In recent years imaging spectrometers have advanced to such a stage that they are light enough to be mounted on unoccupied aerial vehicles (UAVs), yet they still have many of the advantages that field spectrometers have, and more. Such advantages include large spectral coverage (400 nm – 2500 nm) and cooled detectors. This technology has been noticed by the satellite validation community, particularly for surface reflectance, because UAVs overcome many of the disadvantages present in operating field spectrometers, such as damage to the field site during data collection and limited area coverage.

The Fiducial Reference Measurements for Vegetation (FRM4VEG) ESA-founded project is focused developing better methods to validate satellite surface reflectance products, with UAV-mounted hyperspectral imagers being the dominant technology. This project is applying the metrological techniques developed for field spectrometer based Fiducial Reference Measurements (FRMs) to UAV platforms, as well as protocols to collect FRM data from UAVs. The former looks at understanding how to propagate uncertainty in the raw imager and UAV platform data through to the final validation data product, incorporating the correlation structure caused by the pushbroom sensor (amongst other sources).

Our surface reflectance FRM data was collected using a Headwall Co-Aligned hyperspectral imager and LiDAR mounted on a DJI Matrice 600 Pro. We utilised an ASD spectrometer to transfer the reflectance calibration of a Spectralon reference target (which was calibrated in the laboratory at NPL) to a larger tarpaulin, as well as a Microptops sunphotometer to collect a detailed temporal profile of aerosol optical thickness (AOT). The data collection, conducted at Wytham Woods (UK), was timed so that the UAV flight was coincident with a Sentinel-2A overpass.

This presentation will discuss the results of the validation activity at Wytham Woods, including the propagation of uncertainties through the orthorectification post-processing and measurement considerations for matching the in situ data to the Sentinel-2A pixels. Additionally, it will discuss the Committee for Earth Observation Satellites (CEOS) endorsed Surface Reflectance Intercomparison eXercise for VEGetation (SRIX4VEG) which is being organised by the FRM4VEG team. SRIX4VEG brings participants with UAV-mounted hyperspectral imagers from across the world together with the aim of testing the requirement for a common protocol for surface reflectance validation using UAV-mounted imagers. The exercise will first assess the variability caused by different teams with different UAVs and payloads in collecting surface reflectance data. A draft protocol will then be implemented by all teams to assess any reductions in the variability, with feedback from the participants helping to shape an internationally agreed protocol.

Precision Viticulture (PV) is a concept which is becoming increasingly important in the wine-growing sector. It aims to improve the yield and quality of grapes while also minimizing environmental impacts and costs. Currently the potential grape yield and quality is often forecasted by trained operators who monitor the vineyard several times during its developmental growth. Such vigour assessments are time and labour intensive and expensive to undertake. The adoption of PV technologies could have the potential to reduce time and effort spent on manual labour. Remote sensing can be a powerful tool in PV to characterise the in-field variability of a vineyard.
Recently, UAVs have emerged in agricultural applications, providing flexibility and efficiency in diverse environments such as heterogenous vineyards. The inter-row component of the vineyard makes up a large proportion of the architecture. This often leads to the inter-row component dominating the spectral signature of a mixed pixel, in case of coarse resolution satellite imagery. As a result, UAV images are often preferable because pure canopy pixels can be filtered out to focus only on the vines. Structure-from-motion (SfM) algorithms can produce vegetation height maps from point clouds, using RGB imagery, provided that enough soil area is visible. From these height maps, only the grape vines can be retained after filtering out soil, shadows or inter-row vegetation. By applying vegetation indices, vine vigour can be determined. With information on differences in vigour, selective vintage can take place. The employment of selective vintage could be valuable as differences in vigour have been found to influence grape must factors such as acidity and sugar content. With this knowledge, more similar quality wines can be produced. Nevertheless, UAVs are expensive to operate, and the pre-processing is very time intensive. Therefore, the value of high-resolution satellite imagery, such as PlanetScope, should be assessed in determining their use in estimating variability within vineyards.
During the summer of 2021, three UAV flight campaigns were undertaken over an experimental vineyard of the DLR Mosel in Bernkastel-Kues, Germany. The UAV images were processed, then methods to isolate the vine canopy (supervised classification Spectral Angle Mapper i.e., SAM and SfM) and different vegetation indices (NDRE, NDVI and OSAVI), applied. Each of the combinations were evaluated, for each flight, to determine which one best discriminates vigour classes. Before each flight, a trained operator conducted a vigour assessment of the entire vineyard which was used as validation data. During the harvesting season grape sampling took place, based on vines of different vigour. Grape samples were collected from healthy vines, which showed no symptoms of disease. Sampling was based on the different vigour classes determined before harvest by the trained operator.
Our results showed that UAV imagery has the potential to discriminate vigour classes and can predict yield. The combination which was able to differentiate vigour best was the SAM with the OSAVI applied. Other interesting findings were that low vigour vines resulted in higher sugar content and lower acidity and vice versa. Therefore, by separating grapes of different vigour classes, more similar and higher quality wines could be produced. The canopy mask resulting from the Sfm method was not able to isolate all vine canopy pixels, especially at the lower end of the slope. This is likely due to the fact that the flight took place at a constant height on a very steep slope. An effective solution to this problem could be to adjust the UAV height according to the terrain elevation each time an image is captured.
Not all PlanetScope bands overlapped with the UAV sensor (Micasense RedEdge-MX) bands used in this research and the correlations between UAV and PlanetScope data were not as expected when compared to other published papers. Therefore, we were not able to determine the effectiveness of PlanetScope satellites in discriminating vigour at our study site. It can also be assumed that UAVs are more adequate to be applied for most vineyards around the Mosel. This is because the inter-row is often covered with vegetation which is necessary between the rows to reduce erosion and pests. This would result in a bias towards the inter-row vegetation of the mixed pixel.
Keywords: precision viticulture (PV), vigour, UAVs, PlanetScope, grape production parameters

Remote sensing of solar-induced chlorophyll fluorescence (SIF) measured with remote sensing sensors is a key parameter to better understand plant functioning at different spatial and temporal scales. Due to the direct relationship between SIF and photosynthetic activity SIF is important for the monitoring of gross primary productivity (GPP) and the early detection of vegetation stress before it becomes measureable with conventional reflectance-based remote sensing proxies (e.g., vegetation indices) (Ač et al., 2015, Cheng et al., 2013).

The SIF signal is immediately released from chloroplasts after the absorption of sun light and emitted as a continuous spectrum in the range of red and far-red light (650–850 nm). Since SIF is only a small part of reflected radiance (1-5%), its detection is challenging and requires precisely calibrated spectrometers providing spectral high-resolution data and a high signal-to-noise-ratio (SNR) (Porcar-Castell et al., 2021). In previous years, several studies have demonstrated the potential of proximal (Pinto et al., 2016), airborne (Rascher et al., 2015), and satellite sensors (Köhler et al., 2018) measuring SIF at different spatial scales and temporal resolutions.

Besides ground, air- and spaceborne platforms, unmanned aerial vehicles (UAVs) also have the capacity to carry sensors measuring SIF at an intermediate spatial scale to close the gap between proximal and airborne/satellite measurements. Recent progress in the development of commercial UAVs, in terms of payload capacity and flight safety features, allowed for the development of several point spectrometers used to measure SIF (Quiros et al., 2020). Some of those studies were primarily focused on sensor characterization (e.g., etaloning, platform motion, cosine correction of measured irradiance) (Bendig et al., 2018, 2020), while others showed the potential of those spectrometers to track the diurnal dynamics in SIF of different crop canopies (Wang et al., 2021, Campell et al., 2021). However, SIF observations from point spectrometers are challenging because i) geometric accuracy of the projected footprint of the spectrometer has to be determined with considerably more effort than in imaging data, ii) switching mechanisms that are needed to measure both downwelling and upwelling radiance may reduce the total signal throughput and iii) the radiometric quality of the signal is influenced by the UAV platform dynamics in flight for example through tilting, atmospheric effects, and temperature fluctuations since active cooling is difficult to obtain. Data interpretation of point spectrometer signals requires auxiliary imaging data. These data can be obtained again from UAVs, and both spectral as well as structural information can assist signal interpretation.

Pioneering work in measuring SIF image data from UAV platforms has already been published by Zarco-Tejada et al. in 2012. The authors used an airborne imaging spectrometer (Micro-Hyperspec VNIR, Headwall Photonics, USA) mounted on a fixed-wing UAV platforms to measure SIF of a citrus orchard. Thus, in contrast to point spectrometers, they were able to overcome the problem of pointing accuracy and limited spatial information content. Although these studies showed first promising results, the used camera had only 6.4 nm full width at half maximum (FWHM) spectral resolution, which was not ideal for SIF retrieval.

In this study we aimed to use a commercial off-the-shelf and easy to control rotary-wing UAV platform (DJI Matric 600, SZ DJI Technology Co., Ltd, China) to acquire SIF image data. In order to realize this, Forschungszentrum Jülich in cooperation with the University of Applied Sciences Koblenz developed a lightweight and fully integrated dual-camera system, explicitly designed to measure SIF. The dual-camera system consists of two scientific CMOS cameras equipped with ultra-narrow bandpass interference filters (each with 1 nm FWHM). To guarantee a precise wavelength location of the passband and bandwidth of the filters, the optical properties of the lenses were of particular importance. In order to retrieve SIF, using the Fraunhofer Line Discriminator (FLD) principle, one camera is measuring within the O2A absorption feature at 760.7 nm and the other one is measuring at the left shoulder outside the absorption feature at 757.9 nm. Both cameras are connected to a single-board computer with an integrated microcontroller coprocessor, which controls and triggers the cameras and stores the recorded image data. The device is mounted on a DJI Ronin gimbal system, which ensures nadir observations and supplies the required power.

This study shows the first SIF760 map of the newly developed dual-camera system recorded at the agricultural research station of Bonn University 'Campus Klein-Altendorf' in summer 2021 (Fig. 1). On 13 June at midday, image data of a mixed-crop (wheat and bean) breeding experiment, comprised of numerous plots, were recorded with the dual-camera system. Data was processed with a photogrammetric structure from motion workflow for multi camera arrays to produce a mosaiced orthophoto. The derived SIF values, ranging from 0 to 2.3 mWm-2nm-1sr-1, are in a reliable value range for the observed crops at midday during the observed growth stage. SIF values of different plots derived from the UAV data will be compared to simultaneous SIF measurements collected with a mobile FloX system (JB Hyperspectral Devices GmbH, Germany) on the ground, and SIF maps collected by the imaging spectrometer HyPlant, which is the airborne demonstrator of the FLuorescence EXplorer (FLEX) satellite mission of the European Space Agency (ESA). Both devices, FloX and HyPlant, are established SIF measurement instruments providing reliable reference data, which will be used to verify the performance of the dual-camera system and the absolute accuracy of retrieved SIF.

The new dual-camera system, which for the first time provides spatial high-resolution SIF maps recorded from an off-the-shelf rotary-wing UAV platform is of high potential for different applications in breeding and precision agriculture, such as the early detection of stress or the improvement of yield estimates. Furthermore, including a UAV SIF imaging system into a future cal/val concept of the FLEX satellite mission will contribute to close the spatial gap between ground-based and airborne measurements of photosynthetic activity.

References
Ač, A., Malenovský, Z., Olejníčková, J., Gallé, A., Rascher, U., Mohammed, G., 2015. Meta-analysis assessing potential of steady-state chlorophyll fluorescence for remote sensing detection of plant water, temperature and nitrogen stress. Remote Sens. Environ. 168, 420–436. https://doi.org/10.1016/j.agrformet.2020.108145.

Bendig, J., Gautam, D., Malenovský, Z., Lucieer, A. Influence of Cosine Corrector and UAS Platform Dynamics on Airborne Spectral Irradiance Measurements. IEEE International Geoscience and Remote Sensing Symposium IGARSS. 2018, 8822–8825, https://doi.org/10.1109/IGARSS.2018.8518864.

Bendig, J, Malenovský, Z., Gautam, D., Lucieer, A. 2020. Solar-Induced Chlorophyll Fluorescence Measured From an Unmanned Aircraft System: Sensor Etaloning and Platform Motion Correction. IEEE Transactions on Geoscience and Remote Sensing. 58(5), 3437–3444, https://doi.org/10.1109/TGRS.2019.2956194.

P. Campbell, P. Townsend, D. Mandl, MacKinnon, J. 2021. Automated UAS Measurements of Reflectance and Solar Induced Florescence (SIF) for Assessment Of the Dynamics in Photosynthetic Function, Application for Maze (Zea Mays L.) in Greenbelt, Maryland, US. IEEE International Geoscience and Remote Sensing Symposium IGARSS, 2021, pp. 8265–8268, https://doi.org/10.1109/IGARSS47720.2021.9554902.

Köhler, P., Frankenberg, C., Magney, T.S., Guanter, L., Joiner, J., Landgraf, J., 2018. Global retrievals of solar-induced chlorophyll fluorescence with TROPOMI: first results and intersensor comparison to OCO-2. Geophys. Res. Lett. 45 (19), 10456–10463. https://doi.org/10.1029/2018GL079031.

Pinto, F., Damm, A., Schickling, A., Panigada, C., Cogliati, S., Müller-Linow, M., Balvora, A., Rascher, U., 2016. Sun-induced chlorophyll fluorescence from high-resolution imaging spectroscopy data to quantify spatio-temporal patterns of photosynthetic function in crop canopies. Plant Cell Environ. 39 (7), 1500–1512. https://doi.org/10.1111/pce.12710.

Porcar-Castell, A., Malenovský, Z., Magney, T. et al. 2021. Chlorophyll a fluorescence illuminates a path connecting plant molecular biology to Earth-system science. Nat. Plants. 7, 998–1009. https://doi.org/10.1038/s41477-021-00980-4.

Quiros Vargas, J., Bendig, J., Mac Arthur, A., Burkart, A., Julitta, T., Maseyk, K., Thomas, R., Siegmann, B., Rossini, M., Celesti, M., Schüttemeyer, D., Kraska, T., Muller, O., Rascher, U. Unmanned Aerial Systems (UAS)-Based Methods for Solar Induced Chlorophyll Fluorescence (SIF) Retrieval with Non-Imaging Spectrometers: State of the Art. Remote Sens. 2020 (12), 1624. https://doi.org/10.3390/rs12101624.

Rascher, U., Alonso, L., Burkhart, A., Cilia, C., Cogliati, S., Colombo, R., Damm, A., Drusch, M., Guanter, L., Hanus, J., Hyv¨arinen, T., Julitta, T., Jussila, J., Katajak, K., Kokkalis, P., Kraft, S., Kraska, T., Matveeva, M., Moreno, J., Muller, O., Panigada, C., Pikl, M., Pinto, F., Prey, L., Pude, R., Rossini, M., Schickling, A., Schurr, U., Schüttemeyer, D., Verrelst, J., Zemek, F., 2015. Sun-induced fluorescence - a new probe of photosynthesis: first maps from the imaging spectrometer HyPlant. Glob. Chang. Biol. 21, 4673–4684. https://doi.org/10.1111/gcb.13017.

Wang, N., Suomalainen, J., Bartholomeus, H., Kooistra, L., Masiliūnas, D., Clevers, J.P.W. 2021. Diurnal variation of sun-induced chlorophyll fluorescence of agricultural crops observed from a point-based spectrometer on a UAV. International Journal of Applied Earth Observation and Geoinformation. 96, 102276. https://doi.org/10.1016/j.jag.2020.102276.

Zarco-Tejada, P.J., González-Dugo, V., Berni, J.A.J. 2012. Fluorescence, temperature and narrow-band indices acquired from a UAV platform for water stress detection using a micro-hyperspectral imager and a thermal camera. Remote Sens. Environ. 117, 322–337. https://doi.org/10.1016/j.rse.2011.10.007.

Sea ice varies at a range of scales, and needs a range of approaches to understand its variability at each scale. Here we show an approach to investigating spatial relationships and variability of morphological features at an ‘ice floe’ scale (sub-meter to hundreds of meters) using imagery acquired with a small commercially-available RPA (Parrot ANAFI USA) and walked surveys using an electromagnetic ice thickness sounder (Geophex GEM2) and snow depth probe (Snowhydro Magnaprobe). With this ensemble we aim to cover a spatial scale appropriate for connecting point observations on the ground to larger-scale data, from airborne or spaceborne instruments - for example Synthetic Aperture Radar (SAR) imagery or helicopter-towed electromagnetic soundings and imagery. The basic data collection concept is to create a coarse grid of snow and sea ice properties using ground surveys, and fill in the detail with high resolution imagery and structure-from-motion derived terrain. Both data types help to cross-check each other - imagery is used to verify ice types in the ground survey, and ground data help to cross check (for example) sea ice thicknesses and snow depth modeled from imagery-derived terrain. This approach also lends itself to better understanding of how sampling sites are laid out, adding context about ‘where’ data of any kind were collected relative to each other. In turn, this assists interpretation of results, and promotes understanding of how well point sampling sites represent the local area. Using data collected in the Norwegian Nansen Legacy project on expeditions into the northern Barents Sea and Arctic Basin, our preliminary results show that we gain a much better contextual picture of how sea ice thickness, for example, is distributed at a sampling site by combining data from ground sampling and RPA surveys. We also show how different walking patterns can help avoid selective sampling bias when we aim to gather data representative of a larger site or region. We will present the methods used for data collection and coregistration, initial results, and a summary of how it went: what we would change in future, when to apply this approach, and how others could benefit from this style of lightweight, relatively uncomplicated approach using ‘off the shelf’ tools.