|GeoImaging & GeoInformatics|
|Your location:||GIGI||>||Modules||>||Remote Sensing and Digital Image Analysis|
|Introduction||Objectives||Getting Started||Essential Learning||Hungry Minds||Summary|
In this topic you have described the source-target-sensor configuration of remote sensing systems and identified three sets of techniques for digital image analysis, namely pre-processing, image enhancement and information extraction. Let us briefly review these elements in terms of their contribution to problem-solving with remote sensing whilst remebering that studies in remote sensing can adopt any one of three methodologies - inductive, deductive or technologic.
The energy available from the sun is described by the electromagnetic spectrum. Not all of this energy is available for use by remote sensing technologies. The atmosphere absorbs large portions of the spectrum leaving only specific 'bands' of energy, which are able to penetrate to the Earth's surface. Equally, the amount of energy available from the sun is not evenly distributed by wavelength. In fact, the amount of energy emitted by the sun reaches a peak in the range 0.4 Ám to 0.7 Ám, better known as visible wavelengths.
What does this mean for remote sensing? Knowing the amount of solar energy available tells us if there is enough energy available for a passive sensor to function or if an active sensor (one which provides its own energy source) is required. Sensors operating in the visible, near-infrared and thermal infrared regions of the electromagnetic spectrum can usually function effectively by using the solar energy that illuminates the Earth's surface. Active sensors, such as Radarsat which operates in the microwave portion of the spectrum, create their own energy and are, therefore, not dependent on solar energy from the sun which means they can operate at any time of day or night. Measurements at microwave wavelengths are also less affected by atmospheric absorption.
Energy Interaction and Target Detection
Surfaces can often be characterized in terms of their spectral response, or how they appear at a particular wavelength. For example, our eyes observe healthy vegetation as green, due to the greater absorption of blue and red energy compared to green energy. In the near- infrared part of the spectrum, an area commonly used in remote sensing to assess the health or vigour of vegetation, healthy plant leaves are highly reflective. Unfortunately, our eyes are not sensitive enough to electromagnetic radiation at these wavelengths.
Specific objects of interest, commonly referred to as targets in remote sensing jargon, can be identified and separated on the basis of their spectral response (among other things) if there is enough contrast with the background of the scene. A complex set of factors including size, shape, and contrast affect the visibility and detectability of targets. Image interpreters are most interested in those characteristics that differentiate a target from its' surroundings.
Exploiting these benefits through digital image analysis is best realised by building and inverting a model for image interpretation and then running the inverted model. This can be a conceptual model - for example, the choice of three wavebands to optimise the information presented in a colour composite - or a quantitive model such as those that model vegetation canopy reflectance. Models first need to be parameterised - what are the factors that control the signal recorded by a sensor? Knowing what factors affect the characteristics of an image supports forward modelling. Forward modelling is used to assess the likely effect of a change in one or more factor on the signal recorded. An analyst inverts a model to estimate those factors from variations in image characteristics. So far we have not explored this issue in any great depth and you will need to better understand the range of models available before choosing an appropriate model to apply. To apply any model you must also appreciate the need to correct and calibrate an image.
These outputs of remote sensing and digital image analysis can contribute to Earth observation in several forms:
As with anything worthwhile, planning makes all the difference. In the use of technologies such as remote sensing and/or Geographic Information Systems (GIS), effective project planning is critical due in part to the relative newness of operational uses.
Remotely-sensed data are a natural input to GIS because it provides spatially consistent information over large areas. Remote sensing and GIS are among many tools available to resource management professionals today. These tools vary widely in their sophistication, cost, effectiveness, availability, and familiarity. These factors add to the potential risks associated with successful implementation. Planning must be systematic and realistic and must focus on problem solving rather than a technology push. With proper planning, remote sensing can be a powerful tool.