Login 
Landfills and drones – two words that do not easily lend themselves to an image of environmental responsibility; and two words that, when put together, perhaps invoke images of dystopian futures such as those portrayed by Arnold Schwarzenegger.
However, the rise of the machines (or more accurately, remotely piloted semi-autonomous aircraft in this case – some may take issue with the use of the “D” word, but I will use it here to avoid acronyms) has opened up an exciting frontier in the science and technology surrounding the difficult measurement challenge that is the quantification of greenhouse gas flux, that is, the total mass per unit time emitted from some point source or area.
Greenhouse gas concentrations
As we are all acutely aware, greenhouse gas concentrations in the atmosphere are on the rise; and have been on the rise since the dawn of the Industrial Revolution. When averaged over decadal timescales, the rate of increase in globally-averaged atmospheric concentrations of the major greenhouse gases (CO2, CH4, and N2O) is seen to be ever-accelerating as our continued dependence on fossil fuels interacts with various biogeochemical and meteorological feedbacks to result in what we know and love as ‘climate change’.
And climate change is already beginning to bite at the bottom lines of big business and the Earth’s natural capital, as what was once a prediction becomes a reality, manifested through extreme weather events and shifts in taken-for-granted regional and seasonal regimes. However, while the science is abundantly clear that climate change is linked to increased greenhouse gas concentrations, and that increased greenhouse gas concentrations are linked to man-made activity, the minutiae of quantifying and apportioning emissions to individual sources or source-types remains incredibly difficult and the subject of an enormous global academic and technological effort.
To be able to predict the greenhouse gas concentrations of the future, we first need to know what is being emitted today and to understand the Earth system in truly incomprehensible detail. This is especially true for natural (or biogenic) sources and sinks of carbon dioxide and methane that do not care about man’s vain efforts to control their emissions, and instead respond directly to the environment in which they find themselves at the time.
Earth system climate models that attempt to simulate the biogeochemical response to induced perturbations are incredibly complex, built from a vast understanding of physics, chemistry and biology, requiring huge resources of super-computing ability to arrive at predictions about the future. However, these models all rely on the quality and accuracy of initial-state data inputs available to them. And, when it comes to climate change prediction, greenhouse gas emission inventories are one of the largest a priori uncertainties. And as it happens, landfill gas emission is arguably one of the least certain (and not insignificant) man-made component terms.
National emission inventories
National greenhouse gas emission inventories have gradually been introduced as an accounting tool by national governments in response to various iterations of UN climate change agreements to reduce carbon emissions. Their problematic (statistical) compilation is the subject of much acknowledged bias and uncertainty, relying on ‘bottom-up’ assumptions and estimates of emissions from various industries and activities, often using proxies such as fuel burned or energy efficiency calculations. The 2015 UNFCC Conference of the Parties in Paris, however, has set in motion a long overdue requirement for signatory states to validate their emission inventories with real measurement –the so-called ‘top-down’ approach.
So now the world is scrambling with an age-old conundrum well-known to atmospheric scientists: just how do you accurately measure emission flux rates? Which brings us nicely back to the subject of this article: how can we use direct measurement to quantify greenhouse gas emissions from the landfill industry (or any individual site thereof)?
Greenhouse gas sources
Greenhouse gas sources can be grouped into what I consider two broad categories:
• Point sources, such as chimney stacks and tailpipes
• Diffuse or hotspot sources, such as landfill, wetlands and peatland
The former is arguably easier to measure in principle as it is possible to monitor volumetric flow rate and to measure concentrations of gases in a well-mixed flue. The latter, however, creates an enormous sampling challenge if we are to use measurements to compare to simple estimates for any individual site. Consider the objective – to derive net total greenhouse gas emission over a wide area. Landfills represent a continuous source of methane gas and carbon dioxide (and other gases) resulting from the microbial decay of buried and exposed waste.
Modern landfills are typically capped with impermeable membranes and a methane gas extraction system for burning and electricity generation to mitigate harmful emissions to the atmosphere; however, so-called fugitive emission still occurs and the pathways between the decaying waste and the atmosphere are often unknown and unpredictable and can occur anywhere and everywhere on a typical site.
To complicate matters further, the strength of this emission can vary wildly with time in response to environmental factors such as temperature, soil moisture content and atmospheric pressure (to name a few). Such emissions are also tightly regulated (e.g. by the UK Environment Agency) but regulation relies on accurate knowledge. So, just how can we measure the bulk emissions of methane gas from a wide area such as that of a typical landfill site?
Given this complex scenario, conventional monitoring practices employed by regulators and industry such as walk-over surveys and flux chamber measurements make the best of a very difficult situation. More expert (and expensive) techniques such as LIDAR surveys, eddy covariance towers, and tracer release “experiments” may offer increased accuracy but all remain the subject of academic scrutiny as to their comparability and the nature of potential systematic errors and biases.
Understanding monitoring methods
Understanding the pros and cons, and accuracy, of all of these methods was the subject of a major international field campaign at a landfill in the UK in 2014, which included seven UK universities, led by this author. Among the candidate flux techniques being evaluated was the method of so-called mass balancing – a gas budgeting technique that is built on the simple premise that if we can measure the concentration of methane that blows in to a site on the wind and we can measure what goes out on the wind at the other side, then we can calculate what must have been added in-between.
The simple-to-understand conceptual model of mass balancing is elegant; yet, as ever, the devil is in the detail. And the detail here is spatiotemporal sampling – the ability to measure methane gas concentration at very high precision as rapidly as possible, in plumes that dilute and advect away on the wind from their surface source. Moreover, as plumes of gas are known to move upward as well as sideways, a vertical measurement capability is especially important. This is precisely where drones (or unmanned aerial vehicles, or remotely piloted aircraft systems – take your choice of parlance) come in.
Monitoring using drones
Drones offer a sampling platform from which to traverse the atmosphere in three-dimensions relatively rapidly (at the spatial scale of landfill sites, anyway). When equipped with high precision miniaturised in-situ sensors, this rapid sampling capability can populate the upwind and downwind mapping required by approaches such as mass balancing and Gaussian plume inversion (look the latter one up if you’re interested) in a way that less accurate remote-sensing instrumentation cannot.
Add in simultaneous measurement of the wind speed and direction and you then have all that you need to calculate mass flux with a crucial advantage over other techniques – a quantifiable uncertainty budget that conservatively and transparently places a tolerance on the calculated flux. In this author’s opinion, no other flux technique can as readily or transparently derive a robust flux error budget.
The simplicity of the mass balance concept and an ability to directly measure (or otherwise know) sources of variability and measurement error allow a direct uncertainty constraint to be placed on any calculated flux.
In late 2014 and early 2015, we conducted field trials of both fixed-wing (Figure 1) and rotary drone (Figure 2) platforms equipped with wind, carbon dioxide and methane sensors at a landfill near Manchester. The fixed wing system (a Bormatec Explorer) was equipped with in situ trace gas sensors, while the hexrotor system (a modified DJI F550) carried its own power line and a Teflon inlet attached to a pump to deliver sampled air to precision instrumentation on the ground.
The fixed wing system carries the advantage that it can fly anywhere rapidly, while the tethered rotary system carries the advantage that is powered from the ground offering long endurance and the ability to measure using heavier ground instrumentation, but suffers from its inability to fly laterally, making it akin to a vertically controllable balloon. Each is fit for different purposes: the fixed wing for rapid 3D sampling overall, and the rotary for rapid and continuous vertical profiling on station.
Figures 3 and 4 show some of the data gathered in our field trial. Figure 3 shows the measured methane concentration sampled by the rotary system as the wind shifts direction. We clearly sample the methane plume and winds blow from the direction of the landfill and establish a good background concentration for air measured when wind blows from other directions. Concentrations up to 30ppm were measured, compared with a 2ppm background.
Figure 4 shows some of the vertical profiling conducted by the fixed wing system, showing the strong vertical gradient of the sampled plume, capped at around 100m above ground level on the day of measurement. The data represent only seven minutes of sampling on this day. With slick operational practice and good conditions, it is possible to see the plume morphology that can be mapped with many hours of measurement, and therefore the increasing accuracy of flux calculation and constraint as a direct result.
In the exercise, we learned much about the logistical realities (and hardships) of working on landfill and of flying drones in difficult settings. Strict CAA regulations on safe flight prevent flying near to buildings, people, or animals outside of the pilot’s control, or flying in adverse weather conditions.
This required detailed flight and site safety planning in advance of any sampling to secure a base of operations free from interference and a position downwind of the landfill site to capture the evolving plume. The proximity of nearby roads and motorways further restricted our flight planning such that we were not always able to effectively map the full extent of the plume from a safe distance. However, the user-friendliness of modern drone auto-pilot and flight planning software allowed us to programme our flights in advance based on the 24-hour weather forecast, and to optimise flight patterns to gain the best 3D sampling of the atmosphere for mass balancing purposes.
Therefore, while a trained pilot must be on standby to take control of the drone in the event of any failure of automated systems, the semi-autonomy of the technology offers exciting planning capability and this can only be expected to improve as it evolves further.
Looking ahead
Our field trials prove the concept of drone-based greenhouse flux measurement and hint at the routine and autonomous nature that could follow. Precision sensing technology is now being miniaturised to permit small electrically-powered drone applications. Despite what was undoubtedly an enormous initial effort for our team, it is possible to see how future networks of automated sensing drones could routinely and continuously monitor and measure the atmosphere in 3D to derive spatially-resolved greenhouse gas emissions maps for future emissions reporting.
And with better emissions understanding, more targets, policies and regulatory practices concerning future greenhouse gas emissions (and air quality for example) could be implemented, or operational practices could be changed in real-time to mitigate unwanted effects, e.g. in gas leak detection.
Our experience, however, was not without failure – let’s just say we had more than one safe “forced landing”, highlighting the need to plan to fail in the context of safety and flight planning. This latter need is actually what I consider the principal limitation of drone-based measurement at the current time, and it is an important practical one.
Drones do (and probably always will) fall from the sky. And given that commercial-off-the-shelf platforms can weigh as much as 15 kg and cost less than £5,000 there is a real concern that damage will be done and life will be lost. With this in mind, the UK and other governments are preparing legislation and licensing requirements to ensure that trained users can and will use them responsibly and safely. However, this necessary requirement may limit their potential for truly autonomous use for the time-being – at least for the near future until the technology is much better proven. Perhaps Judgement Day will be delayed for the time-being…
Published: 11th May 2016 in AWE International
