Dr Ross Bannister

Senior Research Scientist (NCEO)

We need to talk about assimilation …

Modelling the real world is never a perfect process. Errors and uncertainties in all models accumulate in time, consequently placing limits on the value of any forecast, whether it represents a prediction of the weather, or of flooding, etc. It is possible though to achieve good quality forecasts by correcting the model as it evolves with information from fresh observations, where and when they are available. In the case of weather forecasting, such observations constrain the model to “today’s weather”, so that “tomorrow’s weather” can be forecast as accurately as possible.

Merging observations with models is called data assimilation (or DA for short). For weather forecasting, DA helps to determine initial conditions of a numerical weather prediction model, whose output is used by weather forecasters, or used as input into other models that predict flooding depending upon the expected rainfall. The practical application of DA takes a forecast field valid for today and computes a sort of correction field based on the observations. The modified field, is generally a more realistic model state. The correction fields are determined from an algorithm, which uses not only today’s forecast and observations, but also information about how likely these pieces of information are correct (remember the model forecast is erroneous, and even observations are never exact measures of reality). This algorithm is based as closely as possible on a set of equations called the Kalman Filter equations, which were originally developed for engineering applications in the 1950s and 60s. The Kalman Filter equations have a wide range of uses, e.g. the control of trajectories of bodies over large distances (think of space flight and warfare), but have since been adapted for weather forecasting. The weather forecasting problem though has vastly more degrees of freedom than the basic Kalman Filter can cope with so the full Kalman Filter equations are not used; instead approximate equations are used which are solved using either variational procedures, or using a vastly reduced number of variables (e.g. the ensemble Kalman Filter).

Today’s forecast is our very best estimate of today’s weather … probably not

Standard DA methods assume that errors in forecasts and observations obey a normal distribution (otherwise known as a Gaussian distribution). Consider a forecast of today’s relative humidity (RH) above, say London, of about 98% RH (i.e. close to saturation), and let the error (standard deviation) of this forecast be of order 2.5% RH. The Gaussian distribution with this standard deviation is shown in Fig. 1(a). One can interpret this plot by imagining that one has access to a very large number of forecasts, and the distribution is a means of showing the frequency of possible forecast outcomes. Knowing that about 2/3 of the area of a Gaussian distribution is within one standard deviation of the mean, then 2/3 of the forecasts would have values of 98 ± 2.5 % RH. In the absence of multiple forecasts, we assume that the single forecast represents the most likely point of this distribution (the mean, or centre point), and observations can make adjustments to this forecast (examples are the arrows in Fig. 1(a)). The assumed Gaussian specifies how the relative humidity forecast is allowed to be modified by the observations. If an observation suggests that the forecast should really be drier by about 6% (purple arrow), then the distribution would inhibit this change as it has such a small probability. An equally large positive change also has the same degree of unlikeliness (red arrow). Smaller changes though – of either sign – are deemed far more likely (blue and green arrows), and so such modifications would be more likely to happen. Specification of this distribution is how DA can control how observations can update the forecast.

Figure 1: Possible probability distributions of errors in a forecast of relative humidity when the forecast value is 98% RH. Panel (a) has an assumed Gaussian form of this distribution, and panel b has a non-Gaussian form. The arrows in panel a serve to illustrate likely (blue and green), and unlikely (purple and red) changes to the forecast value as a result of assimilating observations. The Gaussian has identical probabilities for positive and negative updates, but the particular non-Gaussian shown has larger probabilities of updating negatively than positively.

Actual distributions are rarely Gaussian shaped. Shown in Fig. 1(b) is a distribution whose shape is far more realistic for a forecast close to saturation (surprisingly this distribution has the same standard deviation as the Gaussian in Fig. 1(a)). This non-Gaussian distribution is asymmetric – it says that there is a higher probability of an observation lowering the relative humidity (negative correction) than raising it. This makes sense given the forecast is at 98% RH – it is more more likely that the true atmosphere has 95% RH (3% RH lower than the forecast), than super-saturated at 101% RH (3% RH higher). There is a similar picture for forecasts of very dry air (e.g. 2% RH), where qualitatively the distribution would be the mirror image of that in Fig. 1(b). These represent examples where the distributions are not only non-Gaussian, but are also strongly flow-dependent, which makes them difficult to specify in operational situations.

In reality the picture is even more complicated than this, as the distribution needs to account for relationships between different variables (e.g. between different positions, and between different kinds of fields like temperature, and winds). There is also the issue of corrections being made that involve moisture phase changes (between ice, water, and vapour), between which there are also potentially strongly flow-dependent relationships. Also generally in weather forecasting problems, most observations made are not of the model’s own variables, but of something related to them, which represents another complication in the DA problem as a whole.

Why is this important?

The aim of doing data assimilation (DA) is to set realistic initial conditions for models that will deliver more accurate forecasts of the future weather than if the model was not updated by the latest observations. Such forecasts include accurate predictions of rainfall for use in models that predict flooding. Some off-line results that we did in the FRANC project suggest that if non-Gaussianity of the forecast distribution is not well accounted for, it is possible for DA to actually give worse initial conditions – worse by the fact that the initial conditions can be unphysical (for instance represent negative, or supersaturated humidities).

A pragmatic solution might be to simply set negative humidities to zero, and supersaturated humidities to 100% RH after the assimilation step, and before the model is run. This `cure’ though could have side-effects, since such adjustments do not necessarily obey the subtle balances that might be at play, e.g., between humidity and temperature. As always prevention is better than cure, and in this case, we believe that prevention may well be to do with improving the assimilation.

Dr Rob Thompson, University of Reading

8th December 2017

It was last week, the big gathering, 2 days of flash flooding, a flood of presentations and discussions, and a chance to catch up too. The event was again hosted in the University of Reading’s Meadow suite… a nice location – and if it was summer a nice view too – but it isn’t.

The meeting took a largely normal format, after a quick introduction into talks on the science of each TENDERLY work package, a keynote and then we had posters with nibbles – and finally a tapas dinner (in the new Thames Lido) a sleep and then more talks, on various integration projects. Lunch and early career researchers had a short meeting, with videos describing our work for critique and a discussion on the programme paper that we will soon be writing.

Here’s the crew, with not too many folk hiding in the background (I’ll not call you out).

Blogs past and future really tell you about the science we do in this programme, so this doesn’t feel the right time for me to discuss that. But Sue Mason’s talk, told us about what’s going on at the EA. The fairly large changes to the way they operate sound dramatic, but really important ones that should help us all with future flood risk, warning and management.

These science meetings really are a good chance to catch up with work of others, but also friends, I think they’d be less successful without the friendly catching up. I can use it to bounce ideas and thoughts off others (Thanks Tim!).

The lunches provided were again nice too, but I must say I think the naming of the sandwiches was a little… misleading – thanks to @FloodSkinner for pointing out this beauty:

The poster session was interesting, I think many people find the train difficult, talking about your poster for just 2 minutes, to a quite varied audience. But it is a good way in a short space of time to find the posters you really want to look at in real detail and identify who you want to talk to over the rest of the meeting, and David Flack did a great job leading the train – here’s a shot of it in action.

We also had the VR flash flood “game” running, a few more people fell victim to the flood wave, here we see Brian Golding, a little before he gets washed away – but it certainly left an impression on him! You can these days view this yourself with a simple google cardboard viewer and a smartphone, on youtube… its got more detail too on the static view point than the full VR version.

Overall, I think the meeting was a good success and a busy year is ahead, all leading up to the Royal Society meeting, and I guess I should finish in the same place the conference did, an advert for that very event:

Dr Rob Thompson, University of Reading

11October2017

A summer of Floods

We’ve entered autumn (or maybe winter given how it feels!), so its definitely time to look back on the summer we’ve had (or not as it feels). This summer has certainly seemed to have been quite convective, without being warm after June.

This article though will focus on the day of the most high profile UK event, on July 18^th. This day the headlines were floods in Coverack, Cornwall… but they were far from alone, my home of Reading also got hit by a really very intense storm, and one that produced a huge amount of rain and statistics of note. Coverack suffered as a consequence of a very intense thunderstorm, combined with local orography leading to a lot of water pouring through the village. People had to be airlifted to safety and a huge amount of damage was done when flood waters flowing through the village exceeded a metre in depth. 

Coverack’s case is a fine example of the current state of forecasting… the morning’s weather forecast for the area that day was

“Thundery showers are expected to push north across southern parts of the UK through Tuesday evening. Although many places won’t see these showers, there is a chance of localised flooding of homes, businesses and susceptible roads. Frequent lightning may be an additional hazard with possible disruption to power networks. Similarly, but very locally, hail may cause impacts.”

And I can’t argue that was anything but a good forecast, the thunderstorms certainly materialised and were very intense, but small and hence very localised, perfect to demonstrate a problem for the scientists involved in research in to events such as this, observations are sparse. None of the weather stations in the area received a significant amount of rainfall. This is where radar comes in, radar provides areal coverage over a large range (250km from the radar in the UK operational system), with the UK network having coverage over the whole country, almost all with 1km resolution, we can see the rainfall falling even without the presence of rain gauges.

We can see from the radar images that the intense rainfall in the area of Coverack lasted for a few hours, storms triggering to maintain the rainfall rates for several hours.

But we can also see what was happening further East… that last radar image at 1900 shows a large convective system over the New Forest, heading towards Reading. The next few hours would be interesting for certain.

Where we do have numbers and gauge data is our own University of Reading atmospheric observatory. The rain in Reading was phenomenal, and I personally had the misfortune of being out when it occurred. It was like rain I’d previously only experienced in the tropics, driving conditions were horrendous, with incredibly reduced visibility and water simply unable to clear the roads quickly enough. Little appeared in media about the storm, yet there were certainly local impacts from flooding, as evidenced by this photo I took myself in South Reading:

The rain was of most interest to me, but the lightning was also impressive, both sheet and fork lightning. More than 100,000 strikes over the UK.

So I’ll focus on the rainfall rates, very high rainfall rates are not that uncommon, but lasting more than a few minutes is very unusual – and this storm was very, very unusual. The university’s tipping bucket rain gauge got 38.6 mm in the storm (which was later closely matched by the manual gauge), and 1.6 mm in the little shower 20 minutes before the big storm. The gauge experienced 35 mm of that total in 45 minutes, a phenomenal amount, averaging 47 mm/hr for three quarters of an hour. The University of Reading campus gets approximately 640 mm per year on average, so we had 5.5% of the annual rain in 45 minutes – that’s pretty incredible.

Events like these two are exactly what the FFIR programme is all about, about getting better warnings to the right places and more specifically.

By Hannah Cloke  (University of Reading) 24th March 2017

Although formal funded societies and projects can be very important in advancing research and improving how science is used, the unfunded voluntary community initiative of HEPEX has been one of the most important networks that I have been involved in during my career so far. HEPEX (which stands for Hydrologic Ensemble Prediction Experiment) began in 2004 just as I took up my first post as a University Lecturer. HEPEX aims to advance the science and practice of hydrological ensemble prediction and how it is used for risk-based decision making.

Participation in HEPEX is open to anyone wishing to contribute to its objectives, and so the HEPEX community thrives through organising scientific workshops and sessions at major conferences (such as the European Geosciences Union General Assembly every Spring), coordinating joint experiments, highlighting best practice in hydrologic ensemble prediction systems to help practitioners find out how ensemble prediction is being used around the world in different applications (such as for hydropower or flood forecasting), and through our online community interaction including webinars and blog discussions (www.hepex.org; @hepexorg).  The HEPEX community are also very keen to develop serious games to help communicate best practice and to understand how we can improve forecast communication (Arnal et al, 2016)

It is not always easy to explain what you work on, especially when you have to avoid using jargon specific to your field. Yet, this is something that we all have to do. It is important to be able to explain your research simply in order to communicate effectively with scientists in other fields and, for example, businesses, policy makers and the public.  This week in HEPEX we have been thinking about this with the help of a little competition: using only the 200 most commonly used words of the English dictionary, explain “Ensemble hydrological forecasting”. Please consider having a try, you could win yourself a special mystery prize.

The next HEPEX meeting will be in Melbourne in February 2018 in the height of the gorgeous warm Australian summer. The theme for the workshop is ‘breaking the barriers’ to highlight current challenges facing ensemble forecasting researchers and practitioners and how they can (and have!) been overcome.  How can you resist such a tempting offer?

Want to know more? Want to join our community?

HEPEX website: www.hepex.org

HEPEX twitter: @hepexorg

Arnal, L., Ramos, M.-H., Coughlan de Perez, E., Cloke, H. L., Stephens, E., Wetterhall, F., van Andel, S. J., and Pappenberger, F., 2016. Willingness-to-pay for a probabilistic flood forecast: a risk-based decision-making game, Hydrol. Earth Syst. Sci., 20, 3109-3128, doi:10.5194/hess-20-3109-2016.

By David Flack  (University of Reading) 13th March 2017

April is now rapidly approaching, and with it the UK often experiences showery conditions (April Showers). Throughout the course of my PhD (to be submitted next week) I’ve been examining the forecast of different convective thunders storms in different regimes (see my earlier post explaining about the different regimes), so for this blog will focus on what I have found about the forecasts of showers.

Now, whilst we experience showers throughout the year, we generally expect them to be more frequent in April, as the temperatures start to rise and convection changes from being mainly over the sea to over the land. As with most equilibrium convection it’s difficult to predict the location of these showers, which can (in certain situations) result in flooding. See below for a typical April showers case (NERC Satellite Receiving Station, University of Dundee, Scotland, 2012).

To consider the behaviour of showers in forecasts we use convection-permitting models (grid lengths of around 1 km) and run these multiple times to create and ensemble (see earlier posts by Peter Clark and myself for more). Using ensembles allows us to consider different outcomes of the forecasts in terms of whether there will be rainfall, how heavy will it be and where will it be?

In the latest bit of research I have done (currently under review in Monthly Weather Review) I have shown that forecasting the exact location of showers is very difficult. The research I have done assumes we know everything about the initial conditions for the forecast (so have perfect observations of what the atmosphere is like at the start of the forecast) and assumes that our large-scale models are perfect (so we can generate perfect boundary conditions for our forecast). Even then we can only predict the location of showers to around 10 km or so. Note that this is a best case scenario, it’s likely to be a larger distance in reality – more details can be found in my post on the Meteorology Department’s PhD blog.

So how can we communicate to the general public about this uncertainty? Well, it’s difficult – especially as we are still researching this uncertainty so don’t yet know that much about it. There are however ways we can, and do, communicate the risk:

1. Indicate a region where showers are likely to happen – this is what is currently done on TV weather forecasts
2. Indicate the chance of having a shower pass over a certain location – this is also done “you’ll be unlucky to catch a shower” is a phrase that is often used in local TV weather forecasts – what this means is showers are possible within your region but we don’t know if you will be effected.

The question is, and it will always remain, are the better ways to communicate this uncertainty? How can we communicate this on apps (as more and more people are just glancing at their phones for such forecasts)? This is difficult and I won’t cover it – but what I will suggest is a technique that I often use (other than looking at the radar images online and seeing where the showers are moving).

Consider I want to know if a shower is going to hit Reading – I consider the Reading forecast, and the forecast at different locations near Reading e.g. Basingstoke, Newbury, Maidenhead and Wallingford. If I see a showers icon for any of the five locations considered I then know there is a change that a shower is possible at my location.

As we continue to research into this area we will learn more about the uncertainties associated with predicting showers, and once we know more we can then communicate that better in weather forecasts.

By Dr. Albert Chen (University of Exeter) 16th February 2017

The FFIR researcher Dr Albert Chen from the Centre for Water Systems, University of Exeter, was invited by the APEC Climate Center (APCC) to present at the APCC-ASEAN Disaster Management Symposium.

The event was held on 9-10 February 2017 in Jakarta, Indonesia, aiming to encourage the dialogues between scientists and practitioners that will bridge the gap between science and policy in disaster risk reduction and management. Over 50 delegates from 14 countries, mostly government officials, attended the symposium and shared their knowledge with each other.

Dr Chen shared the work in the ongoing NERC FFIR programme and discussed potential future research to help policy makers. The audience identified that flash flooding as a key area where better science and technology are desperately needed to support decision makings in hazard mitigation. Research outcomes from FFIR programme will benefit ASEAN countries in building the capacity of flood forecasting that consequently will enhance early warning and reduce flood damage.

By Dr. Sarah Dance (University of Reading) 19th December 20146

 Many would say that Numerical Weather Prediction has been using “big data” for decades. Routine forecasts are produced using computational models with billions of variables, and tens of millions of observations, several times a day. Most of these observations come from scientifically designed observing networks, such as satellite instruments, weather radar and carefully sited weather stations. However, urban areas also present rich sources of data, that to date have not been fully explored or exploited (e.g., citizen science, smartphones, internet of things etc.), and could provide significant benefits when forecasting on small scales, at low cost.

In surface scientific networks, point observations are often sited away from buildings, in locations that are intended to be more broadly representative of larger areas and not designed to reflect local urban conditions. These observations lend themselves more naturally to comparison with discretized models, whose grid-lengths may be much larger than the size of a building. For datasets of opportunity, a key problem is to understand the effects of the urban environment on the observations so that uncertainties can be properly attributed and proper quality control procedures established. Furthermore there are complex issues surrounding use of the data, such as personal privacy and data ownership that must be overcome.

For the rest of this article we focus one dataset of opportunity arising from air traffic control radar reports. Mode Selective Enhanced Surveillance (Mode-S EHS) is used by Air Traffic Management to retrieve routine reports on an aircraft’s state vector at a high temporal frequency (every 4 to 12 seconds). The state vector consists of true airspeed, magnetic-heading, ground-speed, ground-heading, altitude and Mach number. Mode-S EHS reports can be used to derive estimates of the ambient air temperature and horizontal wind at the aircraft’s location. These derived observations have the potential to give weather information on fine spatial and temporal scales, especially in the vicinity of airports, where there are millions of reports per day. For example high-frequency reporting of vertical profiles of temperature and wind may provide extra information for use in numerical weather prediction that would have particular value in the forecasting of hazardous weather.  While some of the problems of understanding and using datasets of opportunity are circumvented (the effects of buildings are less relevant to flying aircraft), all measurements during aircraft turns and other manoeuvres have to be discarded. Furthermore, the reports are transmitted in small data packets, with limited precision, with the result that the uncertainty in the derived meteorological observations is very large, particularly at lower altitudes. For more information see

Mirza, A. K., Ballard, S. P., Dance, S. L., Maisey, P., Rooney, G. G. and Stone, E. K. (2016), Comparison of aircraft-derived observations with in situ research aircraft measurements. Q.J.R. Meteorol. Soc., 142: 2949–2967. doi:10.1002/qj.2864

By David Archer  (Visiting Fellow Newcastle university and JBA Trust) 27th October 2016

Over the last four years, I have been compiling chronologies of flash floods and the associated causes from intense rainfall, associated occurrence of hail, and results in terms of drowning, deaths by lightning, destruction of houses and bridges, erosion of hillsides and valleys and flooding of property. The main focus of SINATRA has been on Northeast England, Cumbria and Southwest England but chronologies are now almost complete for Lancashire and Yorkshire; an additional less comprehensive chronology has been prepared for the rest of Britain. The source material has been mainly the online British Newspaper archive, with its 15 million searchable pages, but a wide range of documentary sources has also been used. Given the rapid growth of published newspapers in the mid nineteenth century, the records can be considered comprehensive since at least 1850.

In compiling this chronology, event by event, I was struck by the variability of occurrence by year and by decade, which did not fit with the concept of more intense rainfall in a world warming with climate change (Kendon et al. 2014). The most frequent and really damaging flash floods tended to concentrate in the late nineteenth and early twentieth centuries and there were fewer events in many of the later decades of the twentieth century. Figure 1 shows the decadal chronology for Northeast and Southwest England (Archer et al, 2016)

Figure 1 Time series of flash floods by decade from 1800 to 2010 divided by severity for (a) Northeast England and (b) Southwest England (insets show mapped areas covered by time series).

My first reaction to these findings was: Can I explain them away? Are these patterns of change the result of variable reporting of such events or have they been the result of changing catchment conditions? In response to the first, I am convinced that, except for WWII when such reporting was prohibited, such severe events would be reported and described in the press. With respect to catchment changes, the assessment of the relative magnitude of historical pluvial floods is the most problematic. Urban growth has increased impermeable area (likely to increase flood risk) but sub surface drainage has been improved (likely to decrease flood risk). However, in extreme events such as described, where the rainfall intensity is far in excess of the design capacity of drainage systems, sewers are surcharged and surface flows exceeded gulley capacity in both historical and recent events. A fuller discussion can be found in Archer et al (2016).

The chronology has also assembled a time series showing the decadal variability of large hail in Southwest England and Northeast England (Fig. 2) which shows a similar time distribution to flash floods. It is probable that the less frequent reporting in recent decades of hail causing serious breakage of glass is due to the increased glass strength of standard glass panes but the decline in other reports of large hail must reflect a real decline in occurrence. A similar pattern is reported for the whole of England with decadal declines from a maximum around the turn of the 19th/20th century and a minimum occurrence in the 1970s (Webb et al. 2009).

Figure 2 Number of occurrences of large hail with and without reported extensive glass breakage for Southwest and Northeast England.

Chronologies of historical flash floods and occurrence of large hail for Northeast and Southwest England indicate strong natural variability, with the second half of the twentieth century showing the lowest frequency of such events. Unless we can explain the sources of such variability and incorporate them in models to project future incidence we run the risk is of serious underestimation even without the expected increase in risk due to rising temperatures.
Figure 2 Number of occurrences of large hail with and without reported extensive glass breakage for Southwest and Northeast England.

References

Archer (in press) Hail – historical evidence for influence on flooding, Circulation

Archer, D.R., Parkin, G. and Fowler, H.J. ( In press 2016) Assessing long term flash flooding frequency using historical information, Hydrology Research. doi: 10.2166/nh.2016.031

Kendon, E.J., Roberts, N.M., Fowler, H.J., Roberts, M.J., Chan, S.C. and Senior, C.A. (2014) Heavier summer downpours with climate change revealed by weather forecast resolution model, Nature Climate Change  4, 570–576 doi:10.1038/nclimate2258.

Webb, J.D.C., Elsom, D.M. and Meaden, G.T. (2009) Severe hailstorms in Britain and Ireland, a climatological survey and hazard assessment, Atmospheric Research 93,  587–606.

By Dr Geoff Parkin  (Newcastle University) 17th October 2016

A workshop on modelling flooding from intense rainfall with participants from NERC Franc, Sinatra (and Tenderly) projects as well as local stakeholders with interests in flood risk assessment and response was held in Morpeth, Northumberland on 20-21 Sept 2016. Morpeth has had a long history of flooding, with large events in 1963 following snowmelt, and in 2008 when 1000 properties were affected by a 1:137 year event with a peak flow of 360 m³/s.

The aim of the workshop was to develop an integrated modelling strategy to demonstrate end-to-end forecasting capabilities for a single location, including assessment of different modelling approaches for catchment and urban flood modelling, sensitivity to theoretical patterns of convective and frontal storm event movement of river/stream flows and inundation in the Wansbeck catchment, local tributaries, and town centre, and effects of flooding from multiple sources.

An informative field trip was held on the first day, with attendees inspecting the £27M Morpeth flood alleviation scheme, including new and improved flood barriers in the town, the upstream storage reservoir dam and culverts, and ‘log-catcher’ poles which are designed to prevent impacts of woody debris on infrastructure in the town. This was followed by a visit to the contrasting Dyke Head site in the upper catchment, where a set of Natural Flood Management features have been installed demonstrating an alternative low-cost approach to reducing flood risk.

The main workshop discussions were held on the second day, in the ‘Glass Room’ of the Waterford Lodge Hotel in Morpeth. A structured modelling strategy was agreed, informed by approaches used in the Environment Agency/JBA’s Real-Time Flood Impact Mapping Project. Models used and developed within the research projects and industry-standard models used by consultancies are being applied at the full Wansbeck catchment scale, and at very high resolution in urban areas. Simulations are first being run for the 2008 flood event, with comparison against flood depths reconstructed using crowd-sourced information. We will then assess model performance in simulating flooding from multiple sources (fluvial and pluvial) from hypothetical extreme events with different spatial positioning over the area. Evidence from recent floods in Morpeth support wider understanding that flooding from rivers and from localised rainfall both have significant impact, but their combined effects (e.g. when high river levels restrict discharge from storm drain overflows) can be locally complex. The expected outcomes from the study will be improved understanding of capabilities of models used in flood response in the UK for simulating catchment and urban processes, specifically with respect to end-to-end modelling of flooding from multiple sources.

The afternoon session focussed on understanding more about the needs of communities and organisations for real-time flood risk information, as the first activity in Work Task 3.2 of the Tenderly project. Representatives of first responder organisations (Environment Agency, Northumbrian Water, Northumberland County Council) and flood-affected communities (Morpeth Flood Action Group, Northumberland Community Flood Partnership) provided a range of interesting perspectives on how information is used in during periods leading up to and during flood events. In the Tenderly project, this will help to inform how to make better use of methods developed in Franc and Sinatra and of all sources of information, including improved forecasts of convective as well as frontal rainfall, real-time flood modelling outputs, and crowd-sourced information.

Geoff Parkin, Newcastle University