SYSTEM AND METHOD FOR UNMANNED AERIAL VEHICLE-BASED MAGNETIC SURVEY

Abstract

There are approximately 35,000 abandoned and unplugged oil and gas wells in New York with no known location. Unplugged wells emit methane, a strong greenhouse gas, which has the potential to significantly contribute to global climate change and act as a pollutant chemical. A long-range UAV equipped with methane sensors, MagPike (atomic magnetometer), and LiDAR sensors successfully detected unmarked well sites using characteristic magnetic signals generated by vertical metal piping preserved in the ground. The optimal flight altitude and transect spacing was determined for detection driven by the total field strength of the Earth's magnetic field and the height of tree canopies determined by LiDAR. Traditional methods of identifying oil and gas wells are costly and less powerful in acquisition of data such as using large magnetometers attached to helicopters.

Claims

1. A unmanned aerial system, comprising: a video camera, disposed on a Unmanned Aerial Vehicle (UAV) and configured to view an obstacle ahead of the UAV; an above-ground level sensor, configured to determine an above-ground level of the UAV during flight; and an automated control for the UAV, configured to control the UAV according to a flight plan at a predetermined above-ground level having an altitude of less than 50 meters, with a vertical deviation from the predetermined above-ground level dependent on at least the viewed obstruction ahead of the UAV and a predetermined Digital Obstacle Model (DOM) representing a computational model of obstacles for the UAV at the above-ground level prior to flight.

2. The system according to claim 1, further comprising a survey sensor, configured to generate survey data, and the UAV, wherein the UAV has a self-contained power supply to generate the survey data over at least 20 line kilometers in a continuous survey at a speed of 7 meters per second.

3. The system according to claim 2, wherein the survey sensor comprises a magnetometer.

4. The system according to claim 1, further comprising at least one of a Digital Surface Model (DSM) and a Digital Elevation Model (DEM), wherein the DOM is dependent on the at least one of the DSM and the DEM.

5. The system according to claim 1, further comprising a Global Navigation Satellite System (GNSS), and a survey sensor, wherein survey sensor readings are associated with GNSS-determine geolocation.

6. The system according to claim 1, further comprising an autonomous guidance system, responsive to the viewed obstacle ahead of the UAV, configured to perform an avoidance maneuver based on the viewed obstacle.

7. The system according to claim 1, wherein the UAV has a hybrid power source comprising an internal combustion engine and an electric generator, powering a plurality of electric motors providing lift.

8. The system according to claim 1, further comprising at least one magnetometer within a housing, suspended about 4 meters below the UAV during flight.

9. A method for surveying a region, comprising: providing an Unmanned Aerial Vehicle (UAV), comprising: a video camera, disposed on a UAV and configured to view an obstacle ahead of the UAV; an above-ground level sensor, configured to determine an above-ground level of the UAV during flight; a survey sensor, configured to sense a proximate environmental parameter; and an automated control, configured to control the UAV according to a flight plan at a predetermined above-ground level having an altitude of less than 50 meters, subject to a vertical deviation from the predetermined above-ground level dependent on at least the viewed obstacle ahead of the UAV and a predetermined Digital Obstacle Model (DOM) representing a computational model of obstacles for the UAV at the above-ground level prior to flight; flying the UAV according to the predetermined flight plan, at the predetermined above-ground level, while monitoring the video camera for the obstacle ahead of the UAV; and receiving geotagged data from the survey sensor concurrently with said flying.

10. The method according to claim 9, wherein: the survey sensor comprises at least one magnetometer contained within a housing, wherein the automated control computes whether the viewed obstacle will interfere with the housing; and the UAV has a self-contained power supply to generate the survey data over at least 20 line kilometers in a continuous survey at a speed of 7 meters per second.

11. The method according to claim 9, further comprising autonomously guiding the UAV, responsive to the viewed obstacle ahead of the UAV, and autonomously performing an avoidance maneuver based on the viewed obstacle.

12. The method according to claim 9, wherein the flight plan is dependent on at least one of a Digital Surface Model (DSM) and a Digital Elevation Model (DEM).

13. The method according to claim 9, wherein the UAV comprises a Global Navigation Satellite System (GNSS), and wherein survey sensor readings are associated with GNSS-determine geolocation and the UAV is guided according to the flight plan dependent on the GNSS.

14. The method according to claim 9, further comprising correcting the sensed proximate environmental parameter for a diurnal variation in magnetic field.

15. The method according to claim 9, wherein: the UAV comprises a hybrid electric-internal combustion power train, capable of providing sustained flight in excess of 60 minutes at a rate of 7 meters per second; the above-ground level is 40 meters or less; the above-ground level sensor comprises a LiDAR; the survey sensor comprises at least two separated magnetometers, having a sensitivity of at least 1 pT/Hz, configured to sense magnetic objects at a distance of between 30-50 meters; the flight path comprises a serpentine sequence of magnetic north-south traverses; the predetermined Digital Obstacle Model (DOM) is derived from a Digital Surface Model (DSM) and a Digital Elevation Model (DEM), and comprises at least one obstacle along the flight path taller than the above ground level; the obstacle comprises a tree; further comprising receiving magnetic sensor outputs from the at least two separated magnetometers synchronized with geographic location data obtained from a Global Navigation Satellite System (GNSS), compensated for diurnal ambient magnetic field variations.

16. The method according to claim 15, further comprising acquiring survey sensor at a resolution of less than 2.5 meters to detect at least one well having a magnetically permeable well casing.

17. The method according to claim 9, further comprising: removing dropouts in the related to at least one of sensor errors and polar dead zones; downsampling data from the survey sensor to about 1 Hz, and appending Global Navigation Satellite System (GNSS) geolocation data to a data record; diurnally correcting total field magnetic datasets comprising data records from the survey sensor with a magnetic base station; correcting heading errors with a statistical line leveling algorithm; determining a residual total magnetic intensity (TMI); converting TMI to a raster grid using kriging interpolation; low-pass filtering the raster grid using an unweighted moving average kernel convolution; removing an effect of a local geomagnetic-field direction with a reduction to the pole filter (RTP) to create a TMI RTP raster; creating a TMI RTP map to locate peak amplitudes; and plotting the peak amplitudes over a topographic map.

18. The method according to claim 17, wherein the topographic map shows at least one of locations of wells, locations of pipe, locations of unexploded ordnance, and locations of anthropogenic magnetic anomalies.

19. The method according to claim 9, wherein a set of proximate environmental parameters from a survey flight of the UAV over a parcel is processed according to a supervised trained algorithm to identify predetermined features corresponding to labelled features of a set of training data for the supervised trained algorithm.

20. A nontransitory computer readable medium for controlling a UAV to survey a region, comprising: instructions for controlling the UAV to fly according to a flight plan at a predetermined above-ground level having an altitude of less than 50 meters; instructions for determining an above-ground level of the UAV during flight; instructions for analyzing a video stream representing objects ahead of the UAV; and instructions for vertically deviating in real time from the predetermined above-ground level defined by the flight plan, dependent on at least an obstacle ahead of the UAV present in the video stream, and a predetermined Digital Obstacle Model (DOM) representing a computational model of obstacles for the UAV at the predetermined above-ground level prior to flight.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0153] FIG. 1 shows a georeferenced Pennzoil Lease Map, Olean, N.Y. Outline is three-quarter mile shapefile boundary of the survey parcel.

[0154] FIG. 2 shows a digital Surface Model (DSM) of survey area in Olean, N.Y.

[0155] FIG. 3 shows a Digital Elevation Model (DEM) of survey area in Olean, N.Y.

[0156] FIG. 4 shows a Digital Obstacle Model (DOM) of the survey area in Olean, N.Y.

[0157] FIG. 5 shows a Google Earth image of preprogrammed autonomous flight path terrain awareness.

[0158] FIG. 6 shows a mission plan with elevation waypoints using QGroundControl. Each 600×600 m survey consisted of between 400-500 waypoints due to the variable topography.

[0159] FIG. 7 shows Three 600×600 m missions flown on Dec. 7, 2019.

[0160] FIG. 8 shows raw uncorrected total magnetic field data. Note the ˜15 nT heading errors, and ˜10 nT between grid diurnal errors before base station correction. The northeastern portion of the dataset is over interpolated and also needs to be clipped before advanced processing.

[0161] FIG. 9 shows a diurnal variation in magnetic total field intensity recorded at 1/15 Hz with a Geometrics G-858 magnetometer base station. The total magnetic field intensity F changed by >15 nT throughout the survey. Raw data in red and data smoothed over a 3-minutes (13 point) moving average in blue.

[0162] FIG. 10A shows processed total magnetic intensity (TMI) data reduced to the pole (RTP).

[0163] FIG. 10B shows results of ground verification survey and previously GPS mapped wells by the New York State Department of Environmental Conservation.

[0164] FIG. 11 shows TMI RTP data with 5 nT contours.

[0165] FIG. 12 shows TMI RTP with 5 nT contours and putative well locations as crosses.

[0166] FIG. 13 shows noise structure and coherency.

[0167] FIG. 14 shows a georeferenced map with TMI RTP.

[0168] FIG. 15 shows a two-foot orthoimagery from 2012 over the survey area parcel shapefile outlined in black.

[0169] FIG. 16 shows magnetometry data collected at ground height, 10 m, 20 m, and 40 m AGL. Line spacing is 15 m for the aerial trials.

[0170] FIGS. 17A and 17B show magnetometry data collected at ground height; A is the magnetics data that the top sensor received, and B is the magnetics data that the bottom sensor received.

[0171] FIGS. 18A and 18B show magnetometry data collected at 10 m AGL.

[0172] FIGS. 19A and 19B show magnetometry data collected at 20 m AGL.

[0173] FIGS. 20A and 20B show magnetometry data collected at 40 m AGL.

[0174] FIG. 21A shows Theoretical Predictions.

[0175] FIG. 21B shows Observed Data.

[0176] FIG. 22A shows LiDAR data downloaded from FEMA for all three control sites. The vertical derivative of LiDAR data is shown on the left. Red areas show where land has been artificially flattened.

[0177] FIG. 22B shows a Google Earth image of the same region.

[0178] FIG. 23 shows results of a vertical magnetic sensitivity test for unexploded munitions.

[0179] FIGS. 24A-24D show various simulated UXOs (FIGS. 24A-24C) and a magnetic intensity map of inert simulated UXOs in a controlled field trial site (FIG. 24D) with letters on the map matching the position and type of the planted object.

[0180] FIGS. 25A and 25B show magnetic intensity maps over controlled site, with small black dots indicating GPS tracks of the conducted North-South UAV flights in 25A, and the dots absent in 25B to more clearly show the magnetic intensity. Black letters indicate the anomalies associated with simulated UXO objects.

[0181] FIGS. 26A-26C, shows total magnetic field data at blind test site, with two anomalies associated with UXO objects marked 26A and 26B in FIG. 26C. The concrete road can also clearly be seen to the south in the bottom of the magnetic map, likely because of a large amount of rebar used in its construction.

DETAILED DESCRIPTION OF THE INVENTION

Example 1 Detection of Well Heads

[0182] In the proof of concept study, magnetic datasets collected by a commercially-available battery-powered DJI-Matrice600 hexacopter UAV, equipped with a Geometrics Microfabricated Atomic Magnetometer (MFAM) sensor were employed, over a known Butkowsky 1-A abandoned well location near Binghamton, N.Y. The well was drilled to a total depth of 10150 ft in 2003 and plugged and abandoned in 2017; the well pad was subsequently leveled and at the time of the study area was overgrown with light vegetation that obscures any visible evidence of well presence. A series of terrestrial and UAV aeromagnetic surveys were conducted over the well site to record the magnetic expression of the well at the ground surface and at altitude, and calculated the rate of magnetic anomaly dissipation with altitude. Initially, using a standard terrestrial survey design and a, the magnetic anomaly associated with the well was observed, which was ˜18,000 nT at 0.15 m AGL. The UAV equipped with the MFAM sensor was positioned at the center of the magnetic anomaly and the assembly elevated at a rate of 1 m/s to an altitude of 100 m AGL. The magnetic anomaly at 40 m AGL, which correlated to an elevation slightly above the treeline, was ˜400 nT, nearly double the background magnetic field levels at that elevation and the signal dissipated to background levels at ˜50 m. Thus, at ˜40 m AGL, magnetic anomalies associated with vertical wells featuring metal casing remain pronounced above background levels, allowing their identification in wide-area UAV aeromagnetic surveys.

[0183] The area chosen for follow-up study was Chattarugas County in Western New York, where wide-spread hydrocarbon exploration and production activity occurred in the late nineteenth and early twentieth century. FIG. 1 shows a georeferenced Pennzoil Lease Map, Olean, N.Y. The outline is a three-quarter mile shapefile boundary of the survey parcel. FIG. 14 shows a magnetic map of the region. FIG. 15 shows a two-foot orthoimagery from 2012 over the survey area parcel shapefile outlined. FIG. 16 shows the path for obtaining magnetometry data collected at ground height, 10 m, 20 m, and 40 m. Line spacing is 15 m for the aerial trials. FIGS. 17A and 17B show magnetometry data collected at ground height. A is the magnetics data that the top sensor received and B is the magnetics data that the bottom sensor received. FIGS. 18A and 18B show magnetometry data collected at 10 m above ground level (AGL). FIGS. 19A and 19B show magnetometry data collected at 20 m above ground level (AGL). FIGS. 20A and 20B show magnetometry data collected at 40 m above ground level (AGL).

[0184] FIG. 21A shows theoretical predictions and FIG. 21B shows observed data. FIG. 22A shows LiDAR data downloaded from FEMA for all three control sites; the vertical derivative of LiDAR data is shown on the left, and red areas show where land has been artificially flattened. FIG. 22B shows a Google Earth of the same region.

[0185] New York State (NYS) has high-resolution 1 m LiDAR data coverage throughout most of the state and the survey area in Cattaraugus County was collected in 2017 by the Federal Emergency Management Agency (FEMA). NYS LiDAR data is available as .las files, which were used to generate post-processed derivative data products such as digital elevation models (DEM, en.wikipedia.org/wiki/Digital_elevation_model) and digital surface models (DSM). A DSM is a 3D digital model of the first returns from LiDAR, which includes all natural and anthropogenic objects, like trees and buildings (FIG. 2, Digital Surface Model (DSM) of survey area in Olean, N.Y., www.gisresources.com/confused-dem-dtm-dsm/, Zhou, Qiming. “Digital elevation model and digital surface model.” International Encyclopedia of Geography: People, the Earth, Environment and Technology: People, the Earth, Environment and Technology (2016): 1-17). A DEM encompasses the subsequent returns as well at the earth's surface where vegetation and anthropogenic structures have been removed to produce what is often called a ‘bare earth’ model (FIG. 3, Digital Elevation Model (DEM) of survey area in Olean, N.Y.). Subtracting the DSM from the DEM produces a critical derivative data product, a digital obstacle model (DOM). The DOM is essential to avoid terrain obstructions while planning low-altitude missions (FIG. 4, Digital Obstacle Model (DOM) of the survey area in Olean, N.Y. and FIG. 5, Google Earth image of preprogrammed autonomous flight path terrain awareness), while the DEM is necessary to maintain constant altitude AGL. Maintaining consistent altitude AGL is of the utmost importance as the earth's total magnetic field rapidly decays at 1/r.sup.3 and shifts in altitude AGL will result in poor data quality and increase the difficulty in post-processing.

[0186] One of the limiting factors of the Nikulin and de Smet (2019) was the relatively short total flight time allowed by the battery-powered UAS used in the initial aeromagnetic surveys over the Butkowsky 1-A well site. In follow-up efforts to adapt the developed methodology to wide-area surveys UAS-based magnetic data were collected by a UMT Cicada gas-electric hybrid hexacopter platform (umt.aero/cicada/) equipped with the MFAM sensor. This UAS weighs 16.5 kg (36.38 lb) and has a maximum takeoff weight of 19 kg (41.89 lb). The MFAM development kit including the global navigation satellite system (GNSS) receiver was housed and protected in a non-magnetic, light, and durable UMT MagPike enclosure case, manufactured for this purpose. The MFAM development kit consists of two total field magnetometers that can collect data at a sample rate of 1000 Hz and a sensitivity of 1 pT/Hz. The MFAM was tethered to the UAS with thin, strong, and flexible polypropylene rope braided poly cord at a 4 m fixed offset. The optimal tether distance was previously determined to maintain the highest signal-to-noise (SN) ratio for UAS-based magnetic data acquisition (Nikulin and de Smet 2019).

[0187] Using the high-endurance hybrid UAS platform allowed us to plan wide-area missions that covered ˜100 acres in a single 1 hr UAS flight. In fact, this metric could be further expanded in terms of flight time and aerial coverage and remains constrained by line-of-sight rules imposed on small UAS operators by the US Federal Aviation Administration (FAA).

[0188] QGroundControl mission planning software was used to preprogram GNSS-guided autonomous missions where waypoint navigation allowed the UAS and magnetic sensors to maintain constant altitude AGL (FIG. 5). This was critically important as elevation changes in the survey area approached ˜200 m (FIGS. 2 and 3). There were ˜400-500 waypoints per survey mission depending upon terrain changes specific to the survey (FIG. 6, Mission plan with elevation waypoints using QGroundControl. Each 600×600 m survey consisted of between 400-500 waypoints due to the variable topography). Data were collected at an altitude of 45 m AGL in north-to-south and south-to-north transects spaced 20 m apart. Three missions were flown at a speed of 7 m/s in three 600×600 m grids and over 60 line-kilometers of magnetic data was collected during these surveys (FIG. 7, Three 600×600 m missions). A fixed forward looking visible-light camera was used to monitor potential obstacles along the flight path in addition to visual control of the UAS by the pilot-in-command.

[0189] While the raw aeromagnetic datasets revealed some of the larger anthropogenic anomalies, there were considerable errors and artifacts introduced to the datasets as a result of sensor motion and diurnal magnetic field variations (FIG. 8, Raw uncorrected total magnetic field data. Note the ˜15 nT heading errors, and ˜10 nT between grid diurnal errors before base station correction. The northeastern portion of the dataset is over interpolated and also needs to be clipped before advanced processing). The UAV-based magnetics data underwent a standardized processing routine to highlight anomalies associated with well casings and dim other anthropogenic magnetic anomalies. Initially, raw data files were parsed by downsampling from 1000 Hz to 1 Hz and attaching the GNSS data from the NMEA $GPGGA sentence, which contains time, latitude northing, longitude westing, and altitude information. Next, anomalous dropouts related to sensor errors, or polar dead zones, were removed. The MFAM total field data was diurnally corrected with the aid of a Geometrics G-858 that was used as a base station at 1/15 Hz (FIG. 9, Diurnal variation in magnetic total field intensity recorded at 1/15 Hz with a Geometrics G-858 magnetometer base station. The total magnetic field intensity F changed by >15 nT throughout the survey. Raw data in red and data smoothed over a 3-minutes (13 point) moving average in blue.). Heading errors of up to 15 nT in the raw total magnetic field data can be seen as large stripes and were corrected with a line leveling algorithm (FIG. 10A, processed total magnetic intensity (TMI) data reduced to the pole (RTP)). The 12th generation International Geomagnetic Reference Field (IGRF) regional total magnetic field values were calculated for the date, location, and elevation of the surveys (Thébault et al. 2015), in order to determine the residual total magnetic intensity (TMI). These point data were then converted to a raster grid of 5 m pixels using kriging interpolation. The raster grid was low-pass filtered using a 3×3 unweighted moving average kernel convolution. The effect of the local geomagnetic-field direction at the survey location was removed with a reduction to the pole filter (RTP) to create a TMI RTP raster (FIG. 11, TMI RTP data with 5 nT contours). A final TMI RTP map of 5 nT contours was created to locate the peak amplitudes of potential wells.

[0190] FIG. 10B shows a map showing results of ground verification survey and previously GPS mapped wells by the New York State Department of Environmental Conservation.

[0191] Nikulin and de Smet (2019) compared ground and UAV-based magnetic surveys to determine optimal flight parameters like altitude, speed, and line spacing over a single well. In this follow-up study we successfully conducted a wide-area UAS magnetic survey to detect and map orphaned and abandoned oil and gas wells in Cattaraugus County, New York. 1.08 km.sup.2 (267 acres) of magnetic data were collected in three missions in less than four hours and located 28 previously undocumented well locations (FIG. 12, TMI RTP with 5 nT contours and putative well locations as crosses). Each individual mission covered 0.36 km.sup.2 and lasted 69-82 minutes (Table 1).

TABLE-US-00001 TABLE 1 Total and within transect flight duration. Total within survey flight time 3:22:40 and total flight time 3:49:18 for three flights. All flight times in UTC time. Start End Flight Takeoff transects transects Landing Within survey Total flight 1 17:05:18 17:11:19 18:19:43 18:23:25 1:08:24 1:18:07 2 18:54:22 18:57:29 19:58:09 20:03:32 1:00:40 1:09:10 3 20:17:35 20:19:16 21:32:52 21:39:36 1:13:36 1:22:01

[0192] It should be noted that each survey flight used approximately 2 liters of fuel, or approximately ½ of the fuel tank of the UMT Cicada UAS. Therefore, flights could easily cover a much greater area and we estimate that if we excluded the take-off, approach, and landing sequences and increased flight speeds speed, the entire square kilometer parcel could be covered in a single 2-2.5 hour aerial survey, though FAA Part 107 visual line-of-sight rules are limiting. An approved FAA 107.31 visual line of sight aircraft operator waiver would theoretically allow longer missions.

[0193] A terrestrial survey (ground level) conducted over the area with dimensions and spatial resolution, would take at least 15.66 hours, or approximately two 8 hour work days to conduct, assuming a walking speed of 1 m/s, or 3.6 km/h. While this rate of magnetic data collection is considered standard, it is derived from relatively small surveys and its application to a wide-areas survey is inaccurate, given the terrain, dense vegetation, and the size and weight of a conventional terrestrial magnetometer systems. Furthermore, the 15.66 hour survey time estimate is based on an assumption of constant data acquisition with no breaks for the operator or battery replacement times factored into the calculation. Perhaps a more reasonable maximum estimate of 10-15 km/day for a single operator is in order. Consequently, a terrestrial survey of the size and scope collected in Olean, N.Y. would take at 4-6 working days and could only be conducted in the seasonal transitions between Fall-Winter and Winter-Spring when brush undergrowth is slightly less dense and the air temperature allows for longer working days. During other times of the year, environmental conditions, dense vegetation, or thick snow cover in this region make wide-area terrestrial magnetic surveys time- and cost-prohibitive.

[0194] Hybrid UAS magnetic surveys are more operationally efficient than terrestrial or piloted aeromagnetic surveys to detect and map orphaned and abandoned oil and gas wells. Operation of the UAS systems introduces some magnetic noise, however, a 3-4 m separation between the UAV and the magnetometers is found to be sufficient to suppress the effect of the noise on the desired signal. UAS aeromagnetic surveys can be conducted over hazardous environments and difficult terrain that may otherwise be effectively inaccessible for terrestrial or piloted surveys at low altitudes.

[0195] A supervised trained algorithm may be used to improve the data analysis, to identify wells and lower false alarms. Such an algorithm is trained according to known well data which may be before acquisition, or based on confirmation after a survey. This may account for different types of wells, their age, geologic formations, anthropogenic artifacts other than well casings, and the like.

Example 2 Detection of Unexploded Ordnance

[0196] To ensure data acquisition at the high signal-to-noise ratio needed for effective isolation of relatively weak magnetic signals at elevation, it is necessary to establish the sensitivity of the MFAM unit to the magnetic interference fields generated by the operation of the gas-powered engine and powerful electric rotors of the Cicada-M UAV hybrid gas-electric system were measured. The minimal vertical separation of the MFAM unit, encased in a protective UMT MagPike ballistic foam case, from the UAV base was measured. The sensors of the MFAM were oriented in opposing direction, recording magnetic field gradient to allow for faster in-field processing of the datasets. Depending on the throttle of the engine, magnetic interference from the engine dissipated beyond the limit of detection at vertical separation of 1.7-2.4 meters, with 1.7 m correlating to the lowest engine RPM and 2.4 m correlating to the highest RPM allowed by the engine. Various tether lengths and flight elevations were tested to determine the optimal survey configuration of sensor, UAV, engine, and rotors. A vertical separation of 4 m, suspending the MFAM acquisition system using three soft non-magnetic cords attached to the propeller beams of the UMT Cicada-M were thereafter employed, and no detectable magnetic interference correlating to engine or rotor operations (regardless of throttle levels) were evident. Thus, a properly configured hybrid Cicada-M UAV platform is suitable for accurate magnetic surveying.

[0197] In a second experiment, the vertical sensitivity of the UAV-mounted MFAM unit to a magnetic anomaly generated by a single BM-21 UXO round was determined, and the optimal surveying parameters to target this type of UXO defined. FIGS. 24A-24D show various simulated UXOs (FIGS. 24A-24C) and a magnetic intensity map of inert simulated UXOs in a controlled field trial site (FIG. 24D) with letters on the map matching the position and type of the planted object. A 3 m concrete-filled pipe, similar in mass, length and metal content to a 122 mm composite BM-21 round, commonly consisting of the projectile and a metal fragmentation coil wrapped onto the inner wall of the round (FIG. 24A) The simulated UXO was oriented East-West on a concrete pad away from other similarly-sized metallic objects, as shown in FIG. 24D. The UAV system carrying the attached MFAM unit was manually guided to the center of the simulated UXO and proceeded to rise vertically at a rate of 1 m/s, to a total altitude of 34 meters above ground. Results of the second experiment are presented in FIG. 23, where the horizontal axis represents altitude and the vertical axis records total magnetic intensity. As anticipated, the magnetic anomaly was most pronounced directly above the simulated UXO and its intensity dissipated at a geometric progression, until fully disappearing at an altitude of ˜20 m above ground.

[0198] Following the vertical sensitivity test, a number of additional UAV flights were conducted over a controlled area, seeking to determine optimal survey parameters that allow to highlight the presence of the simulated BM-21 UXO, while filtering out magnetic noise from metallic debris. Experimentally, it was determined that at a sensor altitude of 3 m AGL (flight altitude of 7 m AGL), the optimal balance between ability to resolve a simulated BM-21 UXO and limiting false flags from metallic clutter smaller than the test object were achieved. Similarly, a survey spacing of 3 m permitted consistent identification of magnetic anomalies associated with the targeted UXO type. Finally, an optimal traverse speed of 3 m/s was determined, allowing both provide high sampling density (MFAM samples at 1000 samples/second, while the GPS time stamp is placed only every second) and aerodynamic stabilization of the MagPike platform in flight. In sum, a 3-3-3 formula for UAV-based aeromagnetic acquisition was defined: 3-meter sensor elevation, 3-meter traverse spacing and 3 m/s acquisition speed. The speed of acquisition can be further increased up to 10 m/s, if the GPS time stamp frequency is increased to match that of MFAM sampling density.

[0199] Following controlled site testing and optimization of UAV survey parameters in a relatively small contained area, a series of field trials were conducted to determine if automated UAV aeromagnetic surveys over larger areas would allow successful detection of inert MBRL UXOs. Initial controlled field trials on the grounds of Chernihiv Airfield installation in Northern Ukraine, where a series of inert training UXOs were placed at ˜25 m intervals along a linear East-West transect were conducted. Three types of inert munitions were tested—a) 122 mm BM-21 MBRL round, b) SA 22 type anti-aircraft rocket, c) solid metal core 152 mm artillery shell. From this configuration, it was sought to be determined if the system could not only effectively detect, but also discriminate MBRL UXOs from similarly sized and shaped non-magnetic UXOs and a highly-magnetic, but significantly smaller 152 mm artillery rounds.

[0200] UAV surveys were conducted North-South, perpendicular to the transect in the 3-3-3 configuration, as identified in the controlled trials. Raw magnetic data were parsed and de-striped, with correct GPS time markers (FIG. 25A). A simple line leveling technique was then applied to each of the flight lines for every individual flight. This removed the directional interference in the data. Subsequently, the regional total magnetic field for the controlled site was removed. These values were calculated using the International Geomagnetic Reference Field (IGRF) Model. The residual total magnetic intensity was plotted using the Kriging Interpolation to turn the values into rasters and create an image of the data. From there a low-pass convolution filter was applied to remove image background noise and the raster color scale and inversion was adjusted to most effectively intensify the magnetic variation in the data, thus clarifying anomalies which represent UXOs. It was then easier to observe magnetic anomalies with a higher confidence that detected UXOs were not false positive errors.

[0201] FIG. 25A thus shows magnetic intensity maps over controlled site, with small black dots indicating GPS tracks of the conducted North-South UAV flights in A. Black letters indicate the anomalies associated with simulated UXO objects.

[0202] In the processed and mapped dataset, there were three well-defined dipole anomalies—labeled A, B, and C in FIG. 25B. Anomaly 24A was associated with the inert BM-21 round, anomaly 24B was associated with the SA-22 inert missile, and anomaly 245C was associated with the inert metal-core 152 artillery round. What is immediately obvious is that the largely metal BM-21 (FIG. 24A) round has a considerably more intense magnetic signature than a similar-sized and shaped, but largely non-magnetic SA-22 inert round (FIG. 24B). Conversely, the metal-core, but much smaller 152 mm artillery projectile (FIG. 24C) generated a magnetic intensity anomaly of comparable size to the BM-21 round. This result was instructive, as it allowed calibration of the processing filters to specifically identify the metal content and size of the BM-21. In the presence of massive magnetic UXOs of smaller caliber, there may be false flag alarms associated with the high magnetic intensity fields generated by such objects. The inert 152 mm round was a solid metal core training projectile without an explosive chamber, which represents a very atypical object to be encountered in the field.

[0203] Following successful detection of the inert BM-21 round in controlled trials, survey platform, as well as processing and detection algorithms, were tested at a live test site, the Ukrainian Armed Services Honcharivs'ke proving grounds. An area where two MBRL rounds failed to explode on impact was identified: a 220 mm BM-27 Uragan [“Hurricane”] projectile and a 122 mm BM-21 Grad [“Hail”] projectile. This area allowed safe access via a reinforced concrete road, which in turn served as a staging and take-off area for the UAV.

[0204] A survey of the live site was conducted using the same 3-3-3 parameters as defined in the controlled experiments and the dataset was analyzed relying on the same processing and mapping protocols. The results from the blind test surveys of live unexploded 122 mm rockets are presented in FIGS. 26A-26C. In the magnetic dataset, two large magnetic anomalies (labeled as 26A and 26B in FIG. 26C) were observed. On initial assessment, informed by results of the controlled trials, the weaker anomaly 26A was hypothesized to be associated with the larger-caliber BM-27 round shown in FIG. 26A, which despite its larger size contains more non-magnetic aluminum in its design. Consequently, the larger anomaly 26B was hypothesized to be associated with the BM-21 UXO shown in FIG. 26B. The assessment and the geographical coordinates of the anomalies was relayed to the operators of the testing site, who conducted a visual survey in indicated areas. Anomaly 26A was immediately identified as a BM-27 round, which was submerged at about a relatively low angle of impact (FIG. 26A). As for anomaly 26B, the assessment took significantly longer, as there were no visible signs of the UXO at the surface; however, upon closer inspection, the search revealed the tail section of a 95% submerged BM-21 round, as seen in FIG. 26B. Critically, the BM-21 tail section was below ground level and would have been impossible to identify in a wide-area visual survey conducted without constraining the search area.

[0205] Using the defined parameters for the survey equipment and derived survey air speed and altitude parameters, 27,000 line meters can be surveyed during a single 150 min flight of the hybrid UAV, allowing surveying a 600×600 area in approximately 11 hours. If, the sampling rate of the GPS unit is raised to match the sampling frequency of the magnetometer, the survey can be conducted at 10 m/s air speed and the 600×600 area could be surveyed in 3.5 hours with the same level of accuracy and resulting interpretation confidence.

[0206] Application of this methodology may be limited in the presence of tall vegetation and may be influenced by site conditions, including host soil geology, presence of metallic debris, infrastructure and topography, all of which can impact the depth and angle of UXO burial, as well as their successful identification. Perhaps most importantly, impacted areas with vegetation over 10 m tall will force sensor elevation outside the effective range of the survey system. Large near-surface natural and anthropogenic magnetic anomalies may cause constructive or destructive interference within the magnetic datasets, complicating initial analysis.

[0207] The embodiments illustrated herein are described in sufficient detail to enable those skilled in the art to practice the teachings disclosed. Other embodiments may be used and derived therefrom, such that structural and logical substitutions and changes may be made without departing from the scope of this disclosure. The Detailed Description, therefore, is not to be taken in a limiting sense, and the scope of various embodiments is defined only by the appended claims, along with the full range of equivalents to which such claims are entitled.