The Hyperspectral UAV (HyUAV) a novel UAV-based spectroscopy tool for environmental monitoring

Transcription

1 The Hyperspectral UAV (HyUAV) a novel UAV-based spectroscopy tool for environmental monitoring R. Garzonio 1, S. Cogliati 1, B. Di Mauro 1, A. Zanin 2, B. Tattarletti 2, F. Zacchello 2, P. Marras 2 and R. Colombo 1 1 Remote Sensing of Environmental Dynamics Laboratory, Earth and Environmental Sciences depart., University of Milan-Bicocca, Italy (r.garzonio1@campus.unimib.it) 2 Aermatica S.p.A. Gironico, Como, Italy UAVs in Environmental Research, July 2014, University of Exeter

2 Background & Objective Field spectroscopy is essential for remote sensing studies gaining valuable insights on Earth surface optical properties. Manual measurements represent the common way to collect near surface data. The UAV is a promising platform to support field surveys providing autonomous, multi-scale and low-cost measurements. Development and testing of HyUAV: a novel UAV based system for field spectroscopy

3 The Hyperspectral UAV - HyUAV PLATFORM: ANTEOS - Unmanned Aerial Vehicle Four-rotor platform with hovering capability, maximum payload of 2 Kg and flight time of 20 min GPS/IMU Radio connection to the ground control station Hyperspectral PAYLOAD: + RGB camera (Canon S100) + USB4000-VIS-NIR spectrometer On board controlled Simultaneous measurements

4 Payload - Spectrometer Ocean Optics USB4000-VIS-NIR Miniature Fiber Optic Spectrometer UAV support Physical Dimensions: 89.1 mm x 63.3 mm x 34.4 mm Weight: 190 grams Detector Specifications Detector: Toshiba TCD1304AP CCD Pixels: 3648 pixels Spectroscopic Wavelength range: nm Optical resolution: ~1.5 nm FWHM A/D resolution: 16 bit Optical fiber

5 Flight height Payload - Entrance Optic Receptor Footprint (m) Optical fiber Holder for Neutral Density Filters to avoid saturation for high intensity light levels Shutter Variable length fore-optics with Iris Diaphragms to adapt the spectrometer field of view (FOV) Shutter with Integrated Controller to allow a regular in-flight dark-current measurements FOV 15 Sampled areas (footprint) as function of FOV and flight altitudes FOV degree Footprint UAVs in Environmental Research, July 2014, University Flight of height Exeter (m)

6 Payload Data Collection Basic spectrometer operations: Way-points scheme Integration Time Optimisation Dark Current Acquisition Target Spectrum Acquisition Data collection modes: Way-points: user defines a set of way-points and payload operations Transect scheme Transect: continuous collection of measurements along a defined path

7 A dedicated software developed for mission planning The user is able to: Load base maps (i.e. Google map, Bing aerial map) Define the accessible FLIGHT SPACE Create automatically a FLIGHT PATH in order to generate a series of waypoints defining the best path covering a horizontal/vertical surface Create a MISSION PATH to perform the following operations at each way point : Integration time optimization Dark current acquisition Software - Mission planning Target spectrum acquisition

8 DN Software Mission Control In-flight data collection: autonomous based on pre-defined mission planning manual UAV and USB are manually operated Spectrometer management RGB Camera Image Graphical User Interface Flight parameters

9 Software Data Processing Different software packages are under development for systematic processing of collected RGB and hyperspectral data: 1. Optical data and navigation data matching (GPS, roll, pitch, yaw): Geolocation of data Data quality evaluation based on navigation data 2. Hyperspectral data Dark Current (DC) subtraction Non linearity correction Calibration to spectral radiance Estimation of surface reflectance based on Meroni et al., RGB images Image mosaic Textured 3D model (sfm algorithm) Reconstruction of Digital Surface Model (DSM) Agisoft PhotoScan MicMac

10 Laboratory test A dedicated laboratory test-bed was used to simulate in-flight conditions: to evaluate the impact of UAV-rotors vibrations on spectral data to calibrate USB4000 Analyses were performed to consider the following features: Geometric (FOV dimension and position within RGB image) Radiometric Spectral

11 Laboratory test FOV Lamp Halogen lamp feeds the fore-optic (output to the spectrometer) projecting the spectrometer FOV on a levelled surface RGB images are collected in dark conditions and on a reference chessboard (to extract pixel dimension) FOV diameter and centre position within RGB image are estimated using image based algorithms Impact of vibrations (in-flight conditions) resulted negligible FOV Chessboard Dark condition Footprint

12 ASD reference spectrometer Laboratory test Radiometric USB4000 measurements over reference Spectralon panel before, during and after repeated flight simulations Simultaneous measurements collected with ASD FieldSpec (reference) monitoring changes of lamp intensity USB 4000 Radiometric signal shows limited variations that are not related to the switch-on/off of the rotors Data collected used for radiometric calibration of USB (by cross-comparison with ASD) Source lamp Spectralon

13 Laboratory test Spectral USB 4000 A spectral calibration lamp (Ocean Optics, Cal-2000) pointed through the entrance fore-optics Spectral peaks extracted from data collected In-flight wavelength position is not affected by platform vibrations (max deviation of 0.1 nm) Data used for the spectral calibration Spectral calibration lamp

14 Flight test Overview A flight test was done in order to verify the efficiency of the system and the accuracy of the collected data. The study area is a small pit with different surfaces (gravel, grassland, trees). Data from the different targets were collected from UAV during the flight. Each target reflectance from UAV was compared with that collected simultaneously by ASD FieldSpec spectrometer at ground.

15 Flight test Reflectance (1/2) The reflectance was calculated for the different targets using the formula: ρ(λ)= L(λ) Target L(λ) Solar L solar estimation flying over the calibrated tarp (6x6 m) before and after target measurements The graphs show good agreement between UAV and ASD reflectance for three different targets

16 Light gravel Flight test Reflectance (2/2) Grassland Trees Gray gravel Dark gravel

17 Flight test RGB products Example of products derived from the RGB images: Orthomosaic image Digital Surface Model (DSM) Textured 3D model Orthomosaic image Textured 3D model

18 Conclusions & perspectives A HyUAV has been designed and developed providing systematic collection of RGB images and VNIR hyperspectral data for field spectroscopy surveys; The dedicated software packages have been developed for: i) mission planning; ii) data collection; iii) processing and geolocation of hyperspectral data and RGB images; The packages are planned to be available online soon in order to be helpful to similar UAV based system; The laboratory test on USB4000 showed that typical platform vibrations do not affect the spectral and radiometric quality in a significant way; The reliability of collected data has been tested in-flight conditions; the comparison between reflectance signatures estimated from UAV data and ground truth measured with ASD FieldSpec showed a good agreement; The UAV platform is demonstrating to be a helpful technology supporting common field surveys; further activities will regard the use of the system for different environmental applications (i.e. snow albedo monitoring, agricultural and forest applications and in-land fresh waters);

19 Acknowledgments SINOPIAE project: «System-prototype for multi-source integrative observation techniques of multispectral satellite, aircraft and ground data for multi-scale monitoring of the variations of environmental indicators related to atmospheric constituents and energy dispersion» The Project of Interest NextData: A national system for the retrieval, storage, access and diffusion of environmental and climate data from mountain and marine areas. LTDA Research group

20 Thanks for your kind attention UAVs in Environmental Research, July 2014, University of Exeter Roberto Garzonio