Understanding floods in the UAE.

Flood modelling in the United Arab Emirates (UAE) is hard. I always knew this intuitively, but it took time to realise why. As I write this, the country is in recovery mode from very significant rainfall.

A typical rain event here delivers 20mm in 24 hours. Yesterday we had 127mm in 24 hours at Expo City, at the end of an already wet winter. I suspect it will be the largest rainfall the country has recorded.

This country is very proficient at cleanup, and the UAE will be back to business very shortly I'm sure. My thoughts are with those that have suffered injury or property damage. Living through a flood can be a very traumatic experience and I encourage our communities here to offer support to each other.

As my house is drying out, it seems a good time to share some reflections on what I've learned about flooding during my time here in Dubai.

Where the water flows

In most parts of the world, when we track a flood event we look at water level readings from gauging stations installed along rivers. Water flows downhill. We know where it came from and where it is going. The UAE does not have rivers in the traditional sense. Our terrain is shaped by the wind, not water. There are very few flow paths that are intuitively recognisable.

In the UAE, the water flows along the base of the dune formations, roughly in lines determined by the prevailing wind direction. Natural river mouths discharging floodwaters to the Arabian Gulf don't really exist.

These dune formations vary in scale. Some dunes are ten kilometers across and contain entire suburbs, others are only fifty meters. But the parallel dunes trap the water to form an interconnected network of surface ponds. Overflowing upstream ponds will strongly influence downstream ponds. It's not a "normal" river system.

And this is where management gets hard. Predicting how much flow an upstream dune will absorb can be influenced by:

• groundwater levels
• porosity of soils
• subsurface flow paths

Instead of measuring the flow in a river with surveyed cross-sections like a typical hydraulic modelling problem, we are faced with measuring highly variable infiltration rates across the entire catchment. Our assumptions about how much rain the ground can absorb become very sensitive.

Changing terrain

Terrain shaped by wind is not stable under the flow of water. Every rainfall event transports sediment. This ranges from small scours down a highway embankment, through to overtopping and erosion of large dunes. At the large end of the scale, we can see historic evidence of this, creating the flow paths of today's Wadis.

The challenge for hydraulic modellers in the UAE is understanding when the risk of erosion and scour is going to be significant enough to change the course of surface water flows, creating otherwise unanticipated outcomes that we need to manage.

Design Challenges

Against this backdrop of uncertainty, developing infrastructure here faces some additional hydraulic challenges:

• Urban developments often don't have a stable channel to discharge stormwater to. If they install traditional pit-and-pipe drainage networks, this will concentrate flows and lead to significant downstream erosion.
• The pre-development sandy soils are very effective at absorbing rain. Changing the surface to concrete will create very significant differences between the pre- and post-development runoff volumes.
• Rainfall events are rare, and sand accumulation is high. Traditional pit-and-pipe networks require frequent low flows to maintain a self-cleansing velocity. Without these cleansing flows, many pipes and culverts are prone to blockage as large volumes of sediment are mobilised during the rarer events.

Solutions for these challenges exist, but they are often nuanced and require careful site-specific design, with recognition of the role each site plays within the complexity of the broader catchment. Design strategies that work in other places need careful customisation here to be effective.

About the analysis

Background analysis for this piece was prepared using the ANUGA hydraulic model on the Hydrata.com modelling platform. Interactive results and open-source model code can be viewed here:

https://hydrata.com/catalogue/#/map/3739

The model uses a steady-state, constant rainfall and a two-dimensional dynamic hydraulic calculation.

Elevation data is sourced from FATHOM's FABDEM 1.2 ~30m grid digital terrain model. Hydraulic modelling is carried out at a maximum resolution of either 30sqm or 100sqm triangles and as such is not suitable for any applied purpose.

The modelling is broad-scale and general in nature, intended to improve our collective understanding of flood mechanisms in the UAE. To improve the model, the following steps would be necessary as a minimum:

1. Develop accurate hydrology and design storms
2. Apply an appropriate loss model to account for the complex infiltration in the sandy soils
3. Utilise an appropriate high-resolution digital terrain model, supplemented with ground survey and site visits.