What a summer!

By Ben Cosh

What a summer it has been so far. The data is brilliant when it is like this. Stephen Burt keeps an eye on it and has filled me in on the (nearly) record-breaking numbers we’re seeing.

It’s been hot.

The last three weeks average daily maximum temperature is more than 5 degrees hotter than normal. On the hottest day of the summer so far (1st July) we reached 30.8C and we’ve nudged above 30C a few of times.

But we’re a little way off really big records. Even last year we reached 32.5C on 21st June and the hottest maximum in our records is 36.4C from 3rd August 1990. In 1976 we got to 30C for 14 consecutive days, and in 2003 we did so for 6 consecutive days. I remember that week in 2003 because my mother-in-law was visiting from her home in Atlanta. Although 30C is commonplace where she lives, our lack of ubiquitous air conditioning made London almost unbearable!

It’s also been sunny and dry.

We had the second sunniest May since our records began in 1956, with 265 hours of sunshine. June had a third more sunshine-hours than normal with 253 hours. And in the last three weeks (the 21 days to Wednesday 11th July) we’ve had 261 hours of sunshine. That’s 12.44 hours a day. Only 1976 has a sunnier three-week period on record.

As I write this on 13th July, we’ve had no rain at all since 17th June. That’s 25 days and counting. This is our longest dry-spell for 21 years, since a 31-day period with no rain in 1997. I wonder if we’ll make it to the record of 37 days from… you’ve guessed it… 1976.

I am privileged to be the Head of the University of Reading School which includes our world-leading, globally-important department of Meteorology. But I watch the weather like most people do and post ad hoc forecasts (drawn from the main services) to our village Facebook group. There has been little point for the last six weeks though. There are only so many ways you can say: It’s going to be hot and sunny and dry again.

The Met Office and Meteogroup suggest this trend is going to continue for the next couple of weeks or so.

Perhaps a shower or two will creep in.

At the moment, that would be a big story round our way!


Posted in Climate, Historical climatology, Weather | Leave a comment

A newcomer’s reflections on the fourth Lusaka Learning Lab

By Max Leighton (Social Research Assistant for Professor Ted Shepherd)

Figure 1: Lusaka participants recording a video message for the Maputo Learning Lab team.

The fourth Lusaka Learning Lab took place on the 17-18th April 2018, which is where I joined the Future Resilience for African Cities and Lands (FRACTAL) party. Albeit a little late at number four, a fresh pair of eyes can be useful to point out ‘the wood from the trees’ – so this will be my aim here. The first morning started with a diverse thirty-strong group bustling around a circle; water and energy technocrats, officials from Lusaka City Council, a host of community representatives, academics and a spirited facilitator.

We introduced ourselves with various members of the group giving updates on their work inviting others to get involved with future projects. There seemed to be a real sense of togetherness and enthusiastic Q&A sessions followed each update. The atmosphere was relaxed and informal, but purposeful.

It was hard to tell, however, if the trade-off between consensus building within the group, and allowing tensions to surface, was well struck. That said, two energetic debates arose from the two keynote presentations. The first was in response to the 19th statutory instrument of 2018 (SI 19) which enacts groundwater regulation into Zambian national law for the first time; the smallest plots of land issued by Lusaka City Council are smaller than the new minimum distance between boreholes and pit toilets specified in SI 19. The second between a Community-Based Organisation representative and an official from Lusaka City Council escalated into a lively debate involving a number of participants. All debaters seemed decidedly frank, suggesting there was genuine space for disagreement at the learning lab, despite the commitment to consensus.

As part of the lab outputs, there are four co-produced policy briefs – (1) flooding, (2) groundwater pollution, (3) groundwater levels and (4) water supply – which are in the final stages of completion. Time is never enough, so the briefs on groundwater levels and water supply were selected to be the meat for the two days.

Each participant brought their respective knowledge and experience to interrogate the two policy issues, as well as using the information from the Climate Risk Narratives (shown in Figure 2) as a ‘conversation starter’. The narratives distil what is known from climate model simulations (CMIP5) to inform three plausible narratives of the future climate – a little like the design fiction series Black Mirror which imagines plausible dystopian futures based on current trends and is also met with intense discussions about the way the world will be.

Figure 2: Climate Risk Narratives representing three possible future scenarios derived from CMIP5 over the next few decades for the Lusaka region.

Each narrative starts from the possible changes to the natural system and is then contextualised with the human-oriented knowledge to describe the likely societal impacts and responses required. The participants were receptive to the narratives and many applied their own knowledge intuitively. These discussions co-produced streams of knowledge around the two issues (groundwater levels and water supply) and they could have gone on for a great deal longer – we were finally asked to vacate the conference room.

After being welcomed into the messy process of coproduction, it has been particularly interesting to read through Alice McClure’s previous post on the very first learning lab back in September 2016 – also in Lusaka. Two phrases stand out; that participants called for “freedom of speech, creativity, honesty and trust” and, secondly, about the necessity of having “trust [in] the process.”

The group not only seemed to have trust in the learning labs, but many appeared to be emotionally invested. Phrases like; “progress is slow” and “it’s a messy art” have almost become tropes for co-production projects, however, the principles participants originally called for are now in full swing. Achieving this productive environment, by all accounts, isn’t the work of a moment, but capitalising on it and maintaining momentum beyond a 4-year project cycle appears to be a no-brainer. The point I’ll end on is that whilst the learning labs have been aimed at decision-makers, attempting to further integrate community voices seems like a sensible next step.

Posted in Africa, Climate, Climate change, Hydrology | Leave a comment

DARE to use datasets of opportunity

By Joanne Waller

To accurately forecast the weather, we must first describe what is currently happening in the atmosphere. To determine the current atmospheric state, we could use:

  • Previous forecasts (data from complex computational models of the atmosphere) which provide data at all the locations we are interested in, but do not necessarily describe the current state accurately (particularly if the forecast was bad).
  • Atmospheric observations which provide up-to-date information about the current atmospheric state, but only exist at restricted locations.

But rather than rely solely on one of these datasets, we combine them both using the technique of data assimilation. However, to perform the assimilation we need a sufficient quantity of observations to constrain the model, as well as an understanding of how accurate the data is.

Currently the atmospheric observations used in data assimilation come from a multitude of different instruments, from simple thermometers to complex instruments on board satellites, all designed to observe different atmospheric variables e.g. temperature, wind and humidity (Figure 1).

Figure 1: Ground based observing instruments at the University of Reading. Atmospheric Observatory is used to measure variables including air temperature, surface temperature, humidity, wind speed and direction and atmospheric electricity.

But these observations come at a price; it is estimated that every year (in the UK) approximately £2 billion is spent acquiring observational data. Of course we cannot do without these specifically collected high quality observations, but could we supplement them with other observations that we could collect for free (or very cheaply)?

The Data Assimilation for the REsilient City (DARE) project is investigating the potential of these so called ‘Datasets of opportunity’. Datasets of opportunity to be investigated within the DARE project will be used for both weather and flood forecasting and include:

  • Citizen Science Automatic Weather Station Data collected via the Met Offices Weather Observations Website (WOW);
  • Temperature data from private cars;
  • Data from aircraft;
  • CCTV;
  • Smartphone data.

The first observation that we are beginning to consider is the temperature data from private cars. Most cars are now fitted with an ambient air temperature sensor. Due to sheer number of vehicles being driven each day the collection of car temperature readings could provide tens of thousands of observations. Consequently, the Met Office have developed an App that can be used to collect such data. The data is read by plugging a dongle into a vehicle’s on-board diagnostics port. The dongle then sends the data, via Bluetooth to the App which then uploads the data to the Met Office WOW (Figure 2).

Figure 2: Schematic of car ambient air temperature sensor data collection.

Before we can use this data for forecasting we must understand how good the data is. In order to assess both the data collection feasibility and the data quality a small trial was run by the Met Office to collect data using this method. The data was then analysed as part of a University of Reading Undergraduate Research Opportunity Placement Project. The project highlighted a number of potential issues with the data (some of which were anticipated, and some were not), including:

  • Temperature readings that were too high when vehicle had been stationary. Many people may have noticed that this is an issue, particularly in the warm weather we have recently been experiencing (Figure 3)!
  • Temperature readings that were too low when vehicle was travelling at high speed.
  • Increased temperature readings (compared to model data – which could also be inaccurate) when the vehicle was travelling up hill.
  • A time delay in the response of the car temperature sensor.

Figure 3: A) The temperature reading from my car (which had been stationary for 9 hours) on the afternoon of 27th June. B) The temperature reading from my car five minutes later than in A after driving a couple of miles. Note that the temperature reading dropped 5oc in 5 minutes!

An added complication is that for each temperature observation we know the location and time it was measured and the speed the vehicle was travelling. However, we do not have any other information such as the make of vehicle. This can make understanding the data complex since different vehicles have their temperature sensors in different locations. This means that temperatures from different observations will be subject to different errors (for example the airflow under a wing mirror will differ to the airflow in the front bumper). But, only given the data we have, these are hard to distinguish.

At the beginning of this year a much wider trial was conducted and nearly 40,000 observations were collected. Over the coming months we will analyse this data to gain a better understanding of the data quality and to determine methods to select only the temperature observations which have potential benefit for weather forecasting.


Posted in data assimilation, earth observation, Flooding, University of Reading, Weather, Weather forecasting | Leave a comment

Tibetan Plateau Vortices

By Julia Curio

Tibetan Plateau Vortices (TPVs) are meso-scale cyclones that originate over the Tibetan Plateau and move eastwards steered by the subtropical westerly jet above. These storms can also move off the Tibetan Plateau (TP) and travel as far east as eastern China. There they can trigger extreme precipitation events and flooding, e.g. in the Sichuan province and along the Yangtze River valley.

Feng et al. (2014) describe a TPV event occurring in July 2008 where a TPV moved off the TP and travelled north-eastward all the way to the coast of the Yellow Sea within five days. This TPV caused heavy rainfall over a large area of southern and eastern China, especially in the Sichuan province, where the maximum 24 h accumulated precipitation at one station reached 288 mm, which is about a quarter of the average annual precipitation.

Within the MESETA project (Modelling physical and dynamical processes over the Tibetan Plateau and their regional effects over East Asia) we study TPVs in reanalysis data sets and a state of the art general circulation model (Curio et al. 2018a, in revision). We use the objective feature-tracking algorithm TRACK, developed by Kevin Hodges here in Reading, to identify TPVs and their paths. One aim of the project is to assess how well the model can represent TPVs using a 25 km horizontal resolution. Another is to connect the occurrence of TPVs to large-scale atmospheric circulation, e.g. the subtropical westerly jet and the monsoon system, and to analyse the conditions under which these systems can move off the TP and trigger extreme rainfall.

We discussed that it would be important to validate the results from the reanalyses with an independent observational dataset of TPVs before using the reanalyses to validate the model. We found out that the Institute of Plateau Meteorology (IPM) in Chengdu, the capital of the Sichuan Province, is publishing yearbooks of TPVs. These yearbooks are the only available observational dataset on TPVs, so we decided to get in contact with the IPM to establish a collaboration.

Mike Wong (former post-doc in Reading, now a forecaster at the Hong Kong observatory) and I visited the IPM in spring 2017. We had a great time there with Yongren Chen, which was filled with scientific discussions, great food and some sightseeing (including pandas).

To generate the TPV yearbooks the researchers at the IPM use a network of sounding stations and manually identify TPVs and follow their paths using the wind field at 500 hPa and the geopotential height values from the radiosonde soundings which are available every 12 hours.

Yongren then came to Reading for two months in early summer 2017 and we started evaluating our database. From the yearbooks we selected ten TPV cases which moved off the TP and were related to heavy rainfall. We then looked through our database to find the matching cases in a gridded reanalysis of the recent past, called ERA-Interim. In a second step, we completed the comparison by selecting another ten cases from the ERA-Interim database of TPVs and matched them to cases from the yearbooks.

Figure 1: Researchers in the MESETA project and their visitors (top row: Senfeng Liu (IAP, Beijing), Yongren Chen (IPM, Chengdu), Mike (Kai Chi) Wong, Julia Curio; bottom row: Reinhard Schiemann, Andy Turner).

This comparison (Curio et al. 2018b) shows that there are cases where the two methods agree and cases where they disagree. The study revealed possible reasons for disagreement, which are differences in the data coverage, differences in the data, and differences in the methods.

Figure 2: Path of a TPV selected from the TPV yearbook (black dots) and the matching TPV output from automated tracking using TRACK (blue dots).

We found cases where the automated method detects the systems much further west and therefore earlier during their lifetime (Figure 2), which is simply because the observational network (Figure 3) does not have any radiosonde stations in the western Tibetan Plateau where most of the systems are generated. This makes it difficult for the IPM to issue warnings in a timely manner since they only detect the TPV when they move east of 90°E.

Figure 3: Distribution of sounding stations (black dots) on the TP and its surroundings used for the manual TPV tracking. The colour shading shows the underlying orography.

The biggest advantage of the automated tracking of TPVs using a gridded dataset like ERA-Interim is the complete spatial coverage of the Tibetan Plateau and a relatively high temporal resolution of 6 hr, which leads to more reliable paths of the systems. The automated tracking is also objective and reproducible, while the manual tracking can lead to different results when performed by different forecasters.

Our colleagues at the IMP are interested in using the automated tracking to improve their forecasts, and thereby try to increase the lead-time for warnings of extreme rainfall events and subsequent flooding in the Sichuan province. We will continue our fruitful collaboration with the IPM and Yongren over the next years and try to implement the automated tracking of TPVs in the operational forecasting system used by the IPM.


Feng, X., Liu, C., Rasmussen, R., & Fan, G., 2014. A 10-yr Climatology of Tibetan Plateau vortices with NCEP climate forecast system reanalysis. Journal of Applied Meteorology and Climatology, 53(1), 34–46, https://doi.org/10.1175/JAMC-D-13-014.1.

Curio, J., R. Schiemann, K. I. Hodges, and A. G. Turner, 2018a. Climatology of Tibetan Plateau vortices in reanalysis data and a high-resolution global climate model. J. Climate, in revision.

Curio, J., Chen, Y., Schiemann, R., Turner, A. G., Wong, K. C., Hodges, K. and Li, Y., 2018b. Comparison of a manual and an automated tracking method for Tibetan Plateau vortices. Advances in Atmospheric Sciences, 35(8), 965–980, https://doi.org/10.1007/s00376-018-7278-4.


Posted in China, earth observation, extratropical cyclones, Flooding, Monsoons, Numerical modelling, University of Reading, Weather forecasting | Leave a comment

What’s the secret of coarse dust?

By Claire Ryder

Mineral dust aerosol particles are regularly lifted into the atmosphere in arid regions, such as deserts, and transported over thousands of kilometres by the wind, such as from the Sahara desert to the Caribbean Sea, as shown in the satellite image in Figure 1.


Figure 1: Dust crosses the Atlantic from the Sahara desert on 23 July 2005 captured by the SeaWIFS satellite instrument. credit: NASA 

Recent fieldwork forming part of the Fennec project (Washington et al., 2012; Ryder et al., 2015), where aircraft observations were made of in situ Saharan dust properties over the remote desert (Ryder et al., 2013b) have shown that close to dust sources concentrations are very high, as expected, but that unexpectedly, the concentrations of coarse particles (larger than 1 micron diameter) and giant particles (larger than 10 microns diameter) are also higher than expected (see Figure 2), since giant particles would be expected to fall back to the surface within hours due to their larger weight. However, Ryder et al. (2013b) measured particles sized up to 100 microns up to altitudes of 5 km, indicating that they are not so readily deposited from the atmosphere, and giving them potential to travel large distances.

Figure 2: Dust volume size distributions from various aircraft fieldwork campaigns. Fennec, over remote desert, is shown in black. (Taken from Ryder et al., 2013b).

So why does it matter that coarse and giant particles are present to a such a great extent?

Firstly, dust interacts with both solar and terrestrial radiation, perturbing the atmospheric radiation balance. In the solar spectrum, dust causes cooling at the surface by reflecting and absorbing radiation, warming in the atmospheric column, and either cooling or warming at the top of atmosphere (TOA) depending on the brightness of the surface type beneath and the properties of the dust. In the terrestrial spectrum, dust absorbs outgoing longwave radiation and causes a warming at the surface, a cooling in the atmosphere, and a warming at the TOA. The balance of these processes can impact surface temperatures, sensible and latent heat fluxes, regional atmospheric circulation and precipitation, and lead to a small net cooling or warming of the climate system at the TOA.

Dust size has a major impact on this radiative effect, as illustrated by Figure 3, where we see that for larger particles with an effective radius (re) of 9 microns, at solar wavelengths (e.g. 0.5 microns) the single scattering albedo (controlling the amount of absorption exerted by dust), drops substantially, while at terrestrial wavelengths (e.g. 12 microns) the extinction efficiency increases substantially for larger particles. Thus knowing the sizes of dust particles in the atmosphere is crucial in determining how they will alter the radiative balance of the atmosphere.

Figure 3 – Impacts of dust size on optical properties, from Tegen & Lacis (1996). Left panel: single scattering albedo (controlling how absorbing dust is – low values indicate high absorption). Right panel: extinction efficiency (controlling how much radiation is extinguished by dust).

Another important aspect of dust is its impact on the biosphere, by depositing nutrients to ocean and rainforest ecosystems (Jickells et al., 2005). Deposited dust mass is dominated by the largest particles, and therefore understanding the transport of coarse and giant particles is crucial to pinning down the impacts of dust on the biogeochemical cycles.

Finally, coarse and giant dust particles appear to be transported further in the atmosphere than we would expect (Ryder 2013a). Thus it is not surprising that dust models, both in numerical weather prediction and climate models struggle to adequately represent the transport of both coarse and giant dust particles (Kok et al., 2017; Ansmann et al., 2017) and their associated impacts on weather and climate.


Ansmann, A., Rittmeister, F., Engelmann, R., Basart, S., Jorba, O., Spyrou, C., Remy, S., Skupin, A., Baars, H., Seifert, P., Senf, F., and Kanitz, T.,2017. Profiling of Saharan dust from the Caribbean to western Africa – Part 2: Shipborne lidar measurements versus forecasts. Atmos. Chem. Phys., 17, 14987-15006, https://doi.org/10.5194/acp-17-14987-2017.

Jickells, T., et al., 2005. Global iron connections between dust, ocean biogeochemistry and climate. Science, 308, 67 – 71.

Kok, J. F., Ridley, D. A., Zhou, Q., Miller, R. L., Zhao, C., Heald, C. L., Ward, D. S., Albani, S., and Haustein, K.,2017. Smaller desert dust cooling effect estimated from analysis of dust size and abundance. Nat. Geosci.10, 274–278, https://doi.org/10.1038/ngeo2912.

Ryder, C. L., Highwood, E. J., Lai, T. M., Sodemann, H., and Marsham, J. H.,2013a.  Impact of atmospheric transport on the evolution of microphysical and optical properties of Saharan dust. Geophys. Res. Lett., 40, 2433–2438, https://doi.org/10.1002/Grl.50482.

Ryder, C.L., Highwood, E., Rosenberg, P., Trembath, J., Brooke, J., Bart, M., Dean, A., Crosier, J., Dorsey, J., Brindley, H., Banks, J., Marsham, J.H., McQuaid, J.B., Sodemann, H., Washington, R., 2013b. Optical properties of Saharan dust aerosol and contribution from the coarse mode as measured during the Fennec 2011 aircraft campaign. Atmos. Chem. Phys., 13, 303-325, https://doi.org/10.5194/acp-13-303-2013.

Ryder, C.L., J.B. McQuaid, C. Flamant, P.D. Rosenberg, R. Washington, H.E.Brindley, E.J. Highwood, J.H.Marsham,D.J.Parker, M.C.Todd, J.R.Banks, J.K.Brooke, S.Engelstaedter, V.Estelles, P.Formenti, L.Garcia-Carreras, C.Kocha, F.Marenco, P.Rosenberg H.Sodemann, C.J.T.Allen, A.Bourdon, M.Bart, C.Cavazos-Guerra, S.Chevaillier, J.Crosier, E.Darbyshire, A.R.Dean, J.R.Dorsey, J.Kent, D.O’Sullivan, K.Schepanski, K.Szpek, J. Trembath, A.Woolley, 2015.  Advances in understanding mineral dust and boundary layer processes over the Sahara from Fennec aircraft observations. Atmos. Chem. Phys., 15, 8479-8520, https://doi.org/10.5194/acp-15-8479-2015.

Tegen and A. A. Lacis, 1996. Modeling of particle influence on the radiative properties of mineral dust aerosol. J. Geophys. Res.,101, 19 237–19 244.

Washington, R., Flamant, C., Parker, D.J., Marsham, J., McQuaid, J.B., Brindley, H., Todd, M., Highwood, E.J., Ryder, C.L., Chaboreau, J.-P., Kocha, C., Bechir, M., Saci, A.,2012. Fennec – The Saharan Climate System. 2012, No. 60, Vol. 17, No. 3 p31-32, CLIVAR Exchanges.


Posted in Aerosols, Africa, Atmospheric chemistry, Climate modelling, earth observation, Remote sensing | Leave a comment

On the predictability of European summer weather patterns

By Albert Ossό

Forecasts of summer weather patterns months in advance would be of great value for a wide range of applications. However, the current seasonal dynamic model forecasts for European summers have very little skill (Arribas et al. 2011). In a recent work, Ossó et al. 2018 analyse atmospheric and ocean observations and show evidence that a specific pattern of summer atmospheric circulation—the summer East Atlantic pattern (SEA)—is predictable from the previous spring. In particular, the results suggest that a specific pattern of North Atlantic SST anomalies persists until summer influencing the atmospheric circulation and the position of the jet stream in July-August (JA) by changing the background baroclinicity. An index representing this North Atlantic SST pattern in March–April (MA) can predict the SEA pattern in JA with a cross-validated correlation skill above 0.6. Moreover, the SEA pattern has a strong influence on rainfall in the UK, which can also be predicted months ahead with significant skill (~0.56).

An important question is if the current dynamical models are able to simulate this observed spring SST summer circulation relationship. Here, we analyze the representation of this relation in a 120 years long pre-industrial simulation from the Met Office Global Coupled model 2.0 (GC2) (Williams et al. 2015). Figure 1 shows lagged linear regressions between monthly sea level pressure (SLP) and SST anomalies and the preceding MA SST Index in the ERA-Interim reanalysis (left column) and in the GC2 model output (right column). Note that the GC2 boxes used to calculate the SST index have been optimized to capture the SST dipole anomalies in the model (see Ossó et al. 2018). However the results are roughly the same using the same boxes as in the observations.

Figure 1 shows that the SLP anomalies in GC2 are consistent with a significant July atmosphere response to MA SSTs. The spatial pattern of SLP anomalies resembles that of the observations but their magnitude is about half of the size. In contrast with the observations, the simulated anomalies in August are weak and not statistically significant.

The monthly evolution of atmospheric and ocean anomaly observations suggests that the SEA pattern growth in August is driven by a positive feedback between atmospheric circulation and SSTs. Figure 1 (left column) shows that anticyclonic SEA pattern anomalies in July and August are accompanied by warming of SSTs, which increase the amplitude of the SST dipole that forces the SEA pattern, therefore generating a positive feedback. The lack of this mechanism on the model (Figure 1 right column) could explain why the atmosphere response is limited in July. Similar results have been found for large ensembles using the Met Office Decadal Climate Prediction System (DePreSys3) (Nick Dunstone personal communication).  


Arribas A, et al., 2011. The GloSea4 ensemble prediction system for seasonal forecasting. Mon Weather Rev 139:1891–1910.

Posted in Climate, Climate modelling, Seasonal forecasting | Leave a comment

Who discovered the Madden-Julian Oscillation?

By Simon Peatman

The Madden-Julian Oscillation (MJO) is one of the most important meteorological phenomena in the tropics. With a timescale of 30–90 days it bridges the gap between weather and climate (Zhang, 2013), potentially providing predictability over several weeks. It consists of large-scale envelopes of, alternately, organized convection and clear skies, propagating slowly (∼5 m s-1) eastwards from the Indian Ocean, through the Maritime Continent, to the Pacific; and an associated planetary-scale circulation. The MJO interacts with several other phenomena, including the El Niño–Southern Oscillation (Tang and Yu, 2008), monsoons (Lavender and Matthews, 2009; Singh et al., 2017), tropical cyclones (Klotzbach, 2014) and the diurnal cycle (Peatman et al., 2014).

In 1971, Roland Madden and Paul Julian published a study of 10 years of radiosonde data from Canton Island, Kiribati (Madden and Julian, 1971), noting a “very pronounced maximum” in the co-spectrum of zonal wind at 850 and 150 hPa, with period 41–53 days (Figure 1). They hypothesized the cause to be a “large circulation cell oriented in zonal planes and centred in the mid-Pacific”, thousands of kilometres in scale and located near the equator. The following year a further study (Madden and Julian, 1972) analysed data from 25 stations throughout the tropics (and beyond) and determined that the oscillation was indeed a zonally-oriented planetary-scale circulation, confined to 10°N–10°S, which propagates from the Indian Ocean to the Pacific (Figure 2).

Figure 1: Co-spectra of u850 and u150 (dashed, left axis); and surface pressure and u850 (solid, right axis) from radiosonde data from Canton Island, Kiribati. From Madden and Julian (1971).

Figure 2: Schematic diagram of the MJO circulation and convection, with eastward propagation indicated between successive panels. Tropopause height and pressure anomaly are sketched at the top and bottom respectively of each panel. From Madden and Julian (1972).

However, it has recently emerged that although Madden and Julian were unaware of this oscillation before their 1971 paper, they were not the first to have demonstrated its existence. As described by Li et al. (in press), a study published in Chinese eight years earlier (Xie et al., 1963) found a 40–50 day oscillation in radiosonde data from several tropical locations. The Xie study, devoted to typhoon genesis, noted “[t]here is a quite definite relationship between the time, location and frequency of typhoon genesis and the location and strength of the basic flow in the low latitudes”, thus not only discovering the intraseasonal oscillation but also its relation to tropical cyclogenesis in one paper! They further suggested the “oscillation might be helpful for the extended-range forecast of initiation and development of typhoons”, a fact which is still to be fully exploited by forecasters today.

Xie et al. (1963) plotted time series of u700 from three weather stations, shown from east to west down the page in Figure 3: Thiruvananthapuram (Kerala, India), Ho Chi Minh City (Vietnam) and Zamboanga (Philippines). These were hand-drawn and had no temporal filtering applied to the data. They noted “[t]here is a consistent phase change of the zonal wind from Station 43371 [Thiruvananthapuram] to Station 98836 [Zamboanga]. When the westerlies intensified in India, they also intensified in south-east Asia, with a slight temporal delay… The change of zonal wind with time at these stations exhibited a wave-like oscillatory characteristic, with an average oscillatory period of around one-and-a-half months.” We can now recognize this as the MJO circulation, with intraseasonal timescale and eastward propagation.

Figure 3: Time series of u700 against month from three stations (rows; longitudes overlaid in blue — see main text for details) during 1958–1960 (columns). Black dots denote typhoons. From Xie et al. (1963). Red arrows added by Li et al. (in press) to highlight intraseasonal periods.

The Real-Time Multivariate MJO (RMM) indices (Wheeler and Hendon, 2004), computed from OLR and zonal wind, are widely used to plot MJO propagation but date to 1974 only. However, Oliver and Thompson (2012) reconstructed the indices back to 1905 by regressing against surface pressure data. Using the reconstructed indices we can plot the propagation events documented by Xie et al. (1963; Figure 4). There is reasonable agreement – e.g. anomalous low-level westerlies in July 1958 (Figure 3), as seen in Figure 2A, roughly correspond to phases 5–6 (Figure 4b); while anomalous low-level easterlies in late July–August, as seen in Figures 2C–E, roughly correspond to phases 8–2.

The Xie paper is not widely known, mainly because it was published in Chinese. Who knows how many other important papers may exist in non-English journals, unknown to the wider academic world? The analysis of the intraseasonal oscillation by Xie et al (1963) was less detailed than that by Madden and Julian (1971, 1972), and it was unarguably the latter brace of papers which brought the oscillation to the attention of the tropical community. However, if it should come to be known as the XMJO, in recognition of Xie et al.’s pioneering study of 1963, I for one would not complain.

Figure 4: (a) MJO diagram showing the location of active convection in each phase. (b-d, below) MJO events for the times shown in Figure 3, using the reconstructed RMM indices of Oliver and Thompson (2012).

Klotzbach, P. J., 2014. The Madden–Julian Oscillation’s Impacts on Worldwide Tropical Cyclone Activity. J. Climate, 27, 2317–2330.
Lavender, S. L. and Matthews, A. J., 2009. Response of the West African Monsoon to the Madden–Julian Oscillation. J. Climate, 22, 4097–4116.
Li, T., Wang, L., Peng, M., Wang, B., Zhang, C., Lau, W. and Kuo, H.-C. (in press), A Paper on the Tropical Intraseasonal Oscillation Published in 1963 in a Chinese Journal. Bull. Amer. Meteor. Soc..
Madden, R. A. and Julian, P. R., 1971. Detection of a 40–50 Day Oscillation in the Zonal Wind in the Tropical Pacific. J. Atmos. Sci., 28, 702–708.
Madden, R. A. and Julian, P. R., 1972. Description of Global-Scale Circulation Cells in the Tropics with a 40–50 Day Period. J. Atmos. Sci., 29, 1109–1123.
Oliver, E. C. J. and Thompson, K. R., 2012. A Reconstruction of Madden-Julian Oscillation Variability from 1905 to 2008. J. Climate, 25, 1996–2019.
Peatman, S. C., Matthews, A. J. and Stevens, D. P., 2014. Propagation of the Madden–Julian Oscillation through the Maritime Continent and scale interaction with the diurnal cycle of precipitation. Q. J. R. Meteorol. Soc., 140, 814–825.
Singh, M., Bhatla, R. and Pattanaik, D. R., 2017. An apparent relationship between Madden–Julian Oscillation and the advance of Indian summer monsoon. Int. J. Climatol., 37, 1951–1960.
Tang, Y. and Yu, B., 2013. MJO and its relationship to ENSO. J. Geophys. Res., 113, D14106.
Xie, Y.-B., Chen, S.-J., Zhang, I.-L., and Hung, Y.-L., 1963. A preliminarily statistic and synoptic study about the basic currents over southeastern Asia and the initiation of typhoons. Acta Meteorologica Sinica, 33, 206–217.
Wheeler, M. C. and Hendon, H. H., 2004. An All-Season Real-Time Multivariate MJO Index: Development of an Index for Monitoring and Prediction. Mon. Wea. Rev., 132, 1917–1932.
Zhang, C., 2013. Madden–Julian Oscillation: Bridging Weather and Climate. Bull. Amer. Meteor. Soc., 94, 1849–1870.

Posted in Climate, Madden-Julian Oscillation (MJO) | Leave a comment

Evaluating convective-permitting models over South Africa

By Will Keat

Several operational forecasting centres around the world now run convective-permitting models (CPMs) to forecast rainfall. These kilometre-scale models are sufficiently high  resolution to allow convection to be resolved explicitly (i.e. without the need for parameterisation), and have been shown to lead to improved forecasting skill for convective weather events (Clark et al 2016). This is important for helping to mitigate damaging severe convective weather impacts, such as flooding. A key challenge, however, is developing diagnostic tools to evaluate these models that are appropriate and reliable globally. Since 2016, the South African Weather Service (SAWS) has routinely run convective-scale simulations at 4 and 1.5 km resolutions (named SA4 and SA1.5 respectively hereafter) to assist with forecast operations across southern Africa, allowing the applicability of a variety of diagnostics in challenging convective weather situations to be tested. The domains of these models are shown in Figure 1.

Figure 1: Domains for the SA4 and SA1.5 models. The black dots show radars in the eastern half of South Africa; the grey box is approximately the region used in the analysis shown in Figure 2.

The high temporal and spatial resolution of weather radars makes them an ideal tool for studying convection. In 2010, SAWS invested in a large upgrade to their weather radar network, which provides three-dimensional measurements of backscatter (approximately related to the number and size of hydrometeors) every six minutes over thousands of square kilometres. These radar data provide a unique opportunity for model evaluation on the African continent.

How well do CPMs predict the diurnal cycle over South Africa?
One of the key challenges for CPMs is reproducing accurate timing of convection initiation and the diurnal cycle. One diagnostic used in this project to examine this representation is the fraction of data above a fixed radar reflectivity factor.  Figure 2 shows the diurnal cycle of the cloud-and-precipitation fraction above a 10 dBZ threshold for the radar (top row), SA1.5 (middle row) and SA4 model (bottom row) for 12 (left) and 25 (right) November 2016 during significant convective outbreaks over the Highveld region of South Africa (grey box in Figure 1). The SA4 and SA1.5 models generally compare well in terms of diurnal cycle of cloud profile, reproducing deeper clouds on days when deeper convection is observed with the radars, but some key differences can be observed. The SA1.5 tends to peak 1–2 hours too early, whereas the SA4 tends to peak on time or slightly too late. It appears that the models underestimate the height reached by the various cloud-fraction contours. In terms of cloud amount, generally the SA4 and SA1.5 compare well to each other for individual case days, but whether the models over or underestimate the observed cloud amount is case-dependent.

Figure 2: Diurnal cycle of cloud-and-precipitation fraction for 12 November 2016 (left) and 25 November 2016 (right) using a 10 dBZ threshold, for the SAWS radar observations (top row), SA1.5 (middle row), and SA4 (bottom row).

What next?
Similar behaviour of CPMs (such as incorrect initiation timing and too shallow storms) has been observed in other parts of the world, such as over southern UK (Lean et al., 2008, Stein et al., 2015). Understanding why we observe these differences is essential for continual CPM development. Current work is investigating the realism of vertical atmospheric temperature profiles in the models, since these determine atmospheric stability and are therefore fundamental to convection initiation and evolution. Storm tracking, similar to that used in the DYMECS project over the southern UK (Stein et al., 2015), is also being implemented. It is hoped that analysing differences between storm lifecycles in the radar and models will help to identify reasons for incorrect model behaviour.


Clark, P., Roberts, N. , Lean, H. , Ballard, S. P. and Charlton‐Perez, C., 2016. Convection‐permitting models: a step‐change in rainfall forecasting. Met. Apps, 23: 165-181. doi:10.1002/met.1538

Lean, H.W., P.A. Clark, M. Dixon, N.M. Roberts, A. Fitch, R. Forbes, and C. Halliwell, 2008: Characteristics of High-Resolution Versions of the Met Office Unified Model for Forecasting Convection over the United Kingdom. Mon. Wea. Rev., 136, 3408–3424, https://doi.org/10.1175/2008MWR2332.1 

Stein, T.H., R.J. Hogan, P.A. Clark, C.E. Halliwell, K.E. Hanley, H.W. Lean, J.C. Nicol, and R.S. Plant, 2015: The DYMECS Project: A Statistical Approach for the Evaluation of Convective Storms in High-Resolution NWP Models. Bull. Amer. Meteor. Soc., 96, 939–951, https://doi.org/10.1175/BAMS-D-13-00279.1 

Posted in Climate | Leave a comment

Climate change in the Mediterranean Sea

By Fanny Adloff

The Mediterranean is the largest semi-enclosed sea on our planet. Acting as a miniature ocean, this basin is appropriate to study climate change impact on the ocean. The residence time of the Mediterranean waters – of about a century – is smaller than in the world ocean, so that a quicker response to climate change is expected in this vulnerable basin. Hot-spot of marine biodiversity, the Mediterranean hosts 10,000 to 12,000 marine species, one fourth of them being endemic. The Mediterranean suffers from high anthropic pressure at the shore and at the open sea, which substantially increases its ecosystem vulnerability.

IPCC global climate models do not have enough resolution accurately to represent the complex circulation and the water masses of the Mediterranean basin. High-resolution regional ocean models have been thus developed to study the impact of climate change on this specific region.

A six-member ensemble regional climate simulation covers the period 1961-2099 to study the oceanic response to climate change and its uncertainty (Adloff et al., 2015). The ocean response can vary depending on (i) the socio-economic scenario, (ii) the Atlantic boundary conditions, (iii) the fresh water fluxes from river input and Black Sea inflow, or (iv) the atmospheric surface fluxes. To assess the impact of climate change, the 30 year reference period 1961-1990 is compared to the ‘end of 21st century’ period 2070-2099.

The spatially averaged sea surface temperature increase is predicted to be 1.7°C under low-emission scenario, and up to 3°C under the larger emission scenario (Figure 1). The warming could reach 4°C in some specific regions like the Balearic Islands. The surface warning spreads non-homogenously toward depth, and the deep-water formation dynamics are also affected.

Figure 1: Composite of sea surface temperature anomalies maxima (top) and minima (bottom) for the 2070–2099 period w.r.t. 1961–1990 (°C). The largest (maxima) or smaller (minima) anomaly out of the six scenario simulations is represented at each grid point

Sea surface salinity could increase by 1 g of salt per kilogram of water by the end of the century. These hydrographic changes will lead to substantial modification of the water masses properties of the Mediterranean and of its thermohaline circulation (THC). This three-dimensional circulation is mainly driven by deep water formation processes taking place in winter when surface water becomes denser and sinks towards the bottom of the ocean, bringing oxygen to the deepest layers.

In the present climate, this phenomenon mainly occurs in the western Mediterranean (in the Gulf of Lion and in the Adriatic Sea). In future climates, the simulations show considerable changes of the Mediterranean THC, with a large increase of deep water formation in the Eastern Mediterranean (in the Levantine basin).

Significant modifications of surface currents and of mean sea level are also simulated, the latter being very sensitive to the chosen Atlantic boundary conditions (Figure 2, Slangen et al. 2017).

Figure 2: Cumulative thermosteric sea-level change w.r.t. 1961–1990 (cm), averaged over the Mediterranean Sea from the six-member ensemble scenario simulations from Adloff et al. (2015). In blue, the uncertainties linked to the choice of the prescribed hydrographic conditions of Atlantic waters west of Gibraltar, and in red, the uncertainties linked to the choice of the socio-economic scenario

These changes to the physical characteristics of the Mediterranean Sea could severally affect its marine ecosystems. Because of the warmer waters, many species migrate northward and get trapped by a ‘cul-de-sac’ effect at the north coasts of the Gulf of Lion, the Adriatic Sea and the Aegean Sea. Also, massive extinction of non-migratory species such as Gorgons or Posidonias could take place under climate change.


Adloff F. , S. Somot, F. Sevault, G. Jordà, R. Aznar, M. Déqué, M. Herrmann, M. Marcos, C. Dubois, E. Padorno, E. Alvarez-Fanjul, D. Gomis, 2015. Mediterranean Sea response to climate change in an ensemble of 21st century scenarios. Climate Dynamics, doi:10.1007/s00382-015-2507-3.

Benedetti F., F. Guilhaumon, F. Adloff, S.D. Ayata, 2017. Investigating uncertainties in zooplankton composition shifts under climate change scenarios in the Mediterranean Sea. Ecography, doi:10.1111/ecog.02434.

Slangen A.B.A., F. Adloff, S. Jevrejeva, P. W. Leclercq, B. Marzeion, Y. Wada, R. Winkelmann, 2016. A review of recent updates of sea-level projections at global and regional scales. Surveys in Geophysics, doi:10.1007/s10712-016-9374-2

Posted in Climate change, Climate modelling, Environmental hazards, Oceans | Leave a comment

The “size” of the NWP/DA problem

By Javier Amezcua

There is a professor in the University of Reading that likes to say that the Data Assimilation (DA) problem in Numerical Weather Prediction (NWP) is larger than the size of the universe (estimated to be around 1080 atoms, give or take). I thought it would be interesting to explain what he means by that and how one arrives at such an assertion. While doing this, I will also discuss some basic mathematical concepts that are involved in the combined NWP/DA problem.

First we have to understand NWP. In 1904 Vilhem Bjerknes – one of the fathers of meteorology – stated in a seminal paper that predicting the weather is “simply” a physics problem that can be solved mathematically. After all, the atmosphere is a fluid (made of nitrogen, oxygen, argon, water vapour, etc) that sits on top of a rotating sphere (more or less) in which we happen to live. Motion in this fluid is driven by solar radiation, constrained by the rotation of the Earth (effects such as the Coriolis ‘force’ and Taylor columns come to mind), and subject to boundary conditions (the ocean and land). If we consider the variables: velocity (a 3-dimensional variable), air density, humidity, temperature and pressure, then the evolution of the atmosphere is governed by the equations in Figure 1. Although they seem complicated, these expressions come from simple basic principles: conservation of mass, energy and momentum, as well as an equation of state (perhaps this is the least intuitive one).

Figure 1: Equations governing the evolution of the variables in the atmosphere.

The problem does not seem too big so far, right? Seven equations for seven variables (remember that velocity has three components). These variables, however, depend on one time component and three space components (latitude, longitude and height), i.e. these are spatial fields. Six of them are partial differential equations, i.e. they contain derivatives both in space and time.

“Solving” the equations in Figure 1 means performing the time-integration of the seven spatial fields. This is not an easy task. Simplifications can be done (e.g. by considering limits, equilibrium cases, scale analysis) and one can reach some analytical solutions for the simplified problems. However, in the general case (which is the one used in NWP) the equations have to be solved numerically. This requires both spatial and temporal discretisation. For the spatial part, imagine the atmosphere as a hollow shell, and imagine that we divide this shell in small boxes (recall this is a three-dimensional problem). The smaller the boxes the more precise our solution is.

In a simple latitude-longitude grid, the number of boxes (called gridpoints) is

Nboxes = Nlatitudes x Nlongitudes x Nlevels.

And since in each of the boxes we have seven variables we would have the following number of effective variables:

Nvariables = Nboxes x 7.

Let’s say we discretise every degree in both longitude and latitude, and we assign 20 vertical levels. This already results in Nvariables ~ 107  (in operational centres this number is closer to 107). Suddenly we have gone from 7 fields to 107 grid variables! On top of that, these variables evolve in time.

Now, let us think of the kind of system that we are dealing with. As I have discussed in an older post in this same blog, the atmosphere belongs to a class of dynamical systems known as chaotic. In these systems, a single evolution does not really mean much after a given lead-time, since tiny perturbations in the initial conditions can render completely different trajectories after a time. The dream would be to evolve probability density functions and update them with new observations when these are available. This is the ultimate (yet perhaps unattainable) objective of DA.

As a mental exercise, let us think of constructing the pdf’s empirically (histograms) for the variables of the atmosphere at a given time. For simplicity let’s think we divide the range of each variable in 10 bins (not very high resolution, but this will do). For each bin there is a frequency count. For two variables we would have 102 bins, and it would look something like Figure 2. Can you imagine a histogram in three or more dimensions? (I cannot).

Figure 2: A 2-dimensional (bi-variate) histogram (centre) with 1-dimensional histograms in the marginal axes. (Source: Python documentation)

Now, let us go back to our problem. How many frequency counts do we need to store in order to have an empirical probabilistic view of the atmosphere? Nbins = 10Nvariables. If we go back to our calculations we would get: Nbins = 1010000000 … which is colossal! If we stored a number in each and every atom of the universe, we would still have a huge quantity of numbers without a place to be stored. This is what the original assertion refers to. A more precise way of saying it would be: “an empirical probabilistic representation of the discretised state of the atmosphere is bigger than the size of the universe”.

You may be thinking: can the problem be simplified? Perhaps. For instance, if the distribution of the variables is a joint multi-variate Gaussian (a common assumption), then this distribution is totally determined by a mean vector and a covariance matrix. They are still massive entities: the vector has  109 elements and the covariance matrix has 1018 elements. Imagine storing these elements and doing computations with them … well, this is what we do in supercomputers. But this approximation (Gaussianity) is sometimes out of its depth, so we are always looking for ways to improve our probabilistic representation.

Posted in Climate, data assimilation, Numerical modelling, Weather forecasting | Leave a comment