Data assimilation training at the University of Reading

Data assimilation training at the University of Reading

by Amos Lawless

In March 2019 the Data Assimilation Research Centre at the University of Reading organised a 4-day training course in data assimilation, in collaboration with the National Centre for Earth Observation, ECMWF and the DARE project. The course was attended by 24 early-career researchers from 10 different countries, including scientists from universities, research institutes and industry.

The aim of the course was to give students a solid grounding in the theory of data assimilation methods, as well as the opportunity to apply data assimilation methods to a range of numerical models. The first day of the course saw a general introduction to data assimilation, followed by a more in-depth look at variational methods, both from a theoretical and practical point of view. A computer practical session in the afternoon gave students the opportunity to deepen their understanding by running a variational scheme on a simple numerical model. The day ended with an ice-breaker event, allowing attendees to discuss their particular research projects and their interest in data assimilation over a drink and some nibbles.

The remainder of the course looked at the theory and practice of other data assimilation methods, each supported by computer practical sessions, including the ensemble Kalman filter on day 2, hybrid methods on day 3 and the particle filter on the final day. In between students were treated to two lectures on practical applications: PhD student Jemima Tabeart spoke of her work looking at observation error correlations in the Met Office 1DVar data assimilation system, while research fellow Polly Smith spoke about coupled atmosphere-ocean data assimilation. At the end of day 3 a group meal was organised at the Zerodegrees microbrewery restaurant in the centre of Reading, giving further opportunity for informal discussions of the course material and how students could use the ideas in their own projects. At the end of the final day, after attendees were presented with their course certificates, the staff were also presented with a gift – A 900-piece Lego wind turbine from the Danish attendees. So, if you don’t hear from us for a while, you will know why!

All lecture notes from the course and material for the computer practicals are available to download from the course web site




Ellicott City security cameras could offer useful and real time flood information

Ellicott City security cameras could offer useful and real time flood information

by Sanita Vetra-Carvalho

We have come across an illustrating source of a network of security cameras capturing a flash flood in Ellicott City, Maryland, US on Sunday 27th of May 2018. The video is a collage of 12 cameras all located on or near the Main Street in Ellicott City. The information from these videos would have been useful at the time of the flooding.

The videos clearly show how the Patapsco River and two out of its four tributaries (Tiber River and Hudson River) rapidly swell and overflow. We see how this results in the Main Street also becoming a fast flowing river with water washing away cars, destroying buildings, and accumulating debris. The flood lasted only four hours but caused catastrophic damage to local infrastructure, residents property, and claimed a life of a National Guardsman [1,4].

Figure 1. The map showing Ellicott City, Maryland and adjacent rivers. Image credit USA Today.


The cameras were installed by a local property owner, Ron Peters and can be seen in this YouTube video:


The flooding event

A storm released nearly two months of rain, over 9 inches (24cm) in just two hours (3 to 5pm local time), which swept away several roads, cars and brought more than 10 feet (3.0 m) of rapidly moving water down Main Street in the Old Ellicott City [1]. The old city is a very urban area set in a valley next to Patapsco River and its four tributaries. Due to the urban landscape the rainfall has nowhere to go except for running down the valley to the main river.

This was the second 1-in-1000 flood event within two years in Ellicott City. Both 2016 and 2018 events claimed lives and caused millions of dollars in damage [1,4]. However, flooding is nothing new to this city.  The city officials are looking into introducing green areas in the city to allow the rain water to be absorbed into the ground reducing the surface runoff.


New flood alert system

Associate Professor Nirmalya Roy and his group from University of Maryland Baltimore County (UMBC) are working on using a network of temperature and liquid sensors and have produced a new flood warning system for Ellicott City [2,4]. They are also working on using the local flood related information from social media such as Twitter into the flood alter algorithm which warns public through loudspeakers in the city [3,4].

It is clear that the videos captured from these security cameras provide a rich source of water information of the rivers and the Main Street. Information such as water levels and surface velocities can be extracted from these videos [5] and used as part of an existing flood warning system or an independent one. Further, videos also capture additional information on floods and damage caused that is valuable to rescue teams, insurance companies etc.