## How do ocean and atmospheric heat transports affect sea-ice extent?

Downward trends in Arctic sea-ice extent in recent decades are a striking signal of our warming planet. Loss of sea ice has major implications for future climate because it strongly influences the Earth’s energy budget and plays a dynamic role in the atmosphere and ocean circulation.

Comprehensive numerical models are used to make long-term projections of the future climate state under different greenhouse gas emission scenarios. They estimate that the Arctic ocean will become seasonally ice free by the end of the 21st century, but there is a large uncertainty on the timing due to the spread of estimates across models (Fig. 1).

What causes this spread, and how might it be reduced to better constrain future projections? There are various factors (Notz et al. 2016), but of interest to our work is the large-scale forcing of the atmosphere and ocean. The mean atmospheric circulation transports about 3 PW of heat from lower latitudes into the Arctic, and the oceans transport about a tenth of that (e.g. Trenberth and Fasullo, 2017; 1 PW = 1015 W). Our goal is to understand the relative roles of Ocean and Atmospheric Heat Transports (OHT, AHT) on long timescales. Specifically, how sensitive is the sea-ice cover to deviations in OHT and AHT, and what underlying mechanisms determine the sensitivities?

We developed a highly simplified Energy-Balance Model (EBM) of the climate system (Fig. 2)—it has only latitudinal variations and is described by a few simple equations relating energy transfer between the atmosphere, ocean, and sea ice (Aylmer et al. 2020). The latitude of the sea-ice edge is an analogue for ice extent in the real world. The simplicity of the EBM allows us to isolate the basic physics of the problem, which would not be possible going directly with the complex output of a full climate model.

We generated a set of simulations in which OHT varies and checked the response of the ice edge. This is a measure of the effective sensitivity of the ice cover to OHT (Fig. 3a)—it is not the actual sensitivity because AHT decreases (Fig. 3b), and we are really seeing in Fig. 3a the net response of the ice edge to changes in both OHT and AHT.

This reduction in AHT with increasing OHT is called Bjerknes compensation, and it occurs in full climate models too (Outten et al. 2018). Here, it has a moderating effect on the true impact of increasing OHT. With further analysis, we determined the actual sensitivity to be about 1.5 times the effective sensitivity. The actual sensitivity of the ice edge to AHT turns out to be about half that to the OHT.

What sets the difference in OHT and AHT sensitivities? This is easily answered within the EBM framework. We derived a general expression for the ratio of (actual) ice-edge sensitivities to OHT (so) and AHT (sa):

A higher-order term has been neglected for simplicity here, but the basic point remains: the ratio of sensitivities mainly depends on the parameters BOLR and Bdown. These are bulk representations of atmospheric feedbacks and determine the efficiency of outgoing and downwelling longwave radiation, respectively. They are always positive, so the ice edge is always more sensitive to OHT than AHT.

The interpretation of this equation is simple. AHT converging over the ice pack can either be transferred to the underlying sea ice, or radiated to space, having no impact on the ice, and the partitioning is controlled by Bdown and BOLR. The same amount of OHT converging under the ice pack can only go through the ice and is thus the more efficient driver.

Climate models with larger OHTs tend to have less sea ice (Mahlstein and Knutti, 2011). We have also found strong correlations between OHT and the sea-ice edge in several of the models listed in Fig. 1 individually. Ice-edge sensitivities and B values can be determined per model, and our equation predicts how these should be related. Our work thus provides a way to investigate how much physical biases in OHT and AHT contribute to sea-ice-projection uncertainties.

## An inter-comparison of Arctic synoptic scale storms between four global reanalysis datasets

The Arctic has changed a lot over the last four decades. Arctic September sea ice extent has decreased rapidly from 1980-present by approximately 3.4 million square-kilometres (see Figure 1). This has made the Arctic more accessible for human activities such as shipping, oil exploration and tourism. As Arctic sea ice is expected to continue to decline in the future, human activity in the Arctic is expected to continue to increase. This will increase the exposure to hazardous weather conditions, such as high winds and high waves, which are associated with Arctic storms. However, the characteristics of Arctic storms are currently not well understood.

One way to investigate current Arctic storm characteristics is to analyse storms in global reanalysis datasets. Reanalysis datasets combine past observations with current weather models to produce spatially and temporally homogeneous datasets, that contain atmospheric data at grid-points around the world at constant time intervals (typically every 6-hours) per day from 1979-present (for the modern, satellite-era reanalyses). Typically, a storm tracking algorithm is used to efficiently process all of the 6-hour data in the reanalysis datasets from 1979 (60,088 time steps!) to identify all of the storms that may have occurred in the past. Storms can be identified in the mean sea level pressure (MSLP) field (as low pressure systems), or in the relative vorticity field (as large rotating systems). The relative vorticity field at 850 hPa (higher in the atmosphere than the atmospheric boundary layer) is typically used so that the field is less influenced by boundary layer processes that may produce areas of high relative vorticity.

At the moment, atmospheric scientists are spoilt for choice when it comes to choosing a reanalysis dataset to analyse. There are reanalysis datasets from multiple institutions; the European Centre for Medium Range Weather Forecasts (ECMWF), the Japanese Meteorological Agency (JMA), the National Aeronautics and Space Administration (NASA), and the National Centers for Environmental Prediction (NCEP). Each institution has created their reanalysis dataset in a slightly different way, by using their own numerical weather prediction model and data assimilation systems. Atmospheric scientists also have to choose whether to use the MSLP field or 850 hPa relative vorticity field when applying their storm tracking algorithm to the reanalysis datasets.

In my recent paper, I aimed to assess Arctic storm characteristics in the multiple reanalysis datasets currently available (ERA-Interim, JRA-55, MERRA-2 and NCEP-CFSR), using a storm tracking algorithm based on 850 hPa relative vorticity and MSLP fields. Below is a short summary of some of the results from the paper.

Despite the Arctic environment changing dramatically over the last four decades, we find that there has been no change in the frequency and intensity of Arctic storms in all the reanalysis datasets compared in this study. It was found in preceding, older versions of atmospheric reanalysis datasets that Arctic storm frequency had increased from 1949-2002 (Walsh. 2008 and Sepp & Jaagus. 2011). This is in contrast with results from the modern reanalysis datasets (from this study, and Simmonds et al. 2008, Serreze and Barrett. 2008 and Zahn et al. 2018) which show no increase in Arctic storm frequency.

Across all the reanalysis datasets, some robust characteristics of Arctic storms were found. For example, the spatial distribution of Arctic storms is found to be seasonally dependent. In winter (DJF), Arctic storm track density is highest over the Greenland, Norwegian and Barents Seas, whereas in summer (JJA), Arctic storm track density is highest over and north of the Eurasia coastline (a region known as the Arctic Frontal Zone (Reed & Kunkel. 1960)) (see Figure 2). The number of trans-Arctic ships in summer is much higher than in winter, and these ships typically use the Northern Sea Route to travel between Europe and Asia (along the coastline of Eurasia). Figure 2b shows that this in fact is where most of the summer Arctic storms occur. In addition, the reanalysis datasets show that ~50% of Arctic storms have genesis in mid-latitude regions (south of 65°N) and travel northwards into the Arctic (north of 65°N). This shows that storms are a significant mechanism for transporting air from low to high latitudes.

In general, there is less consistency in Arctic storm characteristics in winter than in summer. This may be because in winter, the occurrence of meteorological conditions such as low level cloud, stable boundary layers and polar night that are more frequent, which are more challenging to represent in numerical weather prediction models, and for the creation of reanalysis datasets. In addition, there is a low density of conventional observations in winter, and difficulties in identifying cloud and estimating emissivity over snow and ice limit the current use of infrared and microwave satellite data in the troposphere (Jung et al. 2016).

The differences between the reanalysis datasets in Arctic storm frequency per season in winter (DJF) and summer (JJA) (1980-2017) were found to be less than 6 storms per season. On the other hand, the differences in Arctic storm frequency per season between storms identified by a storm tracking algorithm based on 850 hPa relative vorticity and MSLP were found to be 55 storms per season in winter, and 33 storms per season in summer. This shows that the decision to use 850 hPa relative vorticity or MSLP for storm tracking can be more important that the choice of reanalysis dataset.

References:

National Snow & Ice Data Centre (2019) Sea ice index. https://nsidc.org. Accessed 4 Mar 2019.

Reed RJ, Kunkel BA (1960) The Arctic circulation in summer. J. Meteorol. 17(5):489–506.

Sepp M, Jaagus J (2011) Changes in the activity and tracks of Arctic cyclones. Clim. Change 105(3–4):577–595.

Simmonds I, Burke C, Keay K (2008) Arctic climate change as manifest in cyclone behavior. J. Clim. 21(22):5777–5796.

Serreze MC, Barrett AP (2008) The summer cyclone maximum over the central Arctic Ocean. J. Clim. 21(5):1048–1065.

Vessey, A.F., Hodges, K.I., Shaffrey, L.C., Day, J.J., (2020) An inter‑comparison of Arctic synoptic scale storms between four global reanalysis datasets. Clim. Dyn., https://doi.org/10.1007/s00382-020-05142-4

Walsh, J.E., Bromwich, D.H., Overland, J.E., Serreze, M.C. and Wood, K.R., 2018. 100 years of progress in polar meteorology. Meteorological Monographs, 59, pp.21-1.

Zahn M, Akperov M, Rinke A, Feser F, Mokhov I I (2018) Trends of cyclone characteristics in the Arctic and their patterns from different reanalysis data. J. Geophys. Res. Atmos., 123(5):2737–2751.

## The (real) butterfly effect: the impact of resolving the mesoscale range

What does the ‘butterfly effect’ exactly mean? Many people would attribute the butterfly effect to the famous 3-dimensional non-linear model of Lorenz (1963) whose attractor looks like a butterfly when viewed from a particular angle. While it serves as an important foundation to chaos theory (by establishing that 3 dimensions are not only necessary for chaos as mandated in the Poincaré-Bendixson Theorem, but are also sufficient), the term ‘butterfly effect’ was not coined until 1972 (Palmer et al. 2014) based on a scientific presentation that Lorenz gave on a more radical, more recent work (Lorenz 1969) on the predictability barrier in multi-scale fluid systems. In this work, Lorenz demonstrated that under certain conditions, small-scale errors grow faster than large-scale errors in such a way that the predictability horizon cannot be extended beyond an absolute limit by reducing the initial error (unless the initial error is perfectly zero). Such limited predictability, or the butterfly effect as understood in this context, has now become a ‘canon in dynamical meteorology’ (Rotunno and Snyder 2008). Recent studies with advanced numerical weather prediction (NWP) models estimate this predictability horizon to be on the order of 2 to 3 weeks (Buizza and Leutbecher 2015; Judt 2018), in agreement with Lorenz’s original result.

The predictability properties of a fluid system primarily depend on the energy spectrum, whereas the nature of the dynamics per se only plays a secondary role (Rotunno and Snyder 2008). It is well-known that a slope shallower than (equal to or steeper than) -3 in the energy spectrum is associated with limited (unlimited) predictability (Lorenz 1969; Rotunno and Snyder 2008), which could be understood through analysing the characteristics of the energy spectrum of the error field. As shown in Figure 1, the error appears to grow uniformly across scales when predictability is indefinite, and appears to ‘cascade’ upscale when predictability is limited. In the latter case, the error spectra peak at the small scale and the growth rate is faster there.

The Earth’s atmospheric energy spectrum consists of a -3 range in the synoptic scale and a $-\frac{5}{3}$ range in the mesoscale (Nastrom and Gage 1985). While the limited predictability of the atmosphere arises from mesoscale physical processes, it would be of interest to understand how errors grow under this hybrid spectrum, and to what extent do global numerical weather prediction (NWP) models, which are just beginning to resolve the mesoscale $-\frac{5}{3}$ range, demonstrate the fast error growth proper to the limited predictability associated with this range.

We use the Lorenz (1969) model at two different resolutions: $K_{max}=11$, corresponding to a maximal wavenumber of $2^{11}=2048$, and $K_{max}=21$. The former represents the approximate resolution of global NWP models (~ 20 km), and the latter represents a resolution about 1000 times finer so that the shallower mesoscale range is much better resolved. Figure 2 shows the growth of a small-scale, small-amplitude initial error under these model settings.

In the $K_{max}=11$ case where the $-\frac{5}{3}$ range is not so much resolved, the error growth remains more or less up-magnitude, and the upscale cascade is not visible. The error is still much influenced by the synoptic-scale -3 range. Such behaviour largely agrees with the results of a recent study using a full-physics global NWP model (Judt 2018). In contrast, with the higher resolution $K_{max}=21$, the upscale propagation of error in the mesoscale is clearly visible. As the error spreads to the synoptic scale, its growth becomes more up-magnitude.

To understand the dependence of the error growth rate on scales, we use the parametric model of Žagar et al. (2017) by fitting the error-versus-time curve for every wavenumber / scale to the equation $E\left ( t \right )=A\tanh\left ( at+b\right )+B$, so that the parameters $A, B, a$ and $b$ are functions of the wavenumber / scale. Among the parameters, a describes the rate of error growth, the larger the quicker. A dimensional argument suggests that $a \sim (k^3 E(k))^{1/2}$, so that $a$ should be constant for a $-3$ range $(E(k) \sim k^{-3})$, and should grow $10^{2/3}>4.5$-fold for every decade of wavenumbers in the case of a $-\frac{5}{3}$ range. These scalings are indeed observed in the model simulations, except that the sharp increase pertaining to the $-\frac{5}{3}$ range only kicks in at $K \sim 15$ (1 to 2 km), much smaller in scale than the transition between the $-3$ and $-\frac{5}{3}$ ranges at $K \sim 7$ (300 to 600 km). See Figure 3 for details.

This explains the absence of the upscale cascade in the $K_{max}=11$ simulation. As models go into very high resolution in the future, the strong predictability constraints proper to the mesoscale $-\frac{5}{3}$ range will emerge, but only when it is sufficiently resolved. Our idealised study with the Lorenz model shows that this will happen only if $K_{max} >15$. In other words, motions at 1 to 2 km have to be fully resolved in order for error growth in the small scales be correctly represented. This would mean a grid resolution of ~ 250 m after accounting for the need of a dissipation range in a numerical model (Skamarock 2004).

While this seems to be a pessimistic statement, we have observed that the sensitivity of the error growth behaviour to the model resolution is itself sensitive to the initial error profile. The results presented above are for an initial error confined to a single small scale. When the initial error distribution is changed, the qualitative picture of error growth may not present such a contrast between the two resolutions. Thus, we highlight the need of further research to assess the potential gains of resolving more scales in the mesoscale, especially for the case of a realistic distribution of error that initiates the integrations of operational NWP models.

A manuscript on this work has been submitted and is currently under review.

This work is supported by a PhD scholarship awarded by the EPSRC Centre for Doctoral Training in the Mathematics of Planet Earth, with additional funding support from the ERC Advanced Grant ‘Understanding the Atmospheric Circulation Response to Climate Change’ and the Deutsche Forschungsgemeinschaft (DFG) Grant ‘Scaling Cascades in Complex Systems’.

References

Buizza, R. and Leutbecher, M. (2015). The forecast skill horizon. Quart. J. Roy. Meteor. Soc. 141, 3366—3382. https://doi.org/10.1002/qj.2619

Judt, F. (2018). Insights into atmospheric predictability through global convection-permitting model simulations. J. Atmos. Sci. 75, 1477—1497. https://doi.org/10.1175/JAS-D-17-0343.1

Leung, T. Y., Leutbecher, M., Reich, S. and Shepherd, T. G. (2019). Impact of the mesoscale range on error growth and the limits to atmospheric predictability. Submitted.

Lorenz, E. N. (1963). Deterministic Nonperiodic Flow. J. Atmos. Sci. 20, 130—141. https://doi.org/10.1175/1520-0469(1963)020<0130:DNF>2.0.CO;2

Lorenz, E. N. (1969). The predictability of a flow which possesses many scales of motion. Tellus 21, 289—307. https://doi.org/10.3402/tellusa.v21i3.10086

Nastrom, G. D. and Gage, K. S. (1985). A climatology of atmospheric wavenumber spectra of wind and temperature observed by commercial aircraft. J. Atmos. Sci. 42, 950—960. https://doi.org/10.1175/1520-0469(1985)042<0950:ACOAWS>2.0.CO;2

Palmer, T. N., Döring, A. and Seregin, G. (2014). The real butterfly effect. Nonlinearity 27, R123—R141. https://doi.org/10.1088/0951-7715/27/9/R123

Rotunno, R. and Snyder, C. (2008). A generalization of Lorenz’s model for the predictability of flows with many scales of motion. J. Atmos. Sci. 65, 1063—1076. https://doi.org/10.1175/2007JAS2449.1

Skamarock, W. C. (2004). Evaluating mesoscale NWP models using kinetic energy spectra. Mon. Wea. Rev. 132, 3019—3032. https://doi.org/10.1175/MWR2830.1

Žagar, N., Horvat, M., Zaplotnik, Ž. and Magnusson, L. (2017). Scale-dependent estimates of the growth of forecast uncertainties in a global prediction system. Tellus A 69:1, 1287492. https://doi.org/10.1080/16000870.2017.1287492

## Evaluating ocean eddies in coupled climate simulations on a global scale

Despite being only between ~10-100 km in spatial scale, mesoscale ocean eddies are important for their role in global heat transport, responding to climate change as well as fluxing heat, momentum and freshwater between the ocean and overlying atmosphere.

As climate models move towards higher resolution, their ocean components are now able to begin to resolve mesoscale eddies. A high resolution ocean is typically defined as ‘eddy-present’ (EP, ¼ degree) where some eddies are permitted at low- to mid- latitudes, or ‘eddy-rich’ (ER, 1/12 degree) where eddies are presented at most latitudes, excluding the Arctic basin and the continental shelf around Antarctica. The benefits of the increased computational expense, associated with running global climate models with a high-resolution ocean, need to be clearly identified [Hewitt et al., 2017]. Many modelling centres have not yet developed an operational version of their climate models with a high resolution ocean component. The benefits of an EP resolution ocean (where some, but not all, eddies are resolved) is not necessarily superior to a coarser resolution ocean with full eddy parameterization.

As part of my PhD, we present the first global assessment of mesoscale surface eddy properties (e.g. distribution, size, speed and lifetime) in two versions of a high-resolution coupled model, with either an EP or an ER resolution ocean. The model results are validated against a gridded satellite altimeter dataset (called AVISO) with a resolution of ¼ degree [Ducet et al., 2000]. We identify and track closed coherent mesoscale eddies, which are defined by their sea surface height (SSH) contours, each day over a 20-year period . Our tracking algorithm is based on Chelton et al. [2011] and Mason et al. [2014]. Our two immediate questions are: how does the representation of mesoscale eddies change between EP and ER resolution? And how do these properties compare to observations and theoretical predictions?

For a full description and evaluation of the results the reader is referred to Moreton et al. [2020], instead key results are highlighted as following:

• Relative to EP, ER resolution simulates more (+60%) and longer-lasting (+23%) eddies, in better agreement with observations. This is shown in the probability density function and zonal average of eddy lifetime for each dataset in figure 1, as well as in the maps of eddy genesis in Figure 2. Both model resolutions represent eddies at the Western Boundary Currents (WBCs) and in the Southern Ocean well, however both fail to capture as many eddies in subtropical gyre interiors, as found in observations. This reflects model biases at the Eastern Boundary Upwelling Systems, and at the Indonesian outflow.
• Eddies are not expected to be able to be resolved when model grid spacing is larger than the Rossby radius of deformation (i.e. at high latitudes as the model grid spacing converges towards the poles ) [Hallberg et al., 2013]. Interestingly, EP resolution does allow for some eddy growth in these regions, although admittedly less than in ER resolution and observations, as shown in the eddy genesis maps in Figure 2.
• A particularly striking outcome of our analysis was the large differences in eddy size across the two resolutions and in observations, as demonstrated by the probability density functions in Figure 3. Note in the figure a speed-based radius is shown (Lspd): a radius typically used to define eddy size [Chelton et al., 2011]. As expected, small eddies in the finer ER resolution are able to be resolved, but interestingly less larger eddies are represented, in comparison to EP resolution and observations. In addition, the increased eddy size in observations compared to EP resolution is noteworthy, despite both having the same apparent resolution of ¼ degree. It is likely observed eddy radii are biased high by the post-processing and interpolation in the creation of the gridded satellite dataset. Caution is advised when using observational eddies, for example in developing eddy parameterization and understanding eddy dynamics.

This work lays the foundation to explore the role of these tracked eddies in mesoscale air-sea coupling within the climate system, something I am currently working on [Moreton et al., in prep].

This work is funded by the NERC CASE studentship with the Met Office, UK.

References:

D. B. Chelton, M. G. Schlax, and R. M. Samelson. Global observations of nonlinear mesoscale eddies. Progress in Oceanography, 91:167 – 216, 2011, https://doi.org/10.1016/j.pocean.2011.01.002

N. Ducet, P. Y. Le Traon, and G. Reverdin. Global high-resolution mapping of ocean circulation from TOPEX/Poseidon and ERS-1 and -2. Journal of Geophysical Research: Oceans, 105(C8):19477–19498, 2000, https://doi.org/10.1029/2000JC900063

R. Hallberg. Using a resolution function to regulate parameterizations of oceanic mesoscale eddy effects. Ocean Modelling, 72:92–103, 2013, https://doi.org/10.1016/j.ocemod.2013.08.007

H. T. Hewitt, M. J. Bell, E. P. Chassignet, A. Czaja, D. Ferreira, S. M. Griffies, P. Hyder, J. L. McClean, A. L. New, and M. J. Roberts. Will high-resolution global ocean models benefit coupled predictions on short-range to climate timescales? Ocean Modelling, 120, 120-136, 2017, https://doi.org/10.1016/j.ocemod.2017.11.002

E. Mason, A. Pascual, and J. C. McWilliams. A new sea surface height-based code for oceanic mesoscale eddy tracking. Journal of Atmospheric and Oceanic Technology, 31(5):1181–1188, 2014, https://doi.org/10.1175/JTECH-D-14-00019.1

S. Moreton, D. Ferreira, M. Roberts and H. Hewitt. Evaluating surface eddy properties in coupled climate simulations with ‘eddy-present’ and ‘eddy-rich’ ocean resolution. Ocean Modelling, 2020, https://doi.org/10.1016/j.ocemod.2020.101567

S. Moreton, D. Ferreira, M. Roberts and H. Hewitt. SST air-sea heat flux feedback over mesoscale eddies in coupled climate models, in prep.

## North American weather regimes and the stratospheric polar vortex

The use of weather regimes offers the ability to categorise the large-scale atmospheric circulation pattern over a region on any given day. One way of doing this is through k-means clustering of the 500 hPa geopotential height anomaly field. Cassou (2008) determined the lagged influence of the Madden-Julian Oscillation (MJO) on four wintertime regimes over the North Atlantic; these regimes have subsequently become commonly used (e.g. they are in use operationally at ECMWF). Charlton-Perez et al. (2018) used the same four regimes to describe the influence of the stratospheric polar vortex on Atlantic circulation patterns.

Stratosphere-troposphere coupling is often described in terms of either the annular modes (the leading principal component (PC) of hemisphere-wide variability, often known as the Arctic and Antarctic Oscillations (AO/AAO) when discussing the lower-troposphere) or regional leading principal components (such as the North Atlantic Oscillation (NAO)). However, by their definition, this doesn’t tell the full story – only some percentage of it (around 1/3 for the NAO). The downward coupling of stratospheric circulation anomalies onto tropospheric weather patterns is an area of active research. For example, not every sudden stratospheric warming (SSW) event exhibits the “canonical” response in the troposphere of a strongly negative NAO-type pattern (Karpechko et al. 2017, Domeisen et al. 2020).

Could regimes tell us something more? Specifically – could they shed light onto the impact of the stratosphere on North America, which has been under-explored compared with Europe? In a recent paper (Lee et al. 2019), we look at just that.

We use 500 hPa geopotential height anomalies in the sector 20-80°N 180-30°W from ERA-Interim reanalysis for December—March 1979—2017. In order to describe only the large-scale variability, we first reduced the dimensionality of the problem by performing the clustering on a filtered dataset – achieved by retaining only the first 12 PCs which explain 80% of the variance in the dataset. We set k a priori to be 4 in the ­k-means clustering, following Vigaud et al. (2018). The number of clusters is somewhat arbitrary, but 4 has been shown to be significant when comparing with a reference noise model (i.e., testing if the clusters are just the result of forcefully clustering noise, or something meaningful). Once the clusters have been determined from analysis of the dataset – the “centroids” – each day in the dataset is assigned to one of the clusters. The patterns produced (Figure 1) are like a similar analysis in Straus et al. (2007) so we adopt their names.

To diagnose how these regimes vary with the state of the stratospheric vortex, we compute some statistics (Figure 2) based on the tercile category of the 100 hPa 60°N zonal-mean zonal wind on the preceding day (“strong”, “neutral”, and “weak”). 100 hPa is used as a lower-stratospheric measure (compared with 10 hPa used for diagnosing major sudden stratospheric warmings) to assess only those anomalies in the stratosphere which have the potential to influence tropospheric weather.

Evidently, the Arctic High regime is strongly sensitive to the strength of the stratospheric winds, being 7 times more likely following a weak vortex versus a strong vortex. The Arctic Low regime displays the opposite sensitivity, being more likely following a strong vortex. A similar but weaker relationship is found for the Pacific Trough. The Alaskan Ridge regime, however, does not display a sensitivity to the vortex strength. This result was somewhat surprising as the Alaskan Ridge regime resembles a pattern which became known as a “polar vortex outbreak”, but we suggest that (a) the similarity of the pattern to the Tropical-Northern Hemisphere pattern may indicate tropospheric forcing exhibits greater control on this regime, and (b) a possible influence through a barotropic anomaly exists from a distortion of the stratospheric vortex (which is not manifest in the zonal-mean zonal wind).

We relate these regimes to impactful real-world weather by computing the probability of an extreme cold temperature (defined as 1.5 standard deviations below normal) in each regime (Figure 3). We find that the greatest likelihood of widespread extreme cold in North America is during the Alaskan Ridge regime, with secondary likelihood of extreme cold for the west coast during the Arctic Low (recall that this pattern is more likely following a strong vortex), and only a low probability during the Arctic High regime (which is strongly associated with extreme cold in Europe).

Our results therefore suggest that the strength of the stratospheric polar vortex does not change the likelihood of the circulation pattern with the greatest potential for driving extreme cold weather in North America (in stark contrast to Europe), and that prediction of this pattern should look elsewhere – either to the tropics, or to changes in the shape of the stratospheric vortex – including wave reflection events (Kodera et al. 2008, Kretschmer et al. 2018).

Further work will investigate how well these regimes and their response to changes in the stratosphere are captured by the extended-range forecasting models which comprise the S2S database.

This work was funded by the NERC SCENARIO doctoral training partnership.

## Sudden Stratospheric Warming does not always equal Sudden Snow Shoveling

During winter, the poles enter permanent darkness (“the polar night”) and undergo strong radiative cooling. In the stratosphere – a dry, stable layer of the atmosphere around 10-50 km above the surface – this cooling is particularly effective. By thermal wind balance, the strong polar cooling leads to the formation of the stratospheric polar vortex (SPV), a planetary scale westerly circulation that sits atop each winter pole (Figure 1).

In the Northern Hemisphere, the SPV is highly variable, thanks to the generation of large planetary waves in the mid-latitude westerly flow (driven primarily by mountains and land-sea contrast around the continents), which can propagate vertically into the stratosphere and break there, decelerating and deforming the SPV and warming the stratosphere.  In the Antarctic, the presence of the Southern Ocean in the mid-to-high latitudes encircling Antarctica means no similar waves are typically produced. The Antarctic SPV is therefore much stronger than its Arctic counterpart, which is why the ozone hole developed there rather than over the Arctic – with the colder temperatures inside the vortex allowing for the formation of polar stratospheric clouds, which catalyse the reactions that deplete ozone.

Now, since all the weather we experience takes place in the troposphere, you might wonder why we should worry about what happens in the layer above that. In the past, numerical weather prediction models did not resolve the stratosphere, because it wasn’t considered worth the extra computational resources. However, it is now known that the state of the SPV can act as a boundary condition to weather forecasts (especially long-range forecasts that extend beyond 2 weeks ahead, e.g. Scaife et al. (2016)) in a similar way to sea surface temperatures (SSTs). One of the reasons for this is the longer timescales present in the stratosphere (also analogous to SSTs) compared with tropospheric weather systems – an anomaly present in the stratosphere has a long persistence time. But how do these stratospheric anomalies influence the weather we experience?

Let’s take one particularly exciting case of SPV variability: major sudden stratospheric warmings (SSWs). SSWs (defined by the 10 hPa 60°N zonal-mean zonal wind reversing from westerlies to easterlies) occur on average 6 times per decade (Butler et al. 2017) and are associated with either a displacement of the SPV off the Pole, or a split of the SPV into two daughter vortices. Coincident with this is a rapid heating of the polar stratosphere (~50°C in a few days) due to adiabatic warming of descending air – hence the name. The most recent major SSW occurred on 2 January 2019 (Figure 2), but one also occurred on 12 February 2018.

Following a major SSW, the easterly winds descend through the stratosphere over the next few weeks and tend to persist for weeks to months in the lower stratosphere. What happens beneath that in the troposphere is then more varied, but on average there is a transition to a negative Northern Annular Mode (NAM). In a negative NAM, the mid-latitude westerlies associated with the tropospheric jet stream weaken and shift equatorward, increasing the likelihood of cold air outbreaks (and, yes, snow!) in places like the UK and northern Europe (Figure 3). However, that’s only the average response!

In February-March 2018, we did indeed see this response following a major SSW – immortalised as the ‘Beast from the East’ with record-breaking cold weather and heavy snowfall in the UK (e.g. Greening and Hodgson 2019). But following the January 2019 SSW, there was no similar weather pattern. Figure 4 shows a cross-section of polar cap geopotential height anomalies (analogous to the NAM). Reds effectively indicate weaker westerly winds, and the major SSW is evident in the centre (second dashed line from the left). However, it doesn’t persistently “drip” down into the troposphere below 200 hPa, with only a brief “drip” in early February 2019. For the most part, the stratosphere and troposphere did not “talk” to each other.

This SSW was thus “non-downward propagating” (Karpechko et al. 2017), which is the case with somewhere close to half of the observed events.

Why?

Some research suggests this may be due to the type of SSW (split vs. displacement, e.g. Mitchell et al. 2013), the tropospheric weather regimes present following the SSW (e.g. Charlton-Perez et al. 2018), the evolution of the SSW (e.g. Karpechko et al. 2017), the interaction of the vertically-propagating waves with the SPV at the time of the SSW (e.g. Kodera et al. 2016), or some combination of those. Perhaps other forcing from the troposphere may dominate over the signal from the stratosphere – such as the teleconnection of the Madden-Julian Oscillation (MJO) to the North Atlantic weather regimes (e.g. Cassou 2008).

Thus, whilst an SSW may make cold weather more likely, it’s by no means guaranteed – and we still don’t fully understand the mechanisms involved with downward coupling. That’s one of the reasons why, regardless of what the tabloids may tell you, sudden stratospheric warming does not always guarantee sudden snow shoveling!

References

Butler, A. H., J. P. Sjoberg, D. J. Seidel, and K. H. Rosenlof, 2017: A sudden stratospheric warming compendium. Earth System Science Data, https://doi.org/10.5194/essd-9-63-2017

Cassou, C., 2008: Intraseasonal interaction between the Madden–Julian Oscillation and the North Atlantic Oscillation. Nature, https://doi.org/10.1038/nature07286

Charlton-Perez, A. J., L. Ferranti, and R. W. Lee, 2018: The influence of the stratospheric state on North Atlantic weather regimes. Quarterly Journal of the Royal Meteorological Society, https://doi.org/10.1002/qj.3280

Greening, K., and A. Hodgson, 2019: Atmospheric analysis of the cold late February and early March 2018 over the UK. Weather, https://doi.org/10.1002/wea.3467

Karpechko, A. Yu., P. Hitchcock, D. H. W. Peters, and A. Schneidereit, 2017: Predictability of downward propagation of major sudden stratospheric warmings. Quarterly Journal of the Royal Meteorological Society, https://doi.org/10.1002/qj.3017

Kodera, K., H. Mukougawa, P. Maury, M. Ueda, and C. Claud, 2016: Absorbing and reflecting sudden stratospheric warming events and their relationship with tropospheric circulation. Journal of Geophysical Research: Atmospheres, https://doi.org/10.1002/2015JD023359

Lee, S. H., and A. H. Butler, 2019: The 2018-2019 Arctic stratospheric polar vortex. Weather, https://doi.org/10.1002/wea.3643

Mitchell, D. M., L. J. Gray, J. Antsey, M. P. Baldwin, and A. J. Charlton-Perez, 2013: The Influence of Stratospheric Vortex Displacements and Splits on Surface Climate. Journal of Climate, https://doi.org/10.1175/JCLI-D-12-00030.1

Scaife, A. A., A. Yu. Karpechko, M. P. Baldwin, A. Brookshaw, A. H. Butler, R. Eade, M. Gordon, C. MacLachlan, N. Martin, N. Dunstone, and D. Smith, 2016: Seasonal winter forecasts and the stratosphere. Atmospheric Science Letters, https://doi.org/10.1002/asl.598

Tripathi, O. P, A. Charlton-Perez, M. Sigmond, and F. Vitart, 2015: Enhanced long-range forecast skill in boreal winter following stratospheric strong vortex conditions. Environmental Research Letters, https://doi.org/10.1088/1748-9326/10/10/104007

## The Variation of Geomagnetic Storm Duration with Intensity

Haines, C., M. J. Owens, L. Barnard, M. Lockwood, and A. Ruffenach, 2019: The Variation of Geomagnetic Storm Duration with Intensity. Solar Physics, 294, https://doi.org/10.1007/s11207-019-1546-z

Variability in the near-Earth solar wind conditions can adversely affect a number of ground- and space-based technologies. Some of these space weather impacts on ground infrastructure are expected to increase primarily with geomagnetic storm intensity, but also storm duration, through time-integrated effects. Forecasting storm duration is also necessary for scheduling the resumption of safe operating of affected infrastructure. It is therefore important to understand the degree to which storm intensity and duration are related.

In this study, we use the recently re-calibrated aa index, aaH to analyse the relationship between geomagnetic storm intensity and storm duration over the past 150 years, further adding to our understanding of the climatology of geomagnetic activity. In particular, we construct and test a simple probabilistic forecast of storm duration based on storm intensity.

Using a peak-above-threshold approach to defining storms, we observe that more intense storms do indeed last longer but with a non-linear relationship (Figure 1).

Next, we analysed the distribution of storm durations in eight different classes of storms dependent on the peak intensity of the storm. We found them to be approximately lognormal with parameters depending on the storm intensity. A lognormal distribution is defined by the mean of the logarithm of the values, μ, and the standard deviation of the logarithm of the values, σ. These parameters were found from the observed durations in each intensity class through Maximum Likelihood Estimation (MLE) and used to create a lognormal distribution, plotted in Figure 2 in dark purple. The light purple distribution shows a histogram of the observed data as an estimate of the probability density function (PDF). By eye, the lognormal distribution provides a reasonable first-order match at all intensity thresholds.

On this basis we created a method to probabilistically predict storm duration given peak intensity. For each of the peak intensity classes, we have calculated the values of μ and σ for the lognormal fits to the duration distributions shown as the black points in Figure 3. It is clear from the points in the left panel of Figure 3 that μ increases as intensity increases, agreeing with the previous results in Figure 1 (i.e., duration increases as intensity increases).

The parameter μ can be approximated as a function of storm intensity by:

μ(intensity) = A ln (B intensity−C)

where A, B and C are free parameters. A least squares fit was implemented, and the coefficients A, B and C were found to be 0.455, 4.632, 283.143 respectively and this curve is plotted, along with uncertainty bars, in Figure 3 (left). Although the fit is based on weighted bin-centres of storm intensity, the equation can be used to interpolate for a given value of intensity. σ can be approximated by a linear fit to give σ as a function of the peak intensity. Figure 3 (right) shows the best fit line which has a shallow gradient of −5.08×10−4 and y-intercept at 0.659.

These equations can be used to find lognormal parameters as a function of storm peak intensity. From these, a distribution of duration can be created and hence a probabilistic estimate of the duration of this storm is available. This can be used to predict the probability a storm will last at least e.g. 24 hours. Figure 4 shows the output of the model for a range of storm peak intensity compared against a test set of the aaH index. The model has good agreement with the observations and provides a robust method for estimating geomagnetic storm duration.

The results demonstrate significant advancements in not only understanding the properties and structure of storms, but also how we can predict and forecast these dynamic and hazardous events.

## Characterising the seasonal and geographical variability in tropospheric ozone, stratospheric influence and recent changes

Williams, R. S., Hegglin, M. I., Kerridge, B. J., Jöckel, P., Latter, B. G., and Plummer, D. A.: Characterising the seasonal and geographical variability in tropospheric ozone, stratospheric influence and recent changes, Atmos. Chem. Phys., 19, 3589–3620, https://doi.org/10.5194/acp-19-3589-2019, 2019.

Approximately 90 % of atmospheric ozone (O3) today resides in the stratosphere, which we know as the ozone layer (extending from ~15-35 km), where it plays a critical role in filtering out most of the harmful ultraviolet (UV) rays from the sun. The gradual formation of the ozone layer from around 600 million years ago was key in Earth’s evolutionary history, as it enabled life to flourish on land. Lesser known is the importance of the remaining ~ 10 % of atmospheric ozone, which is found in the troposphere and has implications for air quality, radiative forcing and the oxidation capacity of the troposphere. Whilst ozone is a pollutant at ground level, contributing to an estimated 6 million premature deaths globally per year, it also acts to cleanse the troposphere by breaking down a large number of pollutants, along with some greenhouse gases. Ozone is however a greenhouse gas in itself – where it has a maximum radiative forcing in the upper troposphere. It is an example of a non-well mixed gas, owing to its spatially and temporally highly varying sources and sinks, as well as its relatively short global mean tropospheric lifetime of about 3 weeks.

A major source of tropospheric ozone is the photochemical reactions of emission precursors such as carbon monoxide (CO), nitrogen oxides (NOx) and volatile organic compounds (VOCs), which have both natural and anthropogenic sources, in addition to the natural influx of ozone-rich air from the stratosphere. The magnitude of these two competing influences has been poorly quantified until the recent advent of satellite observations and the development of comprehensive chemistry-climate models (CCMs), which simulate interactive chemistry and are stratospherically well-resolved.

Our study aimed to update and extend the knowledge of a previous key study (Lamarque et al., 1999), that investigated the role of stratosphere-troposphere exchange (STE) on tropospheric ozone, using two contemporary state-of-the-art CCMs (EMAC and CMAM) with stratospheric-tagged ozone tracers as a diagnostic. We first sought to validate the realism of the model ozone estimates with respect to satellite observations from the Ozone Monitoring Instrument (OMI), together with spatially and temporally limited vertical profile information provided from ozonesondes, which we resolved globally on a seasonal basis for the troposphere (1000-450 hPa) (Figure 1).

Whilst we found broad overall agreement with both sets of observations, an overall systematic bias in EMAC of + 2-8 DU (Dobson Units) and regionally and seasonally varying biases in CMAM (± 4 DU) can be seen in the respective difference panels (Figure 1b and 1c). A height-resolved comparison of the models with respect to regionally aggregated ozonesonde observations helped us to understand the origin of these model biases. We showed that apparent closer agreement in CMAM arises due to compensation of a low bias in photochemically produced ozone in the troposphere, resulting from the omission of a group of emission precursors in this model, by excessive smearing of ozone from the lower stratosphere due to an inherent high bias. This smearing is induced when accounting for the satellite observation geometry of OMI, necessary to ensure a direct comparison with vertically well-resolved models, which has limited vertical resolution due to its nadir field of view. The opposite was found to be the case in EMAC, with a high (low) bias in the troposphere (lower stratosphere) relative to ozonesondes. Given the similarity in the emission inventories used in both models, the high bias in this model indicates that excess in situ photochemical production from emission precursors is simulated within the interactive chemistry scheme. These findings emphasise the importance of understanding the origin of such biases, which can help prevent erroneous interpretations of subsequent model-based evaluations.

Noting these model biases, we next exploited the fine scale vertical resolution offered by the CCMs to investigate the regional and seasonal variability of the stratospheric influence. Analysis of the model stratospheric ozone (O3S) tracers revealed large differences in the burden of ozone in the extratropical upper troposphere-lower stratosphere (UTLS) region, with some 50-100 % more ozone in CMAM compared to EMAC. We postulated that CMAM must simulate a stronger lower branch of the Brewer-Dobson Circulation, the meridional stratospheric overturning circulation, since the stratospheric influence is isolated using these simulations. This has implications for the simulated magnitude and distribution of the downward flux of ozone from the stratosphere in each model. Shown in Figure 2 is the zonal-mean monthly evolution of ozone volume mixing ratio (ppbv) from ozonesondes and EMAC over the period 1980-2013 for the upper (350 hPa), middle (500 hPa) and lower (850 hPa) troposphere, together with the EMAC O3S and derived fraction of ozone of stratospheric origin (O3F) (%) evolution.

We found that the ozonesonde evolution closely resembles that of both EMAC and CMAM (not shown) throughout the troposphere. A clear correspondence in the seasonality of ozone is also evident for the EMAC O3S tracer, and in turn the O3F evolution, particularly towards the upper troposphere. Nonetheless, both models imply that over 50 % of near-surface ozone is derived from the stratosphere during wintertime in the extratropics, which is substantially greater than that estimated by Lamarque et al. (1999) (~ 10-20 %), and still considerably higher than more recent studies (~ 30-50 %) (e.g. Banarjee et al., 2016). This indicates that the stratospheric influence may indeed be larger than previously thought and is thus an important consideration when attempting to understand past, present and future trends in tropospheric ozone.

Finally, we analysed height-resolved seasonal changes in both the model O3 and O3S between 1980-89 and 2001-10. The calculated hemispheric springtime (MAM/SON) changes in ozone are shown in Figure 3, and equivalently for O3S in Figure 4, for the upper and middle troposphere (350 and 500 hPa), as well as for the surface model level. A general increase in tropospheric ozone was found worldwide in all seasons, which is maximised overall during spring in both the Northern Hemisphere (~ 4-6 ppbv) and the Southern Hemisphere subtropics (~ 2-6 ppbv), corresponding to a relative increase of about 5-10 %. Respectively, a significant stratospheric contribution to this change of ~ 3-5 ppbv and ~ 1-4 ppbv is estimated using the model O3S tracers (~ 50-80 % of the total change), although with substantial inter-model disagreement over the magnitude and sometimes the sign of the attributable change for any given region or season from the stratosphere.

Although surface ozone changes are dominated by regional changes in precursor emissions between the two periods – the largest, statistically significant increases (> 6 ppbv) being over south-east Asia – the changing influence from the stratosphere were estimated to be up to 1–2 ppbv between the two periods in the Northern Hemisphere, albeit with high regional, seasonal and inter-model variability. In relative terms, the stratosphere can be seen to typically explain 25-30 % of the surface change over regions such as the Himalayas, although locally it may represent the dominant driver (> 50 %) where changes in emission precursors are negligible or even declining due to the enforcement of more stringent air quality regulations over regions such as western Europe and eastern North America in recent years.

To summarise, our paper highlights some of the shortcomings of the EMAC and CMAM CCMs with respect to observations and we emphasise the importance of understanding model bias origins when performing subsequent model-based evaluations. Additionally, our evaluations highlight the necessity of a well-resolved stratosphere in models for quantifying the stratospheric influence on tropospheric ozone. We find evidence that the stratospheric influence may be larger than previously thought, compared with previous model-based studies, which is a highly significant finding for understanding tropospheric ozone trends.

References:
Lamarque, J. F., Hess, P. G. and Tie, X. X.: Three‐dimensional model study of the influence of stratosphere‐troposphere exchange and its distribution on tropospheric chemistry., J. Geophys. Res. Atmos., 104(D21), 26363-26372, https://doi:10.1029/1999JD900762, 1999.

Banerjee, A., Maycock, A. C., Archibald, A. T., Abraham, N. L., Telford, P., Braesicke, P., and Pyle, J. A.: Drivers of changes in stratospheric and tropospheric ozone between year 2000 and 2100., Atmos. Chem. Phys., 16, 2727-2746, https://doi.org/10.5194/acp-16-2727-2016, 2016.

## The Colour of Climate

Email: Jake.J.Gristey@noaa.gov

Gristey, J.J., J.C. Chiu, R.J. Gurney, K.P. Shine, S. Havemann, J. Thelen, and P.G. Hill, 2019: Shortwave Spectral Radiative Signatures and Their Physical Controls. J. Climate, 32, 4805–4828, https://doi.org/10.1175/JCLI-D-18-0815.1

Sunlight reaching the Earth is comprised of many different colours, or wavelengths. Some of these wavelengths cannot be detected by the human eye, such as the ultraviolet (UV) wavelengths which famously cause sunburn. Fortunately for us, the most intense sunlight is found at harmless visible wavelengths and reaches the surface with relative ease, allowing us to see during the daytime. Sometimes nature aligns to dramatically separate these wavelengths, producing beautiful optical phenomena such as rainbows. More often, however, the properties of the atmosphere and surface lead to intricate differences in the wavelengths of sunlight that get reflected back to space (Fig. 1).

Satellites have observed specific wavelengths of reflected sunlight to infer the properties and evolution of our climate system for decades. Satellites have also independently measured the total amount of reflected sunlight across all wavelengths to track energy flows into and out of the Earth system. It has been less common to make spectrally resolved measurements at many contiguous wavelengths throughout the solar spectrum. In theory, these measurements would simultaneously provide the total energy flow – by integrating over the wavelengths – and the “spectral signature” associated with all atmospheric and surface properties that determined this energy flow. Our recent study puts this theory to the test.

Almost 100,000 spectra of reflected sunlight were computed at the top-of-atmosphere under a diverse variety of conditions. Applying a clustering technique to the computed spectra (which identifies “clusters” in a dataset with similar characteristics) revealed distinct spectral signatures. When we examined the atmospheric and surface properties that were used to compute the spectra belonging to each spectral signature, a remarkable separation of physical properties was found (Fig. 2).

Surprisingly, the separation of physical properties by distinct spectral signatures, as shown in Fig. 2, was found to be robust up to the largest spatial scales tested of 240 km. This is similar to the footprint size of one of the only previous satellite instruments to measure contiguous spectrally resolved reflected sunlight, the SCIAMACHY**, providing an exciting opportunity to investigate spectral signature variability in real observations. We found that the frequency of spectral signatures in real SCIAMACHY observations followed the expected behaviour during the West African monsoon very closely (Fig. 3).

Overall, the separation of physical properties by distinct spectral signatures demonstrates great promise for monitoring evolution of the Earth system directly from spectral reflected sunlight in the future.

Funding acknowledgement: This work was supported by the Natural Environment Research Council (NERC) SCience of the Environment: Natural and Anthropogenic pRocesses, Impacts and Opportunities (SCENARIO) Doctoral Training Partnership (DTP), Grant NE/L002566/ 1, and from the European Union 7th Framework Programme under Grant Agreement 603502 [EU project Dynamics–Aerosol–Chemistry–Cloud Interactions in West Africa (DACCIWA)]

*Note several key simplifications in Fig. 1 for the purposes of visual effect: atmospheric properties are separated, but often occur simultaneously and throughout the atmosphere; the depicted path of sunlight is one option, but sunlight emerging at the top of the atmosphere will come from many different paths; sunlight reflected by the surface will need to travel back through the same gases (and likely other properties) on its way back to the top of the atmosphere, which is not shown. The spectra in Fig. 1 are generated with SBDART using a set of arbitrary but realistic atmospheric and surface properties.

** SCIAMACHY = Scanning Imaging Absorption Spectrometer for Atmospheric Chartography.

Jake completed his PhD at Reading in 2018 and now works at the NOAA Earth System Research Laboratory (ESRL) in Boulder, Colorado.

## Extending the predictability of flood hazard at the global scale

When I started my PhD, there were no global scale operational seasonal forecasts of river flow or flood hazard. Global overviews of upcoming flood events are key for organisations working at the global scale, from water resources management to humanitarian aid, and for regions where no other local or national forecasts are available. While GloFAS (the Global Flood Awareness System, run by the European Centre for Medium-Range Weather Forecasts (ECMWF) and the European Commission Joint Research Centre (JRC) as part of the Copernicus Emergency Management Services) was producing operational, openly-available flood forecasts out to 30 days ahead, there was a need for more extended-range forecast information. Often, due to a lack of hydrological forecasts, seasonal rainfall forecasts are used as a proxy for flood hazard – however, the link between precipitation and floodiness is nonlinear, and recent research has shown that seasonal rainfall forecasts are not necessarily the best indicator of potential flood hazard. The aim of my PhD research was to look into ways in which we could provide earlier warning information, several weeks to months ahead, using hydrological analysis in addition to the meteorology.

Broadly speaking, there are two key ways in which to provide early warning information on seasonal timescales: (1) through statistical analysis based on large-scale climate variability and teleconnections, and (2) by producing dynamical seasonal forecasts using coupled ocean-atmosphere GCMs. Over the past 4.5 years, I worked on providing hydrologically-relevant seasonal forecast products using these two approaches, at the global scale. This blog post will give a quick overview of the two new forecast products we produced as part of this research!

Can we use El Niño to predict flood hazard?

ENSO (the El Niño Southern Oscillation), is known to influence river flow and flooding across much of the globe, and often, statistical historical probabilities of extreme precipitation during El Niño and La Niña (the extremes of ENSO climate variability) are used to provide information on likely flood impacts. Due to its global influence on weather and climate, we decided to assess whether it is possible to use ENSO as a predictor of flood hazard at the global scale, by assessing the links between ENSO and river flow globally, and estimating the equivalent historical probabilities for high and low river flow, to those that are already used for meteorological variables.

With a lack of sufficient river flow observations across much of the globe, we needed to use a reanalysis dataset – but global reanalysis datasets for river flow are few and far between, and none extended beyond ~40 years (which includes a sample of ≤10 El Niños and ≤13 La Niñas). We ended up producing a 20th Century global river flow reconstruction, by forcing the Camaflood hydrological model with ECMWF’s ERA-20CM atmospheric reconstruction, to produce a 10-member river flow dataset covering 1901-2010, which we called ERA-20CM-R.

Using this dataset, we calculated the percentage of past El Niño and La Niña events, during which the monthly mean river flow exceeded a high flow threshold (the 75th percentile of the 110-year climatology) or fell below a low flow threshold (the 25th percentile), for each month of an El Niño / La Niña. This percentage is then taken as the probability that high or low flow will be observed in future El Niño/La Niña events. Maps of these probabilities are shown above, for El Niño, and all maps for both El Niño and La Niña can be found here. When comparing to the same historical probabilities calculated for precipitation, it is evident that additional information can be gained from considering the hydrology. For example, the River Nile in northern Africa is likely to see low river flow, even though the surrounding area is likely to see more precipitation – because it is influenced more by changes in precipitation upstream. In places that are likely to see more precipitation but in the form of snow, there would be no influence on river flow or flood hazard during the time when more precipitation is expected. However, several months later, there may be no additional precipitation expected, but there may be increased flood hazard due to the melting of more snow than normal – so we’re able to see a lagged influence of ENSO on river flow in some regions.

While there are locations where these probabilities are high and can provide a useful forecast of hydrological extremes, across much of the globe, the probabilities are lower and much more uncertain (see here for more info on uncertainty in these forecasts) than might be useful for decision-making purposes.

Providing openly-available seasonal river flow forecasts, globally

For the next ‘chapter’ of my PhD, we looked into the feasibility of providing seasonal forecasts of river flow at the global scale. Providing global-scale flood forecasts in the medium-range has only become possible in recent years, and extended-range flood forecasting was highlighted as a grand challenge and likely future development in hydro-meteorological forecasting.

To do this, I worked with Ervin Zsoter at ECMWF, to drive the GloFAS hydrological model (Lisflood) with reforecasts from ECMWF’s latest seasonal forecasting system, SEAS5, to produce seasonal forecasts of river flow. We also forced Lisflood with the new ERA5 reanalysis, to produce an ERA5-R river flow reanalysis with which to initialise Lisflood, and to provide a climatology. The system set-up is shown in the flowchart below.

I also worked with colleagues at ECMWF to design forecast products for a GloFAS seasonal outlook, based on a combination of features from the GloFAS flood forecasts, and the EFAS (the European Flood Awareness System) seasonal outlook, and incorporating feedback from users of EFAS.

After ~1 year of working on getting the system set up and finalising the forecast products, including a four-month research placement at ECMWF, the first GloFAS -Seasonal forecast was released in November 2017, with the release of SEAS5. GloFAS-Seasonal is now running operationally at ECMWF, providing forecasts of high and low weekly-averaged river flow for the global river network, up to 4 months ahead, with 3 new forecast layers available through the GloFAS interface. These provide a forecast overview for 307 major river basins, a map of the forecast for the entire river network at the sub-basin scale, and ensemble hydrographs at thousands of locations across the globe (which change with each forecast depending on forecast probabilities). New forecasts are produced once per month, and released on the 10th of each month. You can find more information on each of the different forecast layers and the system set-up here, and you can access the (openly available) forecasts here. ERA5-R, ERA-20CM-R and the GloFAS-Seasonal reforecasts are also all freely available – just get in touch! GloFAS-Seasonal will continue to be developed by ECMWF and the JRC, and has already been updated to v2.0, including a calibrated version of the hydrological model.

So, over the course of my PhD, we developed two new seasonal forecasts for hydrological extremes, at the global scale. You may be wondering whether they’re skilful, or in fact, which one provides the most useful forecasts! For information on the skill or ‘potential usefulness’ of GloFAS-Seasonal, head to our paper, and stay tuned for a paper coming soon (hopefully! [update: this paper has just been accepted and can be accessed online here]) on the ‘most useful approach for forecasting hydrological extremes during El Niño’, in which we compare the skill of the two forecasts at predicting observed high and low flow events during El Niño.

With thanks to my PhD supervisors & co-authors:

Hannah Cloke1, Liz Stephens1, Florian Pappenberger2, Steve Woolnough1, Ervin Zsoter2, Peter Salamon3, Louise Arnal1,2, Christel Prudhomme2, Davide Muraro3

1University of Reading, 2ECMWF, 3European Commission Joint Research Centre