ADDED VALUE OF NEW REMOTE SENSING AND UAV FOR RISK … · ADDED VALUE OF NEW REMOTE SENSING AND UAV FOR RISK PROTECTION SURVEY AND MONITORING ASP 462 – RockTheAlps I ROCKTHEALPS

ADDED VALUE OF NEW REMOTE

SENSING AND UAV FOR RISK

FOREST PROTECTION SURVEY AND

MONITORING

ASP 462 – RockTheAlps

WP1 – Activity A.T2.2

Deliverable D.T2.2.3

Coordinated by: doc. dr. Milan Kobal, Barbara Žabota (UL); prof Marco Piras (POLITO)

Ljubljana, September 2018

ADDED VALUE OF NEW REMOTE SENSING AND UAV FOR RISK PROTECTION SURVEY

AND MONITORING


I

ROCKTHEALPS PARTNERSHIP

IRSTEA – National research institute of science and technology for environment and agriculture

BRGM – French Geological Survey

Alp’Géoriskques

SFS – Slovenian Forest Service

UL – University of Ljubljana, Biotechnial Faculty, Department for Forestry and Renewable Forest Resources

SFI – Slovenian Forestry Institute

UNIPD – University of Padova, Department of Land, Environment, Agriculture and Forestry

DISAFA – Department of Agricultural, Forest and Food Science, University of Turin

ERSAF – Regional Agency for Services in Agricultue and Forest - Lombarida Region

PAT-SFF – Autonomous province of Trento, Forest and Wildlife Department

POLITO – Politecnico di Torino, Department of Territory, Land and Infrastructure Engineering

BFW – Federal Research and Training Centre for Forests, Natural Hazards and Management

BMLFUW – Austrian Federal Ministery of Agriculture, Forestry, Environment and Water Management

LWF – Bavarian State Institute of Forestry

BFH – HAFL – Bern University of Applied Sciences, Department of Forest Sciences


AND MONITORING


II

TABLE OF CONTENT

Nessuna voce di sommario trovata.

1 INTRODUCTION ............................................................................................... 1

2 UAV TECHNOLOGY ........................................................................................ 2

3 THE ADVATAGES OF USING UAV FOR FOREST SURVEY AND

MONITORING .................................................................................................................... 9

4 THE LIMITATION OF THE UAV USE IN FORESTRY ............................ 13

5 CASE STUDIES ................................................................................................ 14

5.1 CEVO (BS) – ITALY .......................................................................................... 14

5.2 COLLE SANTA LUCIA (BL) – ITALY ............................................................ 15

5.3 CESANA (TO) – ITALY ................................................................................... 16

6 DATA ANALYSIS AND RESULTS ................................................................ 16

7 CONCLUSIONS ................................................................................................ 21

8 REFERENCES .................................................................................................. 22


AND MONITORING


III

TABLE OF FIGURES

Figure 1. Different airframes. From the left: polystyrene, plastic, aluminium, carbon fibber ........................... 2

Figure 2. Examples of professional, semi-professional and action cameras for UAVs...................................... 3

Figure 3. U-blox GNSS receivers ....................................................................................................................... 5

Figure 4: two different method for image acquisition: nadir (on the left) and oblique (on the right) ............... 7

Figure 5: Using drones a high spatial resolution of orthophoto images can be achieved ................................... 9

Figure 6: With drones a high spatial resolution of data can be achieved as the observed phenomena can be

monitored more frequently. .............................................................................................................................. 10

Figure 7: 3D model of the rockfall and surrounding forest can offer a better visualisation of the rockfall

impact on forest and forest protection function ................................................................................................ 11

Figure 8: 3D model of rockfall and surrounding forest .................................................................................... 11

Figure 9: An example of NDVI index on the location of rockfall Kekec (Slovenia) where good difference can

be absorbed among the rockfall area and its contact with forest ...................................................................... 12

Figure 10: Comparison of canopy height model (CHM) among raster cell size 1×1 m and 0.25×0.25 m. ...... 13

Figure 11: Location (left) and test site (right) .................................................................................................. 14

Figure 12: flight planning ................................................................................................................................. 15

Figure 13: Location (left) and flight planning (right) ....................................................................................... 15

Figure 14: Location (left) and flight planning (right) ....................................................................................... 16

Figure 15: flowchart of UAV data processing .................................................................................................. 17

Figure 16: example of Alignment step ............................................................................................................. 18

Figure 17: example dense point cloud (on the left) and classification (on the right) ........................................ 18

Figure 18: DSM and orthophoto in Cevo ......................................................................................................... 19

Figure 19: DSM and orthophoto in Colle Santa Lucia ..................................................................................... 20

Figure 20: DSM and orthophoto in Cesana ...................................................................................................... 21

ADDED VALUE OF NEW REMOTE SENSING AND UAV FOR RISK FOREST PROTECTION SURVEY

AND MONITORING


1

1 INTRODUCTION

Forest structure is a key parameter in defining the protection function of forest against

natural hazards (Maier et al., 2008). Forests especially in mountainous areas protect people

and their assets against different slope processes e.g. rockfalls, avalanches, landslides,

debris flows etc. Forest cover has a mitigation effect due to standing and lying trees and

their capacity to dissipate energy. Defining forest stand structure is an important task as

based on different parameters, etc. stand density, basal area, DBH, the protective effect of

forest can be evaluated (Belcore et al., 2018).

As already mentioned in D.T1.2.3, since several year, the use of UAV (unmanned aerial

system) aerial surveys by using unmanned aerial vehicles (UAV), especially through

Airborne Laser Scanning (ALS) or photogrammetry, were carried out in the last years to

acquire point clouds to be used in 3D models aimed at achieving an accurate description of

tree crowns and terrain as well as surveying and monitoring natural hazards that occur in

the forests.

The use of drones in forestry has been primarily focused on mapping and monitoring forest

fires, however their use is increasing for applications such as tree crown/gap mapping,

forest stand mapping, volume estimation, wind blow assessment, pest monitoring and

harvest planning (Paneque-Gálvez et al., 2014). Successful implementation of drones in

forestry depends on following features of UAV’s: flexibility of use in flight planning, low

cost, reliability and autonomy, and capacity to produce a high-resolution data (Torresan et

al., 2017). UAVs in forestry can be used for (based on examples in: Vepakomma et al.,

2015; Tang and Shao, 2015; Banu et al., 2016; Torresan et al., 2017; Taddese Berie and

Burud, 2018): tree species classification, quantification of spatial gaps in forests, post-fire

recovery monitoring and forest fire measuring, forest health monitoring and forest diseases

mapping, post-harvest soil displacement estimation and supporting intensive forest

management.

The use of UAV can have many benefits in surveying and monitoring protection forest in

order to maintain protective effect of forest and regenerate it protective effect after the

events that can largely decrease it. Combined with remote sensing techniques UAV offer a

versatile opportunities how to improve forest mapping and monitoring with the regards to

natural hazards like rockfall. The following content is presenting some of the major

advatages of UAV for risk protection survey and monitoring regards to traditional field

methods and also conventional remote sensing techniques.


AND MONITORING


2

This report introduces the UAV in forested sites, analysing the different sensors required

and available results carried out by a data acquisition made by drone. In the second part of

the report focuses on some case studies, stressing advantages and disadvantages of its use

for rockfall monitoring and generally over forested sites.

2 UAV TECHNOLOGY

Aerial vehicles are complex systems made by hardware and software structures. The

improvement of electronics allowed the development of navigation and control systems

more and more available on the market.

The main components required to use the UAV are:

1) the aerial platform, which includes the airframe, the navigation system, the power

system and the payload

2) ground control station (GCS), which allows the human control from a remote

emplacement; and the communication system, which supports the communication

between the other two components.

The UAV could be composed by different components::

airframe: it is the main structure, which has to be designed according to the specific

sensors that have to be installed. Its structure has to consider the weight regarding

in particular the power and the communication and control systems on board.

Moreover, the airframe needs to be properly designed to withstand the forces that

can occur during the flight and not cause deformation and vibration. Fixed wings

on UAVs are mainly made on polystyrene or plastic, while common multi-rotors

airframes are made on aluminium or carbon fibber (in such a way as to be

lightweight and resistant) (Figure 1), and the number of arms is a function of the

expected payload and the number of engines;

Figure 1. Different airframes. From the left: polystyrene, plastic, aluminium, carbon fibber

Navigation system (board): this board allows autonomous or semi-autonomous

flights through hardware and software components. The specification of the

autopilot for a UAV can be summarized as reported in Table 1. The autopilot

includes positioning (Global Navigation Satellite System – GNSS receiver) and

attitude (inertial platform) systems and the data registered from these sensors are

managed by a microcontroller that, after the process of the input signals, can send

updated information to the autopilot to control the stability and the trajectory;


AND MONITORING


3

Physical specifications

Sensor specifications

Autopilot functions

Size Operative temperature

Waypoints navigation

Weight Max angular rate

Auto take-off and landing

Power consumption

Max acceleration

Altitude hold

Required voltage Max angular velocity

Servo control rate

Required CPU Max altitude Telemetry rate

Memory Operative airspeed

Fail safe

Table 1. Autopilot main specifications

Focusing more the attention on the payload, various gimbals and acquisition systems can be mounted on-board. These sensors can mainly include the following:

• digital cameras: the most common use for UAVs is image and video acquisition for monitoring, photogrammetry [0]. Different kind of cameras are now available as professional, semi-professional and action cams, but the lens calibration is always required to ensure a final good result;

Figure 2. Examples of professional, semi-professional and action cameras for UAVs

• thermal detector: these systems can be used in situations with an important gradient of temperature in order to add the thermal information to the visible one [0],

• multispectral cameras: these sensors can be very helpful in environmental applications to classify different kinds of vegetation or to monitoring the growth and variation of some areas [0]. Table 2 summarizes some characteristics of the most used multispectral sensors;

PARROT SEQUOIA

MICASENSE REDEDGE

Tetracam ADC Micro

N° spectral channel

4 5 3

Spectral range 550-790 nm 400-900 nm 520-920 nm

Weight 0.072 Kg 0.15 Kg 0.09 Kg Table 2. Comparison between the most used multispectral sensors


AND MONITORING


4

• LiDAR (Light Detection And Ranging): these devices are very helpful during the night, with clouds or shadows and especially in presence of many trees (Table 3). This is possible because they are active sensors that transmit a signal to the object then they measure the intensity of the returned signal [0];

YELLOWSCAN RIEGL VUX-1UAV LeddarVu 8

Wavelength 905 nm 905 nm 905 nm

Maximum range 50m typ. 100m abs.

from 85 m to 920 m

from 6 m to 61 m

Accuracy 5 cm 1 cm 5 cm

Scanner field of view

360° 330° 100°

Weight 1.5 Kg 3.5 Kg 0.128 Kg Table 3. Comparison between the most used LiDAR sensors

• IMU-MEMS (Inertial Measurement Unit - Micro Electro-Mechanical Systems): useful for real time navigation and post-processing attitude estimation;

• GNSS receivers: these sensors are able to obtain the position of the UAV along the flight. In particular, navigation systems play a fundamental role and these systems use not only one or

more GNSS receivers, but also an inertial navigation system that is necessary to provide the

vehicle set-up information for each epoch, and to assist the GNSS system in estimating the

position of the vehicle.

The use of these sensors, however, requires the resolution of a series of problems, both

theoretical and practical, linked for example to their synchronization and calibration. Typically,

these problems can be solved by developing appropriate hardware and software tools, where

systematic errors (bias) and sensor drift (s) are analyzed and compensated, with the ultimate goal

of obtaining a navigational solution sufficiently accurate, also in relation to the type of application

in which the sensors are used.

Different solutions can be available, mainly based on these possibilities:

• Single/multi frequency receivers: possibility to improve the solution with the usage of the

L2 frequency;

• Single/multi constellation antennas and receivers: the availability of more than one

constellation can improve, also in this case, the final positioning solution;

• RTK (Real Time Kinematic) approach: it is the possibility to improve the position estimated

in real time with correction from a master or a permanent station;

• GNSS/IMU integration: it is also possible to improve the GNSS position with data from the

inertial platform.


AND MONITORING


5

The market offers many low-cost GPS or GNSS receivers, often available as an OEM card. In other

cases, some Devolopper Kit are available, as in the u-blox receivers: this makes them immediately

usable since the card is already combined with a USB serial port, a power supply device and a LAN

network port.

In particular, the u-blox receiver (Figure 3, now available in v.8) is a GNSS receiver, capable of

receiving also 4 single-frequency GNSS constellations. At a cost of about € 300, it is an interesting

solution with minimum overall dimensions and an easy interface. The presence of the

communication port requires only to prepare a data storage on micro PC and SD card (or similar)

and the transfer of raw data to a processing center.

Figure 3. U-blox GNSS receivers

https://www.u-

blox.com/en/product/ma

x-m8-series

On the other hand, some OEM card solutions with a variable cost between € 100 and €1000 are

also available (examples in Erreur ! Source du renvoi introuvable.). Among the "medium-high"

receivers, some multi-frequency and multi-constellation OEM records should be highlighted, not

included in any management and display device. These cards should therefore be engineered and

could constitute a low cost solution but with centimeter precision suitable also for real time.

These receivers are already set up for RTK acquisitions and they may have options that are not

always included in normal receivers.

SkyTra NS-HP TERSUS BX316 TOPCON B125

Dimension 25 x 25 mm 108 x 54 mm 40 x 55 mm

Weight 3 g 50 g 20 g

Frequency Single Dual Dual

Constellation GPS, Beidou GPS, GLONASS, Beidou GPS, GLONASS, Galileo, Beidou, SBAS

Max acquisition rate

20 Hz 20 Hz 100 Hz

Possibility of RTK Yes Yes Yes

Communication protocol

RTCM 3.x RTCM 2.x/3.x/CMR/CMR+ TPS, RTCM SC104 2.x and 3.x, CMR, CMR+

Classification Low-cost ($ 100) Medium-cost ($1,799.00) Medium/high-cost ($3,500.00)

Table 4. Comparison between different rate OEM GNSS receivers

https://www.u-blox.com/en/product/max-m8-series




AND MONITORING


6

About the IMU sensors, an inertial measurement unit consists of a set of sensors, generally

accelerometers and gyroscopes, necessary to allow the estimation of all the navigation states at a

high frequency. Since these are sensors of an electro-mechanical nature, the measurements

carried out are affected by systematic errors (measurement bias, scaling factors or non-

orthogonality of the sensor triplets) and accidental errors (noise).

Usually the height of flight is fixed not considering a real elevation model. In same case, flight

planners are able to consider some global digital elevation models, and define a flight plan with a

constant distance from the ground (figure 3). This is important in particular in steep areas, where

the value of GSD can be very different is the flight height is constant and the surveyed area has a

strong difference of altitude.

Figure 3: example of flight planner that is able to consider also a global terrain. In the image is also possible to see that flight planner have always the possibility to setup a geofence, which represents the limit of the area of operation of the UAV.

Recently, there is a new approach adopted for photogrammetry application which is based on the

acquisition of oblique images. In this case, the planning has to be completely different and the

traditional tools are not more available. The same problem rises when the image acquisition is

made along a vertical façade (e.g. rock façade), where the flight is manually made, using the

experience of the pilot. Moreover in this condition, working along a vertical plane, the height is

not perfectly guarantee and stable due by the position quality estimated by the GNSS receiver.

The oblique acquisition is the typically based on the use of multirotors.


AND MONITORING


7

Figure 4: two different method for image acquisition: nadir (on the left) and oblique (on the right)

Ground Control Points The use of Ground Control Points is an important element that could have a strong impact on the

accuracy of the SfM-based DSM (James et al., 2012; Turner et al., 2015). GCPs are points of known

coordinated that can be clearly recognized in the photo-sequence acquired by UAV. These points

can be elements present in the field or artificial targets placed in the surveyed area before the

UAV flight. The position of GCPs is acquired using high accuracy topographic methods like GNSS or

total stations and then identified in the photo sequence during the SfM procedure (Harwin and

Lucieer, 2012). The use and the number of GCPs depends on the required final accuracy of the

positioning of the DSM and the quality of the UAV positioning system.

As said before, it is possible to have on the UAV GPS only or multi-constellation receiver, but it is

important to evaluate the combination between receiver and antenna, in order to define the final

performances. In some cases, an external GNSS receiver is installed, which is used to collect the

raw data only. This solution is adopted for direct photogrammetry applications (Turner et al.,

2014; Eling et al., 2015; Mian et al., 2015; Gabrlik et al., 2018), where is required a high resolution

GNNS on board that reduce the importance and the impact of GCPs on the final accuracy of the

DSM.

The number of GCPs and their position is hard to define a priori, but some simple operative

suggestions can be useful for a good distribution of these points: i) follow the limit of the area of

interest ii) insert other GCPs inside the area of interest considering also the elevation differences

of the area. Other critical consideration that have to be evaluated during the deployment of GCPs

are, according to James et al. (2017): i) the importance of datum alignment to gravity ( the

distribution of targets can be carefully considered in particular if the final model can be used for

modelling gradient-sensitive processes like, for example, rainfall runoff), ii) the presence of


AND MONITORING


8

vegetation at the scale of the physical control targets (that could hamper the identification of

targets on images) iii) the absolute 3-D positioning.

Another important element that should be carefully considered during the definition of number

and position of GCPs is the real effort that the deployment and survey of these points required.

Nex and Remondino (2014) propose an evaluation of the time effort in a typical UAV-based

photogrammetric workflow. The time effort evaluation proposed by Nex and Remondino (2014)

is: Flight planning 5%; Image acquisition 20%; GCPs field measurement 15%; image triangulation

15%; DSM generation 25%; Ortho mosaicking 10%; Feature extraction 10%.

In our experience, usually the time required for the survey of GCPs is twice or more the time

necessary for the UAV survey. This element should be carefully considered because the

deployment of targets has to be done before the flight, and this activity can constrain the UAV

flight activity that often should be done in the central part of the day to limit the influence of

shadows.


AND MONITORING


9

3 THE ADVATAGES OF USING UAV FOR FOREST SURVEY AND

MONITORING

The operational altitude of small drones usually ranges from 50 to 300 m and consequently

permits the acquisition of high spatial-resolution imagery (with pixels of few centimetres)

which enhances the visual analysis (Figure 1). High-resolution imagery enables the

identification and monitoring of specific trees, canopy gaps, forest loss, damaged and

fallen trees, degradation and regrowth processes (Paneque-Gálvez et al., 2014; Inoue et al.,

2014; Banu et al., 2016).

Figure 5: Using drones a high spatial resolution of orthophoto images can be achieved


AND MONITORING


10

The identification of single tree species it’s an important factor in evaluation of the

protection forest against rockfalls and other natural hazards. Assessing the forest health

condition can be used for evaluating the state of protection forest.

Small drones can be used for detailed mapping over larger areas than ground surveys

(Banu et al., 2016). With their use we save time for obtaining the imagery and on the small

to medium size areas the monitoring can be repeated in shorter time period which can be

practical for monitoring a particular protection forest at for example one rockfall site.

Imagery can also be acquired more frequently compared with conventional remote sensing

techniques (e.g. satellite and aircraft imagery), meaning that assessment of local protection

forest at shorter intervals (Figure 2) and immediately after the rockfall event, which can

offer an insight on which silvicultural or other technical measures need to be taken in order

to restore protection function of the forest stand. Also, small UAVs can fly bellow the

cloud level which gives them an important advantage over conventional remote sensing

platforms as they can offer images without clouds (Paneque-Gálvez et al., 2014). The

smallest UAVs can also fly near the forest canopy or under the forest canopy, while

medium and larger ones can also fly on higher alltitudes (Tang and Shao, 2015).

Figure 6: With drones a high spatial resolution of data can be achieved as the observed phenomena can be

monitored more frequently.

Based on the images that UAV capture, 3D surface models, digital surface models, digital

elevation models and canopy height models can be generated (Figure 3 and 4). Using those

models, we can improve forest monitoring strategies (e.g. detection of the damaged trees

due to rockfall, the regrowth stage of protection forest etc.) and use them for retrieving


AND MONITORING


11

forest structural parameters (e.g. height, basal area, DBH, tree density, tree species

composition etc.) (Banu et al., 2016; Mohan et al., 2017).

Figure 7: 3D model of the rockfall and surrounding forest can offer a better visualisation of the rockfall

impact on forest and forest protection function

Figure 8: 3D model of rockfall and surrounding forest


AND MONITORING


12

Small drones are easy to operate with and can be used fully or semi-autonomously by the

users that have little training and no geomatic knowledge. Also, the cost of small drones

are relatively low compared with classically obtained aircraft imagery or satellites

(Paneque-Gálvez et al., 2014). We can monitor areas that are hard to access or remote, e.g.

steep slopes, rocky terrain, as rockfalls usually are. With drones we can consequently

improve the working conditions for field workers as they become safer.

With the use of multispectral cameras forest health can be monitored using different

vegetation indices. One of the most common is normalized difference vegetation index

(NDVI) that employs the multi-spectral remote sensing data techniques to find vegetation

index, land cover classification, vegetation, water bodies, open area, scrub area, hilly areas,

agricultural areas, thick forest, thin forest with few band combinations of the remote

sensed data (Gandhi et al., 2015). NDVI is a simple metric that indicates the health of

green vegetation. Chlorophyll stongly reflects near infrared light (NIR, around 750 nm)

while red and blue are absorbed. As it reflects strongly the plant appear green to us,

however reflection in NIR is even greater which helps us in rendering precise data for

analysis. NDVI is based on a principle that the leaves, due to spongly layers on their

backsides, reflect a lot of light in the near infrared to the constrast with most non-plant

object. When a plant is dehydrated or stressed, the spongly layer collapses leading to the

reduced reflection of NIR light when the same amount stays in the visible range. Using the

combination of this signals can help differenciate plant from non-plant and healthy tree

from a damaged tree (Mahajan and Bundel, 2017) (Figure 5).

Figure 9: An example of NDVI index on the location of rockfall Kekec (Slovenia) where good difference can

be absorbed among the rockfall area and its contact with forest


AND MONITORING


13

Therefore, using multispectral cameras in combination with UAV can enable the

calculation of NDVI index (and other vegetation indices) for individual investigated trees

as the pixles (Figure 6) for vegetation layer can be relatively small. This way the trees that

have been affected by a rock and are therefore severly damaged, can be recognized.

Figure 10: Comparison of canopy height model (CHM) among raster cell size 1×1 m and 0.25×0.25 m.

4 THE LIMITATION OF THE UAV USE IN FORESTRY

Nevertheless that UAVs have great advatages for the use in forestry, there still some

limitations that hinder their operation. As UAVs are usually small they have a limit of the

equipment that they can carry onboard. Consequently the amount of additional sensors and

GNSS devices in the monitoring process are quite limited. They are also more susceptible

to pitch, roll and yaw distortions that can affect the georeferencing (Paneque-Gálvez et al.,

2014).

The placing of ground control point can be challenging as the point under the canopy

won’t be visible on imagery and will hinder the georeferencing. Even more, because

rockfalls usually occur on steep terrains that are hardly accessible there might be problem

by finding appropriate places for ground control points. In the remote areas we might also

face a bad signal receipt which will disable us to accurately measure the position of ground

control point or it will hinder the GNSS connection on UAV itself and won’t enable us a

direct georeferencing. The duration of the flight itself is usually short and limited and

depends on the carrying capacity of the UAV and the size of batteries (NEWFOR…,

2014).


AND MONITORING


14

During the flight there is always a possibility of collisions with objects on their way – for

example trees or steep slopes. Similar situations may occur in the lost signal among UAV

and its operator and when we lose the sight of its movement. Weather and atmospheric

conditions can also impact the UAV flights, especially heavy rain, strong winds and fog.

5 CASE STUDIES

UAV systems are involnving in several test sites, with purpose to verify the performance

and accuracy available, considering different platform, software, sensors and apprpoaches.

In the following section, three case studies are described, where UAV has been used in

order to generate a DSM and orthophotos for forestry application and adopted in the RTA

project.

5.1 CEVO (BS) – ITALY

In this fisrt case, a fixed wings UAV has been used to collect RBG image, with purpose to

generate a DSM and a DTM, extracting the terrain only and realizing a classification.

The site is Cevo (BS) (see figure), where a LiDAR data collecteion has been realized many

years ago. Now, the data collection has been focused in a specific area, defined in the

figure.

Figure 11: Location (left) and test site (right)

One flight has been carried out, collecting 188 image and covering an area equal to 15,4

ha. The relative flight height was 127,5m. The average GSD (groud sampling distance) has

been 3 cm/px and a duration of data acquisition equal to 14 minutes.


AND MONITORING


15

Figure 12: flight planning

5.2 COLLE SANTA LUCIA (BL) – ITALY

In this case, the data acquisition was similar to the previous one, where one flight has been

carried out, collecting 302 image but covering a largest are, which was equal to 57.61 ha.

The relative flight height was 100m. The average GSD (groud sampling distance) has been

5.47 cm/px and a duration of data acquisition equal to 22 minutes. Location and flight

planning are reported in the following figures.

Figure 13: Location (left) and flight planning (right)

In this case, a terrestrial data acquisition with Lidar has been carried out.


AND MONITORING


16

5.3 CESANA (TO) – ITALY

In this site, three different flights have been carried out: 2 usign RGB camera and 1 with a

NIR camera. In the following table, the main properties of each flight are summarized.

RGB 1 RGB 2 NIR 1

Average H [m] 100 100 100

GSD [m] 5.47 5.47 6.29

Duration (min) 18 12 19

Area [ha] 60 40 76.4

# images 221 137 176

Table 5. Comparison between different rate OEM GNSS receivers

The flight planning is shown hereafter.

Figure 14: Location (left) and flight planning (right)

6 DATA ANALYSIS AND RESULTS

After data acquisition, each dataset has been processd by two different sofwares: Agisoft

photoscan and Pix4D. The approach used to realize the data processing is quite similar in

both cases, and it could be schematize as:


AND MONITORING


17

Figure 15: flowchart of UAV data processing

where the “alignment” is the step where a preliminary “combination and join” between the

images and sparse point cloud is generated.

Mesh generation

DTM/DSM and orthophoto

generation

Import images (and

geoinformation)Alignment of images

Import GCPs coordinates and

markers identification in the

imagesCamera parameters optimization

Dense point cloud generation

Dense point cloud classification


AND MONITORING


18

Figure 16: example of Alignment step

After this step, imposing the GCPs, it is possible to extract the dense point cloud and the

related classification, as shown in the following figures.

Figure 17: example dense point cloud (on the left) and classification (on the right)


AND MONITORING


19

Last step is the generation of the DSM, DTM and the orthophotos, removing the vegetation

using the classification. Some examples are shown in the following, considering the case

studies.

Figure 18: DSM and orthophoto in Cevo


AND MONITORING


20

Figure 19: DSM and orthophoto in Colle Santa Lucia


AND MONITORING


21

ù

Figure 20: DSM and orthophoto in Cesana

7 CONCLUSIONS

UAS system is surely a very interesting and powerful tool for many different application

fields in particular in forestry, where the use of RGB and NIR camera can be help the

expert and professional to generate DSM, DTM, classification and orthophoto.

Moreover, the new technologies allow to solving all kind of survey and evinronment, but it

is fundamental to have some knownledge on data acquisition and data processing, with

purpose to be able to evaluate the results and to select the correct approach.

In particular in forestry application, background is fundamental and mandatory, to be able

to planning correctly the flight trajectory, the sensor and to platform (fixed wings or multi-

rotors).


AND MONITORING


22

8 REFERENCES

Banu T. P., Borlea G.F., Banu C., 2016. The Use of Drones in Forestry. Journal of

Enviromental Science and Engineering, B 5: 557-562.

Belcore E., Berger F., Grasso N., Lingua E., Piras M., 2018. Investigating barrier effects of

Alpine forests: 3D data acquisition methodologies and result analysis. Geophysical

Research Abstracts, Vol. 20, EGU2018-19768, EGU General Assembly

Gandhi M. G., Parthiban S., Thummalu N., Christy A., 2015. Ndvi: Vegetation change

using remote sensing and gis – A case study of Vellore District. Procedia Computer

Science, 57: 1199-1210.

Inoue T., Nagai S., Yamashita S., Fadaei H., Ishii R., Okabe K., Taki H., Honda Y.,

Kajiwara K., Suzuku R., 2014. Unmanned Aerial Survey of Fallen Trees in a Decidious

Broadleaved Forest in Easter Japan. PLOS ONE, 9, 10: e109881.

Mahajan U., Bundel B.R., 2017. Drones for Normalized Difference Vegetation Index

(NDVI), to Estimate Crop Health for Precision Agriculture: A Cheaper Alternative for

Spatial Sattelite Sensors. In: International Cnference on Innovative Research in

Agriculture, Food Science, Forestry, Horticulture, Aquaculture, Animal Science,

Biodiversity, Ecological Sciences and Climate Change (AFHABEC-2016), Ar Jawaharlal

Nehru University

Maier B., Tiede D., Dorren L., 2008. Characterising mountain forest structure using

landscape metrics on LiDAR-based canopy surface models. In: Blaschke T., Lang S., Hay

G.J. (eds.) Object-Based Image Analysis. Lecture Notes in Geoinformation and

Cartography. Springer, Berlin, Heidelberg

Mohan M., Silva A. C., Klauberg C., Jat P., Cattis G., Cardil A., Hudak A.T., Dia M.,

2017. Individual Tree Detection from Unmanned Aerial Vehicle (UAV) Derived Canopy

Height Model in Open Canopy Mixed Confer Forest. Forests, 8, 340: 1-17.

NEWFOR – Interreg Alpine Space project. NEW technologies for a better mountain

FORest timber mobilization. Workpackage 4: Forest resources & LiDAR – High resolution

aerial images from UAV for forest applications. State-of-the-art. 2014.


AND MONITORING


23

Paneque-Gálvez J., McCali M.K., Napoletano B.M., Wich, S.A., Pin Koh L., 2014. Small

Drones for Community-Based Forest Monitoring: An Assessment of Their Feasibility and

Potential in Tropical Areas. Forest, 5: 1481-1507.

Taddese Berie H., Burud I., 2018. Application of unmanned aerial vehicles in earth

resources monitoring: focus on evaluating potentials for forest monitoring in Ethiopia.

European Journal of Remote Sensing, 51, 1: 326-335.

Tang L., Shao G., 2015. Drone remote sensing for forestry research and practices. Journal

for forestry research and practices, 26, 4: 791-797.

Torresan C., Berton A., Carotenuto F., Di Gennaro S. F., Gioli B., Matese A., Miglietta F.,

Vangoli C., Zaldei A., Wallace L., 2017. Forestry applications of UAVs in Europe: a

review. International Journal of Remote Sensing, 38, 8-10: 2427-2447.

Vepakomma U., Cormier D., Thiffault N., 2015. Potential of UAV based convergent

photogrammetry in monitoring regeneration standards. The International Archives of the

Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-

1/Q4,2015, International Conference on Unmanned Aerial Vehicles in Geomatics, 30 Aug-

02 Sep 2015, Toronto, Canada: 281-285.

Aicardi, I., Dabove, P., Lingua, A. M., & Piras, M. (2016). Integration between TLS and

UAV photogrammetry techniques for forestry applications. iForest-Biogeosciences and

Forestry, 10(1), 41.

Aicardi, I., 2017. UAVs for spatial data acquisition - Sensors evaluation, flight design and

planning, multi-temporal solutions. PhD dissertation.

Arnold J.E.M., 1993. Management of forest resources as common property. The

Commonwealth Forestry Review, 72, 3 pp. 157-161.

Baltsavias EP (1999). A comparison between photogrammetry and laser scanning. ISPRS

Journal of photogrammetry and Remote Sensing 54: 83-94. - doi: 10.1016/S0924-

2716(99)0001 4-3

Bienert A, Scheller S, Keane E, Mullooly G, Mohan F (2006). Application of terrestrial

laser scanners for the determination of forest inventory parameters. In: Proceedings of the

ISPRS Commission V Symposium “Image Engineering and Vision Metrology” (Mass H-

G, Schneider D eds). Dresden (Germany) 25-27 Sep 2006. ISPRS, vol. 36, pp. 36.


AND MONITORING


24

Borruso, A., D’Argenio, A. & Spinnato, C. (2011). Telerilevamento ed emergenze.

L’esperienza di ITHACA.Sessione 3-Geologia ambientale, risorse naturali, cartografia

tematica Sessione 4-Monitoraggio in ambiente urbano e tecnologie di precisione, 20-21.

Bornaz L, Dequal S (2003). The solid image: a new concept and its applications. In:

“International Archives of Photogrammetry, Remote Sensing and Spatial Information

Sciences”. ISPRS, Vol. 34, Part 6/W12, pp.78-82.

Bornaz L, Dequal S, Lingua A (2006). L’ortofoto solida di precisione: un prodotto

innovativo per la rappresentazione e la gestione dei dati tridimensionali [The solid

orthophoto: an innovative product for 3D data representation and management]. In: “Atti

Conferenza Nazionale SIFET”. Taranto (Italy) 13-16 un 2006, pp. 8.

Brovelli, A., Cina, A., Crespi, M., Lingua, A., Manzino, A., & Garretti, L. (2012).

Ortoimmagini e modelli altimetrici a grande scala. Linee Guida, CISIS-Centro

Interregionale di Coordinamento e documentazione per le informazioni territoriali.

Buil F., Núñez-Andrés M. A., Lantada N., Prades, A., 2016. Comparison of

Photogrammetric Techniques for Rockfalls Monitoring. World Multidisciplinary Earth

Sciences Symposium (WMESS 2016), IOP Conf. Series: Earth and Environmental

Science, 44, 042023, pp. 1-7.

Car M., Jurić Kaćunić D., Linrić L., 2016. Volume measurements of rockfalls using

Unmanned Aerial Vechicles. CETRA 2016, 4th

International Conference on Road and Rail

Infrastructure.

Chiabrando, F., De Bernardi, M., & Roggero, M. (2010). Impiego di aerei ultraleggeri e

camere di medio formato per la fotogrammetria aerea a grande scala. BOLLETTINO

DELLA SOCIETÀ ITALIANA DI FOTOGRAMMETRIA E TOPOGRAFIA, 4, 75-95.

Chiabrando, F., Lingua, A. M., Maschio, P. F., Rinaudo, F., & Spanò, A. (2012). Mezzi

aerei non convenzionali a volo autonomo per il rilievo fotogrammetrico in ambito

archeologico. Una giornata informale per i 70 anni del Prof. Carlo Monti - 3 Maggio 2012,

Milano, 3 maggio 2012. pp. 1-12

Chisholm R.A., Cui J., Lum S., Chen B. 2013. UAV LiDAR for below-canopy forest

surveys. Journal of Unmanned Vehicle Systems, Vol. 01(01): 61-68, 10.1139/juvs-2013-

0017.

[14] Boccardo P., Chiabrando F., Dutto F., Tonolo F. G., Lingua A., 2015. “UAV deployment exercise for mapping purposes: Evaluation of emergency response applications”. Sensors, Vol. 15(7), pp. 15717 - 15737.


AND MONITORING


25

Bulusu N., Heidemann J., Estrin D., 2000. “GPS-less low-cost outdoor localization for very small devices”. In: IEEE personal communications, Vol. 7.5, pp. 28 - 34.

[21] Chisholm R.A., Cui J., Lum S., Chen B. 2013. “UAV LiDAR for below-canopy forest surveys”. In: Journal of Unmanned Vehicle Systems, Vol. 01(01): 61-68, 10.1139/juvs-2013-0017.

[23] Colomina I. and Molina P., 2014. “Unmanned aerial systems for photogrammetry and remote sensing: A review”. In: ISPRS Journal of Photogrammetry and Remote Sensing, Vol. 92, pp. 79-97.

European Parliament, 2015. “Civil drones in the European Union”. Briefing.

Gao Y., Wojciechowski A., 2004. “High precision kinematic positioning using single dual-frequency GPS receiver”. In: The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences.

Kalman R. E., 1960. “A new approach to linear filtering and prediction problems”. In: Journal of basic Engineering, Vol. 82(1), pp. 35 - 45.

NASA, 2015. “A report overview of the civil UAV capability assessment”.

[97] Nebiker S., Annen A., Scherrer M., Oesch D., 2008. “A Light-weight Multispectral Sensor for Micro UAV - Opportunities for Very High Resolution Airborne Remote Sensing”. In: Proceedings of XXI ISPRS Congress. Beijing, China.

[103] Pech K., Stelling N., Karrasch P., Maas H.G., 2013. “Generation of Multitemporal Thermal Orthophotos from UAV Data”. In: ISPRS, International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, UAVg13, Rostock, Germany Volume XL-1/W2, pp.305-310.

Piras M., Dabove, P., 2016. “Comparison of two different mass-market IMU generations: Bias analyses and real time applications”. In 2016 IEEE/ION Position, Location and Navigation Symposium (PLANS), pp. 34 - 41.

Pupillo G., Naldi G., Mattana A., Monari J., Poloni M., Perini F., Schiaffino M., Bianchi G., Bolli P., Lingua A., Aicardi I., Bendea H., Maschio P., Piras M., Virone G., Paonessa F., Farooqui Z., Tibaldi A., Addamo G., Peverini O.A., Tascone R., Wijnholds S.J., 2015. “Medicina Array Demonstrator: calibration and radiation pattern characterization using a UAV-mounted radio-frequency source”. Experimental Astronomy, Vol. 39, pp. 405 - 421.

[109] Ramsey C.L., Griffiths P.A., Stokes T.R., 2014. “Multi-rotor Unmanned Aerial Vehicles (UAVs) and high-resolution compact digital cameras: a promising new method for monitoring changes to surface karst resources”. In: Acta Carsologica, Vol. 43.2/3, pp. 269.

[118] Saari H., Antila T., Holmlund C., Makynen J., Ojala K., Toivanen H., Pellikka I.,Tuominen S., Pesonen L., Heikkila J., 2011. “Unmanned Aerial Vehicle (UAV) operated


AND MONITORING


26

spectral camera system for forest and agriculture applications”. In: Proceedings of SPIE, Vol. 8174.

[123] Semsch E., Jakob M., Pavlicek D., Pechoucek M., 2009. “Autonomous UAV surveillance in complex urban environments”. In: Web Intelligence and Intelligent Agent Technologies, WI-IAT’09. IEEE/WIC/ACM International Joint Conferences, Vol. 2, pp. 82-85.

StempfhuberW., Buchholz M., 2011. “A precise, Low-cost rtk gnss system for uav Applications”. In: International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Zurich, Volume XXXVIII-1/C22. Available online: http://www.int-arch-photogramm-remote-sens-spatialinf-sci.net/XXXVIII-1-C22/289/2011/isprsarchives-XXXVIII-1-C22-289-2011.pdf.

[149] Wefelscheid C., Hansch R., Hellwich O., 2011. “Three-dimensional building reconstruction using im-ages obtained by unmanned aerial vehicles”. In: Proceedings of the International Conference on Unmanned Aerial Vehicle in Geomatics (UAV-g), Zurich, Switzerland.

Wisniewski B., Bruniecki K., Moszynski M., 2013. “Evaluation of RTKLIB’s Positioning Accuracy Using low-cost GNSS Receiver and ASG-EUPOS”. In: TransNav: International Journal on Marine Navigation and Safety of Sea Transportation, Vol. 7(1), pp. 79 - 85. Available online: http://www.transnav.eu/Article_Evaluation_of_RTKLIB%27s_Positioning_Wi %C5%9Bniewski,25,412.html.

[155] Zhang J., Liu L., Wang B., Chen X., Wang Q., Zheng T., 2012. “High speed automatic power line detection and tracking for a UAV-based inspection”. In: International Conference on Industrial Control and Electronics Engineering (ICICEE), pp. 266 - 269.

http://www.int-arch-photogramm-remote-sens-spatialinf-sci.net/XXXVIII-1-C22/289/2011/isprsarchives-XXXVIII-1-C22-289-2011.pdf

http://www.int-arch-photogramm-remote-sens-spatialinf-sci.net/XXXVIII-1-C22/289/2011/isprsarchives-XXXVIII-1-C22-289-2011.pdf

http://www.transnav.eu/Article_Evaluation_of_RTKLIB%27s_Positioning_Wi

ADDED VALUE OF NEW REMOTE SENSING AND UAV FOR RISK … · ADDED VALUE OF NEW REMOTE SENSING AND UAV FOR RISK PROTECTION SURVEY AND MONITORING ASP 462 – RockTheAlps I ROCKTHEALPS

Documents