Small unmanned aerial systems (UAS) equipped with an optical camera represent a cost-effective strategy for topographic surveys. These low-cost drones can provide useful information for 3D reconstruction even if they are equipped with a low-quality navigation system. To ensure the production of high-quality topographic models, careful consideration of flight mode and proper distribution of ground control points is required. To this end a commercial drone has been adopted to monitor a small earthen dam using different combinations of flight configurations and adopting a variable number of ground control points (GCPs). Results highlighted that both choice and combination of flight plans can reduce the relative error of the 3D model up to a few meters without the need of including GCPs. The use of GCPs allows the quality of topographic survey to be greatly improved, reducing error to the order of a few centimeters. In particular, the combined use of images extracted from two flights, one with a camera mounted at nadir and the second with a 20° angle, proves extremely beneficial to increase the overall accuracy of the 3D model and especially of the vertical precision.

The acquisition of reliable information in coastal and marine areas is crucial for mapping, monitoring, and management of marine environments. Unmanned Aerial Vehicles (UAVs) are widely used for mapping in coastal areas and marine environments since they can collect aerial data with a spatial resolution of some centimetres. However, there are several limitations that UAVs have to overcome for collecting accurate data. These limitations are related to i) the weather conditions (wind, temperature, clouds etc.) and oceanographic parameters (waves, turbidity, sun glint etc.), ii) the flight mission, iii) the system of the UAV (payloads and vehicle), and iv) the morphology of the study area.
A theoretical protocol that summarizes the parameters that affect the quality of aerial data acquisition from a light UAV, in a coastal area of interest, was developed. This paper presents the testing and validation of the UAV theoretical protocol in the real world. Prediction data used as input values in the protocol for resulting the optimal flight windows on the marine environment. A coastal area with a variety of marine habitats and bathymetric differences has been chosen as a testing area. Flights at different times and in different conditions were performed to test the parameters and the thresholds of the UAV model. Flight missions were created in different flight heights and with different payloads (RGB cameras, multispectral cameras). The aerial data processed in photogrammetric software and 3D models and ortho mosaics produced. These ortho mosaics were then compared to conclude on the optimal conditions of UAV data acquisition for the marine environment.
Results released significant differences on the quality of the produced ortho mosaics. The local weather condition and the water clarity, during the optimal flight windows, are the most important parameters for the quality of the produced data. Specifically, the sun glint effect on the sea surface as well as the waves make the collection of benthic information impossible. Clouds and humidity prevent the sunlight from reaching the seabed, making difficult the separation of benthic features. These conditions should be avoided during the data acquisition for desired results.
Considering the results, the optimal flight windows are important parameters of a cost-effectiveness methodology for marine mapping. The UAV protocol can be used for the acquisition of reliable and accurate data in the marine environment which is essential for many marine applications like marine habitats, coastal bathymetry, and coastal morphology. The ultimate target is to combine the UAV ortho mosaics with satellite images for deriving accurate geoinformation on the coastal environment.

Recent development of UAV (Unmanned Aerial Vehicles) technology has provided opportunity of mounting several sensors for data acquisition of forest areas. This study reflects utilization of LiDAR (Light Detection and Ranging) ULS (Unmanned Laser Scanning) technology providing high resolution 3D point clouds for forest inventory purpose. The studied stand is located in northern part of Czech Republic. For data acquisition was used UAS (Unmanned Aerial System) VUX-SYS completely integrated laser scanning system RICOPTER and VUX-1UAV multi return compact laser scanner. Data acquisition was performed by autonomous flight with perpendicular zig-zag pattern for deriving high resolution dense 3D point cloud of chosen Scots pine (Pinus sylvestris L.) stand. As reference data were used data collected on the ground including trees positions, height, crown projection and DBH (Diameter at Brest Height) that were measured for each single individual tree. To reach high accuracy and precision of 3D point clouds was used RTK (Real Time Kinematics) GPS ground station for precise calculations of trajectory including the information from inertial unit on the GPS. Single scans were linked with using of RIEGL software package for creation of one dense 3D point cloud of chosen forest stand. The spatial analyses were performed in the environment of Arc GIS 10.6.1 using of Las Tools for creating Las dataset.
The work was divided in to several steps. Estimation of positions and heights of each single individual tree was performed with using of 3D Sample tool in the environment of ArcGIS in particular the Canopy peaks tool was used. Crown projections were estimated also with using of 3D Sample tool. In the first step it was single tree delineation by estimation of crown parameters that was used as a form dividing the area of the whole point cloud to single areas of each single individual tree. The second step was to estimate crown projection parameters mainly the mean values of crown radius and DBH from the stems visible in the scans. The results shows that the laser scanning provides comparable accuracy and precision as ground data in these kind of stands.

Palsas are a form of discontinuous permafrost in the circumpolar zone. Palsa are peat mounts with a core of permanently frozen peat, ice and mineral soil. There are several studies which indicate that palsa’s are melting and decaying as a result of climatic warming. In two palsa’s high accuracy Real Time Kinematic (RTK) GPS measurements and active layer depth have been conducted 2007-2017. A measurement grid with two meter interval was defined over both palsa, (approximately 200 points). Measurements were carried out yearly in the end of August (2007-2018). Active layer depth of each point was measured with an active layer probe. With ArcGIS we created 3-D models of palsa and yearly active layer surfaces. Weather data used is from the Kilpisjärvi climate station allocated about 15 km north of the study sites (1951-2018). Both palsa’s has experienced significant decay especially along the edges. With detailed RTK GPS and active layer monitoring of palsa for 10 years we can follow accurately the development of palsa and its correlation to local climatic factors. We have gathered aerial photograph times series from 1950’s to present from 20 palsa mire basins from Enontekiö Lapland. Combining historical aerial photographs the time span of the study can be stretched up to 60 years. In some of the studied palsa mire basins more than 60% of the palsas have disappeared during the past 60 years. In 2015 we began to monitor changes in the palsas with UAS (Unmanned Aerial System) drones. Since 2016 we have had 12 palsa mires under investigation with twice a year data collection (mid June and late August). UAS –data provides accurate and detailed information of the palsa decaying process. With UAS based orthomosaics and DEM’s we can interpret change rate in area and in volume level. This study demonstrates the combination of field measurements, aerial image archives and UAS data collection to provide accurate knowledge on climate change effects on permafrost landscape.

During the last few years, unmanned aerial vehicle (UAV, or drone) technology has evolved tremendously. At the same time, miniaturization of hyperspectral sensors has made it possible to use these sensors on board small drones. This provides completely new possibilities to carry out high spatial and spectral resolution remote sensing of the environment. Most promising applications utilizing hyperspectral drone imagery include precision agriculture and forest management.
Traditional RGB cameras can already provide valuable information for farmers, who can have a quick general view of the crop condition. In addition, almost automated service chains are available, utilizing this technology for farmers. However, the improved availability of small spectral sensors for drone systems enables even more possibilities to farmers for planning crop management and input use on the cereal fields or tools to optimize harvesting time on grass silage swards by based on estimating quantity and quality of the growing grass sward. The ultimate goal for drone-based precision agriculture is a drone providing real-time information of the field or crop to farmer’s tractor, which then immediately adjusts its systems, such as fertilization, accordingly.
Drones can be used for collecting dense point clouds and spectral information over small forested areas, such as single stands or sample plots. If this data could be processed and interpreted automatically, drones could be used for supporting large-area inventories or stand-wise assessments by replacing part of the required field work. Essential forest parameters needed for determining the management actions include determination of the tree species, trees per hectare (TPI) and mean tree height of seedling stands.
Our research group at the Finnish Geospatial Research Institute FGI has been utilizing drones in various remote sensing tasks since 2008. We have been developing quantitative and SI-traceable image processing chains for hyperspectral drone imagery. End results of the processing chain are geometrically accurate 3D point clouds, radiometrically corrected homogenous reflectance image mosaics and bi-directional reflectance distribution function (BRDF) information of the object. Our latest development is methodology and system to measure reflectance directly on board drone. Key instruments of the system are upwards looking horizontally levelled irradiance sensor and downwards looking imaging spectrometer. The reflectance factor R is calculated using the following equation R = PI*L/E, where L is at-sensor radiance reflected by the object and recorded with imaging spectrometer, and E is the irradiance at object. Requirement for both instruments is that they are radiometrically calibrated, the irradiance sensor has to be horizontally stabilized, and the sensors needs to be accurately synchronized. For irradiance measurements we use novel FGI Aerial Imaging Reference System (FGI AIRS). It measures hemispherical spectral irradiance and GNSS/IMU data and it can be easily paired with any imaging sensor providing output signal of the image capture times. FGI AIRS has a novel optical levelling system to compensate tilting of the irradiance sensor up to 15°. Direct reflectance methodology produces calibrated homogenous reflectance image mosaics under varying illumination conditions. FGI AIRS has been successfully used together with commercial Fabry-Pérot Interferometer -based (FPI) hyperspectral Rikola camera. Rikola camera provides up to 200 bands in the visible and near-infrared range and frame format images but each band is captured at slightly different times. We have performed spectral and radiometric calibration of Rikola camera at laboratory.
Another major current research activity aims at developing hyperspectral camera system and methodology that can perform initial image pre-processing on-board drone and transfer acquired calibrated images to ground station or to cloud service in real-time. Key instrument for real time processing is in-house built FGI Real-time Hyperspectral Camera and Processing Unit (FGI REHU). FGI REHU records 41 bands in frame format in single snapshot. System also includes wide angle RGB sensor and Jetson TX1 on-board computer for real-time processing and data transfer to either ground station or to cloud using 4G modem.
In this presentation, we will introduce our current instrument setup and describe latest quantitative data processing workflows used in drone-based hyperspectral imaging. Our systems have been used in various precision agriculture and forestry campaigns during summer 2018, including 2h beyond visual line-of-sight (BVLOS) forestry campaign. Highlights from these campaigns will be presented.

Precision Agriculture (PA) is a concept based on observing, measuring and responding to inter and intra-field variability in crops. The main goal of PA is to define a decision support system for farm management, optimizing returns on inputs while, at the same time, preserving resources. To achieve this, a fast, reliable, cost-effective and easy method to scan the fields is required. The crop’s condition can be assessed by the stage of ripening, water status, pest attacks and nutritional requirements. Remote sensing platforms, such as satellites or manned aircrafts, have been applied for several years in PA. Lately with the appearance of Unmanned Aerial Systems (UAS), that can be remotely piloted or have a programmed route to perform autonomous flight, new possibilities are offered. PA relies on four main tasks: data acquisition; terrain variability mapping; decision making; and application of management practices. Thus, remote sensing can be used with great benefit in the first three tasks. Acquiring the necessary information is one of the key factors of PA to provide correct support in the decision-making process and recognition of temporal variations. Nowadays, direct ground observations are performed to assess the presence of issues. This study proposes a methodology capable of analyzing very high resolution UAV-based data from different types of sensors in a multi-temporal perspective, focusing on three different crops of great economic impact in Portugal, vineyards (Vitis vinifera L.), chestnut trees (Castanea sativa Mill.) and olive trees (Olea europaea L.) The commercial value of these plantations depends on several factors such as: the caste (vineyards), the plant quality, and the climatic conditions of the region. However, in these three crops, there are problems that may interfere with their development, which make effective detection and inspection techniques necessary to decrease the occurrence risk of those problems, thus enabling a viable and sustainable cultivation of these crops. For all these reasons, the detection of problems in earlier stages becomes indispensable, allowing a quicker response capacity to apply the proper treatment, saving up crops and reducing maintenance costs.
UAVs appear as the ideal tool, allowing the monitoring of crops through the analysis of spectral signatures and its vegetative development. In this work, we propose an automatically methodology allowing the evaluation of the temporal evolution of these cultures, determining the probable causes of potential problems, from biotic and/or abiotic origins, thus permitting the application of the most appropriate measures to eliminate/mitigate the identified problems.

Keywords: Precision agriculture, Unmanned aerial vehicle, multi-temporal data analysis, vineyards, chestnut trees, olive trees

Acknowledgments
The authors would like to acknowledge European Regional Development Fund (ERDF) and North 2020 - North Regional Operational Program, as part of project “INNOVINE&WINE - Vineyard and Wine Innovation Platform" (NORTE-01-0145-FEDER-000038) as well as COMPETE 2020 under the PORTUGAL 2020 Partnership Agreement, and through the Portuguese National Innovation Agency (ANI) as a part of project “PARRA - Plataforma integrAda de monitoRização e avaliação da doença da flavescência douRada na vinhA” (Nº 3447).