Successful Integration of UAV Aeromagnetic Mapping with Terrestrial Methane Emissions Surveys in Orphaned Well Remediation

Abstract

Orphaned oil and gas wells represent an important environmental and economic development concern in areas where historical energy exploration and production activity pre-dated regulations on well abandonment documentation practices. From an economic development perspective, these wells pose a subsurface risk to infrastructure development, while the environmental impact of orphaned wells is largely associated with uncontrolled emissions of both fluids and gasses, most notably atmospheric methane. Often neglected in regulatory oversight, methane emissions from orphaned wells contribute to global greenhouse gas concentrations and without proper mitigation, these emissions contribute to and further exacerbate global climate change processes. An inherent challenge of remediation efforts targeting orphaned wells is the consistent observation that only a fraction of located and identified wells produce the majority of methane emissions, yet no methodology exists to effectively separate out heavy emitters without visiting and assessing each and every well. We demonstrate that emitting wells have certain defined characteristics largely pertaining to the presence and integrity of metal casing to the surface, which can be distinguished as an organized high intensity magnetic anomaly in low-altitude UAV-based aeromagnetic surveys. In this paper, we present a database of identified and ground-truthed wellsites correlated to their magnetic signatures and provide a roadmap to initial prioritization of wellsites for subsequent remediation activities that can be implemented in complex environments where other survey options are not feasible. In a regulatory environment where resources dedicated to wellsite remediation are limited, we propose implementing this methodology as a key element of a targeted approach to wellsite prioritization for subsequent remediation activity.

1. Introduction

Wells designated as orphaned are oil and gas wells that were drilled and abandoned prior to the adoption of regulations detailing wellsite remediation, which include detailed documentation of well location and status [1]. As a result, the exact location of orphaned wells needs to be determined prior to any remediation activity taking place, which proves exceedingly difficult as historic maps may prove inaccurate, incomplete, or unavailable. Fortuitously, the vast majority of orphaned wells in areas of historical development feature a vertical metal casing that creates a strong magnetic signature on the surface, making them unambiguous targets in magnetic surveys. To pinpoint the locations of these wells and associated pipelines in regions with minimal plant cover, researchers often rely on ground-based terrestrial magnetic surveys [2]. Similarly, aerial magnetic surveys conducted with piloted aircraft offer an effective alternative to terrestrial surveys in areas devoid of vegetation, human-made obstacles, and gentle landscapes that enable aircraft to fly at low altitudes over extensive areas. Low-altitude helicopter surveys can be used in more environmentally challenging areas [3,4], but are logistically more difficult to deploy and expensive [5]. UAV-based magnetics were largely developed for geological mapping and mineral exploration [6,7,8,9], but they have many more potential applications. In recent years, low-altitude UAV-based aeromagnetic surveys emerged as a potential solution to orphaned well location in areas where terrestrial surveys are impractical due to difficult environmental conditions and where higher altitude piloted surveys do not provide the required sensitivity or resolution for identifying survey targets [10]. Utilizing UAVs for surveys significantly reduces the need for labor and proves more cost-effective compared to traditional ground surveys or piloted aerial surveys.

Once located, orphaned wells can be assessed for possible leaks and gas emissions. Wells that were improperly plugged, abandoned without any plugging, or where the original plug has deteriorated frequently leak formation fluids into the groundwater and emit gasses into the atmosphere, most notably methane. Methane emissions from orphaned wells is of particular concern, as methane is approximately 28 times more effective at trapping heat in the atmosphere over a 100-year period compared to carbon dioxide [11], making it a significant driver of short-term climate change; its sources include natural processes such as wetlands and human activities such as livestock digestion, rice cultivation, and fossil fuel extraction, with agricultural and energy sectors being the primary contributors to the increase in atmospheric methane concentrations. With over 2 million historic wells designated as orphaned in the United States alone [12], they are collectively emerging as both a direct environmental concern via water contamination and an important contributor to global atmospheric methane emissions.

With renewed emphasis on orphaned well remediation, locating and assessing these sites has emerged as a difficult challenge in areas of historic hydrocarbon development where difficult terrain and dense vegetation render piloted aeromagnetic surveys ineffective and terrestrial visual or geophysical surveys operationally infeasible. Initially, automated UAV surveying targeting orphaned wells emerged as an application of ultra-low altitude aeromagnetic surveys targeting unexploded ordnance (UXO) elements featuring a similar diameter to that of well casing [13]. In a proof-of-concept trial, a conventional battery-powered drone integrated with a microfabricated atomic magnetometer (MFAM) sensor system collected magnetic datasets over a known vertical well in Broome County, NY, revealing that the string of well casing produces a single high intensity magnetic anomaly detectable from an altitude of over 40 m, clearing the regional tree line which extended to 32 m [14]. Following successful initial proof-of-concept trials [14], the methodology was further streamlined and tested in large-scale blind trials [15] that featured both battery-powered and hybrid gas–electric UAV systems. Following successful application and detailed description of the methodology, automated UAV-based aeromagnetic surveys rapidly emerged as seemingly the only viable technological solution to orphaned well location in challenging environments, providing both the required resolution and wide area coverage to effectively isolate magnetic anomalies associated with metal casing of orphaned wells.

Currently, the role of UAV-aeromagnetic surveys is largely limited to wellsite location; however, the results of our most recent research efforts suggest that high quality wide-area aeromagnetic datasets provide a reliable pathway to focus on high probability targets and exclude non-emitting uncased wells. We base our findings on a wide-area aeromagnetic survey effort performed near Olean, NY, in December 2019 [15] with a follow-up large-scale ground-truthing effort undertaken in 2021 and 2022. The ground-truthing effort had the goal of correlating detected magnetic field anomalies to surface expressions of orphaned wells and measuring emitted methane concentrations at these sites (Figure 1). In our ground truthing surveys, we largely corroborated the results of Kang et al. [16], which showed that the majority of wellsites emit little to no methane, while high emitters tend to emit methane at consistently high levels through time and that plugging status (unplugged, plugged, plugged/venting), well type (oil, gas, or oil and gas), and coal area designation are the attributes most associated with high emissions. While some of the described attributes are mostly not applicable in this research, as there is no coal mining in the immediate area and the wells are all oil wells, we noted that there is a well-defined correlation between wellsite casing integrity and its potential to provide a conduit to transmit methane from the original target formation to the surface. Critically, we were able to effectively describe the impact of wellsite casing on the magnetic signature detected, and then invert the process to effectively categorize well sites into priority bins.

2. Materials and Methods

This research is driven by a methodological approach that combines best practices in aeromagnetic UAV surveying, ground verification visual surveys, and terrestrial gas sampling surveys to initially constrain and subsequently determine the location of the wellsite, and finally analyze its emission activity. The calibration and deployment of UAV surveys should be done with special attention paid to survey consistency in terms of transect spacing and altitude above ground level (AGL), with UAV and ground structure safety in mind. Once aeromagnetic datasets are collected, processed, visualized in QGIS, and interpreted, a thorough ground-truthing effort can be performed to visually identify any surface expression of an orphaned well or confirm that no such expression exists. Finally, if a well stem or in-ground well location is identified, a gas concentration survey should be performed to quantify any emissions associated with the identified well. Key elements of these steps are outlined below.

In difficult terrain conditions, all UAV surveys must be automatically piloted relying on mission plans that detail the altitude of the UAV, the transect spacing of the survey, and take-off and landing areas, with take-offs and landing performed manually. A key first step for any UAV aeromagnetic project in an area featuring challenging terrain and tall vegetation, such as New York State (NYS), is to utilize LiDAR datasets available as .las/.laz files from the NYS GIS Clearinghouse [17], and process all the returns to produce a digital surface model (DSM) and just the ground returns for a bare earth digital elevation model (DEM), Figure 2a,b. A simple but quick method to assess the tree canopy elevation AGL is to subtract the DSM from the DEM to create what we call a digital obstacle model (DOM), Figure 2c. Importantly, stand-alone LiDAR and most other topographic datasets are of limited use unless put into a GIS program such as QGIS, ENVI, or ESRI’s ArcGIS Pro to model for site location or to locate features with digital signal enhancement filters, object-based image analysis (OBIA) algorithms, or convolutional neural networks (CNN). LiDAR is commonly also used for cultural heritage and environmental sciences and can be used in real time from UAVs [18,19,20,21,22]. A simple but powerful quantitative edge enhancement filter that highlights lateral spatial heterogeneity is the total horizontal derivative (THD) filter, commonly used in magnetics data processing [23], but it can also be co-opted for electromagnetic data [24], or any sensor data that are displayed as a raster. We suggest that relying on processed LiDAR derivative data elevation products should be considered a best practice in UAV mission planning for low-altitude aeromagnetic surveys where maintaining consistent altitude AGL is critical to dataset quality and subsequent interpretation.

The THD is simply the square root of the sum of the squares of the derivatives of the X and Y, here the rate of change of the elevation along the northing and easting. A THD is more robust than common qualitative techniques such as hillshades as it maintains data sensitivity, precision, and units (here m/m), whereas a hillshade takes native 64-bit data and down-samples it to unitless 8-bit 0–255 scale data. For instance, 64-bit depth data have a dynamic range of 385 dB, whereas 8-bit data have a dynamic range of just 48 dB. Hillshades are often used because of the convenience of a one-click solution to highlight contrasts in most commonly used GIS programs, but as noted above, they have several key disadvantages. This might seem like a moot point as 64-bit data have a much greater dynamic range than the human eye. Although this is true, the dynamic range is not greater than computer vision (like deep neural network algorithms), which is why it is an important consideration in data processing and analysis that will only become more significant in the years to come as data processing power and algorithms grow even more powerful. The X and Y directional derivative maps show the rate of change in elevation in the eastings and northings, respectively (Figure 2d,e), and highlight both areas of steep elevation, but also more subtle features such as historical roads associated with oil infrastructure. Notably, historic roads are readily apparent in the THD map of the area extremely close to the locations of located wellsites in Figure 3, which makes sense as the close proximity of these currently abandoned and overgrown roads would have been necessary to drill and service these wells in the early twentieth century.

Once the UAV missions are planned, relying on the accurate and robust LiDAR-derived digital obstacle model (DOM), UAV flights can be executed with a constant AGL altitude as close to the tree canopy as possible, while avoiding both terrain and ground structure collision. In this particular study, we relied on the use of a UMT Cicada gas electric hybrid hexacopter UAV (Figure 1A) platform, which flew three missions and covered approximately 267 acres in a little over three hours of flight time (Figure 1B). This hexacopter weighs 16.5 kg and has a maximum take-off weight of 19 kg. The UAV flew at 45 m above ground level (AGL) with a Geometric MFAM development kit tethered at a 4 m fixed offset from the UAV at 41 m AGL [15]. The 4 m tether distance for optimal signal-to-noise ratio had previously been determined during field trials over UXOs in Ukraine and wellsites in New York State [13,14] and clears the ~35 m tree canopy height (Figure 1). Universal Ground Control Station (UgCS) mission planning software was used to automate the waypoint guided missions and maintain constant AGL in challenging terrain conditions. The MFAM development kit contains two total field laser pumped cesium vapor magnetometers and can collect data at 1000 Hz at 5 pT/√Hz sensitivity. Data were collected in parallel transects at 20 m spacing to avoid spatially aliasing the well signal in the data; the accurate well location should be within one half of the between-line spacing, here <10 m. Data were processed with a fairly standard workflow described in de Smet et al. [15] and summarized here: data reconciliation between raw data files and field notes, parsing data, removing dropouts and spikes, base station corrections for diurnal variations, statistically line leveling the data to remove heading errors, calculating the International Geomagnetic Reference Field, removing local magnetic field direction with a reduction to the pole filter, kriging the data to create a raster, using a low-pass kernel convolution filter on the raster, and reprojecting the data from geographic NAD83 reference to UTM 17N projected coordinate system where well locations were predicted at peak amplitudes > 10 nT.

Traditional terrestrial surveys for orphaned wells are performed periodically due to the tremendous amount of time that is required to complete them. Typically, New York State Department of Environmental Conservation (DEC) performs orphaned well searches at specific times of the year, these being late winter/early spring and late fall/early winter. Because it is at these times that leaf cover and deciduous vegetation goes dormant, DEC inspectors are afforded the greatest opportunity to locate wells. However, despite this being the easiest time of the year to locate orphaned wells, it is by no means a guarantee that inspectors will be successful. For example, in the spring of 2019, prior to the aerial magnetic surveys described previously in this paper and de Smet et al. [15], two DEC inspectors surveyed the same 267-acres using traditional terrestrial survey methods. Historic records indicated that there could be hundreds of orphaned wells on this property; however, after two days of surveying, only 11 orphaned wells were successfully located.

Following identification of the wellsites from the UAV magnetics data, quantification of methane emissions becomes possible during site assessment. While it is certainly instructive to measure methane emissions over a 24 h or longer period, we believe the assessment can be thought of as binary—a well either emits some methane or it does not. Quantitative methane data were measured with the Sensit Portable Methane Detector (PMD) in concentrations in parts per million (ppm), % low explosive limit (LEL), and % volume (Figure 1 and Figure 3). The Sensit PMD instrument has a resolution of 1 ppm and 0.1% for LEL and volume and an accuracy of +/−10% for all three sensing modes. Before collecting data in the field, we turned on the sensor for a minimum of 30 min before calibrating the instrument using 0.1 and 2.5% methane/air mixes and a 100% methane end member. A static-flux chamber was designed and modeled after Kang et al. [16,22] to maintain an airtight seal over the wellheads and congruence between atmospheric and inner chamber pressure using components that degas methane, while allowing the Sensit PMD’s vacuum pump to intake gasses emitted from the wells. Although the methane concentration data are high quality, the GPS location data from the methane sensor’s internal GPS is likely off by several meters because of multipath errors from trees, low sensor quality, and lack of post-processing for a PPK solution.

3. Results and Discussion

It is often implicitly assumed or explicitly stated that UAV-based aeromagnetic surveys are a key first-look component of environmental remediation, integral to initially constraining areas to further search for orphaned and abandoned oil and gas wells to reduce methane emissions, but is this actually true? Several studies demonstrate that orphaned and abandoned oil and gas wells emit methane, although just a few wells are super emitters while most wells emit little to no methane [16,25]. This study largely corroborates these previous findings by showing that unplugged wells can and do emit methane and a minority are rather high emitters. Can UAV-magnetometry surveys be used to identify the location of orphaned and abandoned oil and gas wells so they can be better prioritized for remediation?

The initial wide-area UAV-based magnetic survey by de Smet et al. [15] in December 2019 located 72 potential wells in a 267-acre area in just 3 h of UAV flights, and in just one weekend of ground verification (early October 2021), 31 of these wells were confirmed and their methane concentrations quantitatively measured (Figure 4). The drone magnetometry survey was critical in locating the 31 previously unlocated (UL) wells in such a short period of time. Without the precise magnetics maps of high amplitude anomalies, we would have been searching the forest blind as if looking for needles in a haystack. Of these 31 previously UL wells, 25 were intact and still contained metal casing (in fact, two still had intact pumpjacks), while 6 were uncased and likely stripped for iron during WWII (Figure 5). The six uncased wells emitted on average 0.5 ppm methane and four emitted no methane. Of the 25 cased wells, only seven emitted >2 ppm while six emitted >10 ppm methane, with one well emitting LEL of methane (Figure 1). Kang et al. [16] showed that unplugged wells are more likely to be high emitters than plugged wells. Moreover, high emitters sustained high flow rates throughout their two-year study. This study indicates that we can further constrain likely high emitters to wells that are both unplugged and cased. The data from this study show that uncased—likely stripped—wells are unlikely to emit large quantities of methane.

4. Conclusions

Today, UAV-magnetometry surveys have emerged as a reliable wide-area survey method to efficiently locate orphaned and abandoned oil and gas wells. However, as we show in this work, initially constraining the location of these wells with aeromagnetic surveys should be perceived as only the start of a broader process of orphaned wellsite remediation. Once constrained in aeromagnetic surveys, wells should be targeted by both geophysical and visual pedestrian surveys to identify any remnant surface expression of the wellsite. At that point, the identified wellsite can be assessed with a methane detector to quantify emissions status and the well can be characterized in terms of remediation priority. We propose that the described sequence of precision low-altitude aeromagnetic surveying, followed by terrestrial geophysical and visual confirmation surveys, and culminating with direct geochemical sampling efforts should be seen as a standard workflow in areas where orphaned oil and gas wells present an impediment to economic development and an environmental concern. Importantly, we demonstrate that uncased unplugged wells emit little to no methane and can be initially identified in aeromagnetic datasets as weak magnetic anomalies and can be initially deprioritized for subsequent surveying and sampling efforts. Cased unplugged wells marked by high-intensity magnetic anomalies, on the other hand, have been shown to have the highest potential as probable sources of continued emissions and should be prioritized for remediation, thereby decreasing fugitive methane emissions into the atmosphere. This mitigation and remediation strategy should result in a time- and cost-efficient way to remove methane from the atmosphere compared to efforts relying solely on pedestrian surveys where every well is seen as having an equal emission risk. While our results to date are very encouraging, there are limitations associated with the overall sample size of surveyed sites as well as with the short time of gas sampling. Future research will focus on improving three aspects of methane quantification: (1) collecting longer term time-series data at high and moderate emitters, (2) collecting more quantitative data over more wells in the area, and (3) using the larger database of wells to determine if magnetic amplitude is correlated to and can be used to forecast methane emissions at an individual wellsite level. With continued improvements in available UAV and sensor technologies, we foresee further improvements to both accuracy and effectiveness of remote sensing approaches to the challenging problem of orphaned wellsite identification, classification, and ultimate remediation.

5. Patents

Nikulin, A., and De Smet, T. S. US Patent Application No. 17/321,154.