0

Standardized protocol for UAV data acquisition

TRuStEE - Training on Remote Sensing for

Ecosystem modelling.

Deliverable 1.5

WP1

Milano, 31st August 2018

1

TRuStEE - Training on Remote Sensing for Ecosystem modelling

Deliverable 1.5

WP1

Project title: Training on Remote Sensing for Ecosystem modelling

Project Acronym: TRuStEE

Grant Agreement number: 721995

Topic: MSCA-ITN-2016 Innovative Training Networks

Start date of the project: 01/10/2016

Duration of the project: 48

Dissemination level: Public

Responsible of the deliverable: Stephanie Delalieux

Phone: +32 14 33 67 16

Email: stephanie.delalieux@vito.be

Date of submission: 31st August 2018

2

Table of Contents

Summary .............................................................................................................................. 4

Aims ....................................................................................................................................... 4

1. General .................................................................................................................... 5

1.1 Legal framework........................................................................................................................... 5

1.2 Ethics ............................................................................................................................................ 6

2. Flight planning .............................................................................................................. 7

2.1 Task definition .............................................................................................................................. 7

2.2 Site assessment ............................................................................................................................ 7

2.3 Risk assessment and management .............................................................................................. 7

2.4 Planning of the flight and of the required equipment ................................................................. 8

3. Flight preparation ........................................................................................................ 9

3.1 Weather forecast ......................................................................................................................... 9

3.2 Site permissions and notifications ............................................................................................... 9

3.3 Communications .......................................................................................................................... 9

3.4 Preparation of equipment and clothing ....................................................................................... 9

3.5 Preparation of the crew ............................................................................................................. 10

4. Flight procedure ........................................................................................................ 11

4.1 In situ site assessment ............................................................................................................... 11

4.2 In situ weather assessment ........................................................................................................ 11

4.3 Pre-flight controls....................................................................................................................... 11

4.4 Flight ........................................................................................................................................... 11

4.5 Post-flight controls ..................................................................................................................... 12

5. Ancillary measurements ......................................................................................... 13

5.1 STS measurements ..................................................................................................................... 13

5.2 Spectroradiometric measurements ........................................................................................... 16

5.3 Ground control points ................................................................................................................ 17

5.4 Generic toolbox and other materials ......................................................................................... 18

6. Post-flight tasks.......................................................................................................... 19

6.1 Summary of performed measurements .................................................................................... 19

6.2 Field work report ........................................................................................................................ 19

6.3 Data and product quality control and validation ....................................................................... 19

3

References ........................................................................................................................ 20

4

Summary This protocol is the product of the multiple years of experience in drone data acquisition that the authors

shared and compared to create a unique, complete protocol.

Beside the framework on flight regulation and on ethical issues, the protocol addresses the practical

matters that researchers working with drone have to solve: when to fly? what safety procedures to follow?

what ancillary measurements to take? This protocol guides the reader through the essential steps of the

planning, preparation and realization of a drone flight. In section 2, the criteria employed in the phase of

flight planning are explained, followed, in section 3, by the practical steps needed to be able to perform the

flight: checking the weather forecast, preparing the crew as well as any other person involved in the

operation. Section 4 describes how to carry out the main steps of a flight, from take-off to landing. A

separate section is dedicated to the ancillary measurements that are often collected to correct,

geometrically and spectrally, the acquired drone data. The final paragraph illustrates what post-flight tasks

should be implemented to make sure that significant information is taken note of, in such a way that it can

improve the interpretation of the acquired data and improve the planning of future flights.

Aims The main objective of this deliverable is to make a list of all the essential steps to take in order to perform a

drone flight and to make recommendations on how to execute them.

5

1. General

1.1 Legal framework The regulation of Unmanned Aerial Vehicles (UAV) has been an ever evolving field with constantly ongoing development in recent years to keep pace with technology and the increasing popularity of UAVs with a variety of consumers. Currently, the regulation of UAV operations within Europe is still dependent on weight categories. According to regulation No 216/2008, platforms with maximum take-off mass (MTOM) of 150 kg and above are regulated by the European Aviation Safety Agency (EASA), whereas lighter platforms are to be regulated by the EU member states. This has led to a fragmentation within UAV regulation across Europe for the vast majority of UAV operations (EASA, 2018). Regulations are developed by national bodies such as the Civil Aviation Authority (CAA) in the United Kingdom (UK) for example. While similar, regulations at national level have proved to vary in all compared variables (Stöcker et al., 2017). This limits cross-border collaboration within the EU as well as posing difficulties regarding manufacturing requirements for UAVs. The EU, as well as UAV operators and manufacturers within would greatly benefit from overarching regulations, especially for operating internationally. Currently, a new basic regulation has been proposed (see NPA 2017-05) and is under discussion between the European commission, council and parliament. Besides safe operations of UAVs through competency requirements and standardised risk assessment procedures for specific operations, this regulation should also address privacy and data protection. Member states will reserve the rights to designate no-fly zones and zones with increased restrictions for UAV operations. Alleviations may also be made by designating specific zones for operations and enabling model clubs for recreational operations. The large majority of scientific UAV operations will fall within the <150 kg weight category and therefore, as stated above, their regulation currently differs slightly throughout EU member states. Instead of listing these here we instead provide the UK specific regulations as an example, which has inspired and in turn been inspired by the regulations adopted by other member states. The CAA regulations represent an amalgamation of international, EU and domestic legislation. Key legislation for all aviation is laid down in the Civil Aviation Publication (CAP) 393 “Air Navigation Order and Regulations” . Unmanned aircraft are divided into categories, differentiating between Small Unmanned Aircraft (SUA) with MTOM <20 kg, Light Unmanned Aircraft with MTOM >20kg and <150kg and Unmanned Aircraft with MTOM >150kg (and thus regulated by EASA). It contains only few articles relating to SUAs (article 94) and small unmanned surveillance aircraft (SUSA) (article 95) which includes any aircraft capable of data acquisition and is therefore of relevance to any scientific operation. Article 94 deals with basic safety requirements. It specifies the following:

1. No object may be dropped from an SUA so as to endanger any person or property. 2. The operator of the SUA may only fly if he is reasonably satisfied that the flight can safely be made. 3. Direct, unaided visual contact with the aircraft must be ensured at all times in order to avoid

collisions. 4. If the aircraft weighs more than 7 kg excluding fuel, it must not be flown:

a. In class A, C, D or E airspace, unless air traffic control permission has been acquired. b. Within an aerodrome traffic zone during its period of operation, unless permission has

been acquired. c. At a height of more than 400 feet (~120 m) above the surface. d. For the purpose of commercial operations, unless appropriate certification has been

acquired. Article 95 details further regulations for SUSA, specifically distance constraints. Without permissions, SUSA must not be flown:

a. Within 150 m of a congested area. b. Within 150 m of an assembly of over 1000 people. c. Within 50 m of any vessel, structure or person not under the control of the operator. d. Within 30 m of any person not under the control of the operator during take-off and landing.

6

To supplement CAP 393, the CAA published the CAP 722 “Unmanned Aircraft System Operations in UK Airspace – Guidance”, currently in its 6th edition. It highlights safety requirements for airworthiness and operation standards where CAP 393 is not applicable to UAVs. Further details can be found in the respective documents. Besides the above stated minimum safety distances, there are a variety of permanent or temporary restricted areas in which UAV operation is limited or forbidden outright. It is the operator’s responsibility to inform themselves about no-fly zones, Notifications to Airmen (NOTAMs) and other warnings prior to conducting a flight, using the air navigation charts and online resources available. Any notable occurrence during the flight, such as technical failure, near-miss or collision are subject to the CAP 382 “Mandatory Occurrence Reporting Scheme” and an appropriate report should be filed to aid future preventative measures. In the specific case of near-misses or collisions, an Air Proximity Report must be submitted to the UK Airprox Board. For the purpose of acquiring data from UAVs in the UK it can be noted that as of now no additional certification is required, providing it does not fulfil the criteria of commercial operation. This however strongly varies by country. The legal framework for UAV operation is still in flux and expected to change with new applications of UAVs emerging across different sectors. Keeping up to date with the legislation is central to guarantee any future UAV based activities can be carried out safely and without difficulty.

1.2 Ethics

Ethical issues in regards to UAV operations can emerge as a result of the interaction of how the system is used by a user and those who experience this use (Wilson, 2014). The majority of ethical discussions in regards to UAV use have focused on military applications and thus primarily deal with issues which are not directly applicable to civil operations. For the civilian use of UAVs capable of surveillance, the issue of privacy is the primary concern (Finn and Wright, 2012). The recent General Data Protection Regulation (GDPR) (EU, 2016) and its national implementations provide legislation on how data containing sensitive personal information has to be handled. While this prevents the use and publication of sensitive information, it does not prevent its acquisition. Furthermore, as long as distance constraints as were detailed in the previous section are correctly observed, there are generally no explicit permissions required to operate and acquire data over private property. However, this may not prevent conflict which may arise due to perceived breach of privacy. This is especially likely as few people are informed sufficiently on the legislation relating to UAV use and data acquisition or do not trust the data to be handled adequately. From an ethical perspective and also in the UAV operators best interest to ensure no incidents, any persons whose property is flown over or themselves or their property is imaged should be informed and written permission acquired wherever possible. When using UAVs for environmental research, the acquisition of sensitive personal information can be expected to be minimal. Nevertheless, the requirements for ethical use of UAV systems and measures for avoiding conflict will extend beyond observing regulations, of which the researcher must be aware. Another aspect of particular relevance to environmental research is the disturbance of wildlife. A code for best practice of UAV operations in the vicinity of animals has been outlined by Hodgson and Koh, (2016). Due to a current lack of empirical evidence relating to wildlife reaction to UAVs, they advocate for caution and seeking expert advice on the affected taxa. Where disturbance is likely, an appropriate UAV system and sensor should be selected to minimise audio and visual stimulus while satisfying the study objectives. Launch and recovery sites should be carefully chosen to maintain a reasonable distance to animals at all times. Furthermore, animal response should be assessed in the field and any operations ceased if they prove to be disruptive. Observed responses should also be published in the study along with platform details to benefit further research.

7

2. Flight planning

2.1 Task definition The general objective of the research and thus of the flight is to be defined, in order to determine the best parameters in terms of: · spectral resolution · spatial resolution; · time of the flight during the day: this parameter is essential, for instance, in case the objects to be

surveyed are tall and cast shadows over each other (e.g. trees, tall buildings), or when flying over water (Duffy et al., 2018);

· platform to fly: fixed wing aircrafts can usually survey larger portions of land in comparison to multi-rotors (Duffy et al., 2018);

· other flight parameters: for instance, for many agricultural applications images are acquired at nadir, while for photogrammetric surveys different acquisition angles are required.

2.2 Site assessment The area to be surveyed should be identified with its geographical coordinates. In the first place, restrictions over the surveyed area should be checked: · Permanent restrictions: these regards, for instance, military areas. · Temporary restrictions: they are usually defined in case of a hazardous event (e.g. a wildfire) or in case

of a security-related event (e.g. a G8 meeting). · Restricted airspaces: these are areas including sensitive locations, for instance a nuclear or military

facility. Special permissions are needed to fly over these airspaces. · Other restrictions: these are country-dependent and include, for instance, natural and marine areas or

bird conservation areas. Normally, each country makes a dataset of airspace restrictions available to the public or at least to the authorized pilots. In the second place, geographical features that might affect the flight should be identified: for instance, electricity towers, wires, trees, tall buildings. These may affect the flight planning. Thirdly, the size and morphology of the surveyed area must be evaluated: this will help in determining the best flight lines, the flight time and therefore the number of needed batteries, the necessity of a co-pilot to keep the drone in sight. In the case of a fixed-wing platform that needs a big space in order to take off, it is advisable to check that such a space is available, within or next to the surveyed area. Finally, other site characteristics should be kept in mind, including: the possibility of an event involving a big number of people in or nearby the surveyed area, ongoing at the same time of the flight; the possibility of having public approaching the pilot and co-pilot, to be controlled with a cordon. The site assessment should be done with a preliminary visit on the field and/or with the aid of other virtual instruments (e.g. USGS Earth Explorer, Google Earth).

2.3 Risk assessment and management The procedure of risk assessment helps in identifying the possible hazards involved in the drone flight and, consequently, in managing them. Consequences of ill-managed risks may can involve damage to the aircraft or even to people. It is advised to make a list of the possible hazards (e.g. electromagnetic interference determined by the presence of an electricity tower), together with the potential consequences and the likelihood to happen. Moreover, it is good practice to establish emergency hover zones. It is essential to consider site-specific hazards, as explained in Duffy et al. (2018).

8

2.4 Planning of the flight and of the required equipment Flight planning includes the definition of the following parameters: · Waypoints: these are points on the ground which need to be surveyed. · Flight lines, with image frontlap and sidelap: it depends on the complexity of the surveyed site,

especially with regard to tall objects or complex soil morphology. They will be planned so that all the waypoints are surveyed.

· Flight height: it depends on the platform, the desired spatial resolution and the presence of obstacles. · Flight time: it determines the number of required batteries and the necessity of splitting the flight into

two or more parts. · Take-off and landing zones: their size and characteristics can vary considerably when operating a multi-

rotor or a fixed wing aircraft (Duffy et al., 2018). These parameters must be determined also by keeping into consideration the site assessment: for instance, the presence of a tall tree might impose a certain flight height, just like the presence of an electricity tower could determine a change in the flight lines. If the site assessment cannot be very accurate, a preliminary flight can be useful in providing the necessary information (see the case of the rainforest mentioned in Duffy et al., 2018). Several software packages are available to plan UAV flights by setting the preferential flight parameters (e.g. DJIFlightPlanner (DJI, 2016), eMotion (SenseFly Ltd, 2014)). A list of the required equipment should be compiled, taking into account: platform to fly; camera(s); batteries; repair kit; special clothing for the pilot and co-pilot; protection glasses or sunglasses for the pilot and co-pilot; instruments for ancillary measurements (see §6); toolbox; notebooks and/or logbooks. The flight planning should be discussed with all the members of the crew.

9

3. Flight preparation Following recommendations are largely based on the ESI DroneLab operations manual (Cunliffe et al, 2017).

3.1 Weather forecast Long range weather forecasts will be acquired a week before each planned operation from the most

reliable source, usually national. Ideally, a larger time-frame would be designated within which operators

and field personnel are flexible, and initial adjustments to the planned operation date and time could be

made daily. For practical reasons this is not often possible. Once a date is fixed, the weather forecast is

checked again 24 hours prior to the planned operation time, at which point the definitive go-ahead or

cancellation is given. Operations are cancelled either if a safe flight cannot be guaranteed due to wind or

precipitation, or if weather conditions are unfavourable for data acquisition. Wind speeds at which it is safe

to operate vary strongly between systems, with multirotor UAVs generally able to cope with higher winds

than fixed wing. In both cases, the manufacturer’s recommendations should be followed. Gust speeds

should be used as reference in place of average wind-speeds.

Depending on the type of data acquisition and sensors used, high platform stability or constant, direct

illumination may be required. This can be more difficult to judge and operations may be attempted despite

non-ideal conditions, as long as safe operations are guaranteed.

3.2 Site permissions and notifications To ensure lawful and ethical UAV operations, permissions of the land-owners of any sites flown over and/or

imaged should be acquired. Where possible, written consent should be obtained as a signature or by email

so it is on record. If the site is within 5 km of an aerodrome traffic zone, explicit permission should be

acquired from the air traffic control, for which the approximate coordinates, flight height and time of flight

must be provided.

To notify third persons of UAV operations, warning signs or additional people should be placed along public

footpaths.

Flights within restricted zones should be avoided wherever possible, as costly and time-consuming

applications to the relevant authorities will have to be made.

3.3 Communications Where operations take place close to controlled airspace, communication should be sought with the

responsible air traffic control service, which can update the operator on any unforeseen activity. All

relevant numbers as well as those of emergency services should be identified and recorded at the flight-

planning stage.

Communication must be guaranteed between all members of the team at all times, using either mobile

phones or radios.

3.4 Preparation of crew’s equipment and clothing Prior to the flight, the crew should take care of wearing suitable clothing, also according to the weather

forecast. High visibility clothing indicating the member’s role (e.g. “Pilot”) is preferable. Crew members

should always bring identification documents, as well as any other document proving that necessary flying

permissions.

10

3.5 Preparation of the crew With due anticipation, the crew members should be informed of:

- Objective and location of the flight

- Possible hazards at the time and location of the flight

- Role they are assigned to

- Any other significant detail.

11

4. Flight procedure

4.1 In situ site assessment Once on the site, the crew should check carefully:

· The presence of obstacles or hazards, on the surveyed site or in the neighbouring area. · The conditions of the takeoff and landing sites. · The position of the sun, which should not affect the ability of the pilot and co-pilot to keep the drone

in sight during the flight. · The presence of people who might approach the pilot and co-pilot during the flight. · The presence of animals that might interfere with the flight operations.

If needed, a safety perimeter should be set up and/or changes should be made to the flight plan. Any significant finding should be reported on a logbook.

4.2 In situ weather assessment The crew should check weather conditions on the site, especially with regard to wind speed, visibility and cloud height from the ground. This should be done by using specific instruments (e.g. handheld anemometer) and with a visual assessment if necessary. Indeed, wind speed can change from the point where it is measured (on the ground) to the altitude at which the aircraft flies; looking at the movement of trees around the area or at cloud speed can help in determining how strong is the wind. Usually, the manual of the platform contains information on the maximum wind speed at which it is advisable to fly the UAV (e.g. DJI, 2017; SenseFly Ltd, 2014). When flying at dawn or dusk, it is recommendable to double-check civil twilight hours. Notes about weather conditions should be as well reported on the flight logbook.

4.3 Pre-flight controls Before beginning the flight, the crew should check that all the necessary instruments are taken and working. To make this job more systematic, the use of a checklist is recommended. This should include:

· Check that the drone registration number is legible · Check and that there is no external anomaly (e.g. in the propellers, transmitters, gimbal, etc.) (DJI,

2017; SenseFly Ltd, 2014) · Check that the gimbal clamp and the lens caps have been removed · Check that the SD card and the battery are in place · Check that the batteries of the drone and of the computer are fully charged.

4.4 Flight During the flight, the members of the crew take care of different tasks. The pilot is in charge of piloting the drone from takeoff until landing. Takeoff and landing require different operations in fixed-wing and multi-rotor platforms and whether the operation is manually or automatically conducted. However, there are some common steps to be followed in all the cases (e.g. DJI manual): first, the UAV and the remote controller must be turned on and connected; the Inertial Measurement Unit (IMU) and the compass must be calibrated; the GPS must be receiving signal from at least 6 satellites. The PIC should check that the takeoff can take place safely and announce loudly the beginning of the flight with a standard statement (e.g. “Take off”). When working with a multirotor, at first the UAV should be kept at eye-level altitude so that the PIC can check that the platform is stable and responding to commands. During the flight, the pilot should constantly check the behavior of the drone, as well as the remaining battery life and any event happening in the surroundings; if a co-pilot is present, the pilot should regularly communicate with him/her, to confirm that everything is in order and the flight can continue. Before

12

landing, the PIC should check that the landing area is safe and then announce clearly the landing with a standard statement (e.g. “Landing”). Finally, the PIC can proceed with the landing operations, which differ in case of automatic or manual mode. The co-pilot must maintain a visual line of sight with the drone during the flight to support the PIC, to ensure that even in the short moments when the PIC has to look away the drone is flying as expected. The co-pilot should immediately notify the PIC if the platform is not behaving normally or if a hazard arises. If necessary, the co-pilot should support the PIC in the takeoff and landing operations, ensuring that the site is safe. At the end of the flight, the PIC and the co-pilot need to remove the camera and the batteries from the drone.

4.5 Post-flight controls After the flight, it recommendable to check the quality of the acquired images, so that, if the image quality is not good and if it is possible, the flight can be repeated. There are some programmes available to check image quality on the field (e.g. Postflight Terra 3D (SenseFly Ltd, 2014)).

13

5. Ancillary measurements

5.1 STS measurements Objective: Define the procedure of assembly and configuration of the necessary equipment to accompany the measurements made with a compact STS series Ocenoptics spectroradiometer onboard a UAV to obtain data on vegetation canopies. The system is based on the Specy autonomous spectroscopy system developed by Andreas Burkart (Burkart et al., 2014) (http://www.fz-juelich.de/ibg/ibg-2/EN/methods/specy/specy_node.html), protocols for data acquisiton have been developed in the context of SynerTGE project (http://www.lineas.cchs.csic.es/synertge/). Instruments/materials needed

· ASD FIELDSPEC® 3 spectroradiometer: · Laptop · STS spectroradiometer- air unit · STS spectroradiometer- ground unit · RGB camera (drone) · STS cosine receptor (ground unit) · 99 % Spectralon reference panel · Batteries and/or power supply for all instrumentation · Canvas / umbrella to avoid continued exposure to the Sun from ASD spectroradiometer · Tripods and heads · Toolbox with cutter, scissors, meter, · Clinometers/bubble level · Table and chairs · Cameras · Logs + maps · Pens

Protocols

Field measurements on support to drone acquisitions

Simultaneously to the drone flight, ground measurements will be performed with a STS unit as the one onboard the drone as well as with an ASD Fieldsepc 3 spectroradiometer in order to provide information for calibration of the data acquire from the STS air unit. Spectral processing includes spectral match between STS air and ground units’ bands, temporal match using air and ground units’ timestamps and the computation of Hemispherical Conical Reflectance Factor as:

𝐻𝐶𝑅𝐹=𝑁_𝑎𝑖𝑟/𝑁_𝑔𝑛𝑑 〖∙𝑓〗_𝑋𝑐𝑎𝑙 〖∙𝜌〗_𝑟𝑒𝑓

Instruments for ground measurements will be located in an open area, ideally within the same plot where drone will fly with the air unit. The STS ground unit will be mounted on a tripod with the sensor pointed towards the sky and the tripod deployed at its maximum height. Prior to this operation, a cosine receptor should be attached to the STS lens (do not to touch the white surface of the cosine with your fingers). Check that the STS is securely mounted in the tripod and level the sensor/cosine receptor as much as possible using a bubble level or a clinometer. Make sure the sensor is on and functioning correctly and it has enough battery charge before the STS-Air (drone) starts the measurements. The STS ground unit should be continuously measuring while the drone with the STS air unit is flying.

Before and after each flight the STS-air will be dismounted from the drone to measure on the spectralon panel. The sensor should be manually held horizontally about 15 cm from the surface, centered on the panel and without projecting shadows (figure 1). Sensor will be measuring on the panel to take about 40 spectra. Before that, the foreoptics will be covered to record the Dark Current signal STS ground and air instruments.

14

Figure 1. STS ground unit measuring on the Spectralon panel

The ASD spectroradiometer will be additionally used for sensors’ inter-calibration. The ASD will be placed on its own trunk or a work table. Next to it will be placed a tripod on which the cosine receptor will be mounted and connected to the ASD fiber (figure 2). It is very important that you never touch the diffuser plug on the top of the cosine. Once the cosine receptor is installed, insert the fiber, being careful not to twist or pull. The fiber should never be tightened so the ASD will be always close to the tripod on a stable surface. The cosine of the ASD should be installed at a height similar to the STS-ground. ASD unit will be protected from the excess heat produced by the Sun using an umbrella (figure 3).

Figure 2. ASD Fielspec 3 cosine receptor mounted in the tripod

15

Figure 3. ASD protected from the excess heat produced by the Sun

Ground support team will be dedicated to continuous measurements of direct radiation with the ground STS and at intervals of 50 spectra under direct and 10 under diffuse illumination with the ASD. To perform the diffuse radiation measurements, one of the 2 operators in the control area will use a solid, black object (i.e a plastic folder) to intercept the solar radiation, verifying that the cosine of the ASD is in completely shadow throughout this process (figure 4).

Figure 4. Acquiring diffuse radiation measurements with the ASD cosine receptor

Drone measurements

Do not forget to synchronize the times of all the sensors to be used in the campaign (ground-air STS sensors, ASD sensor and RGB camera) before starting drone measurements.

Measurements from the drone to characterize specific target areas in a natural canopy will be made at an average altitude between 15 and 20 meters (depending on the canopy height) trying to maintain the drone a minimum of 1-1.5 minutes over each target area to obtain a minimum number of spectra (~ 10 spectra). Ideally the drone will perform some slight movements over the target (i.e. tree crown) so to ensure representativeness of the measurmetntes in heterogeneous surfaces.

An operator will follow the evolutions of the drone from the ground and will take note in a log of the sequence of measurements, approximate lapse of time on each target, drone height, and any other incident that occurs during the measurements.

After each flight the operator will check that the measurements are correct (number of files/sequence of measurements) and will copy the files from the memory card to a computer or external drive.

The ground-air measurements sequence would be:

1. In the ground: Measure Dark Current (5 measurements) with both air and ground STS sensors. Cover the foreptic with a black cloth thick enough to block all light and measure with STS-air on Spectralon® (40 spectra)

2. In the air: STS air measurements on the vegetation targets

16

3. In the ground: Measure Dark Current (5 measurements) with both air and ground STS sensors. Cover the foreptic with a black cloth thick enough to block all light and measure with STS-air on Spectralon® (40 spectra).

5.2 Spectroradiometric measurements Especially during flights meant for research purposes, getting accurate values of reflectance is essential. This can be achieved by performing a spectral correction of the acquired images, performed by employing ground spectral data, collected with the aid of a spectroradiometer. The procedure usually requires:

• measurement of a white reference: the white reference is represented by an object with >99% diffuse reflectance, namely a spectralon® , which is a highly Lambertian, diffuse reflectance material (Labsphere, 2006). Spectralons are highly Lambertian, diffuse reflectance materials, chemically inert and available on the market with different % of reflectance values. Beside the white reference, also spectralons with other % of reflectance can be used to improve the radiometric correction of the UAV images. The size of the spectralon(s) depends on the spectral resolution of the employed camera and on the Field of View (FOV) of the pistol carrying the optic fiber.

• measurement of other objects in the image: the targets should be homogeneous, sufficiently large (at least 5x5 pixels) and featureless. In general, objects like asphalt, sand or concrete are considered good targets. it is good to choose potential targets before the flight, however the final choice will be done on the field, because the target should be in full light and not in the shadow.

Before the flight, it is important to make sure that the optic fiber is working and is in place and to check that the battery of the instrument is fully charged. According to the number of measurements foreseen and to the battery life of the instrument, it may be advisable to bring more than one fully charged battery. The spectroradiometer should be turned before beginning the measurements, to allow warming up; the time needed for the warming up depends on the instrument and on the type of measurement to be performed, in general ranging from 15 to 60 minutes (ASD Inc., 2012). The instrument must be set up in order take measurements with the optic fiber. Optic fibers must be handled with care to prevent damage, which would give a weaker signal than normal and therefore affect the measurements. The spectralon(s) should be put on the ground, in a well-illuminated area, without touching its surface with hands. A visual inspection of the area should be done to detect other well-illuminated, homogeneous targets. During the measurements, the operator should wear dark clothing, which can slightly affect the collected spectra (Goetz, 2012); moreover, the operator should keep away from objects that might influence the reflectance of the spectralon (e.g. cars). The operator should always be facing the sun during the measurements, so that his/her shadow is not cast on the measured target. For each measurement, the pistol should be held above the target and perpendicular to it (Goetz, 2012). The procedure should follow these steps:

1. Optimize the spectroradiometer. Beside doing it at the beginning, it’s necessary to optimize again if the sky conditions change or if the instrument is saturating.

2. Take the white reference: save a measurement of the >99% reflective spectralon as “White Reference”. When working outdoors, it is advisable to take the white reference more than once. Consulting the manual of your instrument is recommended. Normally, if the reflectance of the white reference is less than 1, it is recommended to take the white reference again; if the reflectance is more than 1, it is recommended to re-optimize the instrument and take the white reference; if the reflectance is still 1, there is no need to take the white reference again.

3. Measure the chosen reference targets: it is advisable to take more than one spectrum for each target, so as to compute an average spectrum.

4. Measure the spectralon(s): this time, the spectrum should be saved as a target and not as a white reference.

17

For each step, it is recommended to take more than one spectrum of the target/white reference/spectralon, to allow for the computation of an average spectrum which will reduce noise. The number of collected spectra per target is specified on the instrument’s manual; however, it should be a compromise between the optimal number and the number of targets to measure, i.e. the total number of spectra to collect, as this is a time-consuming operation. In general, the number of measurements of the white reference should be double the number of measurements of the targets. The number of collected spectra per target should stay constant within the set of measurements.

5.3 Ground control points Ground Control Points (GCPs) are a necessity to georeference the UAV survey in 3D space and assess the

geometric accuracy of resulting products. Alternative solutions require very high precision geolocation of

the platform itself which provides the capability of more rapid mapping but is challenging and provides

merely decimetre scale accuracy (Fazeli et al., 2016). Predominantly used in photogrammetric processing

workflows, GCPs also allow to constrain the reconstruction. In order for GCPs to be used for these

applications, their location must be recorded with adequate precisions which depend on the desired

geometric accuracy of the result as well as the resolution of the data. In typical UAV applications where

very fine-grained digital elevation models and maps are produced, targeted survey precisions are in the

centimetre range. This can be achieved by utilising differential global navigation satellite system (GNSS)

receivers or by post-processing with a dense network of reference stations available. Level positioning and

sufficient logging over a clearly identifiable GCP centre is crucial for high quality data acquisition.

For the selection of appropriate GCPs for the study, a number of factors should be considered. When using

GCPs, the goal is that the measured point (usually its centre) should be identifiable on an acquired image

with at least 1 pixel precision. Therefore, their size must be appropriate so that they can be clearly

identified at the acquired resolution (dependent on sensor and flight altitude). Additional considerations in

regards to the spectral behaviour must be observed. If the GCP contains bright colours, these may saturate

and contaminate neighbouring pixels, making it harder to identify the precise location of the centre. This is

the case with commonly used white and black checkerboard designs. The use of red and yellow as

contrasting colours has been proposed to avoid this issue but possibly making the GCPs harder to

differentiate from some backgrounds. GCPs can be weatherproofed by using water-resistant paints or

laminating sheets of paper or cardboard. The latter is a low-cost option which however greatly increases

the risk of specular reflections in the acquisition which can obscure the central point and is not advisable.

Instead, matte water-resistant paint on lightweight plastic has proven to be a convenient option. Where

data is acquired in bands beyond the visible spectrum, as is the case with monochromatic infrared (IR)

cameras as well as thermal IR imagers, laboratory tests should be carried out to ensure GCP centres are

easily distinguishable in the respective bands. Alternative GCP designs may be necessary.

Specific GCP designs may be used to enable automatic GCP recognition in photogrammetric software such

as Photoscan (Agisoft, St. Petersburg, Russia), however they must be very large to be identifiable at typical

survey resolutions.

The distribution of ground control points is crucial especially for photogrammetric processing as it

determines how well the reconstruction can be constrained in georeferenced 3D space. As such, a larger

number of high quality GCPs should always improve the geometric accuracy and robustness of the result.

However, GCP placement often represents the most laborious and time intensive task within the UAV

survey workflow. This provided the motivation for a number of studies on optimal GCP placement and the

minimum number of required GCPs for a given survey. To find a survey appropriate design it is advised to

consult the existing literature on this topic.

18

James et al., (2017) specify that the GCPs should cover the entire area of interest of the survey. Specifically

the edges of the survey area should be well constrained, so no extrapolation of coordinate transformations

occurs (James and Robson, 2012). The required GCP density depends on the project accuracy required. In

their presented case study of a bare-soil survey at 100 m altitude and 23 mm pixel resolution, a survey

precision of 50 mm was attained utilising 15 GCPs with ~50 m uniform spacing. They highlight that GCP

requirements can be lowered if the acquired images are of high quality and show strong network geometry.

Minimum GCP density is also influenced by survey flying altitude as it is recommended to image multiple

GCPs within one image for robust reconstruction.

Based on the analysis by James et al., (2017), a triangular, mostly equidistant distribution of GCPs is

advisable, with the distance proportional to the flight altitude (approx.. ½ altitude, ideally adjusted for

camera field-of-view).

In contrast, Tonkin and Midgley, (2016) identified that more than 4 GCPs did not appear to greatly improve

the quality of their survey, though further GCPs were added randomly. They note that additional GCPs are

nevertheless required to assess the quality of the resulting solution in terms of horizontal and vertical

accuracy. Accuracy assessment which is independent of GNSS precision can be achieved by placing a cluster

of GCPs in relatively close proximity and measuring the distance between them using e.g. a tape measure.

For the physical placement of the GCPs in the field, it is advised to securely fasten them using pegs or

weighted objects wherever possible to ensure they remain stationary throughout the acquisition. For

repeat surveys of the same site, it is highly encouraged to install permanent GCP markers to avoid

repeating the time-consuming GNSS measurements and minimizing geometric variations due to different

placement.

5.4 Generic toolbox and other materials There are a number of tools which are helpful to have in the field, some of which are non-essential for successful UAV operation but can prove useful to secure against unplanned incidents. These items along with all UAV equipment and sensors should be listed on a check-list to ensure they are taken into the field. Items include:

a. Tools: - Screwdrivers & Allen keys - Replacement screws & wingnuts - Cable ties - Duct-tape

b. Power: - LiPo-safebags - Powerbank mobile chargers - Car power inverter / generator

c. Misc: - Anemometer - First-aid kits - Two-way radios - Fire extinguisher or blanket - High-vis vests (if necessary) - Notepads & pens - Head torch (if necessary) - Food & water - Safety tape (if necessary)

19

6. Post-flight tasks

6.1 Summary of performed measurements Recording the measurements acquired immediately after each flight is key for good data management and detailed metadata greatly aids in any further use of the data. The measurements performed can be grouped by UAV flight and details added to an existing flight log document, recording the time of start and end of each acquisition. Besides noting the type of measurements acquired, the name of the folder or folder structure containing the specific data, downloaded after each flight, should be specified. Additional metadata which is of relevance to the particular measurement should also be recorded here. For spectral acquisitions, this is either an anecdotal description of illumination variations during the acquisitions or the name of a file including concurrently recorded irradiance data. Further helpful information are wind-speeds during the acquisition, other weather conditions and any observed abnormalities.

6.2 Field work report A fieldwork report should be compiled as soon as possible following fieldwork proper. Most importantly it

should feature a more verbose description of the fragmented information and metadata acquired in the

field. It is necessary as handwriting or short comments which can only be understood in context may cause

confusion when interpreting data at a later date. The report will also include a summary and discussion of

the fieldwork on the whole which should reflect on whether initial requirements were fulfilled, what data

likely needs to be re-acquired and what future strategies could address the issues encountered.

6.3 Data and product quality control and validation With the amount of variables involved in UAV acquisitions and factors outside of the operator’s control, it is

highly likely that not all initial parameter settings were optimal. First controls of acquired data should

therefore be performed visually in the field. Especially if repeat flights can be carried out, this check is vital

as it allows adjustments to acquisition parameters. Typical checks include assessing images for motion blur

and over- or underexposure. These kind of issues can be eliminated by tuning acquisition parameters, given

external conditions remain constant. For spectral acquisitions, it is important to identify oversaturation and

adjust integration times to new illumination conditions (re-calibration).

Further quality control in the field is generally time limited and must be conducted afterwards. This

includes a full inspection of acquired imagery and identification of illumination variations, cloud shadows

and blur which will influence the processed result. Prior to time-consuming photogrammetric processing

for example, unwanted blurry images as well as imagery from ascent and descent should be removed. A

decision must also be made if the overall data quality appears sufficient for product generation. For

example, in terms of illumination variations during acquisition, it may not prove productive to continue

processing and generate products to determine illumination effects are too large at a later stage.

The quality of resulting products such as orthomosaics, digital elevation models (DEM) and spectral data is

finally assessed against field-measured reference data. For orthomosaics, this includes the GCPs withheld

as check-points reporting horizontal RMSEs as well as visual assessment of the orthorectified, stitched

result. DEMs are commonly assessed using check-points as well as georeferenced height validation points,

to report horizontal and vertical RMSE values. Finally, spectral data is compared to reference spectra

measured in the field, usually over uniform surfaces and reference panels, their RMSE and differences

between derived indices assessed.

20

References ASD Inc. (2012). FieldSpec4 User Manual. https://doi.org/10.1007/SpringerReference_28001

Burkart, A. ; Cogliati, S. ; Schickling, A. ; Rascher, U. A Novel UAV-Based Ultra-Light Weight Spectrometer for Field Spectroscopy IEEE sensors journal 14(1), 62 - 67 (2014) [10.1109/JSEN.2013.2279720]

CAA. (2016a). CAP 393: The Air Navigation Order 2016 (ANO) and Regulations. Retrieved from: http://www.legislation.gov.uk/uksi/2016/765/contents/made

CAA. (2016b). CAP382: Mandatory Occurrence Reporting Scheme. Retrieved from: http://publicapps.caa.co.uk/modalapplication.aspx?appid=11&mode=detail&id=214

CAA. (2015). CAP 722: Unmanned Aircraft System Operations in UK Airspace - Guidance. Retrieved from: https://publicapps.caa.co.uk/modalapplication.aspx?appid=11&mode=detail&id=415

Cunliffe, A.M., Anderson, K., DeBell, L., Duffy, J.P., 2017. A UK Civil Aviation Authority (CAA)-approved operations manual for safe deployment of lightweight drones in research. Int. J. Remote Sens. 38, 2737–2744. doi:10.1080/01431161.2017.1286059

DJI. (2016). DJI Flight Planner. Retrieved August 13, 2018, from http://www.djiflightplanner.com/

DJI. (2017). Phantom 4 User Manual, 0–3.

Duffy, J. P., Cunliffe, A. M., DeBell, L., Sandbrook, C., Wich, S. A., Shutler, J. D., … Anderson, K. (2018). Location, location, location: considerations when using lightweight drones in challenging environments. Remote Sensing in Ecology and Conservation, 4(1), 7–19. https://doi.org/10.1002/rse2.58

EASA. (2018). NPA 2017-05: Introduction of a regulatory framework for the operation of drones — Unmanned aircraft system operations in the open and specific category. Retrieved from https://www.easa.europa.eu/document-library/notices-of-proposed-amendment/npa-2017-05#group-easa-downloads

European Union. 2016. General Data Protection Regulation. Retrieved from: https://gdpr-info.eu/

Fazeli, H., Samadzadegan, F., Dadrasjavan, F., 2016. Evaluating the potential of RTK-UAV for automatic point cloud generation in 3D rapid mapping. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. - ISPRS Arch. 41, 221–226. doi:10.5194/isprsarchives-XLI-B6-221-2016

Finn, R.L., Wright, D., 2012. Unmanned aircraft systems: Surveillance, ethics and privacy in civil applications. Comput. Law Secur. Rev. 28, 184–194. doi:10.1016/j.clsr.2012.01.005

Goetz, A. F. H. (2012). Making Accurate Field Spectral Reflectance Measurements. ASD Inc.

Hodgson, J.C., Koh, L.P., 2016. Best practice for minimising unmanned aerial vehicle disturbance to wildlife in biological field research. Curr. Biol. 26, R404–R405. doi:10.1016/j.cub.2016.04.001

James, M.R., Robson, S., 2012. Straightforward reconstruction of 3D surfaces and topography with a camera: Accuracy and geoscience application. J. Geophys. Res. Earth Surf. 117, 1–17. doi:10.1029/2011JF002289

James, M.R., Robson, S., d’Oleire-Oltmanns, S., Niethammer, U., 2017. Optimising UAV topographic surveys processed with structure-from-motion: Ground control quality, quantity and bundle adjustment. Geomorphology 280, 51–66. doi:10.1016/j.geomorph.2016.11.021

Labsphere. (2006). Spectralon® Diffuse Reflectance Standards. Retrieved from https://www.labsphere.com/labsphere-products-solutions/materials-coatings-2/targets-standards/diffuse-reflectance-standards/diffuse-reflectance-standards/

SenseFly Ltd. (2014). eBee senseFly Extended User Manual

21

Stöcker, C., Bennett, R., Nex, F., Gerke, M., Zevenbergen, J., 2017. Review of the current state of UAV regulations. Remote Sens. 9, 33–35. doi:10.3390/rs9050459

Tonkin, T.N., Midgley, N.G., 2016. Ground-control networks for image based surface reconstruction: An investigation of optimum survey designs using UAV derived imagery and structure-from-motion photogrammetry. Remote Sens. 8, 16–19. doi:10.3390/rs8090786

Wilson, R.L., 2014. Ethical issues with use of Drone aircraft. 2014 IEEE Int. Symp. Ethics Sci. Technol. Eng. ETHICS 2014. doi:10.1109/ETHICS.2014.6893424