Going hyperspectral: the 'unseen' captured?
Jusoff, Kamaruzaman (2010) Going hyperspectral: the 'unseen' captured?
All objects, name them soil, water, trees, vegetation, structures, metals, paints or fabrics, create a unique spectral fingerprint. A sensor determines these fingerprints by measuring reflected light, most of which registers in wavelengths, or bands, invisible to humans. This is what the crime scene investigation (CSI) television programs have popularized how DNA or fingerprints can be used to solve crimes. Similarly, forest CSI of “seeing” the trees in the deep high mountain tropical forest is now a major focus in the air and spaceborne hyperspectral sensing technology and in other different applications such as agriculture, environment, geology, transportation, security, and several others. The availability of sub-meter resolution colour imagery from satellites coupled with internet based services like Google Earth and Microsoft Virtual Earth have resulted in an enormous interest in remote sensing among the general public. The ability to see one’s home or familiar landmarks in an image taken from hundreds of kilometers above the earth elicits wonder and awe. Deciding where, when, what and how to sense or measure the DNA of individual trees from the air or space is a crucial question in the sustainable development and management of our Malaysian tropical forest ecosystems. However, to monitor, quantify, map and understand the content and nature of our forest, one would ideally like to monitor it everywhere and all the time too. This is impossible, and consequently, forest engineers must select relatively very high to high near to real time resolution sensors with the ability to transcend boundaries, capabilities, features and interfacing realms for such measurement. The dynamic interplay of these elements is precisely coordinated by signaling networks that orchestrate their interactions. High-throughput experimental and analytical techniques now provide forest engineers with incredibly rich and potentially revealing datasets from both air and spaceborne hyperspectral sensors (also known as imaging spectrometers). However, it is impossible to exhaustively explore the full experimental and operational hyperspectral sensors available in the market out there and so forest engineers must judiciously choose which one is the best to perform and fulfill their project objectives and missions. The complexity and high-dimensionality of these systems makes it incredibly difficult for forest engineers and other users alone to manage and optimize sensing processes. In order to add or derive value from a hyperspectral remotely sensed image several factors such as resolution, swath, and signal to noise ratio, amongst others need to be considered. A grand challenge for the forest engineer’s scientific discovery in the 21st Century is therefore, to devise very high real-time ultra-spatial and spectral air and space borne sensors that automatically measure and adapt sensing operations in large-scale and economical systems with the unseen captured. This lecture therefore focuses on the emerging theory, origin of the hyperspectral sensors, research, practice, limitations and identifies future challenge and outlook of hyperspectral sensing systems in the quest towards a sustainable Malaysian forestry context and other different applications to capture the “unseen”. It is quite certain that advances in hyperspectral remote sensing and more sophisticated analytical methods will resolve any “unseen” issues in time with the best approach of transcending boundaries and interfacing remote sensing data with precise information from the field plots. Unfortunately, as a relatively new analytical technique, the full potential of air and spaceborne hyperspectral imaging has not yet been realized in Malaysia
Repository Staff Only: Edit item detail