FUNDAMENTALS OF REMOTE SENSING FOR RISKS ASSESSMENT FRANÇOIS BECKER International Space University and University Louis Pasteur, Strasbourg, France; E-mail: becker@isu.isunet.edu Abstract. Remote sensing instruments aboard satellites observe the properties (e.g., intensity) of electromagnetic radiation (e.g., from the Sun) backscattered and/or emitted by the surface. Thus, they record some information about the surface. Different information is obtained in the visible/near infrared, thermal infrared and microwave parts of the spectrum. 1. Introduction As has been stated in the previous paper (Rycroft, 2000), environmental risk assessment is a very important task. Indicators of environmental risks which are reliable and measurable must be identified. These indicators must be measured and monitored in a reliable manner at an affordable cost, to estimate the risks and to locate the potential risks with a reliable probability of occurrence. Remote sensing may be a very useful tool to contribute to the achievement of these tasks. In fact, remote sensing allows a continuous monitoring of all the areas where environmental risks may occur, leading to the possibility of implementing warning systems. As an example, considering tornadoes in meteorology, the indicators could be the temperature and structure of the clouds, and their trajectory and velocity. These indicators can be monitored by remote sensing using radiometry from geostationary satellites which allow a global view every half an hour and, very soon, a quarter of an hour, by Meteosat Second Generation (MSG), and even shorter when focussed on a particular tornado, as is done by GOES. The aim of the STER course is to make risk managers aware of the potentialities of remote sensing and to give them the basic tools to identify the most relevant indicators which are measurable and which can be monitored by remote sensing. The course presents how remote sensing works, what can be measured by remote sensing, how to interpret and analyse the recorded data, and what are the constraints in the choice of an indicator measurable by remote sensing. In Section 2, we review the definition and principles of remote sensing; we shall introduce what can be called the remote sensing chain, with its components. Section 2 also introduces the characteristics of the electromagnetic radiation. Once the elements of the remote sensing set-up have been presented, we can consider the flow of information from the ground to the instrument, and the transformation of this information into data by the instruments. Different parameters come into play; Surveys in Geophysics 21: 127 131, 2000. 2001 Kluwer Academic Publishers. Printed in the Netherlands.
128 FRANÇOIS BECKER Figure 1. Remote sensing principles: the Earth is a complex system which is irradiated by natural (solar) and/or artificial (antenna) sources of electromagnetic radiation. The Earth responds to this excitation in two ways: (1) by the emission or re-emission of electromagnetic radiation (the EM response) in three major spectral domains, the Visible/Near InfraRed (V/NIR), Thermal InfraRed (TIR) and MicroWave (MW) parts of the electromagnetic spectrum, and (2) by changing thermodynamic state (thermal response) with the emission of TIR and MW electromagnetic radiation. the recorded signal, the acquisition parameters, the physical parameters measurable by remote sensing and the impact of the properties of these parameters on the images. 2. Remote Sensing Principles 2.1. DEFINITIONS As is seen in Figure 1, the Earth may be considered as a system which is irradiated by natural (solar) or artificial (antenna) sources of electromagnetic (EM) radiation. When remote sensing is performed using only natural sources of radiation, it is referred to as passive remote sensing ; when it includes artificial sources, it is referred to as active remote sensing. Remote sensing for risks assessment is therefore the combination of (i) a measurement by remote means of the temporal evolution of the EM response of the Earth s surface, (ii) the location and mapping of these responses, and (iii) the deduction from these responses of the temporal and spatial evolution of the relevant indicators, and their interpretation in terms of risks assessment. 2.2. REMOTE SENSING CHAIN An instrument is placed on board a satellite which moves around the Earth in a certain orbit. In order to measure the Earth from an altitude which is as constant as possible, this orbit is circular and is characterized by its radius and its inclination (the angle between the equatorial and orbit planes). Two main categories of orbit
FUNDAMENTALS OF REMOTE SENSING 129 are used: Sun synchronous (altitude of the order of 1000 km, inclination of the order of 95, the angle between the orbit plane and the Earth-Sun direction being constant), and geosynchronous (altitude of the order of 36,000 km, inclination zero, the angular velocity of the satellite and that of the Earth being the same, the footprint of the satellite being fixed with respect to the Earth). With remote sensing, three types of coverage which allow three particular types of analysis, characterized by three types of resolution, can be performed. (i) Areal cover Remote sensing from space allows the coverage of large areas in a very small time, thanks to the large Field of View (FoV) of the instruments (linear side from 60 to 3000 km) and the large velocity (of the order of 7 km/s) for a Sun-synchronous satellite (at an altitude of the order of 1000 km). This allows spatial and textural analysis. Such an analysis is limited by the spatial resolution of the instrument. The spatial frequencies of the image may then be analyzed using a spatial Fourier analysis. The dimension of the element of surface measured (the ground pixel ) varies from a few meters to 50 km, depending on the instantaneous Field of View of the instrument and the altitude. It can be difficult to locate the ground pixel if the satellite is not oriented as planned and if its attitude varies. These effects generate the need for geometrical corrections. (ii) Temporal cover With remote sensing the same element of the Earth s surface is observed periodically and therefore its temporal evolution can be studied. This allows a temporal analysis of the characteristics of the surface being measured. Such an analysis is characterized by the repeat time (or frequency of pass) and the time of overpass of the pixel. Depending on the type of orbit, this repeat time may vary from a fraction of an hour to a month, and the time of the pass may be chosen. For Sunsynchronous orbits, each element of the surface is overpassed at the same solar time, which gives almost identical conditions of solar irradiation. A temporal Fourier analysis may be performed to study the temporal variation of the characteristics of the surface. (iii) Electromagnetic characteristics Instruments on board the satellite allow the measurement of several characteristics of the EM radiation. The main characteristics of the EM radiation for remote sensing are: Intensity, spectrum, polarization, angular variation, as determined by the radiometric, spectral and angular resolutions of the instrument. Several types of analysis are possible, namely: Pattern recognition, content analysis (spectral, angular and polarization). Direction of propagation. This allows stereoscopy and therefore terrain elevation models to be constructed.
130 FRANÇOIS BECKER Time of propagation, which is essential for altimetry, whether by radar or laser. 2.3. FLOW OF INFORMATION (i) Information from the ground/pixel The element of the surface which is being measured is called a ground pixel; the acquisition system represents each measured element of the surface on the screen of a computer by a pixel (an abbreviation of picture element). This ground pixel is the location where the relevant surface information is somehow imprinted on the electromagnetic radiation emitted or re-emitted by this element of surface. This is mainly due to three mechanisms of the interaction of the EM radiation with this element of surface: reflection and scattering of the EM radiation coming from an external source (either the Sun or the radar or lidar source), and the direct emission of EM radiation. Part of the incoming EM radiation which is not reflected or scattered is absorbed by the surface, and contributes to the thermal response of the Earth. There are therefore four interaction mechanisms. In the case of passive remote sensing, reflection and scattering takes place mainly in the Visible and Near InfraRed bands, while emission takes place mainly in the Thermal InfraRed and MicroWave bands. (ii) Atmospheric perturbations and atmospheric windows The EM radiation reflected, scattered or emitted by the ground pixel propagates to the detector through the atmosphere and is modified as a result of three competing mechanisms: absorption by air molecules, scattering by molecules and aerosols, and emission by air molecules and aerosols. Absorption by the atmosphere reduces the intensity of the EM radiation; this depends strongly on the wavelength of the radiation. The atmosphere is opaque to EM radiation, except in three major spectral domains called the atmospheric windows: Visible and Near InfraRed (V/NIR): wavelengths of 0.4 2.5 µm corresponding to solar or lidar irradiation. Thermal InfraRed (TIR), comprising two subdomains: wavelength of 3.5 5 µm, corresponding to solar irradiation and surface emission, and 8 13 µm, corresponding to surface emission of lidar irradiation. MicroWave (MW): wavelengths of 1,000 600,000 µm, corresponding to surface emission or radar irradiation. Scattering has two effects: attenuation of the direct beam coming from the ground pixel and scattering into the detector of radiation coming from all other parts of the surface. This second effect is the source of the so-called upward atmospheric radiation. Scattering generally decreases as the wavelength increases. This is why clouds are opaque to visible, near infrared and infrared radiation, while they are transparent to microwaves. Radar and microwave remote sensing are therefore
FUNDAMENTALS OF REMOTE SENSING 131 all-weather, while optical remote sensing (V/NIR and TIR) works only when there are no clouds. References Rycroft, M.J.: 2000, From environmental risks, as problems, to Earth observations as solutions to these problems, Surveys in Geophysics 21, 115 125.