a coordinated fleet of CSP stations can indeed provide fully dispatchable electricity, and in some cases even baseload, at low cost.@en

Solar power in the North African region has the potential to provide electricity for local energy needs and export to Europe. Nevertheless, despite the technical feasibility of solar energy projects, stakeholders still perceive projects in the region as risky because of existing governance issues. Certain areas of solar projects, such as construction, operation and management, are the most prone to governance risks, including lack of transparency and accountability, perceived as barriers for deployment of the projects. It is likely that large-scale foreign direct investment into solar energy will not eliminate existing risks, but might even increase them. Furthermore, the recent political changes in the region have addressed some governance risks but not all of them, especially bureaucratic corruption. Stakeholders recommend a broad set of measures to facilitate development of solar projects in the region, ranging from auditing of individual projects to simplification and unification of bureaucratic procedures.@en

Previous studies have demonstrated the possibility of maintaining a reliable electric power system with high shares of renewables, but only assuming the deployment of specific technologies in precise ratios, careful demand-side management, or grid-scale storage technologies. Any scalable renewable technology that could provide either baseload or dispatchable power would allow greater flexibility in planning a balanced system, and therefore would be especially valuable. Many analysts have suggested that concentrating solar power (CSP) could do just that. Here we systematically test this proposition for the first time. We simulate the operation of CSP plant networks incorporating thermal storage in four world regions where CSP is already being deployed, and optimize their siting, operation and sizing to satisfy a set of realistic demand scenarios. In all four regions, we show that with an optimally designed and operated system, it is possible to guarantee up to half of peak capacity before CSP plant costs substantially increase.@en

Across Europe, national governments have started to strategically plan adaptation to climate change. Making adaptation decisions is difficult in the light of uncertainties and the complexity of adaptation problems. Already large amounts of research results on climate impacts and adaptive measures are available, and more are produced and need to be mediated across the boundary between science and policy. Both researchers and policy-makers have started to intensify efforts to coproduce knowledge that is valuable to both communities, particularly in the context of climate change adaptation. In this paper, we present results from a study of adaptation governance and information needs, comparing eight European countries. We identify sources and means for the retrieval of information as well as gaps and problems with the knowledge provided by scientists and analyzed whether these appear to be contingent on the point in the policy-making cycle where countries are. We find that in this early phase of adaptation planning, the quality of the definition of needs, the way uncertainty is dealt with, and the quality of science-policy interaction are indeed contingent on the stage of adaptation planning, while information needs and sources are not. We conclude that a well-developed science-policy interface is of key importance for effective decision-making for adaptation.@en

This paper reviews the potential vulnerability of solar energy systems to future extreme event risks as a consequence of climate change. We describe the three main technologies likely to be used to harness sunlight-thermal heating, photovoltaic (PV), and concentrating solar power (CSP)-and identify critical climate vulnerabilities for each one. We then compare these vulnerabilities with assessments of future changes in mean conditions and extreme event risk levels. We do not identify any vulnerabilities severe enough to halt development of any of the technologies mentioned, although we do find a potential value in exploring options for making PV cells more heat-resilient and for improving the design of cooling systems for CSP.

@en