DE10113330A1 - Remote multi-sensor process comprises capturing bush and forest fire gas and particle concentration data for analysis - Google Patents

Abstract

A remote infra-red measuring process determines the seat of a fire using the near infra-red and visible spectrum. A multi-sensor process simultaneously captures data including temperature, gas and particle concentration for analysis. An Independent claim is also included for a multi-sensor system comprising a combination of a bi-spectral image generating infra-red sensor operating in the thermal infra-red range, a sensor (VIS/NIR-CCD) operating in the visible and infra-red range, both with high spatial resolution and a spectral high-resolution IR-Fourier transform spectrometer (IRFTS) operating in the thermal infra red range with relatively low resolution.

Description

The invention relates to a method for optical telemetry fire scenes using IR sensors.

The invention also relates to a device for Execution of the procedure.

For many cases, such as when using fire weir and disaster control, environmental protection up to global climate research, is a localization of Brandher the, the determination of the temperature, the area of a fire and the composition and amount of trace gases generated and smoke particles of great importance. With biomass fires of all kinds, for example forest, grass and savanna fires, Fires from coal seams, oil fires or natural disasters, such as Volcanic eruptions, numerous gaseous species and often smoke particles are also produced in large quantities.

The resulting clouds of smoke are in the visual spectrum rich optically largely opaque and so make one Localization of the source of the fire by observation with the human eye or camera impossible. The burns processes often show a very inhomogeneous space structure in temperature and gas and particle Emissions, the quantitative determination with selective in situ measurement methods, for example probes, complicate and make inaccurate. For this is a non-contact optical Telemetry required the entire fire area visualize with good spatial resolution and simul tan can analyze.

The clouds of smoke from fires contain a very large number small particles called aerosols, the diameter of which is 0.05 to 1 µm is significantly smaller than the wavelength of the thermal infrared range from 3 to 14 µm. Therefore, in  Contrary to the visual spectral range, the smoke clouds in the thermal infrared range predominantly transparent. Therefore can in the two atmospheric windows of the infrared from 3 to 5 µm and from 8 to 14 µm thermal imaging devices to the location, temperature, through clouds of smoke and the extent of fire sources and smoldering fires and to depict it pictorially. The remaining absorption and Scattering properties of the aerosols in the atmospheric fen stern can be used to improve the optical properties the aerosols, d. H. the optical thickness and the size distribution to measure in the column.

Numerous combustion gases, such as CO ₂ , CO, NO, NO ₂ , NH ₃ , CH ₄ , HNO ₃ etc., have a number of so-called molecular bands in the infrared spectral range, which consist of countless individual spectral lines of the molecules. These spectral lines of the different molecular species show pronounced spectral absorption and emission properties depending on the atmospheric pressure, the concentration of the gas species and the gas temperature and can therefore be used for the quantitative determination of the composition and concentration of the combustion gases and the gas temperature by means of spectral analysis.

For the simulation of IR radiation from atmospheric gases and Aerosols and for analysis of IR remote measurement data from the sensors there are suitable radiation transport models, databases and Inversion algorithms available. The remote measurement of Smoke particles (aerosols) from the ground are made using an egg a network of automatic sky-scanning son internal photometers (AERONET) that operate at eight wavelengths (340 to 1020 nm) in the visible and near infrared spectral range direct solar irradiance and the angular distribution  the brightness of the sky in a cloudless sky or under Be measure conditions with low clouds.

From this data the spectral optical thickness of the Aerosols, the aerosol size distribution and the scattering phases function of the particles with a fast radiation transfer Calculate inversion code (B. N. Holben, T. F. Eck, I. Slutsker, D. Tanre, J.P. Buis, A. Setzer, E. Vermote, J.A. Reagan, Y. J. Kaufman, T. Nakajima, F. Lavenu, I. Jankowiak and A. Smirnov: "AERONET - a federated instrument network and data archive for aerosol characterization ", remote sensing environment, 66: 1-16, 1998).

The optical thickness of aerosol and gas emissions and the Temperature of biomass fires are monitored from the satellite of an image stationed at an orbit height of 705 km the spectrometer EOS-MODIS (Earth Observing System - mo derate resolution imaging spectroradiometer). MODIS we you. a. specifically for global fire observation and fire analysis used. The spatial resolution of the total 36 spectral channels from MODIS is 250 m in the visible, in near IR 500 m and in thermal IR 1 km.

The corresponding angular resolution is 0.7 mrad in the near IR and 1.4 mrad in thermal IR (M.D. King, Y.J. Kaufman, W. P. Menzel, and D. Tanre: "Remote sensing of cloud, aerosol, and water vapor properties from the Moderate Resolution Ima went Spectrometer (MODIS) ", IEEE Trans. Geosci. Remote Sens., 30: 2-27, 1992 and Y. J. Kaufman, D. Tanre, L. Remer, E. Vermote, A. Chu and B. N. Holben: "Remote sensing of tropospheric aerosol from EOS-MODIS over land using dark targets and dynamic aero sol models ", J. Geophys. Res., 102: 17051-17067, 1997).  

The infrared channels at MODIS are used to detect fires and to determine their position and size as well as the surface temperature of the fires; the spectral resolution is in the range from 30 to 130 cm ^-1 . The surface temperature of the fire is determined using numerical iteration technology from measurements of the radiance of a fire pixel at 4 µm and 11 µm and the blackbody temperature of the background.

The optical thickness of clouds of smoke is derived from the backscatter hold the sunlight on aerosols in the visible spectrum area determined. To determine the optical thickness of Ae Rosolas over land are dark ground surfaces as a background and needed a dynamic aerosol model that turned on the sky measurements with the AERONET solar photometers rests.

To help with the optical thickness of aerosols from space to be able to determine the reflected radiation is Knowledge of the reflectivity of the underlying surface necessary. The reflectivity of the surface of dark objects becomes the spectral relationship in the visible (0.47 µm and 0.66 µm) and in the short-wave infrared (2.13 µm) estimated.

From the optical thickness of the aerosols determined in this way, the column densities of the emitted combustion gases such as CO ₂ , CO, NO, NO ₂ etc. are directly inferred by using known ratios of trace gas to particle emission rates from measurements. The strict correlation between the profile of smoke particles and the profile of CO and CO ₂ is primarily used. The emissions of other trace gases are determined from the corresponding measured trace gases to CO or CO ₂ emission rate. The smoke particles are thus used as tracers for the emissions of trace gases.

Concentrations of trace gases in the atmosphere are measured with op telemetry using a spectrally high resolution directly determines the infrared spectrometer, its spectral Resolution comparable or smaller than the half must be the width of the spectral lines under consideration. The effec tive field of view (FOV) based on interferometer technology ba However, high resolution infrared spectrometer is because of the spectral resolution corresponding to the Jacquinot Criterion limited. With the currently used for this satellite-based spectrometer IMG, the FOV is 0.6 °; this FOV corresponds to one on the earth's surface pro ejected horizontal spatial resolution of 8 km from egg a satellite flight altitude of 800 km.

A known aircraft-borne IR Fourier transform spectrometer AES (Airborne Emission Spectrometer) has a spectral resolution of 0.07 cm ^-1 in the spectral range from 650 to 4250 cm ^-1 (2.35-15.4 µm). Numerous gas species such as water vapor, carbon dioxide, carbon monoxide, ammonia, methanol, formaldehyde and ethylene can be used to determine the column densities and the temperature of the air and the soil from the measured spectra. The gas species detected in the spectra are quantified using a physical model for the fire; A distinction is made between four combustion phases and scenarios, namely flames, smoldering fires, smoke over an undisturbed background and undisturbed background.

The atmospheric temperature and the molecular concentrations from the measured IR spectra by inversion of Radiation transport equations recovered. This inversion process  are mathematically calculated using the known Me least squares method. The problem with The least squares method is generally not linearity. This problem therefore becomes iterative with standard techniques (e.g. Gauss-Newton or Levenberg-Mar quardt algorithm) by an appropriate adjustment of the Pa resolved parameters that describe the state of the atmosphere (e.g. temperature, gas concentrations, etc.) until the Match of a calculated and a measured spec is sufficiently good.

In the case of recovery for vertical observation geometry in the atmosphere reacts to the method of small Most squares are extremely sensitive to disturbances in the measurement data. This difficulty of direct inversion techniques is one As a result of the fact that the measured data are not sufficient can provide information about a solution you are looking for. With the help of regularization methods, some additional brought in information, for example a priori-in formations of meteorological mean values or smoothness conditions.

With the imaging spectroradiometers currently available, sufficient spatial resolution and spectral coverage of the visible and infrared range can be achieved, but far from the spectral resolution of 0.1 cm ^-1 necessary for the spectral analysis of gas emissions. The available interferometric IR spectrometers have the required high spectral resolution, but not sufficient spatial resolution or angular resolution.

The disadvantages of using individual sensors of the type of a spectroradiometer, a high-resolution interferometric spectrometer and a thermal imaging device are as follows. The method used for the imaging spectrometer MODIS requires dark ground surfaces as background to determine the optical thickness of aerosols over land, so that it can only be used in daylight. Furthermore, MODIS has too low a spectral resolution, which in thermal infrared (30 to 130 cm ^-1 ) is only sufficient to determine the average temperature of the atmosphere and of clouds as well as the temperature of surfaces.

Because of the low spectral resolution of the channels of spectroradiometers, individual spectral lines of molecular spectra cannot be measured, so that it is not possible to recover vertical profiles of the atmospheric temperature and molecular concentrations, since this requires a spectral resolution of at least 0.1 cm ^{. 1 is} required. Since it can be concluded indirectly from the optical thickness of the aerosols that the column densities of emitted combustion gases, such as CO ₂ , CO, NO, NO ₂ etc., a direct determination of the column densities or concentrations of the gases and their temperature from infrared measurements is not possible ,

With spectrally high resolution spectrometers the atmo spherical temperature and trace gas concentrations be true. Based on the dimensions of typical fires scenes, however, their spatial resolution is too low. On project onto a fire, for example, on the surface of the earth The field of view of a spectrometer is generally a lot larger than the horizontal spatial extent of the view fire.

Although often a large inhomo within the fire area genicity in temperature, particles and gas concentrations trations can only be roughly spatial with spectrometers  averaged values of temperature and concentrations of Trace gases and aerosols can be determined. In extreme cases it can an interfering background signal from areas outside the Make a spectral analysis impossible.

With the use of spectrally broadband thermal imaging devices can in the atmospheric windows in the thermal IR range from 3 to 14 µm the temperature distribution of solid or liquid surfaces in the fire area with high spatial and radiometric resolution can be determined. The temperature of gases can thus be in suitably chosen spectral ranges (Micro windows), the corresponding absorption bands of the gases included, but only approximately averaged over all gas layers of fire and the atmosphere of the fire become. The spectral resolution is for a determination of the Concentrations of trace gases are not sufficient.

The object of the invention is therefore a non-contact opti telemetry method and a device for carrying it out this method to provide both spatial and spectrally high-resolution image data, especially in thermal Infrared from a sufficiently wide optical face field and analyze it simultaneously.

According to the invention, this object is achieved with the features of Claim 1 solved.

A device for performing this method is in the An pronounced 12.

Advantageous further developments are the subject of claim 1 or 12 claims directly or indirectly related chen.  

To simultaneously the temperature, gas and particle emissions and their concentrations and their spatial distribution over the entire spatial extent of a fire not only to measure, but also to determine simultaneously and directly, According to the invention, an optical sensor in the form of a imaging radiometer (thermal imaging device) and a spectral high-resolution radiometer (IR Fourier transform spectra ter (IRFTS)) linked in a suitable manner. The given up Measured data taken are immediately real-time of a multi subjected to sensory analysis.

For simultaneous measurement of temperature, gas and particle Emissions over the entire spatial extent of a fire are therefore an optical sensor and a telemetry in thermal infrared provided that the capabilities of a imaging radiometer (thermal imaging device) with a spectral high-resolution radiometer (IR spectrometer) in a suitable Form combined.

In addition to an imaging radiometer, an imaging IR camera is expediently a bispectral imaging CCD camera in the visible spectral range and the near IR area (VIS / NIR-CCD), with the same field of view Optics, but higher spatial resolution than a bispectral IR spectrometer (BIRS) used.

According to the invention, the visual fields are the Senso ren adjusted so that the field of view of the spectrome ters IRFTS from the total field of view of the imaging sensors BIRS and VIS / NIR is always completely covered; in this connection is the field of view of the imaging sensors at least equal to or larger than the field of view of the spectrometer. With the imaging sensors in a sufficiently large Field of view covers the entire fire or large areas thereof,  to get an overview, while with the In principle, spectrometers only in a smaller angular range is measured and analyzed.

As a compromise and for practical reasons, a Ver ratio of the diameter of the visual field of imaging Sensors to the diameter of the spectrometer from 10 to 1 be endures. At the same time, the number is also necessary Measuring steps with the spectrometer to cover the entire field of view of the imaging sensors.

According to the invention, the imaging sensors are used first in the larger field of vision the fire scene is visualized and measured the temperature distribution. After that, be inside this large field of vision one or more of interest Sub-areas selected, for example areas with high Temperatures or strong gas and smoke development; hereupon the spectrometer with the smaller field of view is output directed.

In this smaller field of view there is then a detailed one Analysis of the fire with all sensors carried out simultaneously leads. It is therefore necessary that the visual field of the Spek trometer within the field of view of the imaging senso can be aligned as desired and defined (Pointing).

Because the required size of the visual field of the imaging Sensors by the selected measuring distance or the choice of Sensor platform and the typical dimensions of large fires is determined, there are basically three sensor platforms Consideration: satellite, plane and vehicle (ground use).  

Since a bispectral imaging IR sensor BIRS works spectrally broadband in the two atmospheric IR windows from 3.5 to 4.1 µm (MIR) and 8 to 12 µm (TIR), measuring the IR radiation in the two atmospheric The temperature distribution of surfaces and the emission and continuum absorption of gases and particles can be determined. Since the IR Fourier transform spectrometer is sensitive in the thermal IR spectral range from 3.0 to 15 µm and spectrally covers numerous absorption bands of atmospheric and combustion gases, as well as the atmospheric IR windows of the bispectral imaging IR sensor BIRS, the Fourier Spectrometer, the emitted spectral radiance of all radiation sources in the field of view, in particular the emission and absorption lines of the gases, are measured with a high spectral resolution of 0.1 cm ^-1 . The gas temperature and the concentrations of numerous gas species can then be determined from the line spectra of the gases.

Thus, on the one hand, the sensors provided results obtained from the individual sensors, such as the surface temperature using a bispectral imaging IR sensor BIRS and the gas temperature and gas concentrations with the IR Fourier transform spectrometer more precisely or with improved spatial resolution can be determined and for others it is only possible in the thermal IR range the parameters of smoke particles from measurement data forward.

The essential result achieved according to the invention be stands in contrast to those known from the prior art ten solutions in that the optical thickness, the size distribution and the type of aerosols in the column under consideration can be voted.

The invention is based on the drawings in a individual explained. Show it:

Fig. 1 is a schematic non-scale representation of fields of view of an imaging sensor and an IR spectrometer;

Fig. 2 is a schematic non-scale view of a typical fire scene of a forest fire;

Fig. 3 in the form of a block diagram a schematic representation of a fusion of sensor data;

Fig. 4 is a flowchart in a fusion processor to perform spectral parameter analysis, and

FIG. 5a and FIG. 5b is a flow chart for performing Since tenanalyse.

In Fig. 1, the field of view 10 of an imaging IR sensor and the field of view 20 of an IR Fourier spectrometer are shown schematically and to scale. For this purpose, imaging sensors in the infrared and visible spectral range generate a matrix of pixels which are symbolized by a grid, in which a square element represents an image element (pixel) and corresponds to the spatial resolution of an imaging sensor.

Such an image can in this case be generated by a two-dimensional detector array (focal plane array) or by a detector line 11 (pushbroom scanner) scanning in a direction of flight indicated by an arrow. The face field 20 of the IR Fourier spectrometer is shown in a circle and has a low spatial resolution according to its diameter.

Bispectral imaging can be used as sensors BIRS IR sensors for medium infrared MIR (wavelength ranges from 3.5 to 4 µm) and for a thermal infrared TIR (Wavelengths 4 to 2 µm) and an imaging CCD camera VIS / NIR-CCD in the visible (wavelength range 0.4 to 0.7 µm) and in the Near-IR (wavelength range 1.2 to 1.4 µm).

In FIG. 2, a typical fire scene is shown of a forest fire in the fields of view 10 and 20 of an aircraft or satellite-transmitted sensor array of an imaging sensor BIRS and an IR spectrometer IRFTS. The scene includes a line of fire 31 with flames, burned areas 32 with smoldering fires, a smoke and gas cloud 33 over an unburned background and finally an area 34 of the unburned undisturbed background. The observed scenes are classified according to the temperature distribution of the surface in areas with flaming combustion ( 31 ), smoldering fire ( 32 ), smoke ( 33 ) against an unburned background and undisturbed background ( 34 ), which is referred to as generic classes of the scene.

This measurement method is particularly useful for measurements of the tem temperatures and air pollution from large fires for example through forest and savanna fires or oil fires where dense aerosol clouds get into the atmosphere. The Sensors can be from a satellite, an aircraft or from one, for example, on a fire engine assembled soil measuring station.

In the block diagram of FIG. 3, the architecture for a fusion of sensor data is shown schematically in a schematic representation. BIRS, IRFTS and VIS / NIR-CCD sensors generate specific object data in parallel and independently of one another. The output data of these sensors are entered into a fusion processor 50 after an individual data processing in each sensor downstream signal processors 41 to 43 . Here, the object data relate to physical parameters such as surface temperature, area, spectral distribution, geometric structure, temporal changes, as well as the position of objects and a range of confidence in the detection, classification or identification of the objects in the field of view under consideration (FOV).

The fusion processor contains data lanes and a data analysis algorithm. Furthermore, the fusion processor 50 is connected to an input device 52 , for example in the form of a PC which has a keyboard 52 a and a screen 42 b, in which results are entered by the fusion processor 50 and results are returned to the fusion processor 50 become. By a bottom right in Fig. 3 as said given block 53, it is indicated that conditions Meßbedin from the outside as well as a priori knowledge one may be added in the fusion processor 50.

The advantages of this type of fusion, the so-called sensor-level fusion, are:

• - Through individual data processing for discrimination potential objects before entering them into the mergers processor becomes the workload of the fusion processor reduced.
• - The signal processing is also optimized of the individual sensor according to its detector design and operational characteristics.  
• Instructions for signal processing in processors 41 to 43 and the respective sensor can be based on independent data from other sensors.
• There is flexibility in the number and type of sensors without having to change the basic structure of the fusion processor 50 .
• - There is also a cost-effective way to an existing multisensor configuration sion add.

The required spatial resolution at the location of the fire is for using the sensors on the various possible Sensor platforms different. The spatial resolution and the required field of view of optics (FOV) as well as the resulting footprint of the imaging sen sensors are summarized in Table 1.

Table 1

Table 1 shows the spatial resolution of the imaging sensors BIRS and VIS / NIR, their effective field of view FOV and their footprint when the sensors are worn by

• - satellites at an orbit altitude of 400 km, (like Internatio nale space station),
• - An aircraft, the resolution for an altitude of 1 to 10 km, FOV and footprint given for an altitude of 4 km is, and at  
• - A floor insert, the resolution for a slant Distance from 0.1 to 1 km, FOV and footprint for 400 m slope Distance applies.

The radiometric resolution in the thermal IR channels is defined as the noise equivalent radiance NER in W / (cm ² sr) and the noise equivalent temperature difference NE / ΔT, while the radiometric resolution in the visible and near IR is defined as the noise equivalent spectral radiance NESR. These requirements are summarized in Table 2.

Table 2

Because of the very different radiance contrasts in the different spectral ranges for a fire scene Temperatures up to 1800 ° K must be at the required radiome trical resolution a correspondingly large dynamic range of Signal processing must be ensured.

The minimally detectable fire area of the imaging sensors can be estimated from the contrast beam densities encountered in fires and the required spatial and radiometric resolution. The minimally detectable fire area can be significantly smaller than the spatial resolution of the imaging sensors. With a satellite-borne sensor, this area must be ≈4 m ² . In the case of an aircraft-carried system, this minimally detectable fire area is ≈0.2 m ² and is correspondingly smaller for a floor-based sensor, namely only ≈0.002 m ² .

The requirements for the IR Fourier transform spectrometer, thus the field of vision (FOV) and the resulting one Diameter of the visual field at the base, the spectral Resolution and spectral sensitivity (NESR) are in Table 3 listed. The alignment area of the optics (pointing range) must cover the entire field of view cover forming sensors; the corresponding angular range che, within which the IR spectrometer are aligned can also be found in Table 3.

Table 3

With these sensor specifications, a fusion and data analysis algorithm has been developed that combines and analyzes the various sensor data and improves the results and the evaluation of the individual sensor. Based on the results in the fusion processor 50 , instructions are derived from the individual sensors as feedback. These relate to the adjustment of the signal processors 41 to 43 , such as dynamic threshold values, integration times or other parameters of data processing and information about areas for which a more detailed examination has to be carried out, the optical alignment and choice of the field of view as well as the calibration and the operating mode of the sensors.

The functions of the fusion processor 50 sketched in FIG. 3 with the data analysis 51 are described in more detail using the diagram in FIG. 4. The radio metric corrected and calibrated measured data (approximately Prozessie level 1) are in the fusion processor 50 in a Prozes sierungslevel 2 transferred. Here, only those data are fused that come from the smallest common field of view, which is the field of view 20 of the optics of the IR spectrometer IRFTS according to the measurement method described.

The surface temperatures from image data and the IR spectra are combined into a data vector for each defined measurement time and measurement location (block 54 ). With this data vector, a first state estimation and evaluation (block 55 ) of the fire, the atmosphere and the background is carried out with the aid of a physical model of the fire, which is based on the scenario shown in FIG. 2. At the same time, the merged data is used in a further branch (block 56 ) for a first classification of the fire according to the type of combustion and identification of the fuel and for the localization of the high-temperature event.

The state estimate serves as input for a spectral parameter analysis (block 57 ), with the aid of which temperature and concentration profiles of the gases and aerosols are recovered. The results of the parameter analysis (block 57 ) allow in the next step a refinement and improvement of the state estimate of the fire and the discrimination of fire and background (block 55 ).

By repeatedly returning the calculated parameters and fusing the data, the spectral parameter analysis (block 57 ) is continued until there is a minimal deviation between the measurement data and the simulation results. Further results, such as emission ratios of various gases and aerosols and the combustion effectiveness for the individual combustion phases of the fire, can be derived from the results of the spectral parameter analysis.

Together with the parameters of the condition estimation, the fire can be classified, the type of fuel identified and the burned fuel mass quantified. This is the desired result data with a processing level 2 (block 58 ).

In addition to the problem of inversion of measurement data from at recovery is made by means of atmospheric vertical soundings complicated by another problem. Since the optical Ge field of view of the IRFTS spectrometer is significantly larger than the spatial dimensions of a typical fire is determined by the IRFTS considered a highly inhomogeneous fire scene. The observ Eighth spectrum therefore consists of radiation contributions different classes of fire areas with different temperatures and combustion conditions, such as flames, Smoldering fire, smoke, and the undisturbed environment of the fire together.

In order to be able to recover more than just average temperatures and concentration profiles from this spectrum of the observed scene of the IRFTS spectrometer, a combination of the imaging and spectral measurements of BIRS and IRFTS is absolutely necessary. Two approximations are given for a data fusion:

• - Sequential approximation: Assuming that from the Ana lysis of the image data of the BIRS the subdivision of the fire scene is already known in different classes, the right Latin area contributions of the classes with related Soil temperatures as input for the radiation trans port bills used to iteratively do the non-li least squares problem in fitting the to solve spectral data.
• - Approximation through global adjustment: With a recovery algorithm are simultaneously the atmospheric and upper surface properties of the fire determined.

The spectral parameter analysis is made up of individual successive steps. In Fig. 5, the steps and links of the data analysis are shown schematically in a flow chart.

The main components of the spectral data analysis are atmospheric correction models and radiation transport models MODTRAN (Moderate Resolution Transmission) (Block 60 ) and FASCODE (Fast Atmospheric Signature Code) (Block 61 ) with necessary data bases for the molecular species, the natural aerosols and smoke particles as well as inversion algorithms ,

With the radiation transport models in known Parameters such as surface temperature, temperature and con concentration profiles of gas species and particles and optical Parameters of the particles etc., the resulting IR radiation all components of a fire by forward calculations simu lose. By reversing the process, i.e. H. Inversionsrechnun under defined boundary conditions from the IR spectra the physical parameters of the radiation sources,  d. H. Temperature, composition and their profiles, too be recovered.

The radiation transport model MODTRAN 60 simulates the IR spectrum of the fire scene with a low spectral resolution of 1 cm ^-1 in the entire thermal IR range and compares it with the measured spectrum of the IRTFS spectrometer. This spectrum with low spectral resolution is mainly used for the simulation and inversion calculation of the continuous radiation of particles, gases and the soil in the atmospheric windows.

With a "line-to-line" radiation transport model, such as FA-SCODE (Block 1 ), on the other hand, the IR spectrum of the gases with a high spectral resolution of <0.1 cm ^-1 in a number of narrowly defined spectral ranges (micro -Windows) calculated and also compared with the measured spectrum of the spectrometer IRFTS. The HITRAN database (block 62 ) is used for the model calculations with FA-SCODE (block 61 ). This database contains the spectral data required for the model calculations of 32 molecular species in the entire relevant optical spectral range from UV, visible, IR to the millimeter wave range.

An essential component of the data analysis is the aerosol model and the aerosol database (block 63 ) for smoke particles. The smoke clouds transmit undisturbed the majority of the thermal radiation from objects behind, but a small fraction is dampened by absorption and scattering, ie extinction, on the smoke particles. Furthermore, a portion of infrared radiation is added in the direction of observation due to thermal self-emission of the particles. This affects the temperature measurement of the surfaces in the fire. The spectrally dependent absorbance and self-emission of IR radiation by the smoke particles can be used to determine the optical thickness, size distribution and type of the smoke particles in the measured column.

The radiation properties of smoke are calculated using the MODTRAN atmospheric radiation transport model (block 60 ). The data required for this are the optical properties of the particles and the vertical profile of the particle density and its temperature in the smoke cloud. The required optical parameters of the particles are the absorption, scattering, extinction coefficient, single scattering requirement and the scattering phase function as a function of the wavelength in the spectral range 0.2 to 300 µm.

The optical parameters of the smoke particles are direct Measurements not with the necessary quality and quantity available. Therefore, the optical parameters are determined using the Mie scattering theory by the computer aerosol model DBASE with aerosol database from measured microphysical egg properties calculated. The aerosol model DBASE contains numbers rich atmospheric standard aerosols and enables that Users, their own new aerosol components, such as smoke aerosols, and to produce mixtures of these components.

The microphysical parameters that the optical parameters the size distribution, the refractive index, the composition (mainly the type of mixture between absorbent and non-absorbent material) and the shape of the particles. The optical properties of the parties can also be affected by the humidity, which increases the aerosol size and the refractive index takes.  

The microphysical parameters of smoke particles are from various sources, mainly international organic mass fire experiments like AERONET, TRACE-A, SAFARI, SCAR-B adopted and in the smoke aerosol model by Remer et al. to briefly sums. (A. Remer et al. Biomass burning aerosol size distribution and modeled optical properties, J. Geophys. Res., 103: 31879-31891, 1998). This smoke aerosol model ba based on a database of 800 size distributions from Smoke particles from various biomass fires.

The results show that the size distribution of Rauchpar is bimodal and right through two lognormal distributions can be presented. The parameters of the lognormal distribution are the mean mode radius, the standard deviation of the natural log of the radius and the number or that Volume of the particles in the cross section of the atmospheric acid le. The two modes are the accumulation mode (small Particles) and the coarse-grained mode (large particles).

In biomass fires, the optical properties become paramount mainly dominated by the smaller particles. The big ones Particles contribute approx. 30% to the total absorbance. The size The external distribution of the smoke particles consists of an external one Mixing a large number of small particles with diameters from 0.1 to 1.0 µm, and a small number of large particles with diameters from 1 to 30 µm. The little particles of her partly consist of an internal mixture of organic Liquids that consist of a solid spherical core surrounded by black carbon. The big particles with one Diameters <1 µm consist mainly of the material Earth's crust.

Standard aerosol models already available in MODTRAN, such as the urban / industrial aerosol model, one  Do not replace smoke aerosol model. The differences from Smoke aerosols and urban / industrial aerosols mainly in the ability to retain moisture and thereby growing (moisture factor). Smoke particles be sit a lower moisture factor and so it is Particle growth is much less due to water input.

Because of the nonlinear nature of the least-qua method the infrared spectrum must be repeated for a whole series he can be calculated from estimates of the parameter vector. For Successful recovery requires a number narrow spectral ranges, called micro windows be able to choose from the entire thermal IR range, in which the molecular species to be determined is one shows sufficient signature of the absorption and on the other is not disturbed by the absorption of other molecules. of others are for the precise modeling of the spectrum Li never-to-line radiation transport calculations with high spectra indispensable resolution. However, this prohibits for reasons the necessary computing time a model calculation in total thermal IR range.

Therefore, for a recovery scheme, the use of a Set of micro windows provided. The required spec The tral range depends on the parameters resulting from the Measurements are to be recovered. Because water vapor at affects almost the entire infrared range and carbon dioxide and carbon monoxide to the main forms of combustion gases in fires must include the choice of micro windows ultimately the recovery of the concentrations of these gases and in addition the gas temperature as a minimum requirement possible (see Table 4, in which micro window the important most important molecules as well as their meaning and use are).  

Table 4

In the first step, for the imaging sensor BIRS the atmospheric correction for water vapor, the emission correction  of the surfaces and the solar reflection correction in Spectral range 3-4 µm carried out.

With known inversion methods based on known line to-line radiation transport models, such as FA- SCODE, based, can with the spectrometer IRFTS measured line spectra iteratively the profiles of the gas temperature rature and the concentration of different gas species become.

With the radiation transport model, the IR spectrum of the Spectrometer IRFTS, which is classified by the BIRS weighted, simulated and compared with the measured spectrum. The optical thickness the particle column is made up of the differential differences the spectral radiance in the two atmospheric Windows like those of the broadband thermal imaging devices and detected by the spectrometer.

The temperature of the particles required for this is assumed to be the temperature of the surrounding combustion and atmospheric gas. For this purpose, the spectral emission properties of CO ₂ and CO in the region of the molecular bands, which are weighted with the spatial temperature distribution determined by the thermal imaging devices, are simulated and compared with the measured spectra. Individual model parameters such as gas temperature, concentrations of individual gas species and particle concentration are varied so that they result in the best match (best fit) with the spectral data of the spectrometer.

The model parameters determined in this way become transmissions correction of the continuum absorbance of gas and particles set. This enables one in atmospheric windows  improved determination of the surface temperature and in the loading corresponding to the absorption bands of the molecules Offset correction of species concentrations. The procedure enables the determination of the column densities of all essential Chen combustion gases in the optical path.

The results of the data analysis and the quantities derived from it with the help of "a priori" knowledge regarding the combustion processes and known measurement conditions, such as geology, season, time of day, atmosphere are:

• 1. Column densities and profiles of the concentration in the direction of the line of sight of the gas species: H ₂ O, CO ₂ , CO, O ₃ , CH ₄ , NO, NO ₂ , SO ₂ , HCN, NH ₃ , HNO ₃
• 2. Temperature profile of the atmosphere and combustion gases in Direction of the line of sight
• 3. Spatial distribution of the surface temperature of the view line limitation (ground, solid or liquid surfaces) with location coordinates (geolocation)
• 4. Optical thickness of the aerosols and their size distribution in the pillar of line of sight
• 5. Emission ratios of all detected gas species X: ΔX / ΔCO ₂ based on the reference gas CO ₂ or CO and difference Δ to the corresponding atmospheric background concentration
• 6. Emission ratio of the aerosols as particle number N or parti body mass M related to CO: ΔN / ΔCO or ΔM / ΔCO
• 7. Profile of the aerosol particle number N based on ΔCO concentrations tration profile in the direction of the line of sight  
• 8. Combustion efficiency CE (Combustion Efficiency) of carbon C in fire: CE = [C] _CO2 / [C] _CO2 + [C] _CO + [C] _aerosol )
• 9. Classification of fire in surface areas: flame end Combustion, smoldering combustion, smoke from unburned Background, unburned background
• 10. Improved lateral spatial resolution for gas concentrations trations and gas temperature related to the spatial resolution capacity of the IR spectrometer
• 11. Identification of the fuel type

In the following, essential relationships of the Invention summarized.

Clouds of smoke are partially transparent in the thermal IR range rent and enable the use of imaging IR radiometers for determining the position and temperature of the Fires and smoldering fires. Show in this IR area numerous combustion gases also spectral absorption and Emission properties used to determine concentrations can be used with IR spectrometers.

By means of the optical telemetry method according to the invention a multi-sensor system in the thermal IR spectral range, in Near infrared and in the visible spectral range with associated multisensory analysis of the data are the temperature as well as gas and particle concentrations in combustion processes determined simultaneously. The facility's multi-sensor system according to the invention has a bispectral imaging IR sensor (BIRS) (thermal imaging device, IR scanner) in the thermal Infrared range and a CCD camera in the visible and near  Infrared (VIS / NIR-CCD), both of which have a high spatial resolution own solution, as well as a spectrally high-resolution IR Fourier Transform spectrometer (IRFTS) in the thermal infrared range on. The spatial resolution of the IR spectrometer is here low compared to that of the imaging sensors.

By merging the data from the sensors, on the one hand, the Results of the individual sensors such as the surface temperature with the bispectral imaging IR sensor BIRS and the Gas temperature and gas concentrations with the spectrometer IRFTS more accurate or with improved spatial resolution and on the other hand it only makes it possible the parameters of smoke particles in the thermal IR range derive the measurement data. An essential new result is that the optical thickness, the size distribution and the type of Aerosols can be determined with it.

In the method according to the invention there is no lighting source because the thermal radiation of the objects is being used. The process is not specific to optical and thermal properties of the background under consideration in limited to the scene.

Claims (18)

1. A method for the optical remote measurement of fire scenes with means of IR sensors, characterized in that initially with a first bispectral imaging sensor, the spectral broadband in the two atmospheric IR windows of about 3.5 to about 4.1 microns (MIR) and approx. 8 to approx. 12 µm (TIR) works and has a high spatial resolution, and with a second bispectral imaging sensor that is spectrally broadband in the visible spectral range from approx. 0.4 to approx. 0.7 µm and in the near IR range from approx. 1.2 to approx. 1.4 µm (NIR), but has a higher spatial resolution than the first image-forming sensor, visualized the fire scene and measured the temperature distribution, that afterwards inside the field of view of the imaging sensors one or more areas of interest are selected, on which, as the third sensor, a spectrally high-resolution, in the thermal IR spectral range of approx. 3.0 to approx. 15 µm sensitive IR Fourier -Transform spectrometer is aligned, which has a smaller field of view than the two imaging sensors, and that in this smaller field of vision then a detailed analysis of the physical conditions of the fire scene simultaneously with both the imaging sensors and with the IR Fourier- Transform spectrometer is carried out, wherein the field of view of the IR Fourier transform spectrometer can be aligned as desired and defined within the field of view of the two imaging sensors. 2. The method according to claim 1, characterized in that at measuring IR radiation in the two atmospheric Windows with the first bispectral imaging IR sensor the temperature distribution of surfaces as well  Emission and continuum absorption of gases and particles is true. 3. The method according to claim 1 or 2, characterized in that with the IR Fourier transform spectrometer, the emitted spectral radiance of all radiation sources located in the visual field, in particular the emission and Absorpti onslinien of the gases with a high spectral resolution of about 0, 1 cm ^{-1 is} measured and that the gas temperature and the concentration of numerous gas species is determined from the line spectra of the gases. 4. The method according to any one of the preceding claims, characterized characterized in that the three sensors are parallel and independent generate mutually specific output data, which after a Signal processing in each sensor as object data, the phy sical parameters such as B. surface temperature, area, spectral distribution, geometric structure, temporal changes changes as well as the location of the objects and the area of trust the detection, classification or identification of the objects in the visual field under consideration, in a fusion process processor can be entered in which with the help of a fusion and data analysis algorithm the various sensor data combined, analyzed and results are determined. 5. The method according to claim 4, characterized in that on based on the results obtained in the fusion processor as feedback instructions to the individual sensors be directed which the adjustment of the signal processors such as B. dynamic thresholds, integration times or on other parameters of data processing and information about Areas for which an in-depth investigation is required as well as the optical alignment and choice of the visual field,  concern the calibration and the operating mode of the sensors fen. 6. The method according to claim 4 or 5, characterized in that that only those data are merged which result from the smallest common field of view of all sensors which is the field of view of the optics of the IR Fourier Trans form spectrometer. 7. The method according to any one of claims 4 to 6, characterized ge indicates that the surface data from image data and the IR spectra for each defined measurement time and location a data vector are combined that with this data vector a first assessment of the condition and evaluation of the fire, the atmosphere and the background with the help of a physika model of fire is carried out, and that paral To do this, merge the merged data into another branch a first classification of fire by type of burn Fuel identification and identification and localization tion of the high temperature event can be used. 8. The method according to claim 7, characterized in that the Condition estimation as input of the spectral parameter analysis serves by means of which the temperature and concentration Profiles of gases and aerosols from the fire scene are recovered be that with the help of the results of the parameter analysis in next step is a refinement and improvement of the To Status assessment of fire and discrimination against fire and background is made by repeated backs management of the calculated parameters and fusion of the data spectral parameter analysis is continued until ei ne minimal deviation between the measurement data and the simula results, and that from the results the spectral parameter analysis larger quantities like that  the emission ratios of various gases and aerosols and the combustion effectiveness for the individual combustion phases of the fire can be derived. 9. The method according to any one of claims 4 to 8, characterized ge indicates that when the measurements are combined with the imaging sensors and with the IR Fourier transform Spectrometer for data fusion a sequential approximation is carried out, assuming that from the Ana lysis of the image data of the bispectral imaging IR sensor divides the fire scene into different ones Classes are already known, the relative areal Contributions of the classes with associated soil temperatures as Input used for the radiation transport calculations in order to iteratively solve the nonlinear problem of the smallest Solve squares when adjusting the spectral data. 10. The method according to any one of claims 4 to 8, characterized ge indicates that when the measurements are combined with the imaging sensors and with the IR Fourier transform Spectrometer for data fusion an approximation through global Adaptation is carried out with a recovery algorithm simultaneously the atmospheric properties and the surface properties of the fire are determined. 11. The method according to any one of claims 4 to 10, characterized ge indicates that spectral data analysis is essential elements atmospheric correction models and beam lung transport models with necessary associated databases for the molecular species, the natural aerosols and smoke particles and inversion algorithms are used, whereby with the radiation transport models in known Para meters, surface temperature, temperature and concentration on profiles of the gas species and particles and optical parameters  the particle the resulting IR radiation from all compo simulates the fire by forward calculation and by reversing the process, d. H. Inverse modeling, under defined boundary conditions from the infrared spectra the physical parameters of the radiation sources, d. H. Tem temperature, composition and their profiles can be. 12. Device for carrying out the method according to a of the preceding claims, characterized by an off education as a multi-sensor system consisting of a Kombina tion of a first bispectral imaging sensor, the spectral broadband in the two atmospheric IR windows from approx. 3.5 to approx. 4.1 µm (MIR) and approx. 8 to approx. 12 µm (TIR) works and a high spatial resolution has a second bispectral imaging sensor that spectral broadband in the visible spectral range of approx. 0.4 to approx. 0.7 µm and in the near IR range from approx. 1.2 to 1.4 µm (NIR) works, but a higher spatial resolution solution as the first imaging sensor, and a third ter sensor, which as a spectrally high resolution, in therma len IR spectral range from approx. 3.0 to approx. 15 µm sensitive ches IR Fourier transform spectrometer is formed a smaller field of view than the two imaging ones Sensors. 13. The device according to claim 12, characterized in that the first bispectral imaging sensor, the spectral broadband in the two atmospheric IR windows of approx. 3.5 to approx. 4.1 µm (MIR) and approx. 8 to approx. 12 µm (TIR) tet, is a thermal imaging device or an IR scanner. 14. Device according to claim 12, characterized in that the second bispectral imaging sensor, the spectral  broadband in the visible spectral range from approx. 0.4 to approx. 0.7 µm and in the near IR range from approx. 1.2 to approx. 1.4 µm (NIR) works, is a CCD camera. 15. Device according to one of claims 12 to 14, characterized characterized in that the visual fields of the sensors each other are adjusted so that the field of view of the IR Fourier Trans form spectrometer of the total field of view of the two images producing sensors is always completely covered, whereby the matching field of view of the two imaging sen but always larger or equal to that of the IR Fourier transform spectrometer. 16. The device according to claim 15, characterized in that the size of the field of view of the imaging sensors so is sized to cover all or part of the fire are recorded. 17. The device according to claim 16, characterized in that the ratio of the diameter of the visual field of the images generating sensors for the diameter of the IR Fourier transform Spectrometer is about 10: 1. 18. Device according to one of claims 12 to 17, characterized records through an arrangement on a sensor platform a satellite, an airplane or on the ground, in particular on a vehicle.