In the penultimate account in a series of articles on how the mountains of Scotland influence our approach to monitoring and flood forecasting, Mike Reading from the Met Office writes about the challenges of mountain forecasting.

“In Scotland, we face some very challenging upland environments for flood forecasting, as explained in this article.  Forecasting for mountain regions follows a similar process to any other forecast. This starts by looking at observational data, then analysing the initial situation. Computational processes are then applied to create forecast values. Finally, a conclusion is drawn from all the available information.

Observational data largely consists of measurements from weather stations, rain radar and satellite imagery. The analysis of observational data helps to verify model data and influences the final forecast. Observational data is also incorporated into computer models through a process called data assimilation, which acts to bring the model closer to reality at the start of each model run.

The meteorologist will take the available observational and model data and use these, along with their own experience, to create a forecast for precipitation accumulations in the mountains, whether this will be locked-up as snow or fall as rain, and whether there is likely to be melting of lying snow.

What are the challenges around forecasting in the mountains?

Accurate observational data is crucial to creating an accurate forecast. However, official weather observation stations in the mountains are rare, with only four in Scotland and one in England, and these stations only recording wind, temperature and dewpoint. This requires meteorologists to look at surrounding low-level observations then assess how they are likely to relate to conditions at higher altitudes, often a less than straightforward task. Rain radar is also less reliable in mountain regions as the topography often interferes with the radar beam. Algorithms attempt to mitigate against this to an extent by enhancing the rain radar signal over mountains to account for the effect of orographic uplift. However, these algorithms are not perfect and a lack of rain gauges in the mountains to verify rain radar means that there is always a degree of uncertainty over exactly how much precipitation is falling over the hills.

In addition to limited observational data, the complex terrain of mountains is impossible for models to recreate exactly. This is important because mountainous terrain will have a large impact on rainfall accumulations, enhancing it in some areas through orographic uplift and the Seeder-Feeder effect, and acting as a focus for thunderstorm development in the summer months. Other areas will experience much drier weather than they would otherwise experience in flat terrain. While all models will attempt to recreate these terrain effects, the extent to which they achieve this will depend largely on the model resolution.

Past upgrades to the Met Office supercomputer have vastly improved the accuracy with which the models represent the topography. The highest resolution operational model run by the Met Office, the UKV with a resolution of 1.5km, will usually do a good job of picking out areas of orographic enhancement, but even these high resolution models will fail to capture local features and the exact height of hills. This will result in the raw models often underestimating, and occasionally overestimating, precipitation accumulations in the mountains. Post processed model data can often do a better job than raw models due to taking into account local climatology and past observational data, but once again the sparsity of observations in the mountains limits the effectiveness of this post processing.

Another challenge that mountain environments pose to flood forecasting is snowmelt. While models have become much better at predicting whether precipitation will fall as rain, sleet or snow, what happens to snow after it reaches the ground is more challenging. In a mountain environment snow will rarely fall as an evenly distributed layer. Wind usually transports it resulting in relatively bare areas and deeper drifts where wind-blown snow has accumulated. In addition to this, the UK’s maritime climate often results in freeze-thaw processes that can transform loose powder snow into hard snow-ice. Both of these processes can make snow more resilient to melting. These factors are captured poorly, if at all, by the model, and need to be considered by the meteorologist.

One final challenge of mountain regions for flood forecasting is assessing how impactful any potential flooding will be. Due to orographic uplift, rainfall totals in mountainous regions are often considerably higher than in surrounding low-lying regions. Mountain regions tend to be far more resilient to these higher figures due to their sparse population and the frequency with which they experience high rainfall events. However, they will channel accumulations from across a large catchment area into a small number of watercourses at lower altitudes. The time period over which accumulations fall and how quickly this feeds into watercourses at lower levels must be considered. Wind direction also plays an important role; large rainfall events brought in by winds from an unusual direction can often be more impactful than those on prevailing winds.

All of the above results in the mountains being one of the most difficult regions to forecast for. Despite huge advances in model performance over recent years, a considerable amount of experience and local knowledge remains crucial to providing accurate guidance when it comes to forecasting what impact mountainous regions may have on flood forecasting. The Met Office forecasting team at Aberdeen, part of the Scottish Flood Forecasting Service, has been creating mountain forecasts since the early 1980s. This has built up valuable experience in predicting mountain weather and, working alongside the forecasting hydrologist, now has a very important role in forecasting its potential to lead to impactful flooding.”