Drone-borne hyperspectral imaging for sustainable mineral exploration

By Sandra Lorenz

Drone-borne hyperspectral imaging for sustainable mineral exploration

Drone-borne hyperspectral imaging is a new and promising technique for fast and precise acquisition, as well as delivery of high-resolution hyperspectral data to a large variety of end-users. Drones can overcome the scale gap between field and air-borne remote sensing, thus providing high-resolution and multi-temporal data. They are easy to use, flexible and deliver data within cm-scale resolution. So far, however, drone-borne imagery has prominently and successfully been almost solely used in precision agriculture and photogrammetry. Drone technology currently mainly relies on structure-from-motion photogrammetry, aerial photography and agricultural monitoring. Recently, a few hyperspectral sensors became available for drones, but complex geometric and radiometric effects complicate their use for geology-related studies.

Hyperspectral sensors have become a key tool for a large range of applications in remote sensing and are now widely used in geology, mineral mapping and exploration. During the last few years, lightweight hyperspectral imaging (HSI) sensors have been increasingly developed for use on unmanned aerial systems (UAS). These drone-borne sensors are able to close the gap between field- and air- or space-borne data and provide small-scale high-resolution hyperspectral imagery. The acquisition of image data with UAS is fast, easy, targeted and without the need of extensive time- and cost-consuming planning. It is mostly independent of cloud cover conditions and is even applicable in barely accessible areas. The heavily decreased influence of the atmosphere obviates the need for the often difficult and complex atmospheric correction. Nevertheless, geometric and radiometric correction of drone-borne data is challenging mainly due to small, unpredictable platform shifts and the influence of the micro-topography. Thus, high resolution digital elevation models are required for correction. Common and established workflows for the pre-processing of aerial hyperspectral scanner data are not or only partly applicable.

In general, two types of aerial platforms or UAS are available:
(1) fixed-wing systems with the advantage of long flight endurance and fast cruising speeds. So, large areas can be captured within one flight, but their disadvantage is a limited payload capacity. The greater the payload, the bigger is the wingspan. In this context, especially take-off and landing become thrilling with expensive and/or sensitive equipment on-board. Furthermore, a fast shutter speed is needed to get high quality image data.
(2) Multi-copters have the disadvantage of limited range and flight time; however, they can carry heavier payloads. Due to their hovering capacities, they are more stable at the point of image acquisition, and also, landing is more controlled (especially important with expensive or sensitive payloads).

Sensors in the visible and near-infrared spectral range (VNIR) of the electromagnetic spectrum have been preferably used on UAS due to their low weight and size. In contrast, short wave infrared (SWIR) sensors exceed the payload capacity of most lightweight aerial platforms. Larger UAS with higher payload complicate the handling in remote areas and increase the difficulty to obtain flight permission by the local authorities in most countries. The most prominent material absorption features in the VNIR spectral range originate from green vegetation, while only a few mineral groups, mainly iron oxides and some rare earth elements, show typical absorption features. Those mineralogical features can be quite shallow, and their truthful determination depends mainly on an accurate data correction. By contrast, the occurrence and health analysis of vegetation is easily determinable in the VNIR range even with a low spectral resolution. This is a main reason why the development in drone-borne HSI has been mainly applied in and for agricultural and environmental monitoring. Thus, the easy, fast and reliable acquisition and in-time interpretation of drone-borne data increased the accuracy and abilities of precision agriculture. Numerous publications describe the assessment and processing of multispectral and hyperspectral drone-borne data in modern agriculture. They focus mainly on sensor calibration, photogrammetry and illumination correction between single mosaic images. Georeferencing is commonly performed using ground control points (GCPs). Topographic correction, which is essential for geological targets, is not applied for the mostly flat, uniform and smooth agricultural areas. An additional correction to reflectance and a high signal to noise ratio (SNR) are not necessary for most applications in precision agriculture and environmental research, as the commonly-used vegetation indexes can be determined even with low spectral resolution or high noise. Thus, correction algorithms used for vegetation monitoring are not applicable for geological targets. The great diversity of targets in terms of terrain, size and spectral signatures demands a finer spectral resolution, a precise topographic correction and a higher SNR. As most of the minerals show no or weak absorption features in VNIR range, even subtle spectral differences can be important for interpretation and raise the need for a careful data processing. The application of drone-borne data for geological issues therefore is extremely rare and mainly limited to UAV-based photogrammetry and 3D photogrammetry using RGB-sensors for structural geology or landslide mapping. Beside RGB sensors, drone-borne thermal cameras have been used in a first attempt to observe the temperature of mud volcanoes.

This is an excerpt of the journal article: The Need for Accurate Geometric and Radiometric Corrections of Drone-Borne Hyperspectral Data for Mineral Exploration: MEPHySTo—A Toolbox for Pre-Processing Drone-Borne Hyperspectral Data, by Sandra Jakob, Robert Zimmermann and Richard Gloaguen. Published: January 18., 2017 in Remote Sens. 9(1), 88, DOI: 10.3390/rs9010088 under a Creative Commons Attribution License (CC BY 4.0). 

Sandra Lorenz
Researcher

Sandra Lorenz is currently working with the Helmholtz-Institute for Resource Technology, Helmholtz-Zentrum Dresden-Rossendorf in Freiberg, Germany