Our field work

The international project team are using field site locations in the UK, Australia and Norway, and further afield, to test the methodology during high flood flow events. As we all know, one can never accurately predict rainfall, so the team has been agile, working with colleagues across Europe and Asia to ensure that sites are available where high quality, scientifically robust validation can take place. The innovative use of both fluvial and tidal field sites across multiple countries is helping to minimise the risks associated with the capture of individual flood events and allowing the testing of the methodology under multiple conditions.

Firstly, the team selects the site and timing of field-work based on expected flows/rainfall. This is then relayed to Planet to task the collection of video imagery using their SkySat satellite constellation at the same time as in-situ validation measurements of river surface velocity and discharge are being obtained by the Team. Synchronising these events enables the team to compare the ground-based data and the satellite-imagery for that site and time, providing a valuable ground-truthing dataset. Several different image velocimetry approaches currently used in ground-based monitoring are being tested by the Project Team for EO data, to develop recommendations for their operational use and ongoing development.


Validation on the Burdekin River in Queensland, Australian

The first field campaign took place on the Burdekin River in February 2022. Mark Randall and his team travelled over 24 hours in sometimes treacherous conditions (watch the video) to acquire these drone images, as well as Acoustic Doppler current profiler (ADCP) measurements. At the same time Planet acquired three videos of the same stretch of river,

Measurements taken from the ADCP, which is the universally agreed ground-validation method for flow velocities and discharge calculations in rivers, are being compared to flow velocities derived from the drone and satellite video images acquired over the same location and time period. Watch the video to get a feel for field work in the expansive Australian environment.

 

Validation across Scotland

Image
Satellite image over the Falls of Lora

A still from a satellite video of a strong tidal flow at the Falls of Lora, Argyll, 18 March 2022. Image courtesy Planet Labs

In March 2022, Nick and Paul Scholefield (UKCEH) travelled more than 1000km, collecting hydrometric data from four sites in Scotland - river Tweed at Norham, river Spey at Boat O'Brig, river Tay at Ballathie, and at the Falls of Lora, a tidal site, in Loch Etive. Fine skies, with wisps of clouds dominated most skies - a shock for Scotland in March, but in case we got ahead of ourselves, no usable satellite video was acquired at Ballathie thanks to thick cloud cover there.

Cloud cover aside, the project has to contend with river levels. There were no 'good' high flows this year - so best efforts were employed to find the times when there would most likely peak or near-peak flows for this year. Good flows ensure that there are visible 'tracers' (turbulence in most cases) on the surface which are used by the algorithms to calculate flow velocities.

Despite these challenges, initial results are promising. Above is an image taken from a 30-sec, high resolution (50cm pixel) satellite video over Falls of Lora site (in particular the area of white water below and around the bridge).

Below is an example of a drone image that has undergone processing. This image was shot with a £1000 DJI Air 2S consumer grade drone.

Image
A still from a processed drone video of a tidal flow

A still from a processed drone video of the same tidal flow at the Falls of Lora on 18 March 2022. Video shot with a £1000 DJI Air 2S consumer grade drone.

To enable the construction of a floodplain model, Paul also flew his LiDAR-equipped drone over the sites, to generate very high resolution LIDAR maps. These will help improve results by enabling more precise rectification of imagery, as well as providing a detailed cross-section of out-of-bank flows (see image below for an example from Boat O'Brig).

Image
A LiDAR point cloud of a river and fields created with a DJI M300 RTK drone and DJI L1 LiDAR sensor
A LiDAR point cloud created with a DJI M300 RTK drone and DJI L1 LiDAR sensor. River Spey at Boat O’ Brig, 17 March 2022