Skip to main content

Camera Observation and Modelling of 4D Tracer Dispersion in the Atmosphere

Periodic Reporting for period 4 - COMTESSA (Camera Observation and Modelling of 4D Tracer Dispersion in the Atmosphere)

Reporting period: 2020-05-01 to 2021-10-31

Turbulence is one of the long-standing big challenges in the atmospheric sciences. Kinetic energy produced at the largest atmospheric scales cascades down to the molecular scale where it dissipates. A related aspect of turbulence is its effect on tracer dispersion. Turbulence controls the dilution of pollution emitted into the atmospheric boundary layer (ABL). Dispersion modelling is limited by a lack of theoretical understanding as well as experimental data. In past experiments, artificial tracers were released into the atmosphere and resulting atmospheric concentrations measured. However, this was done mainly at discrete sampling locations, not in 3-D.

In COMTESSA, we have executed a set of ground-breaking atmospheric tracer dispersion experiments combined with state-of-the-art data analysis and modelling of turbulent dispersion. The experiments observed sulfur dioxide (SO2) puffs and plumes (both released artificially as well as from a volcano) with nine simultaneously measuring cameras equipped with ultraviolet (UV) and infrared (IR) filters. This provided the basis for a high time- and 3D-space-resolution tomographic imaging of puffs and plumes.
We carried out three artificial tracer release measurement campaigns in the summers of 2017, 2018 and 2019 at the Rena military facility in Eastern Norway (Fig. 1). A fourth campaign in 2020 had to be cancelled because of COVID restrictions. During the campaigns, we released SO2 as an artificial tracer that could be measured downwind from a release mast (10-m high in 2017; 60-m high in 2018 and 2019) that was set up at the site. During the first two experiments, SO2 was blown upward from bottles located at the ground through a pipe. We made both continuous releases over extended time periods (~30 min) as well as near-instantaneous puff releases (Fig. 2). To facilitate even shorter puffs, we filled balloons with SO2, pulled them up the mast, and exploded them there during 2019. The release mast, and another mast, were equipped with eddy covariance systems at different heights to measure turbulent fluctuations of the wind. A drone was used to record temperature profiles throughout the ABL.

To detect SO2 in the atmosphere, we built six UV and three IR SO2 camera systems that were connected to computers running self-developed software to control camera operation and record data. All cameras are also equipped with GPS systems and clinometers to record time, camera position and pose. The SO2 cameras were placed around the mast to observe the SO2. Camera distances from the mast ranged from ~160 m to ~1.2 km.

We also carried out a 3-week campaign at Stromboli volcano in spring of 2019 (Fig. 3). The volcano is a strong natural SO2 source and was selected because of its quasi-continuous activity. Cameras were placed at six positions around the whole island, facilitating observations from different angles (Fig. 4). One week after the end of our campaign, an extraordinarily strong eruption killed one person at exactly the location where we operated one of our cameras.

The measurements were compared with state-of-the-art modelling of turbulent tracer transport. Large-eddy Simulation (LES) models are the most advanced models available to resolve turbulence in the ABL. We adapted the freely available open source code PALM (Parallelized Large-Eddy Simulation Model, Fig. 5) and used it to reproduce the conditions at our field site with an extremely high resolution of ~1 m (Fig. 6 and 7). Comparisons with the turbulence measurements and tomographic data are not yet finished but will be published soon.

The LES simulations were also used as input to a 3D Monte Carlo radiative transfer model (RTM) which simulated the signal recorded by UV cameras to assess if images of SO2 plumes may be used to derive plume statistics of relevance for the study of atmospheric turbulent dispersion.

To reconstruct the 3-D tracer concentration distribution based on camera measurements, we applied two methods, one relatively simple, the other more complicated. The more simple one applies simple triangulation and allows determining the total mass of a tracer puff, the trajectory of its mass centre, and the spread of the puff mass around its centre, as a measure of turbulent dispersion. The more complicated method allows a full 3-D reconstruction of the tracer distribution. Figure 8 shows puff observations from five UV cameras and Figure 9 the corresponding 3-D tomographic concentration reconstruction.

Scientific results of our project were so far mostly communicated via scientific publications, and a few more publications are currently in preparation. The results also led to the improvement of PALM and FLEXPART, models that are widely used in the scientific community. The project has considerably advanced our understanding of turbulence, and has also made large technological advances in camera observations and tomography.
In a novel synthetic set-up, where we used LES output as input for a 3-D radiative transfer model, we created artificial camera images for a variety of conditions (e.g. different solar zenith angles, aerosol loadings). A comparison between artificial images and original LES data confirmed that the cameras can indeed provide quantitative and accurate data that allows determining tracer concentration statistics. This was an important step demonstrating the feasibility of the suggested approach and the first time this was shown.

We performed completely novel tracer release experiments with instantaneous puff and continuous plume releases, where up to nine cameras were used simultaneously to quantitatively record tracer concentrations around the release site. A simple tomographic algorithm allowed to track the centre-of-mass of the puffs and the tracer dispersion around it. Already with this simple set-up, we could separate tracer meandering and relative dispersion and test the scaling laws for relative dispersion in the early phase of puff dispersion.

More complex tomographic algorithms that allow a fully 3-D reconstruction of the tracer concentrations were found to be extremely sensitive to even small inaccuracies in camera position, pose, time synchronization, and lens distortion characteristics. It took a long time until we found a robust-enough algorithm that is now showing promising results. This is currently evaluated in more detail and will be described in a forthcoming publication.

With respect to dispersion modelling, a literature study showed that no previous study could clearly demonstrate the ability of LES to model higher-order moments of the concentration fluctuations generated from a small source in the highly turbulent conditions encountered in the ABL. We could, for the first time, show that the LES approach can simulate higher-order concentration moments (up to the fourth) but only when run at very high resolution.

LES model simulations of the field experiments were done in an unprecedented nesting set-up, where first 3 nesting levels of a mesoscale model were used to scale down meteorological reanalysis data. Four more nesting levels of the LES itself brought the model resolution around our experimental site to ~1 m. A publication, in which the model simulations are compared with eddy covariance measurements at our tower, is currently being prepared. Another publication will compare the model results with the tomographic tracer reconstruction.
Figure 4: Geometry of camera set-up (red symbols) on Stromboli island (shown is volcano topography).
Figure 9: Tomographic 3-D reconstruction of SO2 tracer concentration from observations shown in Fig8
Figure 8: Simultaneous observations of an SO2 tracer puff from five different cameras.
Figure 6: Modeled wind direction (arrows) and speed (contours) at Rena military facility (red mark).
Figure 3: UV and IR camera heads set up 11 June 2019 on Stromboli Island.
Figure 2: Examples of camera observations of an SO2 plume (left) and puff (right).
Figure 7: LES simulation showing northerly component of wind speed for several zoom regions at Rena.
Figure 5: A plume dispersing in a wind tunnel, simulated with LES.
Figure 1: Release tower and camera system at the Rena field site.