Publications by Duncan Watson-Parris


Biomass burning aerosols in most climate models are too absorbing

Nature Communications Nature Research 12 (2021) 277

H Brown, X Liu, R Pokhrel, S Murphy, Z Lu, R Saleh, T Mielonen, H Kokkola, T Bergman, G Myhre, R Skeie, D Watson-Parris, P Stier, B Johnson, N Bellouin, M Schulz, V Vakkari, JP Beukes, PG van Zyl, S Liu, D Chand

Uncertainty in the representation of biomass burning (BB) aerosol composition and optical properties in climate models contributes to a range in modeled aerosol effects on incoming solar radiation. Depending on the model, the top-of-the-atmosphere BB aerosol effect can range from cooling to warming. By relating aerosol absorption relative to extinction and carbonaceous aerosol composition from 12 observational datasets to nine state-of-the-art Earth system models/chemical transport models, we identify varying degrees of overestimation in BB aerosol absorptivity by these models. Modifications to BB aerosol refractive index, size, and mixing state improve the Community Atmosphere Model version 5 (CAM5) agreement with observations, leading to a global change in BB direct radiative effect of −0.07 W m−2, and regional changes of −2 W m−2 (Africa) and −0.5 W m−2 (South America/Temperate). Our findings suggest that current modeled BB contributes less to warming than previously thought, largely due to treatments of aerosol mixing state.


Machine learning for weather and climate are worlds apart.

Philos Trans A Math Phys Eng Sci 379 (2021) 20200098-

D Watson-Parris

Modern weather and climate models share a common heritage and often even components; however, they are used in different ways to answer fundamentally different questions. As such, attempts to emulate them using machine learning should reflect this. While the use of machine learning to emulate weather forecast models is a relatively new endeavour, there is a rich history of climate model emulation. This is primarily because while weather modelling is an initial condition problem, which intimately depends on the current state of the atmosphere, climate modelling is predominantly a boundary condition problem. To emulate the response of the climate to different drivers therefore, representation of the full dynamical evolution of the atmosphere is neither necessary, or in many cases, desirable. Climate scientists are typically interested in different questions also. Indeed emulating the steady-state climate response has been possible for many years and provides significant speed increases that allow solving inverse problems for e.g. parameter estimation. Nevertheless, the large datasets, non-linear relationships and limited training data make climate a domain which is rich in interesting machine learning challenges. Here, I seek to set out the current state of climate model emulation and demonstrate how, despite some challenges, recent advances in machine learning provide new opportunities for creating useful statistical models of the climate. This article is part of the theme issue 'Machine learning for weather and climate modelling'.


Climate impacts of COVID‐19 induced emission changes

Geophysical Research Letters Wiley (2020) e2020GL091805

A Gettelman, R Lamboll, C Bardeen, P Forster

The COVID‐19 pandemic led to dramatic changes in economic activity in 2020. We use estimates of emissions changes for 2020 in two Earth System Models (ESMs) to simulate the impacts of the COVID‐19 economic changes. Ensembles of nudged simulations are used to separate small signals from meteorological variability. Reductions in aerosol and precursor emissions, chiefly Black Carbon (BC) and sulfate (SO4), led to reductions in total anthropogenic aerosol cooling through aerosol‐cloud interactions. The average overall Effective Radiative Forcing (ERF) peaks at +0.29±0.15 Wm−2 in spring 2020. Changes in cloud properties are smaller than observed changes during 2020. Impacts of these changes on regional land surface temperature range up to +0.3K. The peak impact of these aerosol changes on global surface temperature is very small (+0.03K). However, the aerosol changes are the largest contribution to radiative forcing and temperature changes as a result of COVID‐19 affected emissions, larger than ozone, CO2 and contrail effects.


The CLoud–Aerosol–Radiation Interaction and Forcing: Year 2017 (CLARIFY-2017) measurement campaign

Atmospheric Chemistry and Physics Copernicus Publications 21 (2021) 1049-1084

SJ Abel, PA Barrett, N Bellouin, A Blyth, KN Bower, M Brooks, MI Cotterell, Z Cui, N Davies, B Dingley, P Field, P Formenti, H Gordon, R Herbert, B Johnson, AC Jones, JM Langridge, F Malavelle, F Peers, J Redemann, P Stier, JW Taylor, R Wood, H Wu

The representations of clouds, aerosols, and cloud–aerosol–radiation impacts remain some of the largest uncertainties in climate change, limiting our ability to accurately reconstruct past climate and predict future climate. The south-east Atlantic is a region where high atmospheric aerosol loadings and semi-permanent stratocumulus clouds are co-located, providing an optimum region for studying the full range of aerosol–radiation and aerosol–cloud interactions and their perturbations of the Earth's radiation budget. While satellite measurements have provided some useful insights into aerosol–radiation and aerosol–cloud interactions over the region, these observations do not have the spatial and temporal resolution, nor the required level of precision to allow for a process-level assessment. Detailed measurements from high spatial and temporal resolution airborne atmospheric measurements in the region are very sparse, limiting their use in assessing the performance of aerosol modelling in numerical weather prediction and climate models. CLARIFY-2017 was a major consortium programme consisting of five principal UK universities with project partners from the UK Met Office and European- and USA-based universities and research centres involved in the complementary ORACLES, LASIC, and AEROCLO-sA projects. The aims of CLARIFY-2017 were fourfold: (1) to improve the representation and reduce uncertainty in model estimates of the direct, semi-direct, and indirect radiative effect of absorbing biomass burning aerosols; (2) to improve our knowledge and representation of the processes determining stratocumulus cloud microphysical and radiative properties and their transition to cumulus regimes; (3) to challenge, validate, and improve satellite retrievals of cloud and aerosol properties and their radiative impacts; (4) to improve the impacts of aerosols in weather and climate numerical models. This paper describes the modelling and measurement strategies central to the CLARIFY-2017 deployment of the FAAM BAe146 instrumented aircraft campaign, summarizes the flight objectives and flight patterns, and highlights some key results from our initial analyses.


Cloud adjustments dominate the overall negative aerosol radiative effects of biomass burning aerosols in UKESM1 climate model simulations over the south-eastern Atlantic

Atmospheric Chemistry and Physics Copernicus Publications 21 (2021) 17-33

H Che, P Stier, H Gordon, D Watson-Parris, L Deaconu

The South-eastern Atlantic Ocean (SEA) is semi-permanently covered by one of the most extensive stratocumulus cloud decks on the planet and experiences about one-third of the global biomass burning emissions from the southern Africa savannah region during the fire season. To get a better understanding of the impact of these biomass burning aerosols on clouds and radiation balance over the SEA, the latest generation of the UK Earth System Model (UKESM1) is employed. Measurements from the CLARIFY and ORACLES flight campaigns are used to evaluate the model, demonstrating that the model has good skill in reproducing the biomass burning plume. To investigate the underlying mechanisms in detail, the effects of biomass burning aerosols on the clouds are decomposed into radiative effects (via absorption and scattering) and microphysical effects (via perturbation of cloud condensation nuclei (CCN) and cloud microphysical processes). The July-August means are used to characterise aerosols, clouds and the radiation balance during the fire season. Results show around 65% of CCN at 0.2% supersaturation in the SEA domain can be attributed to biomass burning. The absorption effect of biomass burning aerosols is the most significant in affecting clouds and radiation. Near the continent, it increases the maximum supersaturation diagnosed by the activation scheme, while further from the continent it reduces the altitude of the maximum supersaturation. As a result, the cloud droplet number concentration responds with a similar pattern to the absorption effect of biomass burning aerosols. The microphysical effect, however, decreases the maximum supersaturation and increases the cloud droplets concentration over the ocean; although this change is relatively small. The liquid water path is also significantly increased over the SEA (mainly caused by the absorption effect of biomass burning aerosols) when biomass burning aerosols are above the stratocumulus cloud deck. The microphysical pathways lead to a slight increase in the liquid water path over the ocean. These changes in cloud properties indicate the significant role of biomass burning aerosols on clouds in this region. Among the effects of biomass burning aerosols on the radiation balance, the semi-direct radiative effects (rapid adjustments induced by biomass burning aerosols radiative effects) have a dominant cooling impact over the SEA, which offset the warming direct radiative effect (radiative forcing from biomass burning aerosol–radiation interactions) and lead to overall net cooling radiative effect in the SEA. However, the magnitude and the sign of the semi-direct effects are sensitive to the relative location of biomass burning aerosols and clouds, reflecting the critical task of the accurate modelling of the biomass burning plume and clouds in this region.


Aerosol forcing masks and delays the formation of the North-Atlantic warming hole by three decades

Geophysical Research Letters American Geophysical Union 47 (2020) e2020GL090778

G Dagan, P Stier, D Watson-Parris

The North-Atlantic warming hole (NAWH) is referred to as a reduced warming, or even cooling, of the North-Atlantic during an anthropogenic-driven global warming. A NAWH is predicted by climate models during the 21st century and its pattern is already emerging in observations. Despite the known key role of the North-Atlantic surface temperatures in setting the Northern-Hemisphere climate, the mechanisms behind the NAWH are still not fully understood. Using state-of-the-art climate models, we show that anthropogenic aerosol forcing opposes the formation of the NAWH (by leading to a local warming) and delays its emergence by about 30 years. In agreement with previous studies, we also demonstrate that the relative warming of the North-Atlantic under aerosol forcing is due to changes in ocean heat fluxes, rather than air-sea fluxes. These results suggest that the predicted reduction in aerosol forcing during the 21st century may accelerate the formation of the NAWH.


The Southern Hemisphere midlatitude circulation response to rapid adjustments and sea surface temperature driven feedbacks

Journal of Climate American Meteorological Society 33 (2020) 9673-9690

T Wood, A Maycock, P Forster, T Richardson, T Andrews, O Boucher, G Myhre, B Samset, A Kirkevåg, J Lamarque, J MüLmenstädt, D Olivié, T Takemura, D Watson-Parris

Rapid adjustments-the response of meteorology to external forcing while sea surface temperatures (SST) and sea ice are held fixed-can affect the midlatitude circulation and contribute to long-term forced circulation responses in climate simulations. This study examines rapid adjustments in the Southern Hemisphere (SH) circulation using nine models from the Precipitation Driver and Response Model Intercomparison Project (PDRMIP), which perform fixed SST and coupled ocean experiments for five perturbations: a doubling of carbon dioxide (2xCO2), a tripling of methane (3xCH4), a fivefold increase in sulfate aerosol (5xSO4), a tenfold increase in black carbon aerosol (10xBC), and a 2% increase in solar constant (2%Sol). In the coupled experiments, the SH eddy-driven jet shifts poleward and strengthens for forcings that produce global warming (and vice versa for 5xSO4), with the strongest response found in austral summer. In austral winter, the responses project more strongly onto a change in jet strength. For 10xBC, which induces strong shortwave absorption, the multimodel mean (MMM) rapid adjustment in DJF jet latitude is ∼75% of the change in the coupled simulations. For the other forcings, which induce larger SST changes, the effect of SST-mediated feedbacks on the SH circulation is larger than the rapid adjustment. Nevertheless, for these perturbations the magnitude of the MMM jet shift due to the rapid adjustment is still around 20%-30% of that in the coupled experiments. The results demonstrate the need to understand the mechanisms for rapid adjustments in the midlatitude circulation, in addition to the effect of changing SSTs.


Overview: The CLoud-Aerosol-Radiation Interaction and Forcing: Year-2017 (CLARIFY-2017) measurement campaign

Atmospheric Chemistry and Physics Discussions European Geosciences Union (2020)

JM Haywood, SJ Abel, P Barrett, N Bellouin, A Blyth, K Bower, M Brooks, K Carslaw, H Che, M Cotterell, N Davies, B Dingley, P Field, H Gordon, M de Graaf, A Jones, J Langridge, F Malavelle, D Partridge, F Peers, J Reedemann, K Szpek, J Taylor, D Watson-Parris, P Zuidema

The representation of clouds, aerosols and cloud-aerosol-radiation impacts remain some of the largest uncertainties in climate change, limiting our ability to accurately reconstruct and predict future climate. The south-east Atlantic is a region where high atmospheric aerosol loadings and semi-permanent stratocumulus clouds are co-located, providing a natural laboratory for studying the full range of aerosol-radiation and aerosol-cloud interactions and their perturbations of the Earth’s radiation budget. While satellite measurements have provided some useful insights into aerosol-radiation and aerosol cloud interactions over the region, these observations do not have the spatial and temporal resolution, nor the required level of precision to allow for a process level assessment. Detailed measurements from high spatial and temporal resolution airborne atmospheric measurements in the region are very sparse, limiting their use in assessing the performance of aerosol modelling in numerical weather prediction and climate models. CLARIFY-2017 was a major consortium programme consisting of 5 principal UK universities with project partners from the UK Met Office and European and USA-based universities and research centres involved in the complementary ORACLES, LASIC and AEROCLO-sA projects. The aims of CLARIFY-2017 were four-fold; (1) to improve the representation and reduce uncertainty in model estimates of the direct, semi-direct and indirect radiative effect of absorbing biomass burning aerosols; (2) improve our knowledge and representation of the processes determining stratocumulus cloud microphysical and radiative properties and their transition to cumulus regimes; (3) challenge, validate and improve satellite retrievals of cloud and aerosol properties and their radiative impacts; (4) improve numerical models of cloud and aerosol and their impacts on radiation, weather and climate. This paper describes the modelling and measurement strategies central to the CLARIFY-2017 deployment of the FAAM BAe146 instrumented aircraft campaign, summarises the flight objectives and flight patterns, and highlights some key results from our initial analyses.


The hemispheric contrast in cloud microphysical properties constrains aerosol forcing

Proceedings of the National Academy of Sciences National Academy of Sciences 117 (2020) 18998-19006

IL McCoy, D McCoy, R Wood, L Regayre, D Watson-Parris, DP Grosvenor, JP Mulcahy, Y Hu, FAM Bender, PR Field, KS Carslaw, H Gordon

The change in planetary albedo due to aerosol−cloud interactions during the industrial era is the leading source of uncertainty in inferring Earth’s climate sensitivity to increased greenhouse gases from the historical record. The variable that controls aerosol−cloud interactions in warm clouds is droplet number concentration. Global climate models demonstrate that the present-day hemispheric contrast in cloud droplet number concentration between the pristine Southern Hemisphere and the polluted Northern Hemisphere oceans can be used as a proxy for anthropogenically driven change in cloud droplet number concentration. Remotely sensed estimates constrain this change in droplet number concentration to be between 8 cm−3 and 24 cm−3. By extension, the radiative forcing since 1850 from aerosol−cloud interactions is constrained to be −1.2 W⋅m−2 to −0.6 W⋅m−2. The robustness of this constraint depends upon the assumption that pristine Southern Ocean droplet number concentration is a suitable proxy for preindustrial concentrations. Droplet number concentrations calculated from satellite data over the Southern Ocean are high in austral summer. Near Antarctica, they reach values typical of Northern Hemisphere polluted outflows. These concentrations are found to agree with several in situ datasets. In contrast, climate models show systematic underpredictions of cloud droplet number concentration across the Southern Ocean. Near Antarctica, where precipitation sinks of aerosol are small, the underestimation by climate models is particularly large. This motivates the need for detailed process studies of aerosol production and aerosol−cloud interactions in pristine environments. The hemispheric difference in satellite estimated cloud droplet number concentration implies preindustrial aerosol concentrations were higher than estimated by most models.


Assessing California wintertime precipitation responses to various climate drivers

Journal of Geophysical Research: Atmospheres American Geophysical Union 125 (2020) e2019JD031736

RJ Allen, J-F Lamarque, D Watson-Parris, D Olivie

Understanding how drivers of climate change affect precipitation remains an important area of research. Although several robust precipitation responses have been identified under continued increases in greenhouse gases (GHGs), considerable uncertainty remains. This is particularly the case at regional scales, including the West Coast of the United States and California. Here, we exploit idealized, single forcing simulations from the Precipitation Driver Response Model Intercomparison Project (PDRMIP) to address how climate drivers impact California wintertime precipitation. Consistent with recent work, GHGs including carbon dioxide and methane, as well as solar forcing, yield a robust increase in California wintertime precipitation. We also find robust California precipitation responses to aerosols but with opposite responses for sulfate versus black carbon aerosol. Sulfate aerosol increases California wintertime precipitation, whereas black carbon reduces it. Moreover, California precipitation is more sensitive to aerosols, particularly regional emissions from Europe and Asia, than to GHGs. These precipitation responses are consistent with shifts in the jet stream and altered moisture fluxes. Although the idealized nature of PDRMIP simulations precludes a formal attribution, our results suggest that aerosols can perturb precipitation and fresh water resources along the West Coast of the United States.


The significant role of biomass burning aerosols in clouds and radiation in the South-eastern Atlantic Ocean

Atmospheric Chemistry and Physics Copernicus Publications (2020)

H Che, P Stier, H Gordon, D Watson-Parris, L Deaconu

The South-eastern Atlantic Ocean (SEA) is semi-permanently covered by one of the most extensive stratocumulus cloud decks on the planet and experiences about one-third of the global biomass burning emissions from the southern Africa savannah region during the fire season. To get a better understanding of the impact of these biomass burning aerosols on clouds and radiation balance over the SEA, the latest generation of the UK Earth System Model (UKESM1) is employed. Measurements from the CLARIFY and ORACLES flight campaigns are used to evaluate the model, demonstrating that the model has good skill in reproducing the biomass burning plume. To investigate the underlying mechanisms in detail, the effects of biomass burning aerosols on the clouds are decomposed into radiative effects (via absorption and scattering) and microphysical effects (via perturbation of cloud condensation nuclei (CCN) and cloud microphysical processes). The July–August means are used to characterise aerosols, clouds and the radiation balance during the fire season. Results show around 68 % of CCN at 0.2 % supersaturation in the SEA domain can be attributed to biomass burning. The absorption effect of biomass burning aerosols is the most significant in affecting clouds and radiation. Near the continent it increases the maximum supersaturation diagnosed by the activation scheme, while further from the continent it reduces the altitude of the maximum supersaturation. As a result, the cloud droplet number concentration shows a similar pattern. The microphysical effect of biomass burning aerosols decreases the maximum supersaturation and increases the cloud droplets concentration over the ocean; however, this change is relatively small. The liquid water path is also significantly increased over the SEA (mainly caused by the absorption effect of biomass burning aerosols) when biomass burning aerosols are above the stratocumulus cloud deck. The microphysical pathways lead to a slight increase in the liquid water path over the ocean. These changes in cloud properties indicate the significant role of biomass burning aerosols on clouds in this region. Among the effects of biomass burning aerosols on radiation balance, the semi-direct radiative effects (rapid adjustments induced by biomass burning aerosols radiative effects) have a dominant cooling impact over the SEA, which offset the warming direct radiative effect (radiative forcing from biomass burning aerosol–radiation interactions). However, the magnitude and the sign of the semi-direct effects are dependent on the relative location of biomass burning aerosols and clouds. The net biomass burning aerosols radiative effect shows a negative cooling effect in the SEA, indicating the significant role of biomass burning aerosols in affecting the regional radiation balance and climate.


Constraining uncertainty in aerosol direct forcing

Geophysical Research Letters American Geophysical Union 47 (2020) e2020GL087141

D Watson-Parris, N Bellouin, L Deaconu, N Schutgens, M Yoshioka, L Regayre, K Pringle, J Johnson, C Smith, K Carslaw, P Stier

The uncertainty in present-day anthropogenic forcing is dominated by uncertainty in the strength of the contribution from aerosol. Much of the uncertainty in the direct aerosol forcing can be attributed to uncertainty in the anthropogenic fraction of aerosol in the present-day atmosphere, due to a lack of historical observations. Here we present a robust relationship between total present-day aerosol optical depth and the anthropogenic contribution across three multi-model ensembles and a large single-model perturbed parameter ensemble. Using observations of aerosol optical depth, we determine a reduced likely range of the anthropogenic component and hence a reduced uncertainty in the direct forcing of aerosol.


Bounding global aerosol radiative forcing of climate change

Reviews of Geophysics American Geophysical Union 58 (2019) e2019RG000660

N Bellouin, J Quaas, S Kinne, P Stier, D Watson-Parris, O Boucher, KS Carslaw, M Christensen, A-L Daniau, JL Dufresne, G Feingold, S Fiedler, P Forster, A Gettelman, JM Haywood, U Lohmann, F Malavelle, T Mauritsen, DT McCoy, G Myhre, J Muelmenstaedt, A Possner, M Rugenstein, O Sourdeval, V Toll

Aerosols interact with radiation and clouds. Substantial progress made over the past 40 years in observing, understanding, and modeling these processes helped quantify the imbalance in the Earth's radiation budget caused by anthropogenic aerosols, called aerosol radiative forcing, but uncertainties remain large. This review provides a new range of aerosol radiative forcing over the industrial era based on multiple, traceable, and arguable lines of evidence, including modeling approaches, theoretical considerations, and observations. Improved understanding of aerosol absorption and the causes of trends in surface radiative fluxes constrain the forcing from aerosol‐radiation interactions. A robust theoretical foundation and convincing evidence constrain the forcing caused by aerosol‐driven increases in liquid cloud droplet number concentration. However, the influence of anthropogenic aerosols on cloud liquid water content and cloud fraction is less clear, and the influence on mixed‐phase and ice clouds remains poorly constrained. Observed changes in surface temperature and radiative fluxes provide additional constraints. These multiple lines of evidence lead to a 68% confidence interval for the total aerosol effective radiative forcing of ‐1.6 to ‐0.6 W m−2, or ‐2.0 to ‐0.4 W m−2 with a 90% likelihood. Those intervals are of similar width to the last Intergovernmental Panel on Climate Change assessment but shifted toward more negative values. The uncertainty will narrow in the future by continuing to critically combine multiple lines of evidence, especially those addressing industrial‐era changes in aerosol sources and aerosol effects on liquid cloud amount and on ice clouds.


Up to two billion times acceleration of scientific simulations with deep neural architecture search

CoRR abs/2001.08055 (2020)

MF Kasim, D Watson-Parris, L Deaconu, S Oliver, P Hatfield, DH Froula, G Gregori, M Jarvis, S Khatiwala, J Korenaga, J Topp-Mugglestone, E Viezzer, SM Vinko

Computer simulations are invaluable tools for scientific discovery. However, accurate simulations are often slow to execute, which limits their applicability to extensive parameter exploration, large-scale data analysis, and uncertainty quantification. A promising route to accelerate simulations by building fast emulators with machine learning requires large training datasets, which can be prohibitively expensive to obtain with slow simulations. Here we present a method based on neural architecture search to build accurate emulators even with a limited number of training data. The method successfully accelerates simulations by up to 2 billion times in 10 scientific cases including astrophysics, climate science, biogeochemistry, high energy density physics, fusion energy, and seismology, using the same super-architecture, algorithm, and hyperparameters. Our approach also inherently provides emulator uncertainty estimation, adding further confidence in their use. We anticipate this work will accelerate research involving expensive simulations, allow more extensive parameters exploration, and enable new, previously unfeasible computational discovery.


Efficacy of climate forcings in PDRMIP models

Journal of Geophysical Research: Atmospheres American Geophysical Union 124 (2019) 12824-12844

TB Richardson, PM Forster, CJ Smith, AC Maycock, T Wood, T Andrews, O Boucher, G Faluvegi, D Flaeschner, O Hodnebrog, M Kasoar, A Kirkevåg, J-F Lamarque, J Mülmenstädt, G Myhre, D Olivié, RW Portmann, BH Samset, D Shawki, D Shindell, P Stier, T Takemura, A Voulgarakis, D Watson-Parris

Quantifying the efficacy of different climate forcings is important for understanding the real‐world climate sensitivity. This study presents a systematic multi‐model analysis of different climate driver efficacies using simulations from the Precipitation Driver and Response Model Intercomparison Project (PDRMIP). Efficacies calculated from instantaneous radiative forcing deviate considerably from unity across forcing agents and models. Effective radiative forcing (ERF) is a better predictor of global mean near‐surface air temperature (GSAT) change. Efficacies are closest to one when ERF is computed using fixed sea surface temperature experiments and adjusted for land surface temperature changes using radiative kernels. Multi‐model mean efficacies based on ERF are close to one for global perturbations of methane, sulphate, black carbon and insolation, but there is notable inter‐model spread. We do not find robust evidence that the geographic location of sulphate aerosol affects its efficacy. GSAT is found to respond more slowly to aerosol forcing than CO2 in the early stages of simulations. Despite these differences, we find that there is no evidence for an efficacy effect on historical GSAT trend estimates based on simulations with an impulse response model, nor on the resulting estimates of climate sensitivity derived from the historical period. However, the considerable intermodel spread in the computed efficacies means that we cannot rule out an efficacy‐induced bias of ±0.4 K in equilibrium climate sensitivity to CO2 doubling (ECS) when estimated using the historical GSAT trend.


Cumulo: a dataset for learning cloud classes

Proceedings of the NeurIPS 2019 Workshop Tackling Climate Change with Machine Learning Climate Change AI (2019) 11

V Zantedeschi, F Falasca, A Douglas, R Strange, MJ Kusner, D Watson-Parris


tobac 1.2: towards a flexible framework for tracking and analysis of clouds in diverse datasets

Geoscientific Model Development Copernicus GmbH (2019)

M Heikenfeld, PJ Marinescu, MATTHEW Christensen, DUNCAN Watson-Parris, F Senf, SC van den Heever, PHILIP Stier

<jats:p>Abstract. We introduce tobac (Tracking and Object-Based Analysis of Clouds), a newly developed framework for tracking and analysing individual clouds in different types of datasets, such as cloud-resolving model simulations and geostationary satellite retrievals. The software has been designed to be used flexibly with any two- or three-dimensional time-varying input. The application of high-level data formats, such as Iris cubes or xarray arrays, for input and output allows for convenient use of metadata in the tracking analysis and visualisation. Comprehensive analysis routines are provided to derive properties like cloud lifetimes or statistics of cloud properties along with tools to visualise the results in a convenient way. The application of tobac is presented in two examples. We first track and analyse scattered deep convective cells based on maximum vertical velocity and the three-dimensional condensate mixing ratio field in cloud-resolving model simulations. We also investigate the performance of the tracking algorithm for different choices of time resolution of the model output. In the second application, we show how the framework can be used to effectively combine information from two different types of datasets by simultaneously tracking convective clouds in model simulations and in geostationary satellite images based on outgoing longwave radiation. The tobac framework provides a flexible new way to include the evolution of the characteristics of individual clouds in a range of important analyses like model intercomparison studies or model assessment based on observational data. </jats:p>


Water vapour adjustments and responses differ between climate drivers

Atmospheric Chemistry and Physics Copernicus Publications 19 (2019) 12887-12899

O Hodnebrog, G Myhre, B Samset, K Alterskjaer, T Andrews, O Boucher, G Faluvegi, D Fläschner, P Forster, M Kasoar, A Kirkevag, J-F Lamarque, D Olivie, T Richardson, D Shawki, D Shindell, KP Shine, P Stier, T Takemura, A Voulgarikis, D Watson-Parris

Water vapour in the atmosphere is the source of a major climate feedback mechanism and potential increases in the availability of water vapour could have important consequences for mean and extreme precipitation. Future precipitation changes further depend on how the hydrological cycle responds to different drivers of climate change, such as greenhouse gases and aerosols. Currently, neither the total anthropogenic influence on the hydrological cycle nor that from individual drivers is constrained sufficiently to make solid projections. We investigate how integrated water vapour (IWV) responds to different drivers of climate change. Results from 11 global climate models have been used, based on simulations where CO2, methane, solar irradiance, black carbon (BC), and sulfate have been perturbed separately. While the global-mean IWV is usually assumed to increase by ∼7 % per kelvin of surface temperature change, we find that the feedback response of IWV differs somewhat between drivers. Fast responses, which include the initial radiative effect and rapid adjustments to an external forcing, amplify these differences. The resulting net changes in IWV range from 6.4±0.9 % K−1 for sulfate to 9.8±2 % K−1 for BC. We further calculate the relationship between global changes in IWV and precipitation, which can be characterized by quantifying changes in atmospheric water vapour lifetime. Global climate models simulate a substantial increase in the lifetime, from 8.2±0.5 to 9.9±0.7 d between 1986–2005 and 2081–2100 under a high-emission scenario, and we discuss to what extent the water vapour lifetime provides additional information compared to analysis of IWV and precipitation separately. We conclude that water vapour lifetime changes are an important indicator of changes in precipitation patterns and that BC is particularly efficient in prolonging the mean time, and therefore likely the distance, between evaporation and precipitation.


Analysis of the atmospheric water budget for elucidating the spatial scale of precipitation changes under climate change

Geophysical Research Letters American Geophysical Union 46 (2019) 10504-10511

G Dagan, P Stier, D Watson-Parris

Global mean precipitation changes due to climate change were previously shown to be relatively small and well constrained by the energy budget. However, local precipitation changes can be much more significant. In this paper we propose that for large enough scales, for which the water budget is closed (precipitation [P] roughly equals evaporation [E]), changes in P approach the small global mean value. However, for smaller scales, for which P and E are not necessarily equal and convergence of water vapor still plays a role, changes in P could be much larger due to dynamical contributions. Using 40 years of two reanalysis data sets, 39 CMIP5 models and additional numerical simulations, we identify the scale of transition in the importance of the different terms in the water budget to precipitation to be ~3500-4000 km and demonstrate its relation to the spatial scale of precipitation changes under climate change.


Evaluation of global simulations of aerosol particle and cloud condensation nuclei number, with implications for cloud droplet formation

Atmospheric Chemistry and Physics Copernicus GmbH (2019)

GS Fanourgakis, M Kanakidou, A Nenes, T Bergman, KS Carslaw, A Grini, DS Hamilton, JS Johnson, VA Karydis, A Kirkevåg, JK Kodros, U Lohmann, G Luo, R Makkonen, H Matsui, D Neubauer, J Schmale, PHILIP Stier, K Tsigaridis, T van Noije, H Wang, DUNCAN Watson-Parris, DM Westervelt

<jats:p>&lt;p&gt;&lt;strong&gt;Abstract.&lt;/strong&gt; A total of 16 global chemistry transport models and general circulation models have participated in this study; 14 models have been evaluated with regard to their ability to reproduce the near-surface observed number concentration of aerosol particles and cloud condensation nuclei (CCN), as well as derived cloud droplet number concentration (CDNC). Model results for the period 2011–2015 are compared with aerosol measurements (aerosol particle number, CCN and aerosol particle composition in the submicron fraction) from nine surface stations located in Europe and Japan. The evaluation focuses on the ability of models to simulate the average across time state in diverse environments and on the seasonal and short-term variability in the aerosol properties.&lt;/p&gt; &lt;p&gt;There is no single model that systematically performs best across all environments represented by the observations. Models tend to underestimate the observed aerosol particle and CCN number concentrations, with average normalized mean bias (NMB) of all models and for all stations, where data are available, of &lt;span class="inline-formula"&gt;−24&lt;/span&gt;&amp;thinsp;% and &lt;span class="inline-formula"&gt;−35&lt;/span&gt;&amp;thinsp;% for particles with dry diameters &lt;span class="inline-formula"&gt;&amp;gt;50&lt;/span&gt; and &lt;span class="inline-formula"&gt;&amp;gt;120&lt;/span&gt;&amp;thinsp;nm, as well as &lt;span class="inline-formula"&gt;−36&lt;/span&gt;&amp;thinsp;% and &lt;span class="inline-formula"&gt;−34&lt;/span&gt;&amp;thinsp;% for CCN at supersaturations of 0.2&amp;thinsp;% and 1.0&amp;thinsp;%, respectively. However, they seem to behave differently for particles activating at very low supersaturations (&lt;span class="inline-formula"&gt;&amp;lt;0.1&lt;/span&gt;&amp;thinsp;%) than at higher ones. A total of 15 models have been used to produce ensemble annual median distributions of relevant parameters. The model diversity (defined as the ratio of standard deviation to mean) is up to about 3 for simulated &lt;span class="inline-formula"&gt;N&lt;sub&gt;3&lt;/sub&gt;&lt;/span&gt; (number concentration of particles with dry diameters larger than 3&amp;thinsp;nm) and up to about 1 for simulated CCN in the extra-polar regions. A global mean reduction of a factor of about 2 is found in the model diversity for CCN at a supersaturation of &lt;span class="inline-formula"&gt;0.2&lt;/span&gt;&amp;thinsp;% (CCN&lt;span class="inline-formula"&gt;&lt;sub&gt;0.2&lt;/sub&gt;&lt;/span&gt;) compared to that for &lt;span class="inline-formula"&gt;N&lt;sub&gt;3&lt;/sub&gt;&lt;/span&gt;, maximizing over regions where new particle formation is important.&lt;/p&gt; &lt;p&gt;An additional model has been used to investigate potential causes of model diversity in CCN and bias compared to the observations by performing a perturbed parameter ensemble (PPE) accounting for uncertainties in 26 aerosol-related model input parameters. This PPE suggests that biogenic secondary organic aerosol formation and the hygroscopic properties of the organic material are likely to be the major sources of CCN uncertainty in summer, with dry deposition and cloud processing being dominant in winter.&lt;/p&gt; &lt;p&gt;Models capture the relative amplitude of the seasonal variability of the aerosol particle number concentration for all studied particle sizes with available observations (dry diameters larger than 50, 80 and 120&amp;thinsp;nm). The short-term persistence time (on the order of a few days) of CCN concentrations, which is a measure of aerosol dynamic behavior in the models, is underestimated on average by the models by 40&amp;thinsp;% during winter and 20&amp;thinsp;% in summer.&lt;/p&gt; &lt;p&gt;In contrast to the large spread in simulated aerosol particle and CCN number concentrations, the CDNC derived from simulated CCN spectra is less diverse and in better agreement with CDNC estimates consistently derived from the observations (average NMB &lt;span class="inline-formula"&gt;−13&lt;/span&gt;&amp;thinsp;% and &lt;span class="inline-formula"&gt;−22&lt;/span&gt;&amp;thinsp;% for updraft velocities 0.3 and 0.6&amp;thinsp;m&amp;thinsp;s&lt;span class="inline-formula"&gt;&lt;sup&gt;−1&lt;/sup&gt;&lt;/span&gt;, respectively). In addition, simulated CDNC is in slightly better agreement with observationally derived values at lower than at higher updraft velocities (index of agreement 0.64 vs. 0.65). The reduced spread of CDNC compared to that of CCN is attributed to the sublinear response of CDNC to aerosol particle number variations and the negative correlation between the sensitivities of CDNC to aerosol particle number concentration (&lt;span class="inline-formula"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML" id="M15" display="inline" overflow="scroll" dspmath="mathml"&gt;&lt;mrow&gt;&lt;mo&gt;∂&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;N&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;d&lt;/mi&gt;&lt;/msub&gt;&lt;mo&gt;/&lt;/mo&gt;&lt;mo&gt;∂&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;N&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;a&lt;/mi&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;span&gt;&lt;svg:svg xmlns:svg="http://www.w3.org/2000/svg" width="48pt" height="14pt" class="svg-formula" dspmath="mathimg" md5hash="9faa6b9bc700a00532091cfd69cae419"&gt;&lt;svg:image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="acp-19-8591-2019-ie00001.svg" width="48pt" height="14pt" src="acp-19-8591-2019-ie00001.png"/&gt;&lt;/svg:svg&gt;&lt;/span&gt;&lt;/span&gt;) and to updraft velocity (&lt;span class="inline-formula"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML" id="M16" display="inline" overflow="scroll" dspmath="mathml"&gt;&lt;mrow&gt;&lt;mo&gt;∂&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;N&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;d&lt;/mi&gt;&lt;/msub&gt;&lt;mo&gt;/&lt;/mo&gt;&lt;mo&gt;∂&lt;/mo&gt;&lt;mi&gt;w&lt;/mi&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;span&gt;&lt;svg:svg xmlns:svg="http://www.w3.org/2000/svg" width="43pt" height="14pt" class="svg-formula" dspmath="mathimg" md5hash="9a0c289e263af38b17f8d2715a056c8f"&gt;&lt;svg:image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="acp-19-8591-2019-ie00002.svg" width="43pt" height="14pt" src="acp-19-8591-2019-ie00002.png"/&gt;&lt;/svg:svg&gt;&lt;/span&gt;&lt;/span&gt;). Overall, we find that while CCN is controlled by both aerosol particle number and composition, CDNC is sensitive to CCN at low and moderate CCN concentrations and to the updraft velocity when CCN levels are high. Discrepancies are found in sensitivities &lt;span class="inline-formula"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML" id="M17" display="inline" overflow="scroll" dspmath="mathml"&gt;&lt;mrow&gt;&lt;mo&gt;∂&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;N&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;d&lt;/mi&gt;&lt;/msub&gt;&lt;mo&gt;/&lt;/mo&gt;&lt;mo&gt;∂&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;N&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;a&lt;/mi&gt;&lt;/msub&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;span&gt;&lt;svg:svg xmlns:svg="http://www.w3.org/2000/svg" width="48pt" height="14pt" class="svg-formula" dspmath="mathimg" md5hash="a47c1357bf9f8959859c5d28931197ed"&gt;&lt;svg:image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="acp-19-8591-2019-ie00003.svg" width="48pt" height="14pt" src="acp-19-8591-2019-ie00003.png"/&gt;&lt;/svg:svg&gt;&lt;/span&gt;&lt;/span&gt; and &lt;span class="inline-formula"&gt;&lt;math xmlns="http://www.w3.org/1998/Math/MathML" id="M18" display="inline" overflow="scroll" dspmath="mathml"&gt;&lt;mrow&gt;&lt;mo&gt;∂&lt;/mo&gt;&lt;msub&gt;&lt;mi&gt;N&lt;/mi&gt;&lt;mi mathvariant="normal"&gt;d&lt;/mi&gt;&lt;/msub&gt;&lt;mo&gt;/&lt;/mo&gt;&lt;mo&gt;∂&lt;/mo&gt;&lt;mi&gt;w&lt;/mi&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;span&gt;&lt;svg:svg xmlns:svg="http://www.w3.org/2000/svg" width="43pt" height="14pt" class="svg-formula" dspmath="mathimg" md5hash="2e76a96aacff9b259f027c6bf554be27"&gt;&lt;svg:image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="acp-19-8591-2019-ie00004.svg" width="43pt" height="14pt" src="acp-19-8591-2019-ie00004.png"/&gt;&lt;/svg:svg&gt;&lt;/span&gt;&lt;/span&gt;; models may be predisposed to be too “aerosol sensitive” or “aerosol insensitive” in aerosol–cloud–climate interaction studies, even if they may capture average droplet numbers well. This is a subtle but profound finding that only the sensitivities can clearly reveal and may explain inter-model biases on the aerosol indirect effect.&lt;/p&gt; </jats:p>

Pages