PHYSICAL RELATIONS CONCERNING HYPERSPECTRAL LINESCANNERS DEPLOYED AT A MULTIROTOR B. Altena a, *, L. Cockx a , T. Goedemé a,b a EAVISE, Faculty of Industrial Engineering, KU Leuven, Sint-Katelijne-Waver, Belgium b VISICS, Faculty of Electrical Engineering, KU Leuven, Heverlee, Belgium KEY WORDS: UAV, Hyperspectral imaging, Sensor modelling, Geometric distortion sensitivity. ABSTRACT: Unmanned aerial vehicles, like a quadrotor, have recently gained interest as alternative observation platform. Equipped with a lightweight imaging spectrograph, these configurations can sense an area in the order of several hectares. This study models such an observation system. It highlights the geometric and radiometric relations of importance for an imaging spectrograph, based on diffracting optics for close-range configurations. These relations are used to model such a system, configured from off-the-shelf components. This most favourable configuration results in a product with a spatial resolution just below 5cm, a swath of 36 meter and flight speed of 4 m/s. Furthermore, an assessment focussing on the “cosine fourth law” and the sensitivity of ground truth in respect to projection. These phenomena are of importance for close-range remote sensing solutions. Its concluded with remarks on georeferencing and a possible solution for reduction of the mentioned errors is given through inclusion of second optical sensor. * Corresponding author: Bas Altena (b.altena@spectrocopter.com) 1. INTRODUCTION Airborne hyperspectral remote sensing is a matured technology and has proven its usability in many applications. For example, in ecological monitoring remote sensing can accurately produce quantitative conservation status assessments (Van den Borre et al., 2011). These systems have even the potential to be of great aid for fine-scaled ecological indicators; their measurements can extend manual observations or reduces within class variability (Spanhove et al., 2012). Though these configurations have their downsides; spaceborne imaging has a revisiting time of several days and in addition is plagued by cloud cover. For airborne campaigns, flight time is expensive and limited. Moreover, it may not be possible to extract all wanted information from the data, as monitoring of sparse vegetation, or estimation of coverage on plant level. If such phenomena need to be detected and assessed, an observation system is needed with an even higher spatial resolution. Fortunately, through the advances in microcontrollers it is now possible to use unmanned aerial vehicles. This may be a more economic and mobile alternative than a traditional hyperspectral imaging campaign. The first construction of compact aerial hyperspectral imaging systems became implemented a decade ago (Sigernes et al., 2000; Yang et al., 2003). Through advances in unmanned aerial platforms, these systems have gained more attention. It resulted in several hyperspectral sensor solutions, for example, (Oppelt & Mauser, 2007; Zarco-Tejada et al., 2011; Abd-Elrahman et al., 2012; Hruska et al., 2012). For geometric correction, these observation systems only use the sensor readings of the IMU and GNSS, or additional coarse terrain information is included. However, due to turbulence of the platform during flight and error propagation within flight path calculation (Kalman filtering) the hyperspectral measurements become misaligned on a scan-line to scan-line level. This study examines the potential accuracy and precision which is achievable for a hyperspectral line scanner on board a quadrotor platform. A similar study is done for an aerial platform (Bostater et al., 2011). Still only variations in speed and roll were considered. Our modelling study will include a more advanced assessment to give a clear insight in the close-range mapping capabilities of such a system. It is a combined assessment of spectral and spatial properties, as these are entwined for these close range observation systems. 2. SYSTEM CONFIGURATION 2.1 Mapping platform For applications stretching several hectares a mobile mapping platform is most beneficial. Solutions that rely on natural power sources, such as, balloons, kites and alike, are not considered due to their dependence on weather and immovability on terrain. The resulting selection of remotely piloted aircraft systems (RPAS) can further be simplified into fixed-wing, multi-rotor or a hybrid case. For this study a multirotor system is modelled. This system has a more dynamic character; both speed and elevation have a broader range of choice. A fixed wing solution is designed for a certain speed at a high elevation, starting 100 meters above the terrain. Consequently the application should adapt to the instrument, while a multirotor adapts to the phenomena of interest. Furthermore, most multirotors have a stabilization rig underneath, this helps with the aiming of the instrument to nadir and partly corrects for wind