About
Research in the McKinnon group sits at the nexus of climate science and statistics. We are interested in questions that span the weather-climate divide, shed light on the interactions between natural variability and human-caused climate change, and link to downstream human and biological impacts.
Research topics
- Computer Science
- Environmental science
- Geography
- Geology
- Ecology
- Environmental resource management
- Physical geography
- Economics
- Management science
- Climatology
- Biology
Selected publications
2026-03-10
peer-reviewOpen access<strong class="journal-contentHeaderColor">Abstract.</strong> Parties to the 2015 Paris Agreement agreed to limit the long-term increase in global average temperature to well below 2 °C and pursue efforts to keep temperatures below 1.5 °C relative to pre-industrial levels. As the world is fast approaching the 1.5 °C warming level on a sustained basis, and with 2024 likely the first year that was over 1.5 °C warmer than 1850-1900, there is ever increasing interest in how we will know whether and when 1.5 °C warming since pre-industrial has been reached or exceeded with respect to a long-term average. This paper represents a comprehensive community methodological overview, building on the IPCC 6th assessment. It explains why there is no straightforward answer and proposes clear and reasoned ways forward. Existing challenges are as follows. Firstly, the Paris Agreement text contains definitional ambiguities around 'pre-industrial', 'global average temperature', whether the assessment should be on realised or long-term human-induced warming, and over what time frame the long-term temperature goal applies. Then, there are intrinsic limitations of observational records which get more uncertain further back in time due to data sparsity and measurement heterogeneity. Finally, in a non-stationary climate, multidecadal mean indicators of global temperature change will either lag behind the change or must rely on expected future temperature changes (based on extrapolation, initialized predictions, or scenario-based and constrained projections). Our analysis shows that knowing 'whether we are there yet' is a multifaceted and inherently probabilistic problem that includes information on the definition of a specific level of global warming, temperature changes over multiple timescales, and also potentially includes unpacking the attribution of human-caused changes from observed variations. Given the policy relevance of understanding where the world stands relative to 1.5 °C, or any other level of global warming since pre-industrial, there are a number of practical steps which could be taken to increase specificity in answering this critical question in a timely manner, and inform future monitoring and assessment activities. This paper reviews a broad range of approaches, identifies the most pragmatic, robust and transparent, and clarifies requirements for use in real time including how to handle and represent remaining uncertainties. We show that it is possible by combining lines of evidence and several methodologies to estimate the present long-term warming level without delay in a manner that is robust both in retrospective validation of crossing past warming levels and, critically, to divergent warming futures including potential wildcard impacts of large volcanoes which can mask underlying warming for several years. Results are benchmarked against historical exceedances of 0.5 °C and 1 °C warming. Long-term warming as assessed using the approaches developed herein and data up to and including 2024 stands at 1.40 [1.23–1.58] °C, and underlying human-caused warming stands at 1.34 [1.18–1.50] °C. In IPCC quantified likelihood language this means that it was<em> unlikely</em> that long-term realised warming had exceeded 1.5 °C by the end of 2024 and <em>very unlikely </em>that human-induced warming had exceeded 1.5 °C.
Journal of Applied Meteorology and Climatology · 2026-04-02
articleSenior authorAbstract Central American precipitation is highly variable due to both local and remote ocean–atmosphere controls from the Pacific and Atlantic basins. These complexities present challenges in developing skillful seasonal rainfall predictions. Here, we leverage the connections between Central American precipitation and large-scale features of climate variability including the Caribbean low-level jet (CLLJ), a critical yet relatively understudied driver of regional rainfall, to explore potential advances in seasonal rainfall predictions. Given the importance of the midsummer months for agriculture, we focus on predictions of July–August precipitation totals with lead times beginning in March and assess the skill of two forecasting methods: direct forecasts from seasonal prediction models and indirect forecasts that combine dynamical and statistical approaches. For the indirect forecasts, we statistically translate dynamical model predictions of large-scale climate indices to precipitation based on observed relationships. Both methods perform similarly in regions where forecast skill is relatively high. In these regions, the CLLJ appears to be a key source of information; indirect predictions using only the CLLJ are comparable to or show improvements over the direct and full indirect predictions. However, in regions where the CLLJ does not provide predictive skill, neither approach performs well. The relationship, or lack thereof, between the precipitation and the various large-scale drivers including the CLLJ, is confirmed through observational and model composites of extreme years. Finally, we compare our forecast methods to a commonly used operational tool, the Climate Predictability Tool, and discuss our findings in the context of regional forecasting efforts. Significance Statement Predicting seasonal rainfall can help reduce climate hazards and improve agricultural outcomes in Central America. We evaluate whether the Caribbean low-level jet (CLLJ), an understudied but important regional climate influence, can be used with other common climate features to improve midsummer rainfall forecasts. We compare direct precipitation predictions from seasonal forecast models with our “indirect” forecasts, which convert large-scale climate patterns such as sea surface temperatures and the CLLJ into regional rainfall predictions using a statistical model. We find that including the CLLJ in the indirect model improves predictions in some areas, but that its performance is often similar to direct precipitation forecasts. In regions where the CLLJ does not enhance the statistical model, forecast skill overall remains low.
2026-01-28
articleOpen accessForced Component Estimation Statistical Method Intercomparison Project (ForceSMIP)
Journal of Climate · 2026-01-23
articleOpen accessAbstract Anthropogenic climate change is unfolding rapidly, yet its regional manifestation can be obscured by internal variability. A primary goal of climate science is to identify the externally forced climate response from among the noise of internal variability. Separating the forced response from internal variability can be addressed in climate models by using a large ensemble to average over different possible realizations of internal variability. However, with only one realization of the real world, it is a major challenge to isolate the forced response directly in observations. In the Forced Component Estimation Statistical Method Intercomparison Project (ForceSMIP), contributors used existing and newly developed statistical and machine learning methods to estimate the forced response over 1950–2022 within individual realizations of the climate system. Participants used neural networks, linear inverse models, fingerprinting methods, and low-frequency component analysis, among other approaches. These methods were trained using large ensembles from multiple climate models and then applied to observations. Here, we evaluate method performance within large ensembles and investigate the estimates of the forced response in observations. Our results show that many different types of methods are skillful for estimating the forced response in climate models, though the relative skill of individual methods varies depending on the variable and evaluation metric. Methods with comparable skill in models can give a wide range of estimates of the forced response pattern in observations, illustrating the epistemic uncertainty in forced response estimates. ForceSMIP gives new insights into the forced response in observations, its uncertainty, and methods for its estimation. Significance Statement The Forced Component Estimation Statistical Method Intercomparison Project (ForceSMIP) aims to reduce uncertainty in estimates of the climate response to anthropogenic and other external forcing and to evaluate statistical and machine learning methods designed to estimate the forced response from individual realizations of the climate system. New and existing statistical and machine learning methods are evaluated within climate models, for which the forced response is known. Applying these methods to observations gives an estimate of the real-world forced response. The observational forced response estimate agrees with climate models on the large-scale features, but it also shows discrepancies that give insights into responses that may not be simulated well by climate models. In some regions with large internal variability, such as the North Atlantic Ocean, it remains difficult to determine the relative contributions of anthropogenic forcing and internal variability to historical changes.
How well can we quantify when 1.5 °C of global warming has been exceeded?
2026-01-28
articleOpen accessAbstract. Parties to the 2015 Paris Agreement agreed to limit the long-term increase in global average temperature to well below 2 °C and pursue efforts to keep temperatures below 1.5 °C relative to pre-industrial levels. As the world is fast approaching the 1.5 °C warming level on a sustained basis, and with 2024 likely the first year that was over 1.5 °C warmer than 1850-1900, there is ever increasing interest in how we will know whether and when 1.5 °C warming since pre-industrial has been reached or exceeded with respect to a long-term average. This paper represents a comprehensive community methodological overview, building on the IPCC 6th assessment. It explains why there is no straightforward answer and proposes clear and reasoned ways forward. Existing challenges are as follows. Firstly, the Paris Agreement text contains definitional ambiguities around 'pre-industrial', 'global average temperature', whether the assessment should be on realised or long-term human-induced warming, and over what time frame the long-term temperature goal applies. Then, there are intrinsic limitations of observational records which get more uncertain further back in time due to data sparsity and measurement heterogeneity. Finally, in a non-stationary climate, multidecadal mean indicators of global temperature change will either lag behind the change or must rely on expected future temperature changes (based on extrapolation, initialized predictions, or scenario-based and constrained projections). Our analysis shows that knowing 'whether we are there yet' is a multifaceted and inherently probabilistic problem that includes information on the definition of a specific level of global warming, temperature changes over multiple timescales, and also potentially includes unpacking the attribution of human-caused changes from observed variations. Given the policy relevance of understanding where the world stands relative to 1.5 °C, or any other level of global warming since pre-industrial, there are a number of practical steps which could be taken to increase specificity in answering this critical question in a timely manner, and inform future monitoring and assessment activities. This paper reviews a broad range of approaches, identifies the most pragmatic, robust and transparent, and clarifies requirements for use in real time including how to handle and represent remaining uncertainties. We show that it is possible by combining lines of evidence and several methodologies to estimate the present long-term warming level without delay in a manner that is robust both in retrospective validation of crossing past warming levels and, critically, to divergent warming futures including potential wildcard impacts of large volcanoes which can mask underlying warming for several years. Results are benchmarked against historical exceedances of 0.5 °C and 1 °C warming. Long-term warming as assessed using the approaches developed herein and data up to and including 2024 stands at 1.40 [1.23–1.58] °C, and underlying human-caused warming stands at 1.34 [1.18–1.50] °C. In IPCC quantified likelihood language this means that it was unlikely that long-term realised warming had exceeded 1.5 °C by the end of 2024 and very unlikely that human-induced warming had exceeded 1.5 °C.
Identifying and correcting inhomogeneities in the near-surface specific humidity record
2026-04-13
article1st authorCorrespondingComment on essd-2025-825 section 4.5
2026-02-06
peer-reviewOpen access<strong class="journal-contentHeaderColor">Abstract.</strong> Parties to the 2015 Paris Agreement agreed to limit the long-term increase in global average temperature to well below 2 °C and pursue efforts to keep temperatures below 1.5 °C relative to pre-industrial levels. As the world is fast approaching the 1.5 °C warming level on a sustained basis, and with 2024 likely the first year that was over 1.5 °C warmer than 1850-1900, there is ever increasing interest in how we will know whether and when 1.5 °C warming since pre-industrial has been reached or exceeded with respect to a long-term average. This paper represents a comprehensive community methodological overview, building on the IPCC 6th assessment. It explains why there is no straightforward answer and proposes clear and reasoned ways forward. Existing challenges are as follows. Firstly, the Paris Agreement text contains definitional ambiguities around 'pre-industrial', 'global average temperature', whether the assessment should be on realised or long-term human-induced warming, and over what time frame the long-term temperature goal applies. Then, there are intrinsic limitations of observational records which get more uncertain further back in time due to data sparsity and measurement heterogeneity. Finally, in a non-stationary climate, multidecadal mean indicators of global temperature change will either lag behind the change or must rely on expected future temperature changes (based on extrapolation, initialized predictions, or scenario-based and constrained projections). Our analysis shows that knowing 'whether we are there yet' is a multifaceted and inherently probabilistic problem that includes information on the definition of a specific level of global warming, temperature changes over multiple timescales, and also potentially includes unpacking the attribution of human-caused changes from observed variations. Given the policy relevance of understanding where the world stands relative to 1.5 °C, or any other level of global warming since pre-industrial, there are a number of practical steps which could be taken to increase specificity in answering this critical question in a timely manner, and inform future monitoring and assessment activities. This paper reviews a broad range of approaches, identifies the most pragmatic, robust and transparent, and clarifies requirements for use in real time including how to handle and represent remaining uncertainties. We show that it is possible by combining lines of evidence and several methodologies to estimate the present long-term warming level without delay in a manner that is robust both in retrospective validation of crossing past warming levels and, critically, to divergent warming futures including potential wildcard impacts of large volcanoes which can mask underlying warming for several years. Results are benchmarked against historical exceedances of 0.5 °C and 1 °C warming. Long-term warming as assessed using the approaches developed herein and data up to and including 2024 stands at 1.40 [1.23–1.58] °C, and underlying human-caused warming stands at 1.34 [1.18–1.50] °C. In IPCC quantified likelihood language this means that it was<em> unlikely</em> that long-term realised warming had exceeded 1.5 °C by the end of 2024 and <em>very unlikely </em>that human-induced warming had exceeded 1.5 °C.
Earth s Future · 2025-09-01 · 3 citations
articleOpen accessAbstract The Pacific Northwest (PNW) experienced a record‐breaking heatwave in late June 2021. Previous studies showed that an anomalous upper‐level anticyclone and associated subsidence heating, fueled by upwind latent heat release, were the main drivers. Land‐atmosphere interactions have generally been found to play a secondary but important role; however their temporal evolution and state dependence on prior soil moisture (SM) and evaporative regimes remain largely unexplored. To assess this, we run 100 ensemble members of the heatwave with varying initial land surface conditions in the Community Earth System Model version 2 (CESM2). The circulation outside the PNW is constrained to observations, ensuring that the large‐scale dynamical drivers are reproduced while local land‐atmosphere interactions are free to evolve in the PNW region under differing soil‐moisture states. While circulation largely dictates the heatwave's magnitude (peak day temperatures about 17C above the climatological mean), perturbations of the SM preconditioning across a realistic range for the time of year lead to about 3C spread in CESM2. We demonstrate how the land‐atmosphere interactions evolve as ensemble members fall below a critical SM threshold where evapotranspiration reduces substantially. We also investigate how an antecedent rain event might have affected this heatwave event. Finally, we simulate the same circulation induced heatwave but in a future climate state with higher greenhouse gases and drier soils. Beyond mean warming effects, drier soils increase the probability of shifting from a wet regime into a transitional regime, exacerbating and elongating the heatwave.
Forced Component Estimation Statistical Methods Intercomparison Project (ForceSMIP)
2025-03-15
preprintOpen accessAnthropogenic climate change is unfolding rapidly, yet its regional manifestation is often obscured by atmosphere-ocean internal variability. A primary goal of climate science is to identify the forced response, i.e., spatiotemporal changes in climate in response to greenhouse gases, anthropogenic aerosols, and other external forcing, amongst the noise of internal climate variability. Separating the forced response from internal variability can be addressed in climate models by using a large ensemble to average over different possible realizations of internal variability. However, with only one realization of the real world, it is a major challenge to isolate the forced response in observations, as is needed for attribution of historical climate changes, for characterizing and understanding observed internal variability, and for climate model evaluation.In the Forced Component Estimation Intercomparison Project (ForceSMIP), contributors used existing and newly developed statistical and machine learning methods to estimate the forced response during the historical period within individual ensemble members and observations, across eight key climate variables (SST, surface air temperature, precipitation, SLP, zonal-mean atmospheric temperature, monthly max. and min. temperature, and monthly max. precipitation). Participants could use five CMIP6 large ensembles to train their methods, but they then had to apply their methods to individual evaluation members, the identity of which was hidden. Participants used methods including regression methods, convolutional neural networks, linear inverse models, fingerprinting methods, and low-frequency component analysis. Here we show how the different methods performed on climate models and what they determined to the be the forced response in observations. Our results show that many different types of methods are skillful for estimating the forced response and that the most skillful method depends highly on which variable and metric is evaluated. Furthermore, methods that show comparable skill can give very different estimates of the forced response in observations, illustrating the epistemic uncertainty in estimating the forced climate response from observations. ForceSMIP gives new insights into the forced response in observations across multiple key variables, but also the remaining uncertainty in its estimation.
Detection and attribution of trends of meteorological extremes in Central America
Climatic Change · 2025-04-30 · 5 citations
articleOpen accessWe present an analysis to determine whether historical trends in extreme precipitation and temperature indices, as well as in yearly averages of several climate variables. To achieve this, we use three methodologies: a) a climate model-based approach, b) a hybrid method that combines models and observations (1979–2019), and c) a climate observations-based method (1983–2016). For each methodology, we compare the climate change signal, represented by the historical trends, to the noise generated by simulated climate datasets (using models or statistical methods) that do not include human influence. Overall, the model-based method suggests possible detection of the human influence in most temperature extreme indices and in precipitation-related indices in the northern countries. The hybrid method detects human influence in significantly fewer variables, but in many cases, consistently with those of the model-based approach. Both the hybrid and observation-based methods exhibit similar noise variability to the model-based method. Notably, due to limitations in data availability, our analysis excludes the most recent five years, during which substantial warming and an increase of extreme events have been observed globally.
Recent grants
The factors governing daily near-surface air temperature variability over land
NSF · $620k · 2020–2025
Frequent coauthors
- 45 shared
Isla R. Simpson
Climate and Global Dynamics Laboratory
- 34 shared
Peter Huybers
Harvard University
- 30 shared
Clara Deser
NSF National Center for Atmospheric Research
- 21 shared
A. N. Rhines
Netflix (United States)
- 19 shared
Martin P. Tingley
- 19 shared
Nicole S. Lovenduski
University of Colorado Boulder
- 12 shared
Geneviève Elsworth
Goddard Space Flight Center
- 12 shared
Flavio Lehner
NSF National Center for Atmospheric Research
Labs
McKinnon groupPI
Education
- 2015
PhD, Earth and Planetary Science
Harvard University
- 2014
MA, Earth and Planetary Science
Harvard University
- 2011
MSc, Geophysics
Victoria University of Wellington
- 2010
BA, Earth and Planetary Science
Harvard University
Awards & honors
- NSF CAREER award
- Kavli Fellow
- Donald Ylvisaker Award for the Best Practice of Statistics
- Resume-aware match score
- Save to shortlist
- AI-drafted outreach
See your match with Karen Aline McKinnon
PhdFit ranks faculty by your research interests, methods, and publications — grounded in their actual work, not templates.
- Free to start
- No credit card
- 30-second signup