Andreas Krause

Cloud-to-ground (CG) lightning poses significant risks to infrastructure, people, and ecosystems. However, while many studies have projected changes in total lightning under warming, future trends in CG lightning remain poorly constrained. Here, we combine ground-based CG lightning observations with six daily surface meteorological variables (mean temperature, total precipitation, short-wave radiation, relative humidity, pressure, wind speed) to predict CG lightning density over land at 0.5° resolution using an XGBoost machine learning model. The model accurately captures contemporary spatial and temporal lightning patterns and outperforms an alternative multiple linear regression model as well as three existing parameterizations based on convective precipitation or convective available potential energy times precipitation. Applying the XGBoost model to historical climate shows a 11% increase in global CG lightning frequency over the 20th century. Future projections using outputs from two climate models (MPI-ESM1-2-HR and UKESM1-0-LL) reveal large increases of 76-140% under a high-emission scenario (SSP3-7.0), equivalent to roughly 30% more CG lightning per degree of global warming. These increases are substantially higher than those projected by the linear model and previous studies. Absolute increases are largest in the tropics, while relative increases are largest at high latitudes, where lightning density often more than doubles. The largest uncertainties emerge in the tropics for UKESM1-0-LL climate, particularly in the Amazon and New Guinea, where future climate differs strongly from current conditions. Our results suggest that climate change may substantially amplify CG lightning frequency, underscoring the need for improved risk assessments.

Machine-learning approaches to predicting lightning have achieved remarkable evaluation scores, far exceeding traditional parameterizations. However, high performance, even on unseen years, does not necessarily demonstrate robust physical relationships between predictors and lightning that hold under climate change. Here, we investigate how similarity between training and testing data affects lightning model evaluation. We introduce a simple benchmark predicting the training-data mean lightning density for each grid-cell and day of year, which reproduces 92% of the variance in daily grid-cell LIS/OTD observations and outperforms machine-learning models and traditional parameterizations. This high apparent skill is largely attributable to the strong spatiotemporal smoothing of LIS/OTD. We further show that previously reported model skill in reproducing inter-annual variability results from inconsistent observational coverage and disappears under a common spatiotemporal domain. Our results suggest that common lightning-model evaluations can overstate projection skill and highlight the need to test models under conditions differing from the training distribution.

Chemistry-climate models are our main tool to project how global lightning activity may change in a warmer world. However, the spatial and temporal scales involved in cloud electrification are too small to be explicitly resolved, so lightning is typically simulated using empirical relationships with proxy variables such as cloud top height (CTH), convective precipitation, convective available potential energy times precipitation, or upward ice flux. This review summarizes existing global lightning parameterizations, their skill in reproducing observed lightning patterns, and their projected future changes. We identify 16 parameterizations (including some variants) that have been applied in global modelling studies, with CTH — the earliest developed — remaining the most widely used. Most parameterizations predict total lightning, while cloud-to-ground lightning is rarely computed directly. Several parameterizations, including CTH, reproduce observed spatial lightning patterns reasonably well (r > 0.7), whereas precipitation-based methods perform worse (r often < 0.6). Most approaches underestimate lightning over central Africa and in mid-to-high latitudes, while overestimating activity over the Amazon and Maritime Continent. They also struggle to capture the observed land-ocean contrast in lightning activity and tend to underestimate temporal variability from diurnal to inter-annual scales. Future projections show large uncertainty, ranging from an 11 % decrease to a 44 % increase in lightning frequency per degree global warming. The mean across all parameterizations is +4.4±7.6 % K^-1, while the mean across all projections (which are strongly biased towards CTH) is +7.5±7.9 % K^-1. Although there is a broad agreement that lightning activity will increase in high latitudes, the tropical response remains highly uncertain.

Chemistry-climate models are our main tool to project how global lightning activity may change in a warmer world. However, the spatial and temporal scales involved in cloud electrification are too small to be explicitly resolved, so lightning is typically simulated using empirical relationships with proxy variables such as cloud top height (CTH), convective precipitation, convective available potential energy times precipitation, or upward ice flux. This review summarizes existing global lightning parameterizations, their skill in reproducing observed lightning patterns, and their projected future changes. We identify 16 parameterizations (including some variants) that have been applied in global modelling studies, with CTH — the earliest developed — remaining the most widely used. Most parameterizations predict total lightning, while cloud-to-ground lightning is rarely computed directly. Several parameterizations, including CTH, reproduce observed spatial lightning patterns reasonably well (r > 0.7), whereas precipitation-based methods perform worse (r often < 0.6). Most approaches underestimate lightning over central Africa and in mid-to-high latitudes, while overestimating activity over the Amazon and Maritime Continent. They also struggle to capture the observed land-ocean contrast in lightning activity and tend to underestimate temporal variability from diurnal to inter-annual scales. Future projections show large uncertainty, ranging from an 11 % decrease to a 44 % increase in lightning frequency per degree global warming. The mean across all parameterizations is +4.4±7.6 % K^-1, while the mean across all projections (which are strongly biased towards CTH) is +7.5±7.9 % K^-1. Although there is a broad agreement that lightning activity will increase in high latitudes, the tropical response remains highly uncertain.

Chemistry-climate models are our main tool to project how global lightning activity may change in a warmer world. However, the spatial and temporal scales involved in cloud electrification are too small to be explicitly resolved, so lightning is typically simulated using empirical relationships with proxy variables such as cloud top height (CTH), convective precipitation, convective available potential energy times precipitation, or upward ice flux. This review summarizes existing global lightning parameterizations, their skill in reproducing observed lightning patterns, and their projected future changes. We identify 16 parameterizations (including some variants) that have been applied in global modelling studies, with CTH — the earliest developed — remaining the most widely used. Most parameterizations predict total lightning, while cloud-to-ground lightning is rarely computed directly. Several parameterizations, including CTH, reproduce observed spatial lightning patterns reasonably well (r > 0.7), whereas precipitation-based methods perform worse (r often < 0.6). Most approaches underestimate lightning over central Africa and in mid-to-high latitudes, while overestimating activity over the Amazon and Maritime Continent. They also struggle to capture the observed land-ocean contrast in lightning activity and tend to underestimate temporal variability from diurnal to inter-annual scales. Future projections show large uncertainty, ranging from an 11 % decrease to a 44 % increase in lightning frequency per degree global warming. The mean across all parameterizations is +4.4±7.6 % K^-1, while the mean across all projections (which are strongly biased towards CTH) is +7.5±7.9 % K^-1. Although there is a broad agreement that lightning activity will increase in high latitudes, the tropical response remains highly uncertain.

Chemistry-climate models are our main tool to project how global lightning activity may change in a warmer world. However, the spatial and temporal scales involved in cloud electrification are too small to be explicitly resolved, so lightning is typically simulated using empirical relationships with proxy variables such as cloud top height (CTH), convective precipitation, convective available potential energy times precipitation, or upward ice flux. This review summarizes existing global lightning parameterizations, their skill in reproducing observed lightning patterns, and their projected future changes. We identify 16 parameterizations (including some variants) that have been applied in global modelling studies, with CTH — the earliest developed — remaining the most widely used. Most parameterizations predict total lightning, while cloud-to-ground lightning is rarely computed directly. Several parameterizations, including CTH, reproduce observed spatial lightning patterns reasonably well (r > 0.7), whereas precipitation-based methods perform worse (r often < 0.6). Most approaches underestimate lightning over central Africa and in mid-to-high latitudes, while overestimating activity over the Amazon and Maritime Continent. They also struggle to capture the observed land-ocean contrast in lightning activity and tend to underestimate temporal variability from diurnal to inter-annual scales. Future projections show large uncertainty, ranging from an 11 % decrease to a 44 % increase in lightning frequency per degree global warming. The mean across all parameterizations is +4.4±7.6 % K^-1, while the mean across all projections (which are strongly biased towards CTH) is +7.5±7.9 % K^-1. Although there is a broad agreement that lightning activity will increase in high latitudes, the tropical response remains highly uncertain.

Chemistry-climate models are our main tool to project how global lightning activity may change in a warmer world. However, the spatial and temporal scales involved in cloud electrification are too small to be explicitly resolved, so lightning is typically simulated using empirical relationships with proxy variables such as cloud top height (CTH), convective precipitation, convective available potential energy times precipitation, or upward ice flux. This review summarizes existing global lightning parameterizations, their skill in reproducing observed lightning patterns, and their projected future changes. We identify 16 parameterizations (including some variants) that have been applied in global modelling studies, with CTH — the earliest developed — remaining the most widely used. Most parameterizations predict total lightning, while cloud-to-ground lightning is rarely computed directly. Several parameterizations, including CTH, reproduce observed spatial lightning patterns reasonably well (r > 0.7), whereas precipitation-based methods perform worse (r often < 0.6). Most approaches underestimate lightning over central Africa and in mid-to-high latitudes, while overestimating activity over the Amazon and Maritime Continent. They also struggle to capture the observed land-ocean contrast in lightning activity and tend to underestimate temporal variability from diurnal to inter-annual scales. Future projections show large uncertainty, ranging from an 11 % decrease to a 44 % increase in lightning frequency per degree global warming. The mean across all parameterizations is +4.4±7.6 % K-1, while the mean across all projections (which are strongly biased towards CTH) is +7.5±7.9 % K-1. Although there is a broad agreement that lightning activity will increase in high latitudes, the tropical response remains highly uncertain.

Lightning is an important disturbance process in forest ecosystems, affecting trees both directly—when a strike kills a tree—and indirectly by igniting wildfires. While lightning–fire interactions are widely studied, direct lightning-induced tree mortality is not represented in global Earth System Models, limiting our ability to assess the full impact of lightning on forests under a changing climate.To address this gap, we implement lightning-induced tree mortality in the dynamic global vegetation model LPJ-GUESS, using field-derived relationships from a Panamanian forest where lightning mortality has been systematically quantified. The model successfully reproduces observed lightning-induced tree mortality at several sites but simulates lower mortality than estimated at other locations. Running the model globally, we quantify the number of trees and associated biomass directly lost to lightning and compare these losses to biomass losses from lightning-ignited wildfires, highlighting key uncertainties in both pathways.To place these present-day impacts in a future context, we synthesize existing lightning parameterizations used in global chemistry-climate models and assess their skill and projected changes in lightning activity. Applying projections from several well-performing parameterizations, we explore how future changes in lightning may alter both direct and indirect lightning-induced tree mortality. Together, our results demonstrate that lightning is a multifaceted and potentially growing driver of forest change, and that accurately representing lightning mortality is essential for robust projections of future forest dynamics.

Multi-functional forestry is a central objective of recent European policy frameworks, such as the New EU Forest Strategy for 2030 and the EU Biodiversity Strategy for 2030. These strategies, together with the LULUCF regulation, aim to ensure the continued provision of multiple forest ecosystem services under climate change, such as timber production, biodiversity conservation, and climate regulation.These expectations arise alongside increasing demands for wood products, leading to partially conflicting objectives. The uncertainty of future climate changes further complicates the development of multi-functional forestry strategies.Here, we demonstrate our recent work addressing these issues. We used process-based ecosystem modeling combined with robust multi-criteria optimization to derive forest management portfolios for climate-smart forestry under climate uncertainty. Using simplified management options and simulations across four RCPs, we show that regionally optimized portfolios can support the provision of multiple ecosystem services across a wide range of future climates. In particular, higher shares of broad-leaved and unmanaged forests were beneficial for biodiversity and other regulating services, but entailed clear trade-offs with timber provision.We further examined the effects of additional constraints, such as maintaining stable harvest levels and enforcing strict protection on 10% of the land area. These constraints substantially reduced management flexibility and made inter-regional compensation between wood production and forest protection necessary, often at the expense of multi-functionality within regions. Overall, our results highlight the difficulty of fulfilling all demands simultaneously under climate uncertainty. Nonetheless, they illustrate how the methodology can be helpful to derive forward-looking climate-smart strategies.