Storm Frank-Short Range Forecast Performance

Figure 1

Figure 1: Weather chart for 0000 GMT on 30th December shows storm Frank located to the NW of Scotland.

On the 30th December, 2015, new maximum peak levels were recorded on the Rivers Cree, Dee and upper Tweed due to rainfall which fell overnight from the 29th to the 30th December associated with Storm Frank; an unusually deep area of low pressure which brought weather fronts across Scotland (see figure 1). Around 370 properties were flooded in Ballater, NE Scotland, alone, but the devastation was wide spread across Scotland (see figure 2). In the previous article “Storm Frank – how was the medium range flood guidance?” we discussed the development and 2- 5 day forecast of Storm Frank and its potential impacts, whereas the aim of the current article is to discuss the variations in, and factors affecting, the forecast performance over the 24 hours prior to the event peak flows to give an overview assessment of model performance.

Figure2

Figure 2: (a)The Flood Guidance Statement (FGS) maps issued ahead of the event on the 29th-30th December, and a map (b) of public reported incidents of flooding associated with storm Frank.

Antecedent Conditions

There had been significant snowfall over the high ground of the South West and North East over the Christmas period (see figure 3a; MODIS image for the 27th December,2015). The 27th-28th December had been relatively dry but there was some rainfall overnight on the 28th into the morning of the 29th, some of which will have fallen as snow. However, by the 29th December, much of the snow cover had melted particularly at lower elevations (figure 3b). Grid-to-Grid (G2G), the gridded hydrological model which we use to forecast river flows nationally, has a snow hydrology component, which estimates snowfall, accumulation and melt based on precipitation, air temperature and basic snow pack principles, and it had simulated the observed pattern of melt.

Figure3
Figure 3: MODIS imagery (courtesy of NEODAAS/University of Dundee) for (a) 1300hrs 27th December, 2015, and (b) 1300hrs 29th December, 2015. NB- in the second image the ground extent of snow cover is obscured by cloud cover, however, the reduction in snowcover can be seen by comparing the images for the north of Scotland.

 Storm Frank- Rainfall

Figure5

Figure 4: 24 hour rainfall totals recorded during storm Frank from 0900hrs on 29th December, 2015.

The main bulk of the rainfall associated with Storm Frank occurred between 1800 hrs (+3 hours for NE Scotland) on the 29th December and 0900hrs (+3 hours for the NE) on the 30th December. Maximum recorded rainfall for the event was 130.6mm which fell in 15.5 hours at Lower Black Laggan raingauge in the southwest (Dumfries and Galloway), whereas in the Upper Dee catchment maximum rainfall was recorded at Glenmuick at 95.6mm in 15 hours. In general, 50 to 100mm of rain fell across the Southern Uplands and much of Highland Scotland (figure 4).

Sample forecast Locations

Table 1 is a sample of gauging stations located near to locations where impacts associated with flooding (see figure 2) occurred and/or new maximum recorded levels were observed (indicated by the values highlighted in red in column E). Columns F, G and H indicates the performance of the three G2G forecasts run in the 24 hours prior to the event peaks by comparing the percentage difference of the forecast peak to the observed and the difference in the estimated time of peak. For the difference in timing a negative value indicates that the forecast peak was predicted to occur earlier than it did in reality and the values are fractions of an hour (therefore a value of -0.25 indicates the forecast peak was estimated to occur 15 minutes earlier than in reality). Forecasts 1 and 2 (F1 and F2) were run at 13:45 and 18:30 on the 29/12/ 2015, and F3 at 07:30 on 30/12/2015. These forecasts use deterministic rainfall forecasts from the Met Office UKV runs at 0900hr, 1200hr and 0000hr.

Table1

Table 1: Current best estimate of observed Q and time of peak on the 30th December, 2015, recorded in WISKI at several locations across Scotland, and the performance of the three G2G forecasts at these locations prior to the event peak. Forecasts 1 and 2 (F1 and F2) were run at 13:45 and 18:30 on the 29/12/ 2015, and F3 at 07:30 on 30/12/2015. Difference of the forecast peaks and timings to the observed at given as percentage difference for flow, and fraction of an hour for time. *value for observed peak is based on the current rating and could be subject to review.

The values highlighted in yellow in columns F-H show at which forecast the previous maximum flow (column D) was forecast to be exceeded. For the 8 sites (red highlight) in table 1 where new peaks were recorded, F1 predicted the previous maximum flow being exceeded for 6 of the sites, and by F3 all of the 8 occurrences were predicted.

In general, the forecasts performed well for predicting the timing of the peak flows, and for indicating the severity of the event at least 24 hours ahead (refer to previous blog article for longer lead time identification of the severity of Storm Frank). For most sites the forecasted peak occurred ahead of the observed time (red text in table 1), which is the preferable direction of timing errors, and the forecast which performed the worst in terms of timing was forecast 2. For the upper Dee gauges which are used for assessing risk for Ballater, the estimate of time for the peaks were in the range of 3 hours too early to 15 minutes late.

Figure 6_v2

Figure 5: Forecast and observed discharge and catchment 3 hour rainfall totals for the River Dee at Polhollick. The observed precipitation is composite (gauge and radar) rainfall from G2G, the forecast precipitation are from runs at 0900hr, 1500hr and 0000hr for G2G simulations at 13:55 and 18:35 on the 29-12-15 and 07:30 on the 30-12-15.

In regards to forecast performance for the peak flows, the severity and spatial extent of the impacts were generally predicted with sufficient lead time (+24 hours) to co-ordinate emergency response. However, in terms of whether the peak flows were over- or under- estimated there is an approximate geographical split across Scotland. For the gauges located in the NE, such as those in the River Dee catchment upstream of Ballater, Forecasts 1 and 2 underestimated the peak discharge of the fluvial event, with forecast 2 performing the most poorly in terms of timings and underestimation of the peak flows (see example in figure 5). The performance of forecast 2 can in part be attributed to the underestimation of rainfall in the forecasts for these catchments, as forecast 2 generally predicted the least amount of precipitation. However, forecast 1 overpredicted rainfall for these smaller and higher elevation catchments, but the flow at the gauge locations still underestimated peak flows.

Figure7

Figure 6: Forecast and observed discharge and catchment average 3 hour rainfall totals for the River Cree at Newton Stewart. The observed precipitation is composite (gauge and radar) rainfall from UKV runs at 0900hr, 1500hr and 0000hrs for G2G simulations at 13:55 and 18:35 on the 29-15-15 and 07:30 on the 30-12-15.

Forecast 1 generally overpredicted the amount of rainfall across Scotland, but whereas forecast 2 underpredicted the rainfall for the Dee catchments, it overpredicted for the more southern gauges (and the River Dee at Park). The rainfall totals were however, still less than for forecast 1. The estimates of peak flows were generally improved over the southern catchments compared to those in the NE. Figure 6 shows the variations in predicted precipitation and river flow for the different forecasts for the River Cree at Newton Stewart in Dumfries and Galloway.

Summary

Based on the weather and national river forecasting results on the morning of the 29th December, additional simulations of G2G were run as new NWP model runs became available to the forecaster. In general the combined rainfall and river forecasts at gauge locations across Scotland indicated that this event would cause significant, and possibly severe, impacts at several locations across Scotland with sufficient lead times to enable co-ordination and planning of emergency response. For 6 of the 8 sites analysed in this blog which recorded new maximum peak river levels, the forecast run at 1345hrs on the 29th December estimated that the previous maximum recorded levels would be exceeded. However, the performance of the forecasts varies across Scotland, and is shown to be very sensitive to the rainfall forecast and model structural uncertainties particularly for the smaller catchments. It’s important to remember that the models analysed here are national scale models being used to predict local impacts and the detailing of rainfall intensity at the scale of small catchments is challenging, despite the relatively high resolution of the hydrological model (1km). The spatial variation in rainfall can be significant even in a situation of widespread (or dynamically forced) rainfall, and the highest resolution of the NWP models is 1.5km. Local rainfall totals and intensity can be focussed in certain catchments due to marginal changes in low-level airflow, with the potential of slow moving ‘mountain-waves’.

As science and technology improves, the predictions of rainfall and river flows will better represent local effects and thus the spatial variations. The evolution of such science will require careful assessment of the capability and application of such data to hydrological modelling in near real time as is required for flood forecasting. However, in the mean time, the performance of the forecasting tools and models which are available to us have been tested and proven this winter, during some of the worst events on record, to be invaluable in providing good and timely indications of the likely flood impacts.

Posted in Forecasting, G2G, Hydrometeorology, Risk communication, Weather prediction | 2 Comments

Storm Frank – how was the medium range flood guidance?

The rainfall associated with Storm Frank on 30 December 2015 caused some of the most severe flooding seen in Scotland for many years, particularly along the river Dee in Aberdeenshire and the rivers Cree and Nith in Dumfries and Galloway, along with record river flows and  rainfall amounts in some areas. In the first of a series of articles John Mitchell and Richard Maxey discuss how well the event was predicted, and how the forecasting service reacted in the days leading up to the flooding.

How well did we forecast the severity of Storm Frank?

In addition to the weather forecast information there was a clear signal from the Grid-to-Grid hydrological model that extreme river flows would be reached in several locations. The guidance from the Met Office discussed below helped give the hydrologists confidence in the flow forecasts that were produced.

December 25

The first signs of a major storm were noted on Christmas Day. Met Office Guidance available to the flood forecasters noted developments in the USA, where warm air was moving north from the Gulf of Mexico and cold air moving south across Canada to combine and generate a strong jet, with the potential of generating very deep lows running towards the UK later on the 29th. These developments could produce prolonged heavy rain. In addition a note was made of strong orographic modulation likely over western hills and very significant rainfall totals.

UM broadscale 300hPa gph wind speeds for 0001Z Wednesday 30 Dec Note the buckled jet extending NE’wards towards W UK, with a marked left exit (development area) lying close to NW UK at that time.

UM broadscale 300hPa gph wind speeds for 0001Z Wednesday 30 Dec
Note the buckled jet extending NE’wards towards W UK, with a marked left exit (development area) lying close to NW UK at that time.

MOGREPS Forecast Charts for 1200Z Tuesday 29 Dec Demonstrate a consistent signal for deep low pressure to be located just to the west of the British Isles. Note that whilst the signal is consistent there is greater uncertainty as to the low’s precise location.

MOGREPS Forecast Charts for 1200Z Tuesday 29 Dec
Demonstrate a consistent signal for deep low pressure to be located just to the west of the British Isles. Note that whilst the signal is consistent there is greater uncertainty as to the low’s precise location.

December 26

On Boxing Day the risk of disruptive rainfall was noted and a National Severe Weather Alert for rainfall was issued. Concern that the synoptic setup was conducive to strong orographic enhancement in addition to dynamic forcing was noted.

Boxing Day: 12Z UKPP Forecast 24-hour Precipitation accumulations for Wednesday 30th Shows locally very large forecast rainfall totals over W and SW Scotland. Note the ‘finger’ of intense totals towards the southern Grampian mountains.

Boxing Day: 12Z UKPP Forecast 24-hour Precipitation accumulations for Wednesday 30th
Shows locally very large forecast rainfall totals over W and SW Scotland. Note the ‘finger’ of intense totals towards the southern Grampian mountains.

Using the Met Office MOGREPS system forecasts noted that the majority of output suggested that some very humid air may be entrained into the system with waves developing along the cold front, allowing a broad warm sector twinned with a very strong flow to cover to the UK on Wednesday. This would provide all the ingredients required to give a further bout of prolonged, heavy precipitation across some western areas. Specifically the strongest signal for heavy rain looked to be on the Wednesday (30th), again highlighting SW Scotland in particular as potentially seeing some very significant totals. This was the first signal of a backing of the wind-flow and hence distribution of orographic rain.

The Flood Guidance Statement (FGS) issued on 26 Dec indicated a low (yellow) flood risk for the south of Scotland for 30 Dec (day 5), indicating a very low likelihood of significant impacts.

December 27

27 December: MOGREPS probability of rainfall totals >50mm (left) 24hr to 18Z Wednesday 30th December. Note emphasis for highest rainfall over Western Scotland and Dumfries and Galloway where probabilities over 60% are indicated.

27 December:
MOGREPS probability of rainfall totals >50mm (left) 24hr to 18Z Wednesday 30th December.
Note emphasis for highest rainfall over Western Scotland and Dumfries and Galloway where probabilities over 60% are indicated.

On 27th December there was reasonably high confidence in the broad development – but there was continued uncertainty as weather disturbances were expected to run along the front. Heavy and persistent rain on SSW facing up-slopes was indicated for the 29/30th December. A Weather Warning – a Yellow Alert for Rain for the 30th December was issued to the public on the 27th that indicated ‘ 60-80 mm over some higher ground within this area, and the potential for the most exposed locations to receive in excess of 120 mm.’

The Flood Guidance Statement (FGS) issued on 27 Dec indicated a medium (amber) flood risk for the south of Scotland for 30 Dec (day 4), indicating a low likelihood of severe impacts. A low (yellow) flood risk for Aberdeenshire was also idenitifed, indicating a low likelihood of significant impacts.

December 28

On 28th December guidance was still for heavy and persistent rain affecting many western areas. There was still high confidence in the spell of wet weather, though there was continuing uncertainty in detailed distribution and timing of the weather systems crossing the UK. Amber warnings of rain were issued – ‘Rainfall accumulations through this period expected to be widely 20-40mm, with 80mm over some higher ground within the area, and the potential for some exposed locations in southwest Scotland to receive 100-150 mm.’

The Flood Guidance Statement (FGS) issued on 28 Dec indicated a medium (amber) flood risk for the south of Scotland for 30 Dec (day 3), indicating a low likelihood of severe impacts. A low (yellow) flood risk for Aberdeenshire was also idenitifed, indicating a low likelihood of significant impacts.Summary Flood Guidance Statement maps for the days leading up to the event on Dec 30.

Summary Flood Guidance Statement maps for the days leading up to the event on Dec 30.

Summary

The potential for heavy, orographic rain on Dec 30 associated with Storm Frank was identified on Christmas Day. Ensemble forecast models furnished the forecasters with invaluable information to determine and communicate the likelihood of the event.
The possibility of a significant event in the south of the country was seen four days in advance (day 5 yellow) with possible severe impacts identified 3 days in advance (day 4 amber). As will be seen, the long lead time forecasts for this area were very good.

The Aberdeenshire event was flagged three days in advance (as a day 4 yellow), but as discussed above the uncertainty in the precise distribution and detailed timing of the event meant that severe impacts for this area were not identified at this early stage. Subtle alteration of wind-direction (and speed) can have marked impact on the distribution of orographic rainfall. In this case, the backing of winds to the south brought greater exposure to the south Cairngorm Mountains and Upper Deeside catchments. At the broad-scale the weather models performed well. It will always be challenging to match capability for wider-scale guidance against the demand for detailed local information.

The next article in this series will look at the forecasts for rainfall and river flow in the immediate run up to the event, how they informed the flood guidance the day before and during the event, and how they performed compared to what happened on the ground.

Posted in Flood, Forecasting, Risk communication, Uncategorized, Weather prediction | 2 Comments

RAINS: Radar Applications in Northern Scotland

Storm Frank rainfall radar (2)

Caption: 24 hour quantitative precipitation estimates from radar across Scotland for Storm Frank (29th – 30th December 2015). Typical catchment average rainfall exceeded 80 mm in 24 hours for the Upper Dee, with some point estimates of 150 mm. All background radar data © Met Office.

The past two months have brought significant rainfall across much of Scotland resulting in some severe disruption due to the impacts of flooding.  Impacts have been seen in early December with Storm Desmond, through to Storm Frank which brought major flooding along the Dee Valley in Northern Scotland.

The use of radar is essential to increase our understanding of the spatial representation and real time estimation of rainfall for flood forecasting across Scotland.  However, whilst this is a vital tool there are some unique challenges with doing so in Scotland not least the effects of the upland environment and distance from the radar.  To this end the flood forecasting service are working with the National Centre for Atmospheric Science (NCAS) and the University of Leeds to deploy a mobile X-band radar to research the use of high resolution radar measurements for improved flood forecasting.

The RAINS project (Radar Applications In Northern Scotland) will run from January to July 2016 and will be making detailed measurements of the atmosphere from a radar based in Kinloss, Morayshire, specifically looking to measure clouds and rainfall as they form across Northern Scotland.  The NCAS radar is capable of measuring clouds and precipitation in high detail and is the only one of its type in the United Kingdom.

X Band deployment

The mobile X Band radar deployed at Kinloss Barracks in Moray (left) and initial quantitative precipitation estimates on the 2nd February (right) when flooding occurred across Ross-shire.

The collaborative project involving the Met Office and Scottish Water will specifically look to improve the understanding of rainfall and the prediction of floods through comparative hydrometeorological modelling using the quantitative precipitation estimates of rainfall generated by the Kinloss radar.

Posted in Flood, Forecasting, Hydrometeorology, Radar, Research | Leave a comment

The 2005 Teviot floods: then and now

As previously reported Storm Desmond caused severe flooding over the weekend of the 5th and 6th December in parts of the country including the Teviot Valley. The River Teviot in Hawick recorded its second highest river flow on record and close to the peak levels recorded in the damaging 2005 flood. However, levels at Ormiston Mill 20km downstream recorded their highest on a record which dates back to 1960. In this article we take a look at what has changed in our capabilities in flood forecasting in the intervening ten years between these two events.

Teviot@CrailingToftsFarmNT69692522

The River Teviot in Crailing in Roxburghshire recording its highest river level on a record spanning 55 years. Copyright Ali Graham Photography.

How do the flood predictions compare?

Meteorological forecasts supporting flooding operations ahead of the 2005 flood did provide early indications with national severe weather warnings activated several days in advance. However for hydrologists what has changed is the access to more detailed spatial and temporal meteorological forecasting tools to help in the decision making process, such as the use of probabilistic rainfall forecasts based on depth and duration. What this supports is the provision of guidance in terms of what is likely, but also what could be expected as a worst case scenario.

Meteorological forecasts then and now

An example of one of the 2005 severe weather warnings for the Scottish Borders, at this time the service was based on fixed rainfall triggers (left). By comparison tools such as MOGREPS-UK that are used now provide guidance on predicted likelihood of a range of rainfall depth-duration, in this case 40mm in 6 hours, with 70% likelihood across the Southern Borders for Storm Desmond (right). Copyright Met Office.

In hydrological terms, a review following the 2005 event noted that warning lead times for this flood on the Teviot were at or around 3 hours and it recommended the development of rainfall-runoff forecasting to increase future lead times. Since then systems have been developed on a catchment scale (PDM) and national scale (Grid-to-Grid) for coupling meteorological and hydrological forecasts and as reported previously this led to flood guidance being provided several days in advance of the recent event. More crucially in the 2015 flood these predictions supported community warnings being issued some 24-hours in advance of the main flooding impacts being witnessed.

Were we more prepared for the 2015 flood?

The 2007 Scottish Government study into the flooding in Hawick, Elgin, Glasgow and other communities looked at societal impacts of these events and specifically one aspect for us how communities respond to flood warnings. At the time the study found that only 33% of those that had previously been flooded used Floodline; only 39.7% moved a car on receipt of warning; and crucially the study found that the effectiveness of a warning largely depends on the time it provides to take appropriate action. The study also noted that although local automatic voice messaging systems were in place (through Scottish Borders Council), there were issues around take up, warning messaging being reliant on two systems and speed and efficiency of warning messaging.

Rainfall for the Teviot floods

Hourly rainfall recorded across the Southern Uplands. 48 hour accumulation for the 2005 flood (111mm) and the 2015 flood (96mm). Data courtesy of SEPA.

Since 2005, SEPA has introduced a direct warning service of which over 300 individuals are now registered in Hawick, helped with the awareness of the flood risk and active community engagement through initiatives such as the Hawick Flood Group. Although the full extent of the flood is still to be assessed the response to the recent flood from Emergency Planners was that ‘we were much better prepared’ which aided the proactive evacuation of 600 people from the community. Notably whilst reports of the 2005 flood mention cars being washed away, amongst the 100 properties affected initial reports are that not one car was flooded in the 2015 event as a result of the precautionary actions.

What lessons can we take away from the 2015 floods?

Whilst the key points from the 2015 Teviot flood might be that forecasts were reasonably accurate, warnings were timely and the response was proportionate to the scale of the flood, the same might not be true about the neighbouring community of Newcastleton which was equally devastated in 2005 yet escaped major flooding this time.

Newcastleton G2G prediction for Storm Desmond

The Grid-to-Grid flow predictions for the Liddel at Newcastleton with a observed peak being recorded at 254 cumecs being closest to the minimum flow ensemble member in these forecasts.

Short range flood predictions for the Liddel Water (see figure) were similar to that of Hawick in that some levels were predicted to be close to those of 2005. However, spatial variation in the predicted and observed rainfall meant that Newcastleton narrowly missed the more severe rainfall accumulation observed across other parts of the southern uplands. This case highlights that flood forecasting continues to be challenging, however the changes over the past decade has enabled us to be more prepared for future flooding.

 

 

 

Posted in Flood, Forecasting, G2G, Hydrometeorology, Probabilistic, Risk communication | Leave a comment

Storm Desmond – Scotland’s first red flood risk

Over the weekend of 5 – 6 December 2015 Storm Desmond – the fourth UK named storm of the winter season – brought damaging winds and heavy rain across Scotland, which experienced its most severe flood event since the advent of the forecasting service nearly 5 years ago. Impacts were widespread, and particularly severe in the southern Borders area. We issued our first ever Flood Guidance Statement containing a red (high) risk area, and also issued the first severe flood warnings since the service began. This is the first of two articles looking at the Desmond flooding. This article looks at the forecasting for this event; the second one will look at how this compared with a similar event in 2005.

floodpix

Areas of Hawick affected by flooding during 5 – 6 December 2015                                                                             Photo credits clockwise from top left: bbc.co.uk; Hawick Councillor Ron Smith; The Southern Reporter; Hawick RFC

Exceptional rainfall totals were recorded in a 48hour period from the 4th to the 6th of December as its associated weather fronts wriggled southward across Scotland and northern England.

Forecasters at the Met Office Aberdeen watched the developments of Desmond closely. The ground was already saturated after a wet and gloomy November – so the challenge was to predict as accurately as possible the intensity and distribution of the rainfall. Forecast models tend to provide more reliable guidance for the distribution of rainfall that is dynamic (frontal) rather than locally generated (e.g. thundery showers).

chartsIt was recognised at an early stage that the system would have a moist ‘river’ or ‘conveyer’ of winds ahead of the cold front. It was also important to consider the orographic enhancement of rainfall totals over windward mountain ranges.

Whilst the frontal structure of Desmond was complex, the initial weather predictions supported the view that the cold weather front would clear southward into England; and that the primary rainfall totals would be associated with the leading weather fronts. As Desmond moved towards Iceland, subsequent weather calculations indicated a trend to a slower and more erratic clearance of the cold front southward over the weekend.

201512060000_u1096_ng_ume4_precip2520_accum1440_2km_1_1[1]

24 hour rainfall totals forecast on Friday for Saturday

Then finally; the cold front almost became stationary across the Borders, as further weather disturbances ran along it bringing further pulses of heavy rain.

The greater persistence and intensity of the moist conveyor, as well as the slower clearance of the dynamic rain associated with the cold front – lead to the prediction of exceptional rainfall accumulations across NW England and the Scottish Borders.

High flows were forecast throughout Scotland, as can be seen on the Grid-to-Grid (G2G) river flow model spatial plot, below. The widespread and prolonged nature of the expected rainfall meant that almost unprecendented quantities of run off were expected.

g2gspatial

4/12/15 07:30 Grid-to-Grid forecast for 5/12/15 18:30

Of particular concern were the Scottish Borders and Tayside, the G2G forecasts for which can be seen below. The forecasts for Hawick in particular were very consistent from two days before the event and closely matched what did occur. The forecast flows were comparable with those experienced in the extreme event of 2005, when there was extensive property flooding. The river Tay forecasts, whilst showing quite a range, all consistently indicated an expected flow approaching at least as great as that seen during the flood event of 2005, and possibly even the event of 1993, which was the largest in recent history. In the event it is clear that the model overestimated the Tay flows somewhat, possibly due to the non-inclusion of hydro schemes upstream which may have attenuated the peak of the flow.

desmond flowsThe heightened flood risk for the weekend was first signalled on the Flood Guidance Statement on Thursday, with a very low likelihood of significant impacts indicated. As confidence increased, on Friday the likelihood was increased to high, giving amber status to a large part of the country. Weather forecast models and G2G were giving consistent forecasts of very high rainfall totals and exceptional river flows. During Saturday it became clear that areas of the Scottish Borders, particularly Hawick and Newcastleton, were expected to experience severe impacts. These areas are not very far from the parts of northern England that were also going through extreme conditions.

fgs

Risk maps from successive Flood Guidance Statements (left). Area of Concern map for peak of event (right)

A large amount of rainfall fell across Scotland from Friday through to Sunday. As can be seen below, the main areas affected in Scotland were the Borders and the centre of the country (Tayside area) plus the north west.

radar_animation

Rainfall intensity from RADAR for the event

raintable

Totals in mm recorded at SEPA raingauges during the event (note Teviothead is an underestimate due to the site flooding)

 

 

 

 

 

The impacts of this event were considerable. 15 flood alerts, 75 flood warnings and 2 severe flood warnings were issued across the country.  Over 150 properties were flooded in Hawick and other locations across the country. Significant travel disruption occurred. The Perth flood defence scheme was activated, entailing the closure of the main town bridge. Several railway lines were closed. Travel impacts were also exacerbated by the (unrelated) recent closure of the Forth Road Bridge for emergency maintenance.

This event proved a major challenge for the forecasting service. For the first time we had to decide on issuing a Flood Guidance Statement with an area on high (red) status. We also input to the decision to issue two severe flood warnings, also an unprecedented situation. Rainfall predictions, particularly the UKV 36 hour high resolution forecasts, proved to be very reliable for the most part. G2G forecasts were invaluable. They were particularly in demand for locations where no local model currently exists, and for lead times longer than that used for local models. Timely forecasts and warnings enabled mitigating actions to be taken. As will be shown in the next blog, improvements in forecasting and warning have led to a much improved outcome since the last time we saw this magnitude of flooding in Scotland.

By Richard Maxey and John Mitchell

Posted in Flood, Forecasting, G2G, Uncategorized, Weather prediction | 2 Comments

Decision making in times of uncertainty: the crying wolf

Nikéh Booister presents the second and final part of this series on decision making in times of uncertainty.

As part of her M.Sc. thesis research in Flood Risk Management Nikéh did a part of her research in SEPA. The research focussed on decision making under uncertainty with an outlook on improving decision making based on the cost-loss method. Nikéh continues working in the field of Flood Risk Management with her company FloodCom.

As part of her M.Sc. thesis research in Flood Risk Management Nikéh did a part of her research in SEPA. The research focussed on decision making under uncertainty with an outlook on improving decision making based on the cost-loss method. Nikéh continues working in the field of Flood Risk Management with her company FloodCom.

“The forecaster has to be the expert on knowing what uncertainties are and where they come from as we could have read in the previous post. The forecaster has to decide on what is relevant and not and to a certain extent already has to make a decision on the quantification of the uncertainty. Whether or not a forecaster understands and interprets correctly the uncertainty, the decision maker has a final call and responsibility to “pull a trigger” for action.

Currently most of the forecasts are presented in a deterministic way, which means a single valued number of a possible future. A part of the uncertainty is presented in a description of the future system state, nevertheless this is not quantified. A possibility is to present a forecast in a probabilistic way, where uncertainties are presented.

As mentioned in the blogpost of Paul Ryles; longer lead time makes the prediction of a future situation more challenging; the rate of uncertainty is higher. Probabilistic forecasts indicate possible future outcomes of a system and give the decision maker and responder more possibilities of responding (or at least the consideration of responding and with this more time to respond). This makes it possible to respond earlier in time and make decisions with a lower regret. Making a decision based on a deterministic forecast seems easier because it presents a binary solution (yes or no decisions). Nevertheless research has proven that decision making based on a probabilistic forecast gives better results (Ramos et al. 2013).

In possible flood situations Aberdeenshire Council can pull out the demountable defences to protect Stonehaven from flooding. Nevertheless at the moment a decision has to be made it is not sure if the event will actually materialise but due to time limitations the decision has to be made.  (Source: Aberdeenshire Council, 2014)

In possible flood situations Aberdeenshire Council can pull out the demountable defences to protect Stonehaven from flooding. Nevertheless at the moment a decision has to be made it is not sure if the event will actually materialise but due to time limitations the decision has to be made. (Source: Aberdeenshire Council, 2014)

When decision makers are introduced to the problem of uncertainty in forecasting it can appear to be too much; too complicated; and out of the scope of the decision maker to respond properly to this. This interacts with the fact that the decision makers and responders are the ones that have to deal with for example the uncertainty of citizen response. Thus, the responsibility is bigger. When a wrong decision is made, there is a threat of a “crying wolf” principle. Despite the efforts, there will always be dissatisfaction with the decision taken or damage occurred. And so, someone to be blamed. When a decision is based on a deterministic one, it is the model or the forecast to be blamed and not the decision maker – which makes the situation more acceptable.

The ultimate question in forecasting and decision making is who is responsible in the end? And who is to blame in the case of a wrong decision being taken? Don’t we all try as hard as possible?

Timing and level uncertainty in flood prediction makes it hard to decide. The longer the lead-time the higher the uncertainty (Source: N.Booister M.Sc. thesis)

Timing and level uncertainty in flood prediction makes it hard to decide. The longer the lead-time the higher the uncertainty (Source: N.Booister M.Sc. thesis)

There are approaches that provide an ‘answer’ and provide the forecaster or the decision maker with a decision making perspective, but when we would like to provide the decision maker with a longer lead-time, as mentioned above, probabilistic forecast might be the solution. In order to avoid blaming of the decision maker; a decision making tool or method needs to be in place to support the decision. One of these methods is the cost-loss method. This method compares the probability of an event to occur with the cost of certain mitigation measures and the losses that they avoid. Although it presents a solution method, it seems too static in time and due to its economic point of view, does not take into account intangible effects.

Models are improving, more lead-time is being provided, but uncertainty will always remain. Decision making stays a difficult and also a sensitive topic.”

 

Posted in Forecasting, Probabilistic, Risk communication | Leave a comment

Decision making in times of uncertainty: the forecaster

In a two-part feature, we cover various roles involved in decision making at times of uncertainty. The first is provided by Paul Ryles, Senior Scientist and operational flood forecaster at SEPA.

PR pic

Paul manages SEPA’s flood risk management and operational hydrometeorology trainee programmes and has been in flood warning service development since 2002

“Flood prediction is a science, especially in challenging situations such as providing long lead times (for example through our five day flood guidance statement) or flash flooding (surface water alerting). One of the primary roles of the flood forecaster in these situations is in communicating uncertainty in a way that is useful for responders. The forecaster must quantify uncertainties in the weather forecast (in terms of location, timing, intensities and amounts of rainfall) and any potential uncertainties in the resultant flood forecasting predictions and hazard footprint.

Meteorological and hydrological tools such as MOGREPS and Grid-to-Grid provide the forecaster with the opportunity to quantify uncertainty in flood predictions. Image of MOGREPS-G and probability of greater than 25mm in 24 hours ahead of ex-Hurricane Kate (Copyright Met Office)

Meteorological and hydrological tools such as MOGREPS and Grid-to-Grid provide the forecaster with the opportunity to quantify uncertainty in flood predictions. Image of MOGREPS-G and probability of greater than 25mm in 24 hours ahead of ex-Hurricane Kate (Copyright Met Office)

Understanding what type of information customers require is vital, as is an appreciation of how this information will be used and what type of decision will be made. The flood forecast will often be used to make crucial ‘yes or no’ decisions, such as committing resources in responding to a flood event, evacuations or deploying temporary flood prevention measures. Understanding these customer requirements is as vital as knowledge of meteorology and hydrology in producing successful flood forecasts.

2015 Sep 23_7043

Flood Forecasters participating in Exercise Big Water, testing response and preparedness to new coastal flood forecasting capabilities.

Having competent and experienced flood forecasters is essential in delivering this service. To this end, the Flood Forecasting Service, with support from the Met Office College, has developed a formal qualification in Operational Hydrometeorology. The aim is for all of our flood forecasters to achieve this qualification, which assesses all aspects of producing and communicating our flood forecasts. The forecasters, who will already have completed training in flood forecasting, warning and response, will undertake this qualification after a period of on-the-job training.

Forecasting some flood situations remains a challenge, however ensuring all flood forecasters are trained and competent (and assessed) in accounting for, quantifying and communicating risk at times of uncertainty should maximise the benefit of flood predictions for responders when making difficult decisions.”

Posted in Forecasting, Hydrometeorology, Risk communication | 1 Comment