ESMValTool : An Evaluation Tool for Earth System Models

By: Ranjini Swaminathan

Global Climate Models are computer programs that simulate the atmosphere, ocean, sea ice, land surface, and the effects of these components on one another. Earth System Models (ESMs) are global climate models that include biogeochemical cycling in terrestrial and marine ecosystems and explicitly model the movement of carbon through the earth system. Output from ESMs are estimates of variables such as precipitation and temperature at different spatial and temporal resolutions and are typically available as three dimensional gridded values.

The Coupled Model Intercomparison Project (CMIP), now in its sixth phase (CMIP6) provides the climate and impact assessment community with a wide range of simulation output from Earth System Models (Eyring et al., 2016b). These simulations help us better understand past climate and also provide estimates for the future under different climate scenarios. It is therefore important to evaluate model simulation output in order to better understand systemic biases in the models and the magnitude of uncertainty in estimates of future projections before extensively using them for analyses and impact assessment studies. There is also a need for a comprehensive framework that performs baseline aspects of model evaluation in a consistent and efficient manner. This will prevent scientists wasting valuable time and effort individually re-implementing well-established evaluation diagnostics instead of just sharing them. A tool or framework that implements basic metrics and diagnostics for model output evaluation with the flexibility of adding newly-researched,  science-based evaluation measures is therefore the need of the hour. We present the Earth System Model Evaluation Tool (ESMValTool) as a tool compatible with current and future phases of the CMIP and many desirable features for comprehensive model evaluation (Eyring et al., 2016a).

Figure 1: Overview and basic working of the ESMValTool

The ESMValTool  is an open-source community diagnostics and performance metrics tool for the evaluation of ESMs. It allows comparisons between multiple models against previous versions and observations. The tool provides a versatile preprocessor to standardize data with regridding, interpolation and masking as well as temporal and spatial extraction facilities. Standard recipes for scientific topics reproduce specific diagnostics and performance metrics considered important for ESM evaluation from peer-reviewed literature (Lauer et al., 2017). The tool is currently at version 2.0 and the most recent release allows new analyses and recipes to be written in one of many open-source languages such as Python, R and NCL. Figure 1 shows a high-level overview of the ESMValTool.

Some key features of the tool are:

  1. Simultaneous multimodel comparison with observed data. Figure 2 shows an example of seasonal standard deviations of surface temperature anomalies for four different CMIP5 models and observed data as generated by the ESMValTool. We see that three of the models show a greater variability than observed data.
  2. Extensive documentation with log files for better traceability and reproducibility.
  3. Wide scope with standard recipes covering different aspects of ESMs such as dynamics, radiation, aerosols and sea ice.
  4. Central installation on the Centre for Environmental Data Analysis’s (CEDA’s) JASMIN infrastructure with CMIP model data retrieval being automatically facilitated.
  5. Diagnostic outputs produced in convenient formats such as plots (pdf, png etc.) and netCDF files.

The ESMValTool v2.0 continues to evolve with diagnostics being ported from the previous release, more relevant diagnostics ported from other assessment tools as well as new ones being added. The tool can be accessed from the ESMValTool GitHub page and a sample of CMIP results produced with the ESMValTool is available for the German CMIP6 project. We encourage users to access the tool and try out the various diagnostics and metrics available.

Figure 2: Standard deviation of global surface temperature anomalies for the period 1982-2005. Figures (a) through (d) are  for different CMIP5 climate model temperature outputs and (e) shows the standard deviation for observations (in this case, from HadCRUT4). Missing data in the observations is represented in white.


Eyring et al. 2016a Towards improved and more routine Earth System Model Evaluation in CMIP5. Earth System Dynamics, 7, 813-830, http://doi:10.5194/esd-7-813-2016

Lauer et al. 2017 Benchmarking CMIP5 models with a subset of ESA CCI Phase 2 data using the ESMValTool, Remote Sensing Environment, 203, 9-39, http://doi:10.1016/j.rse.2017.01.007

The ESMValTool Webpage:,

Eyring et al. 2016b Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization, Geoscientific Model Development, 9, 1937-1958, 

ESMValTool GitHub Page:

German CMIP6 Project evaluation results with the ESMValTool:  


Posted in Climate, Climate modelling, Data processing | Leave a comment

North Atlantic post-tropical cyclones

By Alexander Baker

Figure 1: The 2017 North Atlantic hurricane season. Ophelia’s location stands out from the typical tracks of North Atlantic tropical cyclones during the active 2017 season. Selected major hurricanes occurring during 2017 – Irma, Jose and Maria – are labelled. Track data are from National Hurricane Center. (Figure courtesy of Jo Camp, Met Office.)

In the early hours of the 16th October 2017, an unusual storm made landfall in Ireland and continued to track north-eastwards, causing damage across the northern United Kingdom and Scandinavia over a two-day period. That storm – Ex-hurricane Ophelia – was the easternmost major hurricane (i.e., category 3 or higher) observed in the Atlantic during the satellite era (Figure 1). Ophelia had a tropical origin and structure and transitioned to an extratropical cyclone prior to landfall. It was not the first such ‘post-tropical’ Atlantic storm to make landfall in the mid-latitudes. To give just a few examples, Hurricane Lili (1996) hit the UK and Hurricanes Vince (2005) and Leslie (2018) struck the Iberian Peninsula. ‘Superstorm’ Sandy (2012) – the 4th most costly Atlantic hurricane on record – caused widespread damage across the Northeast United States.

Figure 2: Interannual variability in North Atlantic post-tropical cyclones. (left) Annual total post-tropical cyclone count timeseries for Europe (upper panel) and the Northeast US (lower panel) and (right) non-parametric (Spearman’s rho) inter-reanalysis correlation matrices (significant correlations, where p < 0.01, are in bold type).

Post-tropical cyclones expose populous mid-latitude communities, which may lack the preparedness of regions more commonly struck by tropical storms, to hurricane-force wind speeds and extreme precipitation. The risk of such events is projected to increase in response to anthropogenic climate change because warming may induce poleward and eastward expansion of tropical cyclone genesis areas, allowing more storms to propagate to the mid-latitudes, where they may undergo extratropical transition and re-intensify (Haarsma et al., 2013). A quantitative survey of historical post-tropical cyclone variability – against which climate model fidelity may be evaluated and climate change projections put into context – is therefore required.

To understand historical post-tropical cyclone activity across the North Atlantic, we used an objective storm-tracking algorithm (Hodges, 1995) to identify post-tropical cyclones in four reanalysis datasets (ERA-Interim, JRA-55, MERRA2 and NCEP-CFSR). Here, we briefly discuss interannual variability in post-tropical cyclones, their intensity prior to and following landfall, and their associated precipitation, focussing on those systems which impact Europe and the Northeast US.

So, what have we found? Let us first consider year-to-year variability. The reanalyses show no significant trends in total annual count of post-tropical cyclones impacting Europe or the Northeast US since 1979, although pronounced interannual variability exists (Figure 2, left). Overall, Europe and the Northeast US experience around five and ten post-tropical cyclones per year, respectively, including relatively weak systems. Interestingly, highly active Atlantic hurricane seasons, such as 2005 (the most active ever recorded), do not stand out as peaks in post-tropical cyclone activity. Moreover, there is little co-variability between Europe and the US: that is to say, peaks/troughs in Northeast US activity generally do not correspond with peaks/troughs in Europe. Inter-reanalysis correlations are higher for Europe than for the US (Figure 2, right), suggesting that the different model configurations and resolutions as well as the differing observational data assimilation schemes employed by the reanalyses impact the representation of historical post-tropical cyclone activity over the Northeast US more than that over Europe.

Figure 3: Composite post-tropical cyclone lifecycles. Lifecycles of difference post-tropical cyclone types centred (i.e., where t=0) on landfall within the Northeast US. Legend entries: ‘sym’/‘asym’ = symmetrical/asymmetrical; ‘WC’/‘CC’ = warm-core/cold-core. Legend numbers give the number of cyclones in each Hart category. Shading shows one standard deviation. Example results for the Northeast US from JRA-55.

We now turn briefly to storm structure and the two primary hazards posed by storms: wind and precipitation.

Hart (2003) devised a way to describe cyclone structure and its evolution based on three parameters: cyclone symmetry and lower- and upper-level thermal winds. These parameters allow cyclones to be classified as warm- or cold-core and symmetrical or asymmetrical (frontal). Typically, tropical cyclones are symmetrical and warm-core; extratropical cyclones are asymmetrical and cold-core. Post-tropical cyclones exhibit these – as well as hybrid – structures. By classifying post-tropical cyclones in this way, we found that the majority transition to a typical extratropical structure, but an appreciable number retain structural aspects of their tropical origins (Figure 3). Crucially, warm-core post-tropical cyclones possess the highest wind speeds upon landfall.

Figure 4: Post-tropical contributions to total precipitation. Average post-tropical cyclone precipitation contributions during August for the Northeast US (left) and Europe (right). TRMM precipitation observations are available up to 50°N and the black boxes outline the cyclone landfall domains within the domain of TRMM observational coverage (although both landfall domains extend north of this latitude).

To establish the importance of post-tropical cyclones are for precipitation over Europe and the Northeast US, we sampled satellite-based precipitation estimates (from NASA Tropical Rainfall Measuring Mission) in the vicinity of each cyclone track and calculated post-tropical cyclones’ percentage contribution to the total precipitation. These contributions were mapped across our regions of interest (Figure 4). Post-tropical cyclones are responsible for up to 10% of Europe’s total summer precipitation. This is significant given that, on average, only a few such storms impact southern Europe and the Mediterranean each summer (Figure 2). Across the US East Coast, climatological contributions of up to 50% are seen in August, and contributions of up to 20% occur across New England and parts of southern Atlantic Canada. Collectively, these results highlight the contemporary importance of post-tropical cyclone precipitation.

This work is currently in preparation for publication (Baker et al., 2018). We hope these analyses stimulate further discussion of post-tropical systems, particularly their tracks, tropical-to-extratropical transition, and representation in global climate models. Future work within the PRIMAVERA project will (i) perform comparisons with multiple high-resolution, tropical-cyclone-permitting global climate model simulations and (ii) assess the spread of post-tropical storm risk across the mid-latitudes under anthropogenic climate change.

Finally, an exciting PhD opportunity in the Department of Meteorology, exploring this topic is currently open for applications, and more information is available here.


Baker, A. J., Hodges, K., Schiemann, S., Haarsma, R., and Vidale, P. L. Historical variability of North Atlantic post-tropical cyclones and their importance for extratropical precipitation. (in prep.)

Haarsma, R. J., Hazeleger, W., Severijns, C., Vries, H., Sterl, A., Bintanja, R., Oldenborgh, G. J., and Brink, H. W., 2013. More hurricanes to hit western Europe due to global warming. Geophysical Research Letters 40, 1783-1788.

Hart, R. E., 2003. A Cyclone Phase Space Derived from Thermal Wind and Thermal  Asymmetry. Monthly Weather Review 131, 585-616.<0585:ACPSDF>2.0.CO;2 

Hodges, K. I., 1995. Feature Tracking on the Unit Sphere. Monthly Weather Review 123, 3458-3465.<3458:FTOTUS>2.0.CO;2 


I thank the EU Horizon-2020-funded PRIMAVERA project for financial support of this research and colleagues at the Royal Netherlands Meteorological Institute (KNMI) for helpful feedback.

Posted in Climate, Climate change, Climate modelling, extratropical cyclones, Rainfall, Tropical cyclones | Leave a comment

Later Wet Seasons with More Intense Rainfall over Africa under Future Climate Change

By Caroline Dunning

Rainfall is projected to change as the planet warms in response to rising concentrations of atmospheric greenhouse gases. New research indicates future changes in the timing and characteristics of the rainy seasons over Africa with important implications for impacts of climate change on vulnerable societies.

The fourth IPCC (Intergovernmental Panel on Climate Change) Assessment Report (AR4) states that “Africa is one of the most vulnerable continents to climate change and climate variability”1. This is partly related to low capacity for adaptation across much of the continent but is also related to the high dependence on climate, in particular wet season rainfall.

Most of Africa experiences one or two main wet seasons per year, when the majority of the annual rainfall occurs. For the large proportion of the population dependent upon rain-fed agriculture for their income and subsistence, the timing of this wet season and amount of rainfall is of high importance. Furthermore, wet season rainfall also impacts the recharge of reservoirs, the supply of electricity from hydro-power and the lifecycle of vectors (e.g. mosquitoes) responsible for the transmission of diseases such as malaria. It is therefore important to understand how future climate change will affect wet season rainfall across Africa.

Figure 1: Schematic showing changes in timing of the wet season (left) and changes in seasonal rainfall totals (right).

In a recently published paper, we use a novel method to investigate changes in the seasonal progression of rainfall across Africa. This methodology determines the start (‘onset’) and end (‘cessation’) of the wet season and is applicable across continental Africa. By applying this methodology to climate model simulations (RCP 4.5 and RCP 8.5 scenarios), we produced projections of changes in wet season onset, cessation, length and rainfall totals under future climate change (Figure 1).

The projections show a delay in the onset of the wet season across much of western and southern Africa under future climate change. In regions where there is little change in the end date (‘cessation’, e.g. western Africa) or regions where the end of the wet season gets earlier (e.g. southern Africa) a later onset means the wet season shortens. This may be problematic for crops, as a shorter wet season can lead to a shorter growing season and result in crops not reaching full maturity.

Across much of central Africa projections suggest that the amount of rainfall occurring during the wet season will increase. However, over southern Africa projections show lower rainfall totals. In such regions agricultural adaptation (e.g. to alternative crop varieties) may be required.

The Horn of Africa (Somalia, southern Ethiopia, Kenya and Uganda) and equatorial regions experience two wet seasons per year; one in the Northern Hemisphere spring (known as the long rains) and one in the autumn (known as the short rains). Our results show the long rains ending earlier, and the short rains ending later. The most notable result, however, is the large increase in the amount of rainfall occurring during the short rains.

We linked these changes in timing of the progression of the seasonal rainfall with changes in the summer heat low over the Saharan Desert (see paper for full details).

Another important finding is that within the wet season rainfall intensity will increase, whilst frequency will decrease; heavier rainfall is damaging to crops with delicate flowers such as coffee and cocoa. In addition, long dry periods can reduce soil moisture and harden the surface layer; thus, when heavy rainfall events do occur a smaller fraction infiltrates into the soil and increased runoff leads to soil erosion and flooding.

Overall, we found that future climate change will lead to changes in the timing of wet seasons over Africa, with shorter wet seasons over southern Africa. Seasonal rainfall totals are projected to increase over central regions, but decrease over southern Africa.  Suitable adaptation may be required; for example, using crop varieties that can cope with a shorter growing season or altering farming practices to minimise soil erosion. Other adaptation may be required to protect crops from increasing intensity of rainfall.


Dunning, C.M., Black, E., Allan, R.P., 2018: Later Wet Seasons with More Intense Rainfall over Africa under Future Climate Change. J. Climate, 31, 9719–9738, 

Posted in Africa, Climate, Climate change, Climate modelling, Rainfall | Leave a comment

Weather, Climate and Energy

By David Brayshaw

 A secure, reliable and relatively affordable electricity supply is an essential part of modern life in many parts of the world.  When charging an iPad, for example, one expects that power will be available at the flick of a switch and rarely considers the complex chain of events involved in producing and conveying the electricity to the device.

The production of energy – including electricity – accounts for a large part of national and global greenhouse gas emissions.  Decarbonizing the energy sector is therefore a key ingredient in meeting climate change targets both in the UK and worldwide.  Electricity – widely seen as the easiest form of energy to decarbonize – has seen a rapid shift towards renewables, particularly wind and solar photovoltaic (PV).

In a national-scale electricity system it is difficult to store large volumes of power efficiently and as such a near-instantaneous balance between supply and demand must be maintained.  Thuswhile the growing use of weather-sensitive renewable generation is clearly beneficial in terms of reducing carbon emissions, it presents new challenges for electricity system management as electricity production cannot be controlled to the same extent as that from, for example a traditional gas-fired power station. 

Figure 1: Simulated changes in weather sensitivity of the electricity system in Great Britain as increasing amounts of wind power capacity is installed.  The red line shows the strength of the negative relationship between temperature and annual total demand-net-wind (primarily associated with cold periods driving high demand), which decreases as installed wind capacity increases.  The blue line, in contrast, shows the increasing strength of the relationship between wind-speed and demand-net-wind as installed wind capacity increases (the sign is also negative because more wind implies a lower demand-net-wind).  Recent installed wind capacities for the GB electricity system are marked in grey lines, suggesting that the system’s dominant weather sensitivity changed from being temperature to wind sometime between 2009 and 2015.  The analysis is based on a 36-year modelled “reconstruction” of the power system’s behaviour derived from weather records.  Adapted from Bloomfield et al. (2018)1

This can be illustrated with an example.  It has long been known that electricity demand is strongly sensitive to temperature: as temperatures fall in winter, electricity demand rises.  Recent research, however, suggests that in the UK the impact of year-to-year variations in temperature via demand are now typically smaller than year-to-year variations in wind power supply (as shown in the figure 1).  That is, in terms of the amount of electricity that must be supplied by coal, gas and nuclear power stations, the impact of a calm winter now exceeds a cold winter1 while winter-time weather patterns associated with both cold and still conditions lead to a double whammy (both high demand and low supply2,3,4).  This suggest an important but subtle shift in how the weather risk should be viewed.  To assess how much non-renewable generation capacity is required, it makes more sense to ask what the maximum demand-net-renewables is (i.e., the demand remaining once the contribution from renewables is removed).  Such a calculation is easier than quantifying the minimum renewable generation available during times of peak demand (the so called “capacity credit” of renewables).

The need for high-quality tools for assessing and understanding the impact of weather and a changing climate on the energy system has never been greater.  The Energy-Meteorology group5, working closely with its industrial and academic partners, is leading the development of many such tools: from the first use of multi-decadal “energy system data reconstructions”6, to the development of new European climate services for energy (weeks-to-months ahead7,8,9), and the exploitation of state-of-the-art high-resolution climate data for informing the design of climate-resilient power systems in the coming decade10,11.


  1. Bloomfield, H. C., Brayshaw, D. J., Shaffrey, L. C., Coker, P. J., Thornton H. E., 2018: The changing sensitivity of power systems to meteorological drivers: a case study of Great Britain. Res. Lett., 13 (5), 054028,
  2. Thornton, H. E., Scaife, A. A., Hoskins, B. J., Brayshaw, D. J., 2017: The relationship between wind power, electricity demand and winter weather patterns in Great Britain. Res. Lett., 12 (6), 064017,
  3. Ely, C. R, Brayshaw, D. J., Methven, J., Cox, J., Pearce, O., 2013: Implications of the North Atlantic Oscillation for a UK–Norway renewable power system. Energy Policy, 62, 1420-1427,
  4. Brayshaw, D. J., Dent, C., Zacharay, S., 2012: Wind generation’s contribution to supporting peak electricity demand: meteorological insights. Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability, 226 (1), 44-50,
  5. University of Reading, 2018: Research Themes – Energy Meteorology. Accessed 6 December 2018,
  6. Cannon, D. J., Brayshaw, D. J., Methven, J., Coker, P. J., Lenaghan, D., 2015: Using reanalysis data to quantify extreme wind power generation statistics: a 33 year case study in Great Britain. Renewable Energy, 75, 767-778,
  7. Thornton, H. E., Scaife, A., Hoskins, B. J., Brayshaw, D. J., Smith, D., Dunstone, N., Stringer, N., Bett, P. E., 2018: Skilful seasonal prediction of winter gas demand. Environmental Research Letters. Res. Lett., (in press).
  8. C3S, 2018: European Climatic Energy Mixes. Accessed 6 December 2018,
  9. Earth System Services, 2018: S2S4E Climate Services for Clean Energy. Accessed 6 December 2018,
  10. Santos-Alamillos, F. J, Brayshaw, D. J., Methven, J., Thomaidis, N. S., Ruiz-Arias, J. A., Pozo-Vázquez, D., 2017: Exploring the meteorological potential for planning a high performance European Electricity Super-grid: optimal power capacity distribution among countries. Res. Lett., 12 (11), 114030,
  11. PRIMAVERA, 2018: Primavera: User Interface Platform. Accessed 6 December 2018,


Posted in Climate, Energy meteorology, Renewable energy | Leave a comment

Steps To Make Sea Ice Projections More Robust

By David Schroeder

Climate model projections are our best approach to make predictions about future sea ice in both hemispheres for the whole 21st century. Given the large spread between individual model projections and distinct discrepancies between model results and satellite-based observations for the last decades, the question arises how accurate these projections are. The reliability of such simulations depends on how well key processes are captured in the climate model. CMIP5 (Coupled Model Intercomparison Project Phase 5) models generally simulate the observed sea ice decline better than the older CMIP3 models indicating that, for example, the improved representation of sea ice processes is important (Stroeve et al., 2012).

In Danny Feltham’s sea ice group we have added and improved several sea ice processes in the Los Alamos sea ice model CICE which is used in many climate models (e.g. the current U.K. Met Office model HadGEM3). The most important contributions are a physical melt pond model describing the evolution of melt ponds (Flocco et al., 2012; Schroeder et al., 2014), an elastic anisotropic plastic rheology accounting for the subcontinuum anisotropy of the sea ice cover (Tsamados et al., 2013) and a spatially and temporally variable form drag parameterization for the momentum exchange between ice and atmosphere as well as ice and ocean (Tsamados et al., 2014).

These new developments increase the realism of the sea ice model. However, incorporating these developments in a climate model does not result in better climate simulations as a matter of course. Many physical and empirical parameters are not very well constrained in a sea ice model requiring a tuning process applying observational data. In the past, most of the tuning has been based on sea ice concentration data derived from passive microwave imagery. This is a problem because realistic sea ice thickness is crucial for simulating variability and trends of the sea ice cover. Using radar altimetry, sea ice thickness estimates are available from CryoSat-2 (CS2) (e.g. from the Centre for Polar Observation and Modelling) during the ice growth seasons since 2010. However, the discrimination of ice and open water, the discrimination of ice type, retracking radar waveforms to obtain height estimates, constructing of sea surface height beneath the ice, and estimating the depth of the snow cover result in a grid cell ice thickness uncertainty of about 25 %.

Figure 1:  Mean effective sea ice thickness over read region shown in inlet above. Verification of CICE-default and CICE-best with Cryosat CS2 CPOM. Results from the ocean sea-ice model PIOMAS are added for comparison. Red area is defined as region where impact of winter growth on thickness change dominates other factors and where CS2 data are most accurate.

In a current study we selected an optimal region for comparing sea ice thickness between simulations with the sea ice model CICE and CS2 data (see red region in Figure 1) by taking into account the strengths and weaknesses of both approaches excluding locations where the number of CS2 observations is limited and the sea ice is thin (Schroeder et al., 2018). Figure 1 shows that a default CICE simulation (blue) strongly underestimates ice thickness, but a new CICE setup (red) agrees well. The most important impact comes from implementing a new process: accounting for the loss of drifting snow. Our optimal model configuration does not only improve the simulated sea ice thickness, but also summer sea ice concentration, melt pond fraction, and length of the melt season. In spite of its uncertainties we successfully applied CS2 sea ice thickness data to improve sea ice model physics.

Reducing uncertainty in sea ice projections requires better representation of key processes, not only of the sea ice component, but also in the atmosphere, the ocean and regarding the coupling. Polar cloud formation and the interaction with the sea ice surface has been identified as a key process which is currently not represented satisfactorily.


Flocco, D., D. Schroeder, Feltham D. L., Hunke E. C., 2012: Impact of melt ponds on Arctic sea ice simulations from 1990 to 2007. J. Geophys. Res., 117, C09032,

Schroeder, D., Feltham D. L., Flocco D., Tsamados M., 2014: September Arctic sea-ice minimum predicted by spring melt-pond fraction. Nat. Climate Change, 4, 353-357,

Schröder, D., Feltham D. L., Tsamados M., Ridout A., Tilling R., 2018: New insight from CryoSat-2 sea ice thickness for sea ice modelling. The Cryosphere Discuss. (in review),

Stroeve, J. C., Kattsov V., Barrett A., Serreze M., Pavlova T., Holland M., Meier W. N., 2012: Trends in Arctic sea ice extent from CMIP5, CMIP3 and observations. Geophys. Res. Lett., 39, L16502,

Tsamados, M., Feltham D. L., Schroeder D., Flocco D., Farrell S. L., Kurtz N., Laxon S. W., Bacon S., 2014: Impact of Variable Atmospheric and Oceanic Form Drag on Simulations of Arctic Sea Ice. J. Phys. Oceanog., 44, 1329-1353,

Tsamados, M., Feltham D. L., Wilchinsky A., 2013: Impact of a new anisotropic rheology on simulations of Arctic sea ice. J. Geophys. Res.: Oceans, 118, 91-107,


Posted in antarctica, Arctic, Climate, Climate change, Climate modelling | Leave a comment

Seasonal Forecasting and the 2018 European Heatwave

By Len Shaffrey

The summer of 2018 has been one of the warmest on record in the UK and Europe. Warm temperatures over the summer led to impacts on agriculture, water resources and human health. One interesting question is how predictable was the 2018 European summer heatwave?

The skill of wintertime seasonal forecasts has dramatically improved in the past few years, especially for forecasts of the wintertime North Atlantic Oscillation (e.g. Scaife et al. 2014, Baker et al. 2018). However, there is much less skill for summertime seasonal forecasts for Europe. Interestingly, there was a substantial degree of convergence in the seasonal forecasts for summer 2018, with most forecast systems predicting warmer than average European temperatures. Figure 1 shows the multi-model mean seasonal forecasts of 2018 July-August-September(JAS) 2m temperature anomalies (forecasts were initialised around the 1 June 2018). The seasonal forecasts were predicting temperature anomalies of approx 1C over Eastern Europe. This led the Met Office to make the unusual step of issuing a statement on this summer’s forecast, which was picked up by the media.

Figure 1. Multi-model mean seasonal forecast of 2018 JAS 2m temperature anomalies from the Copernicus C3S seasonal forecast service.

It’s not yet clear why the summer of 2018 was more predictable than usual. This is an active research area which is being addressed in NERC projects such as SUMMERTIME and IMPETUS. Recent studies have highlighted how springtime North Atlantic seas surface temperature drive summertime European atmospheric circulation (Osso et al. 2018), the relationship between springtime Tropical Atlantic rainfall anomalies and summertime circulation over the North Atlantic (Wulff et al. 2017) and suggested that some seasonal forecasting systems can capture interannual variations in European summertime rainfall (Dunstone et al. 2018). Overall, these studies have suggested that there might be much more seasonal predictability for summertime European climate than previously thought. The studies also raise the possibility that deeper understanding of these processes may lead to substantially improved seasonal forecasts of summertime European climate.


Baker, L. H., Shaffrey, L. C., Sutton, R. T., Weisheimer, A. and Scaife, A. A, 2018: An intercomparison of skill and overconfidence/underconfidence of the wintertime North Atlantic Oscillation in multimodel seasonal forecasts. Geophysical Research Letters, 45 (15), 7808-7817.

Dunstone, N., Smith, D., Scaife, A., Hermanson, L., Fereday, D., O’Reilly, C., et al. 2018:. Skilful seasonal predictions of summer European rainfall. Geophysical Research Letters, 45, 3246–3254,

Osso, A., Sutton, R., Shaffrey, L. and Dong, B. 2018: Observational evidence of European summer weather patterns predictable from spring. Proceedings of the National Academy of Sciences of the United States of America, 115, 59-63. 

Scaife, A. A., et al. 2014: Skillful long‐range prediction of European and North American winters. Geophys. Res. Lett.,41, 2514–2519. doi: 10.1002/2014GL059637

Wulff, C. O., Greatbatch, R. J., Domeisen, D. I. V., Gollan, G., & Hansen, F. 2017:. Tropical forcing of the Summer East Atlantic pattern. Geophysical Research Letters, 44, 11,166–11,173.

Posted in Atlantic, Atmospheric circulation, Climate, Climate change, Climate modelling, Environmental hazards, Historical climatology, Hydrology, Numerical modelling, Seasonal forecasting, Waves | Leave a comment

Is it a normal season this year for tropical cyclones in the Western North Pacific?

By Xiangbo Feng

 The Western North Pacific (WNP) is the most active area for tropical cyclones (TCs).  The number of TCs occurred in the WNP so far (end of October) this year is 26 – just the average number of annual TCs over 1980-2017 (Figure 1 upper). As TC season in the north Hemisphere usually finishes up in October, we thus might think we are having another normal or slightly above-normal TC season for TC occurrence in the WNP.

Figure 1: Top: Western North Pacific (WNP) (0-60°N, 100-180°E) tropical cyclone (TC) count during the period 1980-2018. The 2018 TC record is updated to 31st October. Black, blue and green bars show annual, early summer (June-August, JJA) and late summer (September-October, SO) TC counts, respectively, while red bar is for the number of TCs that occur just in the northwestern sector (20-40°N, 120-140°E) of the WNP. Dashed lines are the long-term averages over 1980-2017. Bottom: Monthly Nino3.4 index anomaly during 1980-2018. Nino3.4 is the average sea surface temperature anomaly in the region of 5°N-5°S, 170°-120°W. Data sources: TC data are the Best Track observational TC record, retrieved on the 5th October 2018, from the Japan Meteorological Agency. Nino3.4 is obtained from NOAA PSD , retrieved on the 5th November 2018.

But, statistics from observations show a few interesting points that make this year not normal for TC activities(Only the TCs of tropical storm intensity or greater are considered here, i.e. maximum sustained wind speed ≥ O34 knots). 

  • First, it’s an active early summer (June-August, JJA) for TCs in the WNP (Figure 1 upper). In JJA, the long-term average number of named TCs over this area during 1980-2017 is 11, whilst the past early summer saw 17 tropical storms being named, 6 above the average, making it the second largest (it’s just 1 TC less than in 1994) on record since 1980.
  • Second, it then turned into a quiescent late summer (September-October) for TC occurrence. Late summer is usually a busy period when strong TCs (e.g. super typhoons) occur. The average number of TCs occurred during late summer is 8. There were only 6 TCs declared in the past late summer. More interestingly, the past October was very quiet in the WNP, featured with only one TC (Typhoon Yutu, formed on 21st October) declared. So far, it seems that some TCs, which were supposed to occur in a later time, were formed in an earlier time, making it a very busy early summer.

Figure 2: TC occurrence anomaly (with long-term average removed) in JJA for each year in the Western North Pacific. Bottom right shows the average of TC occurrence in JJA over 1980-2018.

  • Third, at regional scale (Figure 2), we see the past early summer is actually an extremely busy period for TCs especially in the northwestern parts (20-40°N, 120-140°E). The climatology of the number of TCs passing through this region in early summer is 6 (also see Figure 1 upper). But, there were 12 TCs coming through this region in JJA this year, double the average number. In relation to the normal path of TCs, e.g. through the Philippine Sea and Taiwan (Figure 2 bottom right), this year’s TCs tend to shift northward heading to the north of China and the west of Japan. Similar path years are 1994, 2002 and 2004, but they have smaller positive anomalies.

Seasonal and sub-seasonal TC forecasts

Figure 3: Correlations between JMA observed TC occurrence anomaly and Nino3.4 index, for annual values (top), early summer (June-August, JJA) values (bottom left) and late summer (September-October, SO) values (bottom right), over 1951-2018. Dot areas are where the correlations pass the 0.1 significance test (90% confidence level).

It has been well documented that the El Niño–Southern Oscillation (ENSO) plays an important role in the inter-annual variability of TC activity (especially for the TC lifetime) via modulating the large-scale atmospheric environments (e.g. wind shear and relative humidity). As the effect on number of TCs, for example (Figure 1), in 1998, when a strong El Nino event was developed, fewer TCs are formed in the WNP over the whole year; in the 2015-2016 El Nino event, TCs are more likely to recurve to north rather than straightforwardly head to west. The teleconnections of TC occurrence with ENSO phases can be simply described by the correlation coefficient map (Figure 3 top).

Some meteorological services and agencies have included ENSO as one main source of predictability for the WNP TC activities in their seasonal forecast outlooks. Among these services are the GCACIC forecast at the City University of Hong Kong and the TSR forecast at the University College London. Before TC season started this year, they had forecasted that this year would be a normal or slightly above-normal year for the WNP TC occurrence. This is mainly due to the cold ENSO-neutral condition (Figure 1 bottom). In the meantime, the Met Office’s dynamical seasonal forecast model GloSea5 also predicts a normal or slightly below-normal season for the WNP TC occurrence (contact me if you want to see this result). It seems like that those seasonal forecasts did a reasonably good job for predicting the total number of TCs for the whole season.

However, none of those forecasts, as far as I know, had predicted either an active early summer or a quiescent TC late summer for TCs as observed, even with a short forecast lead time. This means that the present seasonal forecast models (both statistical and dynamical) have less ability to predict the TC activities at sub-seasonal timescale (a time range between two weeks to two months). One of reasons for lack of such predictability might be due to the TC activities, which are ENSO-dependent in late summer, but are ENSO-independent in early summer (Figure 3 bottom). Even though the ENSO states are reasonably estimated, it does not necessarily mean that the most part of TC occurrence variability at sub-seasonal timescale can be well predicted.

The apparent sub-seasonal variations of TC activities, at least this year in the WNP, indicate the necessity of developing sub-seasonal TC forecast approaches on top of the normal seasonal forecast. Including some factors at sub-seasonal timescales, such as the Madden–Julian Oscillation and the equatorial waves, may help us to build up such forecast skills. This indicates an area where more methodological work needs to be carried out in the future. Recently, some efforts have been made to address this challenge, such as the WWRP/WCRP Sub-seasonal to Seasonal Prediction Project (S2S)

Here in Reading, the FASCINATE project, one of the Weather and Climate Science for Service Partnership (WCSSP) Southeast Asia projects funded by the Newton Fund (Nick Klingaman is the PI), is carrying out a detailed analysis of TC activities in the WNP. One aim of our project is to identify the sources of predictability for the WNP TCs, from synoptic to seasonal timescales, and evaluate the present forecast systems, such as the Met Office’s Numerical Weather Prediction and GloSea5, on the representations of such prediction sources.

Posted in Atmospheric circulation, Climate, Climate modelling, earth observation, ENSO, Equatorial waves, Historical climatology, Madden-Julian Oscillation (MJO), Numerical modelling, Seasonal forecasting, Tropical cyclones, Waves, Weather forecasting, Western North Pacific, Wind | Leave a comment

Characteristics and enhanced quality control of drifting buoy observations of sea surface temperature from the International Comprehensive Ocean Atmosphere Data Set (ICOADS)

By Simone Morak-Bozzo

Over the last two decades drifting buoys have become the most prevalent in situ measurement method for sea surface temperature(SST). Drifting buoy data are particularly popular because of their high spatial and temporal coverage. Their freedom of movement allows them to provide data offside shipping routes and moorings and the high temporal resolution gives insight not only into seasonal but also diurnal variations of SSTs, as shown in Morak-Bozzo et al, 2015.

Drifting buoy observations need to be verified before they can be used for quantitative studies involving SST data and cleaned from artefacts linked to erroneous observations or faulty instruments. The main archive of drifting buoy observations is the International Comprehensive Ocean Atmosphere Data Set (ICOADS). In our current project we investigate the life story of approximately 26000 buoys from 1979 to 2014, from the time of their deployment to their “death”. We also observe and document how the network of drifters evolves over time and we assess the quality of the data.

Figure 1: For the subperiod 1982-2012 and a total of 22678 buoys, the position of deployment is shown on the left and the position of their last signal on the right, gridded on a 2-by-2 degree grid.

Figure 2: This figure is showing the distribution of the buoy lifetime for all the buoys in the ICOADS archive. We can see that less than 40% of the buoys make it past the first year after deployment and less than 20% survives more than two years.

Figure 3: The maps are showing the cumulative number of drifting buoy observations in each 2-by-2 degree box for seven 5-year periods (1980-1984, 1985-1989, etc.). In the top right corner of each subfigure we show the sum of all observations in that 5-year period. The majority of buoys reports at least every 3 hours, if not hourly.

An important step to determine the reliability of the data is the creation of an automated protocol for quality control (QC) to complement the limited QC product provided by ICOADS. The set of algorithms we developed can detect e.g. stranded and misplaced buoys or non-plausible SST reports.

Compared to an independent data set of SSTs, such as Reynolds AVHRR-only (Reynolds et al. 2007, Banzon et al. 2016) the ensemble of QC flags shows a clear improvement of the quality of the drifting buoy data set.

The resulting data and associated uncertainty will be contributing to a new SST reconstruction dating back to 1850, under the project HOSTACE (Historical Ocean Surface Temperatures: Accuracy, Characterisation and Evaluation).


Morak-­Bozzo, S., Merchant, C. J., Kent, E. C., Berry, D. I. and Carella, G. ,2016:Climatological diurnal variability in sea surface temperature characterized from drifting buoy data. Geoscience Data Journal, 3 (1),  20­-28, doi:

Banzon, V., Smith, T. M., Chin, T. M., Liu, C., and Hankins, W., 2016: A long-term record of blended satellite and in situ sea-surface temperature for climate monitoring, modeling and environmental studies. Earth Syst. Sci. Data8, 165–176, doi:10.5194/essd-8-165-2016.

Reynolds, R. W., T. M. Smith, C. Liu, D. B. Chelton, K. S. Casey, and M. G. Schlax, 2007: Daily high-resolution-blended analyses for sea surface temperature. Journal of Climate20, 5473–5496, doi:10.1175/JCLI-D-14-00293.1.

Posted in Climate, Climate change, Data collection, Data processing, earth observation, Measurements and instrumentation | Leave a comment

Smoke, science, and sharks

By Ross Herbert

In the August of 2017 the Cloud-Aerosol-Radiation Interactions and Forcing – Year 2017 (CLARIFY) measurement campaign took place on a tiny island in the middle of the southeast Atlantic Ocean where we were surrounded by whales, sharks, and most importantly, stratocumulus clouds. During August and September dense layers of strongly absorbing smoke from biomass burning in central Africa were transported over these semi-permanent clouds. These events provided us with a unique opportunity to understand the interactions between the cloud, smoke layers, and radiation; processes which remain key uncertainties within our understanding of global aerosol radiative forcing, cloud feedbacks, and, ultimately, climate change. In this blog I will discuss my experience of participating in the measurement campaign and also outline the high-resolution cloud modelling work that I am currently doing.

The island (with Bear Grylls)

The Ascension Island is situated at 8°S 14°W in the middle of the Atlantic Ocean, 1600km from the coast of Africa and 2250km from Brazil. The island, with a diameter ~ 10km, is a volcanic island natively populated by nesting turtles, colourful land crabs, and (once) huge bird colonies, and non-natively by rats (that appeared with the boats), cats (that were let loose on the island to control the rats but decided to eat birds instead and are now outlawed), and donkeys (that are tolerated). In the mid-nineteenth century, following encouragement from Charles Darwin, the royal navy began bold plans to increase precipitation on the island by ‘greening’ the upland reaches of the largest mountain with anything that would grow. The result: bamboo, bananas, wild ginger, and guava, and the birth of the aptly named ‘Green Mountain’; however, most of the island remains volcanic and devoid of vegetation.

Figure 1:On top of Sister’s peak looking towards Green Mountain

The measurement campaign

My role on the month-long campaign was a mission scientist. Along with several others we worked as a team alongside the pilots to plan the sorties and then join the crew onboard the FAAM BAe-146 research aircraft. For the sorties we would use forecasts and satellite observations to plan flights depending on what our objectives were; this might include incoming plumes of smoke, co-ordinated satellite overpasses, inter-comparison flights, and interesting cloud or convective features.

Figure 2: The FAAM BAe-146 (right) and NASA P3 (left) on Wideawake Airfield

The pilots required detailed flight plans of bearings, altitudes, distances, and times, so the planning would often take several hours. On flight days we would be up at 6am to check the most recent observations and make any last-minute changes to the sortie plan before heading to the airfield. During flights, the mission scientists used real-time data and the expertise of the instrument operators to fine-tune the sortie so that we were focusing on the correct feature. Smoke layers and stratocumulus-to-cumulus transitions are very poorly forecast so we had to make rapid decisions based on what we observed during flight. This information would be relayed to the mission scientist who sat in the cockpit and would ultimately make the final decision before informing the pilots.

This slideshow requires JavaScript.

The campaign was a huge success; we flew 28 sorties totalling 99 hours and collected data in every cloud-aerosol-radiation regime we hoped for and more. As well as the standard atmospheric state measurements (temperature, pressure, relative humidity etc..) we made detailed measurements of the clouds, the particles, and the gases within the atmosphere and smoke layers. This included number, size, and composition, and also radiative properties such as ability to scatter and absorb radiation. The measurements will help towards improving our understanding of how the smoke layers are formed, transported, evolve, and also how they affect radiative fluxes and interact with the stratocumulus clouds. We can also use the measurements to improve our representation of smoke layers and related processes in the computer models that help us predict the climate and forecast the weather. FYI the aircraft data will be available on the Centre for Environmental Data Analysis (CEDA)archive (ask if you want more specifics).

End of campaign. Time to go back to the office

So enough of the campaign, what am I actually doing on the project? I am working with Nicolas Bellouin and Ellie Highwood to investigate what happens to the cloud radiative properties (i.e., how much energy from the sun and the earth’s surface passes through and out of the cloud) when the layer of smoke is elevated above the cloud; this is a scenario that we encountered numerous times during the campaign and that we observe in satellite observations. Previous studies suggest that the smoke may act to thicken the clouds, which makes them brighter and more reflective to incoming solar radiation (sunlight), thus acting to cool the climate. The stratocumulus clouds and smoke layers cover vast areas of the ocean, therefore small changes to the radiative properties of the cloud may have important impacts. My work will help us understand this effect in more detail and understand how sensitive it is to the properties of the smoke layer and cloud.

Smoke above cloud + Sunlight = Heat = Cloud response = Semi-direct radiative effect = Cooling? or Warming?

In my work I am using the MetOffice Large Eddy Model (the LEM) to simulate the evolution of a stratocumulus cloud deck with an elevated layer of absorbing smoke. The LEM has very high spatial resolution (imagine dividing the atmosphere into boxes – high resolution means lots and lots of little boxes) that allows us to simulate the main turbulent motions that drive the movement of energy, winds, and moisture in the atmosphere (typical climate-scale models have spatial resolutions ~ 200 times greater). It is also coupled to a radiation scheme that allows us to represent upward and downward fluxes of radiation through the atmosphere, smoke layer, and cloud. From this we can represent the additional heating that is caused by the smoke and any subsequent changes to the cloud field and radiative properties; we can then finally determine whether the smoke is acting to cool or warm the climate. This specific radiative effect is commonly referred to as the semi-direct effect and in the context of the IPCC AR5 report can be seen as a rapid adjustment to the instantaneous radiative effect of aerosols.

These simulations are giving us very interesting results that are currently being written up for publication (sneak preview: it’s not as simple as we previously thought). Watch this space!

A longer version of this blog can be found at

Posted in Africa, Atlantic, Atmospheric chemistry, Atmospheric circulation, Atmospheric optics, Climate, Climate change, Climate modelling, Clouds, Data collection, earth observation, Energy budget, Environmental hazards, Greenhouse gases, Measurements and instrumentation, Microphysics, Numerical modelling, Solar radiation, Weather forecasting, Wind | Leave a comment

Determining the Earth’s energy and water cycles

By Christopher Thomas

The Earth’s energy and water cycles govern the distribution and movement of energy and water in the atmosphere, oceans and land. Both energy and water are constantly being transported between different regions of the globe, and the two cycles are closely linked (‘coupled’) together. It is very important to determine the size and variability of these transports as well as the long-term average distributions of each quantity. This will enable us to detect trends due to climate change, as well as to provide insight into short term climatic variability such as the El Niño Southern Oscillation.

The Sun provides energy to the Earth via solar radiation; simultaneously, some energy is emitted back into space. It is therefore quite common to consider the energy imbalance at the top-of-atmosphere (TOA). There is currently a small downward imbalance which means the Earth is gaining heat. The incoming radiation is distributed between the atmosphere, land and oceans both vertically and horizontally. Water is transported around the globe by (e.g.) ocean currents, rivers, evaporation, and precipitation. The last two processes provide the crucial link between the energy and water cycles. When water changes state from vapour to liquid or vice versa it either releases or absorbs energy in quantities which are significant enough that they must be included in the energy cycle.

Our aim is to combine a variety of Earth observation (EO) data sets in order to determine the energy and water cycles with as much precision as possible. This work was pioneered by the NASA NEWS team (L’Ecuyer et al. 2015, Rodell et al. 2015) and as a first step we have reproduced their work. The Earth is divided into 16 regions (seven land and nine ocean) as shown in Figure 1. In each region we consider the energy and water content both in the atmosphere and at the surface, as well as the TOA radiation imbalance (no water is lost to space).

Figure 1: The regions used in this study.

Several complementary EO measurements of energy and water transport are combined in each of these regions. These measurements are expressed as fluxes (the rate of flow of energy through an area). The larger the flux in a particular region, the more energy is being added to (or removed from) that region. Both vertical and horizontal fluxes are exploited in this method. Figure 2 shows the initial net vertical energy flux at the Earth’s surface in each of the 16 regions. The initial net flux is obtained by simply adding the downward fluxes together and subtracting the upward fluxes from the total. The values shown in the figure are the average of about a decade’s worth of observations.

Figure 2: Initial values of net surface flux in each region. Positive values indicate heat gain from the atmosphere and negative values indicate heat loss to the atmosphere.

Now, we know that there are various physical balances that should be respected (such as conservation of energy). The measured fluxes don’t satisfy these balances so we have to modify each flux by adding or subtracting a particular amount to it. When doing this it is particularly important to consider the uncertainty on each flux, which quantifies how well we (think we) have measured it. Observations that are poorly measured have large uncertainties and will be allowed to move more than those that are well known. For example, the TOA radiation has been extremely precisely determined, so has a small uncertainty, but the horizontal movement of water in the atmosphere is poorly understood and has a large uncertainty. The latter can therefore be moved relatively further from its initial value when trying to ensure all the physical balances are respected.

The net surface fluxes obtained in this procedure are reproduced in Figure 3. Some features of the solution indicate that it may be possible to make improvements. For example, the North Atlantic (region 13 in Figure 1) is losing less heat than may be expected. A large amount of heat is transported northwards from the South Atlantic by a process known as the Atlantic Meridional Overturning Circulation (see e.g. Buckley and Marshall 2016). This sort of discrepancy has motivated us to look into ways to incorporate extra information into the solution and hopefully mitigate the problems. One method which seems promising is to account for large-scale correlations in satellite measurements of heat loss from the ocean to the atmosphere (technically known as latent and sensible heat fluxes). If the flux measurements over the oceans are positively correlated then they are more likely to all be measured too high or low at once. This could arise due to the data processing methods used to determine the fluxes from satellite data, for example.

Figure 3: Output values of net surface flux in each region. Positive values indicate heat gain from the atmosphere and negative values indicate heat loss to the atmosphere.

We have estimated the heat flux correlations over the oceans and if we include them in the fit we find that more heat is transported from the South to the North Atlantic, which consequently loses more energy to the atmosphere. This is a promising result and we are currently preparing a paper to describe it more fully. We also intend to explore additional possibilities such as using additional data sets, including reanalysis constraints, and enhancing the spatial resolution of the solution by dividing the globe into more regions. In the long term we have ambitions to include the carbon cycle alongside the energy and water cycles.


L’Ecuyer et al., 2015: The Observed State of the Energy Budget in the Early Twenty-First Century. J. Clim. 28(21), 8319–8346, 10.1175/JCLI-D-14-00556.1

Rodell et al., 2015:The Observed State of the Water Cycle in the Early Twenty-First Century. J. Clim. 28(21), 8289–8318,  10.1175/JCLI-D-14-00555.1

Buckley and Marshall, 2016: Observations, inferences, and mechanisms of the Atlantic Meridional Overturning Circulation: A review. Rev. Geophys. 54(1), 5–63, 10.1002/2015RG000493

Posted in Climate, Climate change, Data processing, earth observation, Energy budget, ENSO, Hydrology, Solar radiation, Water cycle | Leave a comment