Summer Baby Monitor No Signal Nextto Ea H Other

Remote Sensing

Kirk J. Havens , Edward J. Sharp , in Thermal Imaging Techniques to Survey and Monitor Animals in the Wild, 2016

Thermal imaging

Thermal imaging is a very powerful remote sensing technique for a number of reasons, particularly when used to elucidate field studies relating to animal ecology. Thermal imaging data is collected at the speed of light in real time from a wide variety of platforms, including land, water, and air-based vehicles. It is superior to visible imaging technologies because thermal radiation can penetrate smokes, aerosols, dust, and mists more effectively than visible radiation so that animals can be detected over a wide range of normally troublesome atmospheric conditions. It is a completely passive technique capable of imaging under both daytime and night-time conditions. This minimizes disruptions and stressful disturbances to wildlife during data collection activities. It is capable of detecting animals which are colder, warmer, or the same as their background temperature because it does not compare temperatures but rather the emissivity of the animal against its background.

While the emphasis of this book is on the counting and observation of wildlife there are other very important applications where remote sensing via thermal imaging can be of use. For example, using thermal imagers in aerial surveys of the landscape for mapping purposes can provide some unique capabilities that cannot be gained any other way. From aircraft heights and at aircraft speeds there are no fundamental problems in achieving ground resolutions down to a fraction of a meter (Stewart, 1988). The main advantage of thermal images over visible aerial photography is that they can sense heat. For example, soil types that are absorbing differing amounts of solar radiation can be mapped as well as shading effects on north/south facing slopes on hilly or mountainous terrain. Shading can also be used to help map features of dry washes, forest edges, fence lines, agriculture fields, drainage ditches, variations in soil moisture, and evaporation and even to determine wind direction in many cases (see Figure 1.2, Chapter 1).

It is interesting to note that Quattrochi and Luvall (1999) identify a similar reluctance on the part of remote sensing scientists to adopt the powerful resources offered by thermal imaging as do we on the part of wildlife scientists engaged in studying and monitoring wildlife populations. Although numerous articles have appeared in the professional literature that have employed thermal infrared (TIR) data for the use in studying specific aspects of landscape-related processes (e.g., evapotranspiration), the direct application of TIR data for assessment of landscape processes and patterns within a landscape ecological purview is lacking. They argue that the use of TIR data from airborne and satellite sensors could be very useful for parameterizing surface moisture conditions and developing better simulations of landscape energy exchange over a variety of conditions and space and time scales. They postulate that TIR remote sensing data can significantly contribute to the observation, measurement, and analysis of energy balance characteristics (i.e., the fluxes and redistribution of thermal energy within and across the land surface) as an implicit and important aspect of landscape dynamics and landscape function.

There are three primary reasons for the lack of enthusiasm to use TIR remote sensing data for landscape ecological studies. First, TIR data are little understood from both a theoretical and applications perspective within the landscape ecological community. Second, TIR data are perceived as being difficult to obtain and work with to those researchers who are uninitiated to the characteristics and attributes of these data for applications in landscape ecological research. Finally, the spatial resolution of TIR data, primarily from satellites, is viewed as being too coarse for landscape ecological applications (e.g., Landsat Thermatic Mapper data at 120 m spatial resolution) and calibration of these data for deriving measurements of landscape thermal energy fluxes is seen as problematic. Interestingly, these reasons are very similar to those given for the limited use of thermal imagers by wildlife scientists in the preface of this book. Quattrochi and Luvall (1999) proposed ways to overcome these misconceptions regarding the use of TIR remote sensing data in landscape ecological research by providing supporting evidence from a sampling of work that has employed TIR remote sensing data for analysis of landscape characteristics.

Thermal imaging technology (see Chapter 7) developed by the military is now available from a number of commercial vendors at reasonable costs. For example, thermal imaging systems, both handheld and airborne units, are now available with sensitivities more than an order-of-magnitude better than the units used in the early experiments devoted to large mammal surveys (Croon et al., 1968; Parker and Driscoll, 1972). Thermal imaging technology provides a method for obtaining complete counts of animals with little risk of behavioral or sampling bias. Chapter 10 provides an extensive review of past thermal imaging studies conducted in the field and laboratory for a number of different applications.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780128033845000038

Imager Selection

Kirk J. Havens , Edward J. Sharp , in Thermal Imaging Techniques to Survey and Monitor Animals in the Wild, 2016

Introduction

Thermal imaging is simply the process of converting infrared (IR) radiation (heat) into visible images that depict the spatial distribution of temperature differences in a scene viewed by a thermal camera. The imaging camera is fitted with an infrared detector, usually in a focal plane array, of micron-size detecting elements or "pixels." The detector array may be cooled or uncooled, depending on the materials comprising the array and the camera's intended use. A lens system focuses scene radiation onto the detector array and appropriate processing electronics display the imagery. Infrared radiation is attenuated by the atmosphere and the degree of attenuation depends greatly on the local atmospheric conditions at the time the imagery is collected. It is important to match the detector response with either of the two atmospheric windows: midwave IR band (MWIR) or long wavelength IR band (LWIR), as shown in Figure 7.2. Remember that for most survey work and field observations we are not concerned with the measurement of temperatures related to objects in the scene but only the apparent temperature differences between objects in the scene. Note that this is a big step back from the level of difficulty that most thermographers face when measuring temperatures in the field, but nonetheless it comes with its own set of demanding requirements.

It is important to create the best images possible to extract meaningful data regarding the detection, recognition, and identification of animals of interest in the field. This is exactly the purpose of surveillance applications, which the military has been laboring over for years. The first thermal imaging cameras were developed in the 1950s by the military; they were large, heavy, and very expensive. The camera technology at that time required that they be cooled with liquid nitrogen. Improvements in camera development continued over time and new detector materials, array fabrication techniques, coolers, optics, electronics, software, and packaging have resulted in reliable high-performance thermal imaging cameras. By the mid-1990s focal plane arrays were mostly of the cooled variety. Uncooled arrays were beginning to make their appearance and were incorporated into new imagers. The technology today encompasses both cooled and uncooled focal plane arrays. Most books devoted to thermal imaging technology are focused on the development of thermal imagers and detector arrays for uses in commercial applications rather than those of concern to the military such as long range surveillance and target identification/acquisition. Civilian uses of thermal imagers are usually devoted to temperature measurements of components used in industrial applications, building inspections, surveillance for security, police, and fire applications, and robotic vision. Wildlife ecologists planning to use thermal imagers in the field will be more interested in long range acquisition and identification of different animal species. Other researchers may be interested in using infrared imagers to study different aspects of animal physiology such as thermoregulation. By locating and monitoring thermal abnormalities in different parts of an animal's anatomy one can infer underlying circulation that may be related to physiology, behavior, or disease. McCafferty (2007) has reviewed 71 empirical studies that used infrared thermography for research on mammals. In either case the advancements in detector array technology and the improvements gained in the infrared imagery for both cooled and uncooled imagers will find use in both applications.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780128033845000087

Using Thermal Imagers for Animal Ecology

Kirk J. Havens , Edward J. Sharp , in Thermal Imaging Techniques to Survey and Monitor Animals in the Wild, 2016

Thermal Imaging and Automation

Several problems must be addressed when considering the design of automated thermal imaging detection applications. Thermal imagers collect radiation from animals and their background, and if there are sufficient differences in the apparent temperatures between the two then quality imagery can be obtained. This imagery can contain sufficient information to count and identify a large number of species and, in many cases, the user is able to make accurate evaluations regarding the activity, age, sex, and physical condition of the animal. These evaluations are not automatically determined and require the detailed examination of the thermographer. This being the case, we assume that there is adequate information obtained from the imagery, but the data processing was carried out in a human brain. The complexity and difficulty of processing thermal images so that robots can see well enough to make decisions in outdoor surroundings is truly a difficult thing to even imagine.

Going from the raw imagery to just counting the number of individual animals in the imagery has taken a significant effort on the part of many researchers and field scientists. It has been realized that the only way to count large numbers of animals during a short period of time is through automating the data extraction from the imagery. The basic problem that needs to be solved is how the number of animals in the collected imagery can be counted so that none are missed. We reviewed a number of efforts in Chapter 10 that used digital image processing, computer vision analysis, superposition of detection and tracking algorithms, and automated motion detection to solve some of these problems.

Automated Detection of Animals

Automated detection of animals using thermal imaging is complicated by a number of factors. Assuming that the thermal images are good enough to unambiguously separate animal signatures from the background and that they are easily visualized in the imagery then detection is easily accomplished. Is there more than one species present? The more species there are the more complex the automation of the detection task becomes because now the individual signatures must be identified as belonging to the species of interest. This in turn means that an algorithm of some sort based on shape, size, or intensity must be developed to perform the identification to complete the task of detection.

A number of efforts that do not require identification of species but instead require the detection of any and all animals present such as in agricultural mowing operations have seen marked improvements. Steen et al. (2012) used imagery collected with an uncooled bolometer in the LWIR with a 640 × 480 pixel FPA mounted on a tractor. The tractor was driven at different speeds to test an algorithm for the automatic detection of caged animals (rabbits, chickens), which were placed in grass ready for mowing. They concluded that the method has potential for automated detection of animals during mowing operations. They reported that under most circumstances detectivities were close to 100%, with the caution that dense crops may hamper the detection of animals.

When the work requires the automated detection and identification or classification of animals collected in the thermal images, the problem becomes more difficult. Nonetheless, Christiansen et al. (2014) have developed methods to detect and classify via a new thermal feature extraction algorithm used on imagery collected from a lift to simulate the platform of an unmanned aerial vehicle (UAV)-based detection and recognition system. The thermal signatures of detected objects are calculated using morphological operations, which are partly invariant to rotation, size, and posture. UAVs are an emerging technology (see Chapter 12), and in modern agriculture, they can be used for many purposes. UAVs can be preprogrammed to navigate flight paths commensurate with mowing operations, and they can be equipped with the required thermal imagers and data processing equipment to detect and enumerate animals in the path of agricultural machinery, thereby reducing wildlife mortality and promoting wildlife-friendly farming.

A system has been demonstrated by Israel (2011) that can detect roe deer fawns (Capreolus capreolus) in meadows prior to or during mowing operations. A UAV-mounted thermal imaging camera was tested in the field to demonstrate the application. Thermal images were collected from the UAV and transmitted as an analog video stream to a ground station, where the user followed the camera live stream on a monitor for manual animal detection in real time. In this system the camera was flown approximately 10 m agl with a view angle of θ _C = 0° with respect to the nadir. The thermal imager was a FLIR Tau 640^® using 32° × 26° FOV optics with an IFOV = 0.89 mrad. The success demonstrated by the manual operation of this detection system provides the impetus to go forward with automation of the sensor platform. Sensor data fusion between the thermal and the visual camera should extend the utility of the system to detect fawns in a wider variety of weather situations, including sunny days.

Detecting cryptic and nocturnal species places more stringent requirements on the data-gathering efforts. Brawata et al. (2013) designed an automated thermal video recording system to monitor cryptic mammalian predators—dingo (Canis lupus dingo), red fox (Vulpes vulpes), and the feral cat (Felis catus)—at food and water resources in Australia. The thermal camera used was an FLIR ThermaCAM S45^® with a 320 × 240 FPA microbolometer sensor operating in the LWIR. The 35 mm lens afforded a 24° × 18° FOV and a spatial resolution of 1.3 mrad. Since the automatic video capture system monitored wild carnivores, it was left unattended for extended periods of time to minimize the impact of human presence. The system remained on at all times but only recorded video when target species were identified in the thermal video frame when large temperature changes were detected between portions of an incoming frame and the average "background" frame. Observers using binoculars and image intensifiers monitored sites and correlated animal sightings with the recorded video. By using their equipment at focal lures (food and water) they were able to monitor three target predator species and determined that the optimal sampling distance for detection, identification, and for collecting behavioral data was 30–40 m. This range limitation may be due to the performance of the imager more than anything else, although there is very little information provided on the local environmental conditions at the time of the data collection, which could also have played a role.

Automated Enumeration of Animals

Again, if there is only a single species present and this is known before the imagery is collected, the problem is reduced to detection if they can be distinguished from one another in the imagery. If the grouping of the animals is too dense or too closely packed to count individuals within the group then a statistically-based algorithm is required to assist with the counting. This problem has been addressed by a number of workers who used thermal imagery to count avian species and bats. Significant advancements were made for animals present in large numbers in the dark when there is a reduced probability of interference from excessive background clutter. Advanced infrared detection and image processing for automated bat and bird censusing has been the subject of much work. Sabol and Hudson (1995) collected thermal imagery on videotape and subjected it to digital image processing routines to extract bat numbers from the imagery. Frank et al. (2003) demonstrated a real-time automated censusing system to make accurate and repeatable estimates of the number of Brazilian free-tailed bats (Tadariada brasiliensis) present independent of colony size, ambient light, or weather conditions and without causing disturbance to the colony. Melton et al. (2005) developed a thermal infrared detection and tracking system for bats in flight. Desholm et al. (2006) used LWIR thermal imagers to construct an automatic system for detecting avian collisions at offshore wind-energy facilities in Denmark. This system could be controled remotely and triggered automatically when a collision was eminent. Betke et al. (2008) studied the evening emergence of Brazilian free-tailed bats using thermal imaging and computer vision analysis. Their image analysis method allowed them to conduct censuses with an accurate and reproducible counting methodology, which was based on the total temporal record of colony emergences. Further, they suggested that similar image analysis methods could be developed to expand census capability to other crepuscular or nocturnal species of bats and birds. Hristov et al. (2008) present new areas of aeroecology research that highlight the use of thermal imaging and computer vision analysis, including population estimates, behavioral observations, thermoregulatory behavior, and bioenergetics (metabolic cost of flight, awaking from torpor, and foraging activities). Hristov et al. (2010) used a combination of thermal infrared imaging and computer vision analysis to provide an effective method for estimating colony size and emergence behavior of Brazilian free-tailed bats. Many of these papers are reviewed in Chapter 10 and are mentioned here because of the effort to bring automation to the detection and/or counting tasks.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780128033845000117

Thermal Imagers and System Considerations

Kirk J. Havens , Edward J. Sharp , in Thermal Imaging Techniques to Survey and Monitor Animals in the Wild, 2016

Abstract

In this chapter the history of thermal imaging is briefly reviewed and a number of applications are identified before a discussion of how a thermal image is formed and what a thermal image looks like and why it looks that way. As a first step in choosing an infrared imager for use in animal studies and population surveys in the wild we examine the performance parameters provided by both the mid-wave IR band (MWIR) and long wavelength IR band (LWIR) thermal imagers, which feature high thermal sensitivity, spatial resolution, and thermal resolution. Another parameter of importance (signal-to-noise ratio) can be used to give physical insight into the significance of these parameters and how they relate to features we see in images.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780128033845000075

Optical Radiation Models

Robert A. Schowengerdt Professor , in Remote Sensing (Third edition), 2007

Surface-emitted component ( $L_{\lambda }^{eu}$ )

The primary source of energy for thermal imaging is the earth itself, which has a typical temperature of 300K. Different materials on the earth, however, can emit different amounts of thermal energy even if they are at the same temperature. Most materials are not ideal blackbodies with 100% radiative efficiency. The efficiency with which real materials emit thermal radiation at different wavelengths is determined by their emissivity, ε. Emissivity plays a proportionality role in the thermal region much like that of reflectance in the visible; it is defined as the ratio of the spectral radiant exitance of a greybody to that emitted by a blackbody (M _λ in Eq. (2-1)) at the same temperature, and is therefore unitless and between zero and one. The emitted radiance of the earth is therefore,

(2.14) $earth\prime ssurface:L_{\lambda }\left(x,y\right)=\varepsilon \left(x,y,\lambda \right)\frac{M_{\lambda }[T\left(x,y\right)]}{\pi }$

.It is implied in this equation that different objects or materials on the earth's surface can have different temperatures, and hence different spectral radiant exitances, as well as different emissivities. Note the similarity between this relation and that for the solar reflective region, Eq. (2-7).

To separate the effects of emissivity and temperature, scientists usually assume one or the other is spatially constant. In thermal studies, such as aerial thermal scanning for building heat loss, the emissivity of various roof materials might be assumed equal in order to estimate temperatures. On the other hand, in some geologic applications, temperature might be assumed constant in order to estimate emissivity. The fact that both affect the radiance, however, means that one should be able to justify ignoring the variation of either parameter. The situation is particularly complicated because emissivity can vary with wavelength, and even temperature. An example of the type of analysis needed to separate T and ε effects in the MWIR is given in Mushkin et al. (2005).

The relation between emitted radiance and the source temperature is not obvious from Eq. (2-1) and Eq. (2-14). To get a better feeling for that, we plot in Fig. 2-16 the spectral radiance as a function of temperature for three fixed wavelengths in the TIR, assuming constant emissivity. The temperature range, 250K to 320K, includes normal daytime and nighttime temperatures on the earth. We see that spectral radiance is approximately linear with temperature over this range, and for any smaller range, as might actually be encountered in a thermal image, a linear approximation is even better. Thus, for our purposes, we can approximate Eq. (2-14) by,

FIGURE 2-16. The dependence of radiant exitance from a blackbody on its temperature at three wavelengths. Emissivity is held constant at one, whereas it actually can vary with temperature and wavelength for a greybody. The temperature range depicted is that for normal temperatures at the earth's surface.

(2.15) $earth\prime ssurface:L_{\lambda }\left(x,y\right)\approx \varepsilon \left(x,y,\lambda \right)\frac{[a_{\lambda }T\left(x,y\right)+b_{\lambda }]}{\pi }$

where a _λ and b _λ are relatively weak functions of wavelength λ. In practice, these coefficients would be given by their average over the spectral passband of the sensor (Chapter 3). Our intent with Eq. (2-15) is to provide a simple, easily-visualized relationship between radiance and temperature; in effect, a complex system has been approximated by a linear relationship that applies over a limited temperature range. If one's goal is to actually calculate temperatures from remote-sensing imagery, then the accurate form of Eq. (2-14) should be used.

The radiation emitted from the earth is transmitted by the atmosphere along the view path to the sensor,

(2.16) $\begin{array}{cc}at-sensor:& \begin{array}{lll}{L}_{\lambda }^{eu}\left(x,y\right)\hfill & =\hfill & {\tau }_v\left(\lambda \right)L_{\lambda }\left(x,y\right)\hfill \\ \hfill & =\hfill & \varepsilon \left(x,y,\lambda \right)\frac{{\tau }_v\left(\lambda \right)\left[a_{\lambda }T\left(x,y\right)+b_{\lambda }\right]}{\pi }\hfill \end{array}\end{array}$

.

The atmosphere's transmittance from 2.5 μm through the thermal IR is shown in (Fig. 2-17). There are four distinct spectral windows available for remote sensing in this spectral region, as determined by molecular absorption bands. As mentioned earlier, the solar energy contribution in this spectral region is relatively small; the rapid fall-off in solar irradiance above 2.5 μm is shown in Fig. 2-18.

FIGURE 2-17. Atmospheric transmittance (solar elevation = 45°) in the midwave IR and thermal IR spectral regions. This curve, like those in the visible and shortwave IR, is obtained from the atmospheric modeling program MODTRAN. The curve appears more detailed than Fig. 2-4 because the abscissa covers six times the range in wavelength with about the same spectral resolution.

FIGURE 2-18. Solar irradiance in the midwave and thermal IR regions (solar elevation = 45°). The ratio of these two curves is the path transmittance depicted in Fig. 2-17.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B978012369407250005X

Contribution of Remote Sensing for Crop and Water Monitoring

Dominique Courault , ... Amanda Veloso , in Land Surface Remote Sensing in Agriculture and Forest, 2016

4.4.1 Evapotranspiration estimation using a simplified model of the water balance

The methods of AET assessment based on thermal imaging have the advantage of taking into account the real state of the vegetation and the water content of the soil at the time of image acquisition, but they are difficult to extrapolate over time. The surface temperature varies very quickly depending on many different factors (surface humidity, wind, etc.) and there is currently no thermal satellite data in orbit with high spatial and temporal resolution. This constraint is a real issue when applying these methods at farm scale for crop and water monitoring.

Another way to produce a spatial AET map is using transfer models simulating the exchanges between the soil, vegetation and atmosphere (commonly called SVAT models). They require climate forcing, information on the current ground vegetation and the hydrodynamic characteristics of the soil. Among the different modeling approaches, a family of widely used methods for agricultural applications to estimate the water balance for crops is the FAO approach proposed by Allen [ALL 98]. One of these methods (called the double coefficient method) calculates AET by distinguishing transpiration and evaporation from the following equation:

[4.1] $AET=\left(KcbKs+Ke\right)ET0$

where ET0 is the reference evapotranspiration of a standard surface of well watered grass (ET0 is calculated using temperature and humidity, net radiation and wind speed using the Penman–Monteith formula); Kcb is the basic crop coefficient reflecting the potential transpiration of the plant under unstressed water conditions (usually ranges from a minimum of 0 in the absence of vegetation to a maximum of 0.7 to 1.15 depending on the crops); Ks is a stress factor which modulates transpiration according to the water availability in the root zone (Ks varies between 1 in the absence of stress and 0 in case of maximum water stress); Ke represents the evaporation coefficient which controls the evaporation of the bare soil fractions depending on the surface moisture (Ke equals 0 in the absence of evaporation and has a physically constrained maximum value of 1.1 to 1.3). A simplification of the dual coefficient method, called the single coefficient method, includes transpiration and evaporation for unlimited water conditions into a single coefficient Kc (Kc   =   Kcb   +   Ke). In Allen et al's publication [ALL 98], various tables of Kc and Kcb values are given for many crops, which vary depending on their phenological stage (Figure 4.17).

Figure 4.17. Kc variation for the four main crop stages

(from Allen et al. [ALL 98])

This method presents a remarkable complementarity with satellite information in the visible range. It has been shown that the Kcb parameter can be estimated from a vegetation index [BAU 87, NEA 89, CHO 94]. The major limitation of this method is the lack of knowledge of soil moisture because this factor varies considerably in space and is for the time not easily obtained from satellite. It is possible to measure it, more or less precisely only at few punctual locations. Additionally, rainfall data can be obtained from meteorological stations but irrigation amounts cannot be known plot by plot over large areas. Another limitation is the absence of satellite data during the frequent cloudy periods [SAA 15]. It is also important to note that soil evaporation simulations do not take into account variations in surface conditions such as the presence or absence of residue, which could have an effect on both the energy budget and on the resistance to water transfer within the soil. Regarding the problem of irrigation, one solution is to simulate the irrigation water supply using a model simulating the water balance on a daily scale, as proposed in the SAMIR tool [SIM 09]. Assumptions about the farmer's irrigation practices must be defined to apply this approach (threshold to start irrigation, amounts of water brought to each field, end date of irrigation, minimum time between two irrigation events, etc.).

An experiment in the Tensift basin (Morocco) showed that the use of the SAMIR model with a NDVI time series computed from Landsat TM images (seven dates from January to June 2003) could correctly reproduce the total amount of water used when growing wheat through irrigation. Figure 4.18 shows the simulation results between January 2002 and June 2002 for three wheat plots on which measurements were performed to validate the model. Simulated and observed irrigation water supplies were respectively 216, 207, 223   mm and 251, 220, 180   mm, indicating a correct agreement between observations and simulations. However, the simulated irrigation dates were often shifted compared with the observations, because the farmer has many constraints that are not included in this simplified approach. The first plot (Figure 4.18(a)) shows a good response of the model (shown in red) compared with observations (shown in black). The second plot (Figure 4.18(b)) shows a strong discrepancy between simulated and observed evapotranspiration in early February because the observed irrigation water input was not simulated by the model. For low vegetation cover, evaporation of bare soil is the dominant process and evapotranspiration is therefore strongly influenced by the dates of water input. For dense vegetation, the time shift for irrigation dates has less impact, because transpiration is the main process and it remains stable as long as there is no stress. For the third plot (Figure 4.18(c)), a sharp drop was observed between March and April, which corresponds to a 10-day stress period that has not been reproduced by the model which uses an automatic irrigation mode (explaining the significant overestimation of the simulated water amount on this plot).

Figure 4.18. Simulation of water balance of three irrigated wheat plots in the Haouz plain (Morocco) by the SAMIR model used in automatic irrigation mode. On the left (in mm): ET0 – reference evapotranspiration, ETobs (black line) – evapotranspiration measured using the turbulent method, ET (red line) – the simulated evapotranspiration. NDVI interpolated from images is represented by a green line. On the right (mm): green vertical lines represent actual irrigation (Ir_obs) and blue vertical lines represent simulated irrigation (Ir_auto). Rainfall is shown by the brown vertical lines. For a color version of this figure, see www.iste.co.uk/baghdadi/3.zip

These results show that if the model provides overall satisfactory information on evapotranspiration and the amount of water supplied by irrigation after aggregation across several plots or a season, the application of this model for piloting irrigation at field scale requires additional information, in particular to take into account the actual irrigations applied by the farmer.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9781785481031500042

Properties of Thermal Signatures

Kirk J. Havens , Edward J. Sharp , in Thermal Imaging Techniques to Survey and Monitor Animals in the Wild, 2016

Spectral domain

As pointed out in the previous section, thermal imaging systems only respond to a small portion of the spectrum. They only detect radiation from the background and animals within the spectral range of the imager being used. Additionally, the background and animals generally only emit a fraction of the radiation emitted by a blackbody. Recall from Chapter 5 that we assumed that Kirchhoff's law equation (5.2) was wavelength independent and that the transmittance (τ), absorptance (α), and reflectance (ρ) had no spectral characteristics, which in general is not true. We can rewrite equation (5.2) showing the spectral dependence as

(9.1) $\tau (\lambda )+\alpha (\lambda )+\rho (\lambda )=1$

Up until now we have treated the emissivity as a constant value ɛ < 1 for opaque objects τ(λ) = 0, such as animals and objects in the background of a scene, as graybodies. The emissivity of real emitters such as animals and objects in a scene is not constant but depends on wavelength (Figure 9.4). This is due to a number of factors, including selective absorption and reflection of the surfaces, local atmospheric conditions, and the dependence of emissivity on the surface conditions of the animals at the time the imagery is collected. As a result, real emitters can only be considered approximations to graybodies. The closer the emissivity is to a constant in the spectral range of the imager the better the approximation will be.

Figure 9.4. The emitted power density depends on the emissivity of the animals (real bodies), which are only approximated as graybodies.

For fieldwork, the sun is the source of heat for creating temperature differences between objects in a scene and for animals with respect to their backgrounds. The bulk of solar loading occurs in the visible and near IR spectral regions λ < 2 μm (see Figure 9.5). Recall that the sun is a blackbody radiator with a temperature around 6000 K and the earth's ambient temperature (soil, trees, rocks, water, and animals) is about 300 K. The peak of the spectral radiant exitance from the surface of the earth is given by Wien's law (λ _max = 2893/T μm) or 9.64 μm, which lies approximately in the middle of the LWIR spectral range. It is important to note that the peak radiant exitance for the sun is in the visible spectral range at 0.5 μm and that reflected solar radiation in the spectral range from 0.5 μm to approximately 3 μm dominates the radiation leaving the surface of the earth. Beyond 3 μm emission of thermal radiation dominates.

Figure 9.5. The spectral distribution of radiation from blackbodies at various temperatures.

The radiation from the sun peaks in the visible region of the spectrum and the radiation from the earth peaks in the LWIR at 9.64 μm.

The direction and intensity of solar radiation at the surface of the earth is constantly changing throughout the day and naturally occurring objects such as rocks, earth, trees, and surface vegetation are heated by the absorption of this energy according to their absorption coefficients α _SOLAR at these wavelengths. These same objects are constantly remitting this energy in the infrared spectral region with an emissivity ɛ _IR similar to the real emitter in Figure 9.4. Note that this same solar loading will also be occurring for animals but with the caveat that they will also be radiating energy derived from metabolic activity as well. Gates (1980) gives the mean solar reflectance and absorptances for the dorsal and ventral aspects of a wide range of animals including birds, mammals, amphibians, and reptiles. The background and the objects with high absorption in the visible and near infrared (λ < 2 μm) will all heat up from solar loading. The actual temperatures will depend on the mean absorption coefficients at the solar wavelengths α _SOLAR and the emissivity in the infrared ɛ _IR where heat is lost as radiation. The ratio of the absorption to emission determines if an object is heating up or cooling down. For example, if (α _SOLAR/ɛ _IR) is large the object will warm up and if it is low the object will cool down. If solar absorption ceases as a result of sunset or by shading effects from trees or passing clouds then the objects will all cool, regardless of their infrared emission characteristics. Emissivities are usually ɛ > 0.90 for animals in the infrared so they will be cooling or emitting under most conditions in the field.

For the most part thermal images are collected in the far infrared spectral range at either 3–5 μm or 8–12 μm wavelength bands. Wyatt et al. (1980) made a determination from examining the field data gathered by Marble (1967) and Parker (1972) that thermal contrast (ΔT) (the intensity difference between deer and its background) was insufficient to identify to species level. The problem they encountered was that the deer signatures were easily detected (i.e., recorded by the thermal imager) but were imbedded in a background of other thermal signatures of uniform intensity with various shapes that were equally or more intense than those of the deer; this is a situation known as background clutter (see Chapter 11). As a result of a large number of competing signatures of comparable intensity the data was unsuitable for use in an automated data identification system based on the intensity of the signatures gathered but it does not mean that thermal contrast (ΔT) is insufficient to detect, recognize, and identify deer in the field. For the identification of animals, the use of smart and cheap image processing techniques like studying video could work where size and shape might play a role.

The results of these studies formed the basis for the design of a multispectral classification approach for the remote detection of deer ( Trivedi et al., 1982, 1984 Trivedi et al., 1982 Trivedi et al., 1984 ). The effort to develop an automatic recognition system for deer based on spectral reflection data in the near-IR spectral band met with limited success. They reported that when controlled field data collected in a previous study at four wavelengths preselected for a maximum discrimination of deer, green vegetation and dry brush on a snow background was subjected to a multistage pattern recognition algorithm, they obtained a deer detection accuracy of 55.2% with no false counts. The development of a deer detection system is only mentioned here because it was an attempt to improve on a previous method, even though it is not a thermal imaging system but rather a remote sensing system.

Wyatt et al. (1985) determined from emissivity measurements that most biological samples have emission spectra that do not have appreciable wavelength dependent features, whereas reflectance measurements indicated the presence of unique spectral signatures for mule deer (Odocoileus hemionus) and some commonly occurring backgrounds like snow and evergreen. The determination that the emission of biological samples shows little spectral dependence indicates that they are fairly good graybodies in the spectral region measured. It appears that the three sands that were measured show the greatest variation in emissivity (∼20% change). The other objects that were measured would typically be found on the open ranges where the deer would be surveyed either spectrally or thermally and their emissivities were all found to be approximately the same in value (ɛ > 0.9). An open range with brush and juniper provide ideal conditions (Trivedi et al., 1982, Figure 3) for thermal imaging survey since the deer will maintain their body temperature during the diurnal cycle and all of the other objects will not. This sets the stage for thermal imaging surveys where large apparent temperature differences will be the norm.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780128033845000099

Thermal Imaging Applications and Experiments

Kirk J. Havens , Edward J. Sharp , in Thermal Imaging Techniques to Survey and Monitor Animals in the Wild, 2016

Abstract

Research results obtained for surveillance-type applications using infrared thermal imaging cameras are reviewed in this chapter along with a number of papers dedicated to comparing the technique of using thermal cameras with other counting methods. We selected papers that would illustrate ways to optimize certain aspects of thermal imaging when applied to the study of animal ecology. The selection represents a very small segment of research in the area of animal ecology and an even smaller segment of all research efforts utilizing thermal cameras as a tool. The chapter is organized to highlight work carried out to detect, count, and observe mammals, birds, bats, and insects (arthropoda) as well as nests, dens, and lairs. There is also a wealth of literature in other application areas dealing with animal physiology and these will be taken up in Chapter 11.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780128033845000105

Land Surface Temperature

Glynn C. Hulley , ... César Coll , in Taking the Temperature of the Earth, 2019

3.6.2.5 Thermal Imaging of Forest Canopies for Stress Detection

One of the most promising approaches to gain information about the water status of trees is thermal imaging of leaves or forest canopy foliage as a proxy for the energy balance, and thus for concurrent transpiration and plant canopy stress ( Leuzinger and Korner, 2007; Scherrer et al., 2011; Kim et al., 2016; Mildrexler et al., 2016). High-resolution thermal cameras capture fine-scale variations in canopy surface temperatures, showing patterns such as cooler leaves and hotter trunks that are not discernable with coarser resolution data such as the 1-km MODIS LST. This information is important for understanding how carbon assimilation changes with plant stress at local scales, and for assessing how forest structure and physiological processes relate to thermal anomalies.

Scherrer et al. (2011) used high-resolution thermal imagery to track drought stress over a deciduous forest in the NW part of Switzerland during a four-week drought from June 24 to July 22 in bright conditions during noon. The high resolution thermal imagery (0.1   km) allows for identification of individual trees (Fig. 3.21, left panel) and the temperature variation within a given canopy between shaded and sun-exposed canopy fractions (Fig. 3.21, right panel). Multiple thermal images collected over the duration of the drought event captured the increasing canopy temperatures as soil moisture and transpiration decreased and the comparison of species-specific responses to drought between moist and dry sites.

Fig. 3.21. A diverse mixed forest (left) and the same area as a thermal image (right). Note the polygons mark individual tree canopies. Changes in transpiration (latent heat flux) are directly coupled to leaf temperature, and therefore, an increase in canopy foliage temperature at otherwise similar environmental conditions (solar radiation, wind), indicates reduced transpiration.

From Scherrer, D.M., Bader, K.-F., Korner, C., 2011. Drought-sensitivity ranking of deciduous tree species based on thermal imaging of forest canopies. Agric. For. Meteorol. 151, 1632–1640.

Moderate resolution satellite-based thermal imagery does not provide the fine-detail but has the advantage of exhaustive spatial coverage. With 1-km² resolution MODIS Aqua LST data (Mildrexler et al., 2018), computed LSTmax anomalies for the Earth's surface from 2003 to 2014 and found strong spatial association between positive anomalies, heat waves, and droughts, such as the 2003 European heatwave (Fig. 3.22). In 2003 positive LSTmax anomalies extended across the European continent, and under these extreme hot and dry conditions, a variety of land cover types including grasslands, forests, croplands, and urban areas experienced intense anomalies. The summer of 2003 in Europe has been called the hottest summer in Europe in 500   years (Stott et al., 2004), and mean summer temperatures exceeded the 1961–1990 mean by 3°C (Schar et al., 2004). When coupled with drought, heat waves pose significant threats to plant growth and survival (Bréda et al., 2006). The frequency and total land area affected by extreme high-temperatures has increased in recent decades, (Hansen et al., 2012; Barriopedro et al., 2011), and these temperature extremes pose serious threats to human health and life.

Fig. 3.22. The intense positive LSTmax anomalies during the 2003 European heat wave extended across the European continent and affected a variety of land cover types. 2003 MODIS Land Cover data set (Friedl et al., 2010) with classification system abbreviations defined as in Fig. 3.18.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780128144589000034

Integrated Sand Management For Effective Hydrocarbon Flow Assurance

Babs Oyeneyin , in Developments in Petroleum Science, 2015

5.2.3 Sand Sampling from Flowlines, Pipelines, and Process Vessels

Sampling is the 'traditional method' of detecting sand and often uses the same equipment and procedure as a sample for water content. Traditionally, a minimal quantity of liquid is extracted from the flowline through a sample point (side-tapping). The associated gas is vented and the liquid is examined for the presence of sand. Since the quantity of sand that can be expected in a sample is very small, and the grain size is often small (50–250   μm), filtration and weighing plus examination under a microscope provides the best analysis route. If the filtration is conducted as the sample is taken, this may restrict the draw rate from the well to a low flow. Filters are also prone to tearing, rendering the sample worthless.

When sand accumulates in vessels, it reduces the volume available for liquids, which reduces the residence time and reduces the efficiency of oil–water separation. Also, the reduced liquid volume renders liquid level control more difficult as the liquid level becomes more sensitive to changes in inflow rate or outflow rate (valve opening). Accumulated sand may also restrict level control instrumentation off-take lines. Finally, sand deposits may lead to under-deposit corrosion of the vessel wall when the fluids are corrosive (water   +   dissolved CO₂) and the vessel material is not resistant to corrosion (e.g., low-alloy steel).

Historically, visual inspection with the vessel off-line, de-pressurised, and vented was the main method of checking for sand – prior to manual removal of the sand. However, since a number of jet-washing techniques have been devised for removing sand without vessel entry, the requirement to monitor sand accumulation by non-intrusive vessel inspection means has become essential.

Common methods for monitoring sand accumulation in vessels while the vessel is in service are:

•

Thermal imaging: Thermal imaging works by viewing the vessel to be cleaned through a camera, which has been designed to pick up heat signatures. The sand and sludge within the vessel will have a difference in temperature to any other substance in the vessel. The camera will pick up this difference in heat, and a profile of the sand and sludge can be built up, because sand accumulations appear as cold regions at the bottom of the vessel. Thermal imaging is not suitable where the vessel is lagged.

•

Neutron back-scatter: Deploys neutron back-scatter transmission.

•

Gamma transmission: Deploys gamma transmission.

•

Ultrasonics: Ultrasonic methods transmit a pulse of acoustic energy through the vessel wall and into the process fluid. The unit then 'listens' for a reflected signal. The time taken and nature of that signal indicate the nature of the interface that reflected the sound signal.

•

Load cells (weight measurement): Used to monitor the accumulation of sand in the collection vessel beneath de-sanding cyclones.

•

Closed Circuit Television (CCTV): Visual inspection for sand in low-pressure water filled vessels can also be accomplished using a camera deployed through an access fitting.

Read full chapter

URL:

https://www.sciencedirect.com/science/article/pii/B9780444626370000051

Summer Baby Monitor No Signal Nextto Ea H Other

Source: https://www.sciencedirect.com/topics/earth-and-planetary-sciences/thermal-imaging