• Non ci sono risultati.

About the possibility to differentiate between different soils on the basis of infrared thermography

N/A
N/A
Protected

Academic year: 2021

Condividi "About the possibility to differentiate between different soils on the basis of infrared thermography"

Copied!
65
0
0

Testo completo

(1)

SCUOLA DI INGEGNERIA CIVILE, AMBIENTALE E TERRITORIALE

Corso di Laurea Magistrale in Ingegneria per l’Ambiente e il Territorio

Environmental and Land Planning Engineering

About the possibility to differentiate between

different soils on the basis of infrared

thermography

Supervisor:

Prof. Carlo De Michele

Assistant Supervisor:

Prof. Matthias Bernhardt

Master Graduation Thesis by:

Maria Chiara Lippera

Student ID n. 857325

(2)

First I would like to express my sincere gratitude to my supervisor Prof. Carlo De Michele, who supported me in this work realized abroad and showed a deep and truthful understanding of my case.

My gratitude goes as well to Prof. Karsten Schulz, who introduced me and warmly welcomed me at the Institute IWHW, never surrendering to ask also for my progresses with the German language.

A great thanks to my assistant supervisor Prof. Matthias Bernhardt, always motivating my work and giving me trust, as well precious advices.

Especially, I would never stop thanking Claire Brenner, who assisted me every day from the (literally) hard work to my constant questions, showing a great strength and patient, even every time I was walking a step behind her.

Finally I want to express my gratitude to all the people working at the Institute, who grant in giving a really friendly and special atmosphere, which I admire and contributed in making my working period at the Institute, not only an opportunity to learn a lot, moreover a great experience.

(3)

List of Figures iv List of Tables v Abstract 1 Sommario 2 1 Introduction 3 2 Theory 6

2.1 Infrared thermal imaging . . . 6

2.1.1 Infrared spectrum . . . 6

2.1.2 Physical Principles . . . 9

2.1.3 The Radiant Temperature and emissivity . . . 12

2.2 Soil surface temperature . . . 14

2.2.1 Land surface energy balance . . . 14

2.2.2 Variations on soil surface temperature . . . 20

3 Methods 24 3.1 Experimental setup . . . 24

3.1.1 Soil boxes and sensors . . . 24

3.1.2 The lysometers . . . 27

3.1.3 Meteorological station . . . 28

3.1.4 Infrared camera . . . 28

3.2 Collection of data . . . 30

3.3 Temporal and spatial analysis of thermal images. . . 32

3.3.1 Reading out the thermal images . . . 32

3.3.2 Temporal analysis of surface temperature . . . 33

3.3.3 Spatial analysis through principal components analysis . . . 34

4 Results 38 4.1 First analysis on the distribution of data . . . 38

4.2 Temporal and spatial analysis for different soils conditions . . . 40

(4)

4.2.1 Differentiation in dry and wet conditions for soils with vege-tation . . . 41 4.2.2 Differentiation in wet conditions for soils with sparse vegetation 46 4.2.3 Differentiation in wet conditions for bare soils and soils with

vegetation . . . 49

5 Conclusions 55

(5)

2.1 Electromagnetic spectrum, detail of visible and infrared ranges. . . 7

2.2 Atmospheric windows in the electromagnetic spectrum [Jensen and Hall, 2007]. . . 8

2.3 Representation of Planck’s law: spectral specific radiations of the black body depending on the wavelength. . . 10

2.4 Scheme of the energy balance at the land surface. . . 15

2.5 Idealized daily fluctuation of surface soil temperature [Hillel, 1982]. 20 3.1 Photography and model of the instrumental setting for the experi-ment. . . 25

3.2 Soil texture triangle and textural classes for the two type of soil used in the experiment . . . 26

3.3 Sensors set-up within the soil boxes. . . 27

3.4 Camera field of view. . . 29

3.5 Camera display through the program Optris Pi connect. . . 30

3.6 Grass cover for the boxes of sand and loam in the month of August, the layer of organic soil is well visible. . . 31

3.7 Processing of the data from the tir camera. . . 33

3.8 Raster data principal component analysis (PCA) and its spatial visualization [Demšar et al., 2013]. . . 35

3.9 Representation of the matrices computed to perform the principal component maps. . . 36

4.1 Time series of statistics of surface temperature over the sandy and loamy pixels, measuring day 10/08/2017. . . 39

4.2 Time series of statistics of surface temperature over the sandy and loamy pixels, measuring day 10/08/2017. . . 40

4.3 Time series of mean surface temperature for the wet and dried boxes with vegetation, measuring day 18/08/2017. . . 41

4.4 Cumulative evaporation rates for the wet boxes, measuring day 18/08/2017. . . 42

4.5 First four principal component images for the whole TIR image, measuring day 18/08/2017. . . 43

4.6 Graph of the eigenvalues for the first 7 principal components, mea-suring day 18/08/2017. . . 44

(6)

4.8 Observations in the space of the first two principal components, measuring day 18/08/2017. . . 46 4.9 Time series of mean surface temperature for the wet boxes with

sparse vegetation, measuring day 06/07/2017. . . 47 4.10 Cumulative evaporation rates for the sand and loam watered boxes,

measuring day 06/07/2017. . . 47 4.11 Time series of wet boxes with sparse vegetation, first and second

principal component image, measuring day 06/07/2017. . . 48 4.12 Observations in the space of the first two principal components,

measuring day 06/07/2017. . . 49 4.13 Time series of mean surface temperature for the bare soils and soils

with vegetation, measuring day 13/09/2017. . . 50 4.14 Photograph of the boxes used for the measurements of bare soil and

soil with vegetation. . . 50 4.15 Cumulative evaporation rates for the bare soils and the soils with

vegetation, measuring day 13/09/2017. . . 52 4.16 Time series of wet boxes for the bare soils and the soils with

vege-tation, first and second principal component image, measuring day 13/09/2017. . . 53 4.17 Observations of only the bare soils in the space of the first two

principal components, measuring day 13/09/2017. . . 53 4.18 Observations in the space of the first two principal components,

measuring day 13/09/2017. . . 54

List of Tables

2.1 sub-bands of the Infrared region . . . 8 3.1 Sensors depths within every soil box . . . 27 3.2 Specifics of the sensors within the Meteorological station . . . 28 4.1 Eigenvalues and cumulative variance explained for every principal

component, measuring day 18/08/2017 . . . 44

(7)

From the intuition of relating dominant thermal patterns, derived from principal components (PC) maps, to physical properties of soils, the Institute of Water

Management, Hydrology and Hydraulic Engineering (IWHW), University of Natural Resources and Applied Life Sciences Vienna (BOKU), has performed studies to

estimate topsoil texture maps through time series of thermal infrared (TIR) remote sensing. High-resolution soil texture maps at the catchment scale are fundamental for the estimation of hydraulic parameters, affecting subsurface water and solute transport processes.

This master thesis aims at evaluating the effectiveness of this new method at a local scale using a high-temporal-resolution TIR time series. The project consisted of an open air experiment carried out at BOKU. The work is necessary to understand how energy exchange between the soil and atmosphere affects the soil surface temperature signature for different soil textures and to analyse if time series of TIR imagery allow for the differentiation between these soil textures. Starting from the acquisition of TIR images from soil plots with two different soil textures, the temporal and spatial variation of soil surface temperature is analysed under varying soil moisture states and natural covers. In analogy to the preceding work, a principal component analysis was performed on the data set. The results lead to the conclusion that differentiating between different soil textures from TIR images is not possible in presence of dense vegetation, while for the case of bare soils, the two soils appear easily differentiable. For further developments, the experiment could be repeated using longer time series and other vegetation covers, taking also into account the influence of the underlying soil texture on the water stress of the vegetation.

(8)

Partendo dall’ intuizione di legare le proprietà fisiche del suolo ai principali pattern termici derivanti dall’analisi delle componenti principali, l’Institute of Water

Man-agement, Hydrology and Hydraulic Engineering (IWHW), University of Natural Resources and Applied Life Sciences Vienna (BOKU), ha condotto diversi studi per

la stima di mappe della tessitura dello spessore superficiale del suolo, sulla base di serie temporali di immagini telerilevate sull’infrarosso termico (TIR). Mappe della tessitura del suolo ad alta risoluzione spaziale sulla scala di bacino, sono fondamentali per la stima dei parametri idraulici, i quali influenzano i processi di trasporto di acqua e soluti nel primo spessore di terreno.

Questa tesi di laurea si propone di valutare l’efficacia di questo nuovo metodo, a una scala locale e utilizzando un’alta risoluzione temporale per le immagini TIR, essendo il progetto basato sulla realizzazione di un esperimento all’ esterno, condotto presso la BOKU. Questo lavoro è necessario per comprendere come gli scambi di energia tra atmosfera e suolo regolano e caratterizzano la firma della temperatura superficiale del suolo per diverse tessiture di terreno, e analizzare se le serie temporali di immagini TIR consentono di differenziare tra queste. Partendo dall’ acquisizione di immagini TIR, di diversi campioni di terreno, sono analizzate le variazioni spaziali e temporali delle temperature superficiali, testate diversi stati di umidità e coperture naturali. In analogia con studi precedenti, un’analisi delle componenti principali è stata eseguita sul set di dati. I risultati conducono alle conclusioni che non è possibile differenziare tra diverse tessiture di suolo a partire dalle immagini TIR in presenza di una fitta vegetazione, mentre per suoli spogli, questi appaiono facilmente differenziabili. Riguardo i futuri sviluppi, si potrebbero effettuare ulteriori sperimentazioni utilizzando una serie temporale più lunga e altri tipi di vegetazione, tenendo anche conto dell’influenza che il tipo di suolo sottostante ha sullo stress idrico della vegetazione stessa.

(9)

Introduction

Acquisition of territorial data is an essential task for engineering works, as the territorial information constitutes the basic element of knowledge for all land man-agement policies. Geographical information systems are a fundamental support for field decisions such as land evaluation, environment protection, spatial planning, security policies and transport. The importance of remote sensing has grown considerably addressing the task of data acquisition.

Remote sensing represents an innovation in environmental monitoring as its po-tentialities and results are remarkable in obtaining with relative effortlessness the land informations, in particular environmental parameters. The advantages rely on the short-time, as measures can be repeated over time or almost continuously, in the distance, which leads to a large spatial coverage, and on the greater objectivity and overall economy compared to the traditional data survey methods.

Nowadays, one of the still ongoing challenges for environmental engineering is the direct retrieval of soil maps using remotely sensed images.

Soil maps are a geographical representation which display, for the area of interest, the diversity of soil types or soil properties, as soil texture, organic matter, depth of horizons and others. Conventional soil property measurement methods, depend on physical analyses in a laboratory, are expensive, require a large number of samples and involve a long analysis to obtain the spatial distribution of those properties over large areas, containing also an estimation of the uncertainty within interpolation and extrapolation models. Particularly, for hydrologic modeling purposes, as the prediction of (sub-)surface water and solute transport processes at the basin scale, spatially distributed hydraulic parameters are required, to characterize the site of the system being studied.

Soil hydraulic parameters, as for example soil hydraulic conductivity, infiltration rate and water holding capacity, are fundamental for hydrological transport models, in order to predict how the water flux is moving towards soil sub-layers. The estimation of hydraulic properties relies on indirect methods via pedo-transfer

(10)

functions, for which are required soil parameters as soil texture, bulk density, and other soil chemical properties, like organic content. Especially soil texture, which represents the relative percentages of various particle sizes in a soil, has a significant impact on many hydraulic properties, as it can be a first base for assessment of infiltration characteristics. Unfortunately, as already mentioned, generally texture informations are taken from different sources, varying in the method of production, and presenting varying coarse spatial resolutions and quality.

The Institute of Water Management, Hydrology and Hydraulic Engineering, from the University of Natural Resources and Life Sciences in Vienna, in collaboration with the Department of Geography, Ludwig-Maximilians-Universität in Munich, and the Institute of Water and River Basin Management, Karlsruhe Institute of

Technology in Karlsruhe, Germany, in the last years has pursued the improvement

of the generation of spatially distributed topsoil maps via the use of thermal remote sensing products. In the article "Estimating spatially distributed soil texture using

time series of thermal remote sensing - a case study in central Europe" [Müller

et al., 2016] is presented a method to derive high-resolution (15 m) spatial topsoil texture patterns for the meso-scale Attert catchment (Luxembourg, 288 km2) from 28 images of ASTER (advanced spaceborne thermal emission and reflection radiometer) thermal remote sensing.

From the previous work, "Identification of catchment functional units by time series

of thermal remote sensing images" [Müller et al., 2014], a principal component

analysis (PCA) was conducted for a statistical extraction of dominant patterns, within the ASTER thermal infrared time series for the same catchment. From the resulting PCs, a comparison with the available data was made, and due to the similarities it was suggested a strong relationship between the first two principal components and, respectively, the land use data and geological information. In the last work [Müller et al., 2016] was then applied a multilinear regression estimator (MLRE) to establish and estimate a functional relationship between the PCs and the fractional texture information. For the calibration and evaluation of the model were used multiple soil texture samples within the catchment. The MLRE prediction gave an overall root mean squared error of 12.7 percentage points, using the full data set for calibration. The result is considered good if compared to the range of uncertainty of recent studies on soil texture estimation, but further investigations are needed with different setups of spatial and temporal resolution.

Hence the demand to analyse through a field experiment if it is possible to observe the influence of soil properties on dynamic changes in time series of thermal remote images and how soil processes affect the characteristics of the temperature signature. The experiment was set in the terrace of the Muthgasse building, in the University of Natural Resources and Life Sciences in Vienna.

Over the experimental set up, is simulated at the local scale what happens during the acquisition of land surface temperature data, through a thermal infrared camera.

(11)

The installation of a meteorological station and monitoring of the different soil types via lysometers and sensors, helps in assessing the interactions between atmosphere and soil, existing through exchanges of energy fluxes.

The aim of the experimental work is the examination of the effect of soil texture on the surface temperature as well as its variability in space and time, for which two type of soils are taken in exam, the loam and the sand, which present a easily distinguishable soil texture. It is expected that the different soil textures will characterize the heat transfer processes taking place for the same atmospheric conditions, and thus the resultant temperature signatures, visible from the time series of thermal infrared images, even under vegetation cover and for different soil moisture conditions. The described project consisted in the active participation in the planning and practical implementation of the experiment from its new complete set-up, the monitoring of measurements, and the consequent analysis of the gained data, interpreting the spatial and temporal variability of surface temperature for the different soil types.

In the first section regarding the theory, are illustrated the principles of the infrared thermography and the physical processes that regulate the soil temperature and its variation. In Sect. 3 are reported the methods, where is shown the experimental setup and its implementation, the collection of measurements, the data processing through computer elaborations and the temporal and spatial analysis. In Sec. 4 are reported the results of the data analysis, which are accompanied by their discussion. In the end in Sec. 5 are reported the overall conclusions and the suggestions for further developments.

(12)

Theory

2.1

Infrared thermal imaging

Thermographic cameras detect and record the electromagnetic spectrum corre-sponding to the thermal infrared (TIR) to produce images, called thermograms. Thermal infrared imaging cameras detect the radiation in a similar way as op-tical cameras detect visible light. The data obtained from the those particular wavelengths of the TIR region are strongly correlated with the thermal aspects of the body object of study. All bodies with temperature above absolute zero(0 °K; -273.16 °C) emit energy due to their thermal status: the amount of energy emitted depends on the emissivity and on the body temperature in question. An higher surface temperature means a greater vibration of the molecules that constitute the object, which is equivalent to a greater radiating intensity at the lower wavelengths. The images in the TIR spectral window characterize uniquely the materials and in thermal remote sensing, the properties of the the surfaces can be investigated, such as the mineral composition, soil moisture and geothermal abnormalities.

2.1.1

Infrared spectrum

The electromagnetic spectrum is a continuous mono-dimensional distribution of electromagnetic energy ordered for increasing wavelengths λ. The distribution of the varying types of energy, in the physical reality of the electromagnetic spectrum, is a continuum, but conventionally for practical reasons is divided it into different ranges or bands.

Infrared radiation (IR) is the electromagnetic radiation between 700 µm and 1 mm, with wavelengths greater than the ones of the visible and smaller than the microwaves and radio waves. Any object with a temperature higher than the absolute zero spontaneously emits radiation in this band. For an object with

(13)

increasing temperature, the emitted radiation has decreasing wavelengths moving towards the visible range, until the object becomes glowing, with wavelengths in the visible red. As it can be seen in Fig. 2.1 for decreasing wavelengths, the visible region, between 0.39 and 0.75 µm, follows the one of the infrared, which is divided into smaller portions.

Figure 2.1: Electromagnetic spectrum, detail of visible and infrared ranges.

In the infrared band there are five main sub-bands: Near Infrared (NIR) and Shortwave Infrared (SWIR), used in remote sensing for phenomena exclusively of reflection type due to the presence of a strong reflection component together with the part of emitted radiation; Mid-wave Infrared (MWIR) for phenomena of both reflection and emission type; long or thermal IR (TIR) for essentially emissivity phenomena, as it will be explained further, while the sub-band in the Far Infrared (FIR) is less employed beyond the 20 µm wavelength. In Tab. 2.1 are reported the most common sub-bands for the IR range, used in remote sensing, with the corresponding temperatures of black bodies for which spectral peaks fall at the given wavelengths, according to Wien’s displacement law, which will be explained further.

The region between 3–35 µm is called the thermal-infrared region, as the radiation emitted by the Earth due to its thermal state is far more intense than the solar reflected radiation. For this reason all the sensors operating in this region are primarily detecting the thermal radiative properties of ground materials [Gupta, 1991].

In the thermography field, within the thermal-infrared region, cameras are made for mainly two spectral ranges: the MWIR from around 3 to 5 µm, and the LWIR around 8 to 14 µm. The restriction to these wavelengths is due firstly to some

(14)

Abbreviation Wavelength Photon Energy Temperature NIR 0.75–1.4 µm 413–886 meV 3591–1797 °C SWIR 1.4–3 µm 413–886 meV 1797 – 693 °C MWIR 3–8 µm 155–413 meV 693 – 89 °C LWIR 8–15 µm 83–155 meV 89 – -80 °C FIR 15–1000 µm 1.2–83 meV -80.15 – -270.15 °C

Table 2.1: sub-bands of the Infrared region

considerations on the amount of radiation to be expected, secondly on the physics of detectors, and last on the transmission properties of the atmosphere [Vollmer and Möllmann, 2017]. Beyond 14 µm, special detectors cooled to very low temperatures are required, and the atmosphere becomes very opaque. Spectral bands must be in fact located in the so-called atmospheric windows: spectral regions in which the atmosphere is particularly transparent. Generally, through the greenhouse gasses, the atmosphere function as a selective filter of the spectrum frequencies. As it

Figure 2.2: Atmospheric windows in the electromagnetic spectrum [Jensen and Hall,

2007].

can be seen in Fig. 2.2, most of the absorption of the infrared radiation in the atmosphere is done by water molecules and carbon dioxide. Absorption results weak in the regions 7–8.5 µm and 11–14 µm, and non-existent between 8.5 µm and 11 µm, leaving a “hole” in the infrared absorption spectrum [Manahan, 2001]. In this range is also found the peak of the radiation emitted from the Earth’s surface, since at ambient terrestrial temperatures, the peak of the Earth’s blackbody radiation occurs at around 9.7 µm, which indicates the highest energy available for sensing in this region.

(15)

All those factors make the range between 8 and 14 µm an excellent window, the most used for studying the thermal properties of terrestrial surface. Using this wavelengths interval, the remote sensing in the TIR region has been generally passive, which means the sensors collect data on the naturally emitted radiations.

2.1.2

Physical Principles

The definition of the electromagnetic spectrum and the physical laws which es-tablish qualitative and quantitative correlations to describe infrared energy, are related to the names of the highest European physical scientists of the twentieth century: Max Planck, Josef Stefan, Ludwig Boltzmann, Wilhelm Wien and Gustav Kirchhoff:

• Planck’s Law;

• Stephen Boltzmann Law; • Wien’s displacement Law; • Kirchhoff’s radiation Law.

These laws define the interaction between electromagnetic radiation and matter applied to a black body, which is a perfect radiator.

The perfect blackbody radiator is defined as an ideal object, that absorbs all the incident electromagnetic radiation, which is then re-emitted in its maximum amount as a function of solely temperature. This is because the black body is a diffuse radiator, as the energy is radiated isotropically, making the concentration of the radiation not dependent on the direction of the emission, but only on its temperature. In practice, the blackbody is a perfect absorber and a perfect radiator, with the maximum efficiency, for every wavelength. At a particular temperature, the black body would emit the maximum amount of possible energy for that temperature. This value is known as radiation of the black body, for which is possible to define therefore, standard radiation curves of a black body generated for each temperature, showing the correspondent radiant energy for each wavelength, as visible in Fig. 2.3. The radiation law by Planck describes the spectral specific radiation Mλ of the black body depending on its temperature T and the wavelength

λ:

=

2π h c2

λ5(e(λ kThc )− 1) (2.1)

where Mλ is energy flux emitted in a given wavelength range (expressed as

W m−2µm), h is the Planck’s constant and K is the Boltzmann’s constant, while c

is the speed of light.

Integrating the Planck’s law on all wavelengths of the spectrum, it can be expressed the total energy emitted by a black body at all wavelengths. According to the

(16)

Figure 2.3: Representation of Planck’s law: spectral specific radiations of the black body

depending on the wavelength.

Stefan–Boltzmann law, the entire emitted radiation, is proportional to the fourth power of the absolute temperature T of the surface. This law, indicating with Mb

the total emitted radiation of a black body, is formulated as:

Mb = σ T4 (2.2)

where σ = 5.67 × 10−8W m−2K−4 is the Stefan-Boltzmann constant and T is the absolute temperature (K) of the surface.

This equation is valid for blackbodies, while for real object it can be rewritten so that it pertains the total spectral radiant flux of real object:

Mr = ε σ T4 (2.3)

The emissivity ε is used as a material constant factor to describe how well an object radiates energy compared to the perfect blackbody radiator. For a given wavelength range, is the ratio between the radiant flux exiting a real-world selective radiating body at temperature T, Mλ,r(T) and a blackbody at the same temperature

Mλ,b(T):

ελ =

Mλ,r(T )

Mλ,b(T )

(2.4) All selectively radiating bodies have emissivities ranging from 0 to <1 that fluctuate depending upon the wavelengths of energy being considered. A graybody outputs a constant emissivity that is less than one at all wavelengths.

(17)

Going back to Fig. 2.3, representing Planck’s law, can be seen that the wavelength, at which can be found the maximum of the emitted radiation of a black body, shifts when temperatures change. The absolute temperature also determines the wavelength distribution of the emitted energy, for which Wien’s displacement law states that the wavelength of maximal radiation intensity λmax is inversely

proportional to the absolute temperature:

λmax =

2898

T (2.5)

where λmax, the maximum wavelength at fixed temperature, is in micrometers.

From the Wien’s displacement law comes the choice of the best wavelengths and instruments to be used in remote sensing, given the properties of the objects under investigation.

The last law from Kirchhoff, expresses the principle of energy conservation. When an electromagnetic radiation hits a surface, three mechanisms of interaction are developed: one part of the incoming radiation flux is reflected, one part is absorbed, and another part is transmitted. Conservation of energy implies that the amount of incident energy is equal to the sum of the absorbed, reflected, and transmitted energy. When those portions are normalized respect to the total incident radiation, is obtained:

ρ + α + τ = 1 (2.6)

with ρ the coefficient of reflection, α the coefficent of absorption and τ the coefficent of transmission. When bodies are opaque, the value of the transmissivity is negligible, but also those coefficients are accompanied by spectral variability. There are few materials that transmit energy efficiently in the infrared region between 7 and 14µm. Germanium is one of the few good transmitters of infrared energy and thus it is used frequently as lens material in infrared cameras. Therefore generally is applied the following:

α = 1 − ρ (2.7)

Always Kirchoff, stated that in the infrared portion of the spectrum, the spectral emissivity of an object generally equals its spectral absorptance: α ∼ ε. Applying this concept that a good absorber is an excellent emitter, the equation can be rewritten:

ε = 1 − ρ (2.8)

So for objects that do not transmit energy, there is a simple balance between emissivity and reflectivity. If emissivity increases, reflectivity must decrease, and so the contrary. The emissive and reflective behavior of most materials is similar in the visible and in the infrared regions of the electromagnetic spectrum. But also at the contrary, some materials that are good absorbers, transmitters, or reflectors in the visible, may exhibit completely different characteristics in the infrared. Going back to the black body, the reflected energy would be almost non-existence, as it is

(18)

a perfects absorber and also a perfect radiatior, so the relationship is:

ε = α = 1 (2.9)

This expression of Kirchhoff’s law sums up the already seen concept of a black body: surface that absorbs much, emits much and whose reflection is negligible.

2.1.3

The Radiant Temperature and emissivity

At the base of thermal infrared remote sensing, radiometric measurements in the thermal infrared are aimed at the determination of the temperature T, for which is essential to make some distinctions, as different typologies of temperature exist. The particles energy of matter in random motion is called kinetic heat, which is also referred to as internal, real, or true heat. Heat can be measured in calories, and as already said, objects having a temperature above absolute zero exhibit this random motion. When these particles collide they change their energy state and emit electromagnetic radiation. Measures of the true kinetic temperature (Tkin)

or concentration of the heat, can be done physically touching the object using a thermometer.

The object’s internal kinetic heat is also converted to radiant energy, as it has been seen in the previous paragraph, which is called external or apparent energy. The electromagnetic radiation exiting an object is the radiant flux [Wm−2], and the concentration of the amount of radiant flux emitted from the object is indicated as its radiant temperature (Trad). There is a positive correlation between the true

kinetic temperature of an object Tkin and the amount of radiant flux radiated from

the object Trad, that will be shown further.

Therefore, radiometers and all infrared sensors can be used, although placed distant from the object, to measure its radiant temperature which correlates with the object’s true kinetic temperature [Jensen and Hall, 2007].

As already explained, the temperature of an object and its emissivity are the two factor that define how much infrared energy an object will emit, and so they control the apparent temperature detected by thermal infrared sensors. These instrumentations calculate the surface temperature on the basis of the emitted infrared radiation from an object. The intensity of infrared radiation, which is emitted by each body, depends on the temperature as well as on the radiation features of the surface material of the measuring object.

Infrared cameras detect and measure the sum of infrared energy over a range of wavelengths determined by the sensitivity of the camera’s detector. These cameras cannot discriminate energy at 7 µm from energy at 14 µm, in the way the human eye can distinguish various wavelengths of light as colors. They calculate the temperature of the objects by detecting and quantifying the emitted energy over the operational wavelength range of the detector. Temperature is then calculated

(19)

by relating the measured energy to the temperature of the blackbody, seen in Eq. 2.2.

Since the emissivity of a real object affects how much energy an object emits, emissivity influences the infrared camera’s temperature calculation. The emissivity

ε is essential parameter to determine the internal temperature through radiation

measurements.

Assuming that the emissivity characteristics of an object is known, the Stefan-Boltzmann law in Eq.2.3 is considered, pertaining the total spectral radiant flux of the real object, and is related with the emitted energy of the blackbody in Eq. 2.2:

σ T4 = ε σ T4 (2.10)

Therefore, the radiant temperature of an object recorded by an infrared camera is related to its true kinetic temperature and emissivity by the following relation-ship:

Trad = ε1/4Tkin (2.11)

For this reason, thermal cameras when pointing at a specific body requires the setting of the emissivity of the object, in order to have the apparent radiant tem-perature recorded equal the true kinetic temtem-perature of the object. While for the blackbody, the ideal body reported in the left side of Eq. 2.10, ε = 1 and the radiant temperature coincides with the kinetic temperature.

For an example, of two objects at the same temperature, one having high emissivity and the other low, even though the two objects have the same kinetic tempera-ture, the one with the low emissivity will radiate less energy. Consequently, the temperature calculated by the camera will be lower than that calculated for the high emissivity object. Generally, natural bodies do not absorb all the incident energy or emit all that stored, so their values of emissivity are always lower than 1. Since the measured radiant temperature would be always lower than the current kinetic surface temperature, when recording in field measurements, as an example, for vegetated surfaces often an emissivity of 0.98 is assumed.

Summarizing what has been explained in this section, in soil surfaces the factors that most control the absorption of incident radiation and the emission of thermal energy due to their emissivity values are [Jensen and Hall, 2007]:

• color: darker colored soils are usually better absorbers and thus emitters than lighter colored ones, which tend to reflect more of the short-wave incident radiations.

• surface roughness: the greater the surface roughness of the soil, relative to the size of the incident wavelength, the greater is the surface area of the object, and thus its potential for absorption and re-emission of energy, moreover also the degree of soil compaction effect emissivity.

(20)

• moisture content: the more moisture a soil contains, the greater its ability to absorb energy and become a good emitter, thus wet soil particles have a high emissivity similar to water.

2.2

Soil surface temperature

In the previous section has been explained how through thermography is possible to detect the surface temperature of an object, and thus of the soil.

Soil is a complicated system, for which is necessary to understand the physical processes that regulate its temperature and variations in space and time, governed by different parameters. Those parameters that influence the surface temperature are incoming radiation, land use, albedo and available water content. Especially the latter is strongly controlled by soil texture, which subsequently influence the thermal inertia signature as given by the temporal patterns of surface temperature [Müller et al., 2016]. In order to know how the firsts listed parameters are influencing the surface temperature, is necessary a brief narration on the land surface energy balance.

2.2.1

Land surface energy balance

Energy exchange at the soil surface and the temperature conditions that result, are fundamental to understand of how soil properties interact with the physical environment [Wullschleger et al., 1991].

Soil temperature varies in response to changes in the radiant, thermal, and latent energy exchange processes that take place primarily through the soil surface. The effects of these phenomena are propagated into the soil profile by a complex series of transport processes, the rates of which are affected by time-variable and space-variable soil properties. The parameters, which are pertinent to the soil, are the specific heat capacity, thermal conductivity and thermal diffusivity [Hillel, 1982], which will be seen further.

The exchange of energy between the Earth’s surface and the overlying atmosphere involves four important processes: absorption and emission of the incoming radi-ation, thermal conduction, turbulent transfer of sensible and latent heat. Each of these processes can be associated with an energy flux density, defined as the rate of transfer of energy across a surface of unit area. In SI units, those energy flux densities have always units of joules per second per square meter (J s−1m−2), which is the same as Watts per square metre (W m−2).

The absorption and emission of the incoming radiation, is summarized using the term “net radiation”. This transfer of energy is also indicated as "radiative flux", and is given by the balance of incoming and outgoing short-wave and long-wave

(21)

radiation emitted respectively from the Sun and the Earth.

The second process of energy transfer, which has been listed before, is the con-duction: the propagation of heat within the ground, by internal molecular motion. Since temperature is an expression of the kinetic energy of a body’s molecules, the existence of a temperature difference within a body will normally cause the transfer of kinetic energy by the collisions of rapidly moving molecules, from the warmer region of the body to their neighbours in the colder region.

The turbulent transfer of heat energy towards or away from the surface within the atmosphere, is known as sensible heat flux. The energy goes heating the air, and it involves the movement of a heat-carrying masses. Convection occurs when there is vertical circulation and mixing in the atmosphere.

The last process is know as latent heat flux and involves the evaporation of water stored in the soil or the condensation of atmospheric water vapour into the surface. The prime example is the process of distillation, which includes the heat-absorbing stage of evaporation, followed by the convective or diffusive movement of the vapour, and ending with the heat-releasing stage of condensation. This flux is the only one which is present both in the energy balance and in the water balance.

The energy balance at the land surface describes how the energy arrives to the soil and how this energy is partitioned, controlling the surface temperature.

Figure 2.4: Scheme of the energy balance at the land surface.

Starting from the "net radiation", is important to recall the principles seen in Sec.2.1.2. The temperature of the soil surface averages about 300 K, which can range from below 273 K, the freezing point, to 330 K or even higher [Hillel, 1982]. The radiation emitted by the soil surface has its peak intensity at a wavelength of about 10 µm and its wavelength distribution over the range of 3–50 µm. As seen in the previous subsection, this is in the region of the infrared.

The sun emits a very different spectrum, since it behaves as a black body, at an effective surface temperature of about 6000 K. The radiation from the sun includes

(22)

the visible light range of 0.3–0.7 µm, and it starts from the ultraviolet radiation

λ < 0.3µm, till the infrared radiation with greater wavelengths up to about λ <3 µm. Most of the emitted radiation has wavelengths between 0.3 to 2.0 µm, with

the maximum around 0.5 µm .

Since there is very short overlap between the two spectra, is a convention to distinguish between them, by calling the incoming solar spectrum short-wave radiation, and the spectrum emitted by the earth long-wave radiation.

The net radiation balance of a bare surface, can be written thus:

Rn = Sn+ Ln (2.12)

where Rnis the net surface radiance, Snis the net short-wave radiance from the sun

and Ln is the net long-wave radiance from the surface. The net surface radiance, is

given from the sum of all incoming minus outgoing radiant energy fluxes, as it will be seen for the two terms which compose it. The net short-wave radiance from the sun to the surface, can be represented as,

Sn= S− S↑ (2.13)

where Sis the total down-welling short-wave radiation and S↑ is the total

up-welling short-wave radiation. Since the Earth’s surface does not emit short-wave radiation, S↑ is entirely associated with reflection of some of the down-welling

radiation from the sun. Following that,

Sn = (1 − α) S= (1 − α) (Sb+ Sd) (2.14)

where α is the surface albedo, Sb is the direct-beam solar radiance and Sd is the

diffuse solar radiance.

The albedo α is an important characteristic of soil surfaces, is also called reflectivity coefficient, expressed α = S/S↓ , as it is the fraction of incoming short-wave

radiation reflected by the soil surface rather than absorbed by it. It always depends on the surface’s roughness, the inclination of the incident radiation relative to the surface, and for the bare soil it can vary widely in the range of 0.1–0.4, depending on the soil’s basic colour, whether dark or light coloured. Apart from the reflected short-wave radiation, governed by the albedo, the emission of the long-wave radiation is a soil-dependent process. The out-going long-wave radiation from the surface is

L= sσ Ts4+ (1 − s) L↓ (2.15)

The first term on the right-hand side of this equation is the radiation emitted by the soil surface and the second term is the reflected down-welling long wave radiation incident on the surface. The emitted radiance is given by Eq.(2.2) with

s the effective emissivity of the surface. The second term is given from Kirchoff’s

law in Eq. 2.8, for which the surface absorbs a fraction s of any incident long-wave

radiation and reflect a fraction (1 − s). Thus net long-wave irradiance of the

surface is:

(23)

Is important to remember that the parameter s, also depends on soil wetness

and can varies up to a 7%, at 11 and12 µm in the infrared region [Hulley et al., 2010].

Going back to (2.12) , the net radiation received by the soil surface is transformed into heat, which warms the soil and consequently the air and vaporizes water. The surface energy balance is expressed as follows:

Rn = G0+ H + λ E (2.17)

where G0 is the soil heat flux, as the rate at which heat is transferred from the

surface downward into the soil profile, H is the sensible heat flux transmitted from the surface to the air above, and LE is the latent heat flux, the product of the evaporation rate E and the specific latent heat of evaporation λ.

Conventionally, all components of the energy balance are taken as positive if directed toward the surface and negative otherwise. For vegetated area, it must be considered that part of the net radiation received is taken up by the plants in their metabolic processes as photosynthesis. Moreover a major part is generally absorbed as latent heat in the twin processes of evaporation and transpiration. Thus, the previous equation is transformed:

Rn = Go+ H + λ E + M (2.18)

where λE this time is the rate of energy utilized in evapotranspiration, H is the rate at which heat is stored in the soil, water, and vegetation, and M represents other mixed energy terms, such as photosynthesis and respiration.

Where the vegetation is short like in a grass field, the storage of heat in the vegetative biomass is negligible compared with storage in the soil. The heat stored in the soil under sparse vegetation might be a fairly large portion of the net radiation at any particular time during the day, but the net storage over a 24-hr period is usually small, and during the night time, the losses of soil heat tend to negate the daytime gains. Overall, the amount of energy stored in soil, vegetation and fixed photochemically, accounts for a rather small portion of the total daily net radiation, with the greater portion going into latent and sensible heat. [Hillel, 1982] Thus in general, the term M in Eq. (2.18) can be neglected. Recalling again the energy balance of Eq. (2.17), it can be written again dividing the terms:

Rn− Go= H + λ E (2.19)

where the quantity Rn− G0 is known as the available energy.

The term G of the ground heat flux at the surface, subtracts energy from the incoming radiation, since energy is lost by heat conduction through the lower boundaries of the soil. It is a positive number when it is directed away from the surface into ground.

(24)

Conduction. Assuming for the ground surface flux G0 the thermal flux moving

towards the one-dimensional vertical direction, it can be written:

G0 = −K

dT

dz (2.20)

The thermal flux G0, is the amount of heat conducted from the surface across a

unit cross-sectional area in unit time, K is the thermal conductivity (W m−1K−1), and dT /dZ is the gradient of temperature in the vertical direction, with z = 0 being the soil surface. The negative sign in this equation is due to the fact that heat flows from a higher to a lower temperature. Taking in account that the temperature gradient in the soil is not constant with depth, also the heat flux varies with depth. The second law of heat conduction, is obtained through the principle of energy conservation in the form of the continuity equation, so that the time rate of change in heat content for a volume element is equal to the change of flux with distance: ρ cm dT dt = − dG dz (2.21)

where ρ is the mass density and cm is the specific heat capacity per unit mass

(J kg−1K−1). The product ρ cm (often designated C ) is the specific heat capacity

per unit volume, and dT /dt is the time rate of temperature change. Combining Eq. (2.20) and Eq. (2.21), the second law of heat conduction, always considered for the vertical direction is:

dT dt = k ρ cm d2T dz2 = Dt d2T dz2 (2.22)

The ratio of the thermal conductivity k to the volumetric heat capacity ρ cm is

called the thermal diffusivity, designated Dt with units m2s−1. To know how

temperature varies in space and time, must be known the pertinent values of those three parameters just defined, the volumetric heat capacity C, thermal conductivity

k, and thermal diffusivity Dt. Together, they are called the thermal properties of

soils, and further will be discussed upon them in section 2.2.2.

Going back to the land surface energy balance of Eq. (2.19), the left side of the equation is constituted by the turbulent fluxes. As implied by the word “turbulent”, these fluxes are largely driven by turbulences, as the air at the surface-atmosphere interface is replenished by wind. The transport of sensible heat and latent heat, carried by vapour, occurs through several layers. The first is the laminar boundary layer, a relatively quiescent body of air in immediate contact with the surface of the evaporating body. Through this layer, which may be only about 1 mm thick, transport occurs by diffusion. Beyond this layer, turbulent transport becomes predominant in the turbulent boundary layer.

The sensible heat flux H represents the loss of energy by the surface by heat transfer to the atmosphere, and is positive when directed away from the surface into the

(25)

atmosphere. It is proportional to the product of the temperature gradient dT/dz and the turbulent transfer coefficient for heat Kh:

H = −KhρaCpa

dT

dz (2.23)

where Cpa is the specific heat capacity of air at constant pressure, ρa is the density

of air, T is temperature, and z is height above the surface.

The latent heat λE represents the loss of energy from the surface due to evaporation. The rate of latent-heat transfer by water vapour from the field to atmosphere, can be expressed as well as follows:

λ E = −Ke

0.622

p ρa de

dz (2.24)

with p the atmospheric pressure, and again proportional to the product of the vapour-pressure gradient (considering the saturated vapour pressure and the effec-tive vapour-pression at a certain height) and the appropriate turbulent-transfer coefficient for vapour.

If it is assumed that the transfer coefficients for heat and water vapor are similar

Kh ∼ Ke, then the ratio between the transport of sensible heat and latent heat

is:

β = H λ E ≈ γ

∆ T

∆ e (2.25)

Here ∆ T /∆ e is the ratio of the temperature gradient to vapor-pressure gradient in the atmosphere above a field, while γ is the psychrometric constant. The ratio

β, called the Bowen ratio, represents the way in which the available energy is

partitioned between the sensible and latent heat flux, and depends mainly on the interactive temperature and moisture regimes of the field. When the soil is wet, the relative humidity gradients between its surface and the atmosphere tend to be large; so that much of the energy received is taken up as latent heat for the process of evaporation, and the temperature gradients tend to be small, maintaining the surface cooled. Thus, β is rather small when the evaporation rate is high.

When the soil is dry, on the other hand, the relative humidity gradients toward the atmosphere are generally small and much of the received energy goes to warming the surface, increasing the temperature at the surface, so the temperature gradients tend to be steep and the Bowen ratio becomes large.

Transfers through the turbulent atmospheric boundary layer take place primarily by means of eddies, which are irregular, swirling microcurrents of air, whipped up by the wind. Eddies of varying size, duration, and velocity fluctuate up and down at varying frequencies, carrying both heat and vapor. While the instantaneous gradients and vertical fluxes of heat and vapor generally fluctuate, if a sufficiently long averaging period is allowed, the fluxes exhibit a stable statistical relationship over a uniform field [Hillel, 1982].

(26)

2.2.2

Variations on soil surface temperature

As already shown in the previous section through the land surface energy balance, surface wetness, is an important variable which controls evaporation and hence mean surface temperature. The other important variable which is responsible for most observed temperature variability is the thermal inertia, which relates the diurnal excursion of surface temperature to the ground heat flux [Price, 1985]. The ground heat flux at the surface G0, varies approximately sinusoidally through

the day, in response to diurnal variations in the net radiation, and also the soil temperature responds sinusoidally to the changing meteorological regime acting on the soil–atmosphere interface.

The simplest mathematical representation of nature’s fluctuating thermal regime, is to assume that at all depths in the soil the temperature oscillates as a pure harmonic function of time, around an average value. The temperature at the

6 12 18 24 Time (h) Ao Tmax Tmin T Temper ature

Figure 2.5: Idealized daily fluctuation of surface soil temperature [Hillel, 1982].

surface represented in Fig.2.5, is expressed as a function of time:

T (0,t) = Tavg + A0 sin(ω t) (2.26)

where T(0,t) is the temperature at z = 0, which corresponds to the soil surface, as a function of time t, Tavg is the average temperature of the surface, and A0 is

the amplitude of the surface-temperature fluctuation, as the range from maximum, or from minimum, to the average temperature, which can be indicated with ∆ T .

(27)

Through the term ω is indicated the radial frequency, which is 2π times the actual frequency. In the case of diurnal variation, the period considered are the 24h of the length of the day, which is reported as 86400 sec, so the radial frequency is

ω = 2π/86,400 = 7.27 · 10−5/sec.

In the below layers, at any depth the amplitude of the temperature fluctuation becomes smaller than A0 and there is a phase shift, which brings to a time delay of

the temperature peak. Those behaviours are expressed in function of the depth z. The physical reason for the damping and retarding of the temperature waves with depth is that a certain amount of heat is absorbed or released along the path of heat propagation, when the temperature of the conducting soil increases or decreases, respectively [Hillel, 1982].

The mathematical formulation won’t be shown here, as the paragraph focuses on the variation of temperature at the soil surface. Assuming the general solution of equation (2.22) under boundary condition, and relating the ground heat flux G0 to

surface temperature Ts(t) = T (0, t), can be obtained [Sellers, 1965] the equation

:

Gt= I ∆ T

ω sin(ω t + π

4) (2.27)

Gt is also a sinusoidal function relative to Ts, through I the thermal inertia, and

with a phase shift of π/4. With a practical example, this mean that G leads Ts

by 3 hours for diurnal cycle, and 1.5 months for annual cycle [Wang et al., 2010]. According to equation (2.27), the diurnal amplitude of the ground heat flux, ∆ G, relates to that of the surface temperature ∆ T , as anticipated, through:

∆ G = I ∆ Tω (2.28)

the thermal inertia is expressed as a proportional coefficient in a linear equation, relating the amplitudes of surface ground heat flux and temperature. The equation (2.28) is exact only for sinusoidal Ts with a constant I [Wang et al., 2010]. In

reality the soil temperature time series acquired in-situ and through remote sensing are not sinusoidal, and also the thermal inertia parameter I is not always constant, due to variable soil moisture, as it will be explained further.

Regardless, for many authors I can be identified as the parameter of primary importance in the Earth and extraterrestrial remote sensing, being the key property controlling the diurnal and seasonal surface temperature variations. It describes the impedance of soil to temperature variations, and is a measure of the thermal properties of soils already shown on section 2.2.1: the thermal conductivity and volume heat capacity of the combined physical components of the soil. This composite parameter is formulated as:

I =qk ρ cm (2.29)

(J m−2K−1s1/2) where k is always the soil thermal conductivity, ρ is the actual bulk density of the soil including water, and cm is the soil specific heat per unit

(28)

of mass. Of these components, the air content in the soil, which can be varied by either compaction or addition of water, has the greatest bearing on soil thermal properties.

The variation of the water content in pores, moreover brings to changes in the soil thermal conductivity and specific heat capacity [Minacapilli et al., 2012]. On the base of the various soil textural classes, the thermal conductivity and volume heat capacity differ on three accounts.

First of all the different total porosities, the percent of total volume, as the dry bulk density of a soil is inversely related to the porosity of the same soil: the more pore space in a soil the lower the value for bulk density.

Secondly, the different soil water distributions in the pores, as thermal conductivity increases much more rapidly by addition of water in sand than clay.

Moreover different soil mineral compositions gives different values for thermal properties. Consequently, thermal inertia will vary with differing water content, porosity, textural class, and mineral composition [KING, 1983].

As it will be seen further, is important to make a distinction also for different soil covers. Soils high in organics may have a bulk density well below 1 g/cm3. The reason why soils rich in organic carbon do have lower bulk density is due to the low density of organic materials. Thus, a layer of organic soil will present a lower thermal inertia, and will be more subject to external temperature variations, than mineral soils, which presents an higher bulk density. While for vegetation cover, fresh leaves have a specific heat capacity comparable to that of water, but the thermal inertia of fresh plant leaves is much smaller than the water or the sand, due to their low thermal conductivities. This means that a high amount of input heat energy is required to raise the temperature of plant leaves, but also once heated to a higher temperature, they retain the heat absorbed for a longer time [Jayalakshmy and Philip, 2010].

Through the use of available time series of thermal remote sensing images, the concept of thermal inertia was applied in different studies for estimations of thermal properties to relate with changes in composition, porosity, and moisture content, especially in arid regions [Pratt and Ellyett, 1979]. The real thermal inertia usually cannot be measured by remote sensing methods, since conductivity, density and thermal capacity can be measured only by contact methods, so that in those studies is applied the apparent thermal inertia (ATI) as a proxy.

Maps of the ATI can be obtained through information on the albedo, derivable from multispectral VIS-NIR, and the surface temperature difference computed by multitemporal TIR images.

The thermal inertia ATI is defined as:

AT I = (1 − α)

∆ T (2.30)

where α is the albedo in the visible range and ∆ T is the difference of the maximum and the minimum temperature, which is usually computed at the daily base. The

(29)

albedo is used to compensate the various absorptivity values of the rocks, while ∆ T is low for the materials with high thermal inertia and high for those with low thermal inertia.

Land surface diurnal temperature range (DTR), apparent thermal inertia and real thermal inertia are the three forms of thermal inertia which in the last decade have been also used to estimate soil water content. A large number of recent studies have shown that near-surface soil moisture can exhibit a direct linkage with remotely sensed information. In this way thermal inertia and thermal infrared information-based remote sensing are sources used to monitor soil moisture, as it has been attempted in many studies through microwaves [Wang et al., 2015]. Several studies derive models that relate soil thermal inertia as a function of water content [Murray and Verhoef, 2007], accompanied by the information on the soil type, in order to estimate soil water content from TIR data by remote sensing. Those models requires readily available soil characteristics, to estimate soil thermal inertia as a function of the water content, across the information of soil texture and bulk density [Lu et al., 2009]. For the authors [Murray and Verhoef, 2007], some distinctions can be seen in different groups of soil textural type, looking at those the variations in the thermal inertia corresponding to the relative saturation. However, few studies have examined directly the relationship between land surface temperature and soil texture, including day temperature, night temperature and diurnal temperature range, and estimated soil texture directly from land surface temperature [Wang et al., 2015].

For this study and its experiment, it won’t be investigated how the content and changes in soil water, and hence land surface temperature, are related to soil texture. The information on the water content in soils is used to see how the water content and the evaporation processes mask the soil temperature response to incoming radiation and consequently its variation in time, for which a differentiation between the different soil textures should be visible.

(30)

Methods

3.1

Experimental setup

The experimental setup took place in the terrace of the University of Natural Resources and Life Science BOKU in Vienna, more specifically in the building Muthgasse 18, where is also situated the Institute of Water Management, Hydrology and Hydraulic Engineering. During the first months the instrumentation has been installed, as shown in Fig. 3.1, composed by:

• boxes of soil; • soil sensors; • lysometers;

• meteorological station; • infrared camera.

3.1.1

Soil boxes and sensors

The four boxes used for the experiment set-up, present a cubic shape of the dimensions 0.5x0.5x0.5m, and are made in wood with a thickness of 2 cm. They are filled up with 2 different soil type with different soil texture, respectively sand and loam, for a total weight around 125 kg for each box. The textural class is reported below at figure 3.2.

Moreover, within each box are positioned different soil sensors. All the soil sensors are connected to the data logging system (CR6 by Campbell Sc.) and detect the soil temperature, soil moisture and ground heat flux within every soil plot. The soil moisture sensors (5TM by Decagon devices) test the temperature, measured by an

(31)

Figure 3.1: Photography and model of the instrumental setting for the experiment.

on-board thermistor, and determine volumetric water content (m3/m3), through the dielectric constant of the soil. Even tough those sensors collect measures of temperature, temperature sensors (107 by Campbell Sc.) (°C) were positioned in the soil as well. The temperature probes use a thermistor to measure temperature, and can be used in soil, water and air, between -35°C and +50°C, with an accuracy

(32)

Figure 3.2: Soil texture triangle and textural classes for the two type of soil used in the

experiment

of ±0.4°C. The heat flux sensors (HFP01 by Hukseflux) measure the heat flux (W/m2) up going or down going through the soil. The sensor is a thermopile, which measures the temperature difference across its ceramics-plastic body. The output is a voltage signal that is proportional to the heat flux of the surrounding medium. It is typically used for energy-balance or Bowen-ratio flux systems, and it has an accuracy within -15% to +5%.

The samplings are conducted at different depths and so every probe was buried horizontally at a certain distance from the surface, paying attention to not let the caves to be overheated by the sun radiation close to the surface. In the table below are reported the depths at which every sensor was positioned.

(33)

Figure 3.3: Sensors set-up within the soil boxes.

Distance from surface Sensor

2 cm Temperature

3 cm Soil moisture

6 cm Temperature

8 cm Heat Flux

12 cm Soil moisture

Table 3.1: Sensors depths within every soil box

3.1.2

The lysometers

The lysometer is a instrument used to measure the amount of actual evapotran-spiration which is released by plants in crops and fields. Is a tool to understand water balance and it can represent field conditions well since it can be also used outside laboratories. A traditional lysometer is composed by a large soil tank where vegetation is grown, and it allows to measure the water lost through the soil after a precipitation input. The amount of water lost by evapotranspiration is computed by calculating the difference between the weight before and after the precipitation input.

For this experiment 4 platform scales (PCE-SD 300SST C ) were installed with a weighting surface of 600 x 500 cm, suitable for outdoor environment. Their weighting range is 300 kg and they have 100g of resolution. The difference from a standard lysometer is that using those scales they don’t account for infiltration, as there is not an outlet bottom for the exceeding water. For this reason during the experiment is not provided an amount of water that exceed the saturation

(34)

conditions, so that only the fluxes of precipitation and evapotranspiration are taken in account.

3.1.3

Meteorological station

The weather transmitter (WXT520 by Campbell Sc.), is installed around 2 m above the ground near by the soil plots. It measures wind speed and direction, precipitation, barometric pressure, temperature, and relative humidity.

The wind sensor consist of three equally spaced transducers that produce ultrasonic signals. Wind speed and direction are determined by measuring the time it takes for the ultrasonic signal of one transducer to travel to the other transducers. The RAINCAP sensor measures the accumulated rainfall, rain intensity, and rain duration, measuring one raindrop at a time.

The meteorological station has also a PTU module that contains a capacitive silicon

BAROCAP sensor for barometric pressure measurements, a capacitive ceramic THERMOCAP sensor for air temperature measurements, and a capacitive thin

film polymer HUMICAP sensor for relative humidity measurements.

In the table below are reported the specifics for all the meteorological mea-sures.

Measuring Range Accuracy Output Resolution

Air Temperature -52° to +60°C ±0.3°C 0.1°C

Wind Speed 0 to 60 m/s ±0.3 m/s

Wind Direction 0° to 360°1° ±3° 1°

Precipitation Coll. area 60 cm2 5% 0.01 mm

Barom. Pressure 600 to 1100 hPa ±0.5 hPa 0.1 hPa

Relative Humidity 0 to 100% RH ±3% 0.1% RH

Table 3.2: Specifics of the sensors within the Meteorological station

3.1.4

Infrared camera

For the collections of the thermal images in the infrared spectrum is used the infrared camera (Pi 400 / Pi 450 by Optris).

Is a thermographic camera of small dimensions with a measurement speed of 80 Hz and a resolution of 382 x 288 pixels. The thermographic images can be taken within a temperature range of -20 °C to 900 °C, as the camera works in the spectral range of 7.5 to 13 µm.

The camera has an objective of 38°, which corresponds to the optical aperture angle of the instrument.

Since the camera is lifted 2.5 m above the plots, the HFOV, the horizontal enlarge-ment of the total measuring at object level, is 1.73 m, while the VFOV, the vertical

(35)

enlargement of the total measuring at object level, is 1.27 m.

This area includes well all the four boxes of soil, leaving some space at the margins. The IFOV, Instantaneous Field Of View, corresponds to the size of the single pixel at object level and is 4.55x4.55 mm. It defines the spatial resolution, or the size of smallest object that can be resolved at the specific distance from the camera.

Figure 3.4: Camera field of view.

The thermal camera is connect to a computer system that enable the use of the program Optris Pi Connect.

This program, associated to the camera, is an infrared analysis software that offers recording and real-time analysis options.

(36)

Figure 3.5: Camera display through the program Optris Pi connect.

An average value of the emissivity for the camera was left as for default equal to 1, used for the calculation of the surface temperature. This value is not too far from the soil emissivities, as they range from 0.90 to 0.95 for dry soils up to 0.97 for wet soils.

Moreover the differentiation between the different soil emissivities could be neglected as in the experiment, the interest is more upon the variations of surface temperature than on the absolute values corresponding to the kinetic energies.

3.2

Collection of data

The data collection took place in the summer term 2017. In a first decision due to the start of the warming season, were settled the conditions to test the vegetation cover over the boxes.

In the start of July the boxes of soil were filled at the top with a layer of 2 cm of organic soil, than together with the grass seeds, fertilizer was applied. During this period, when there was still only the cover of organic soil, 2 measures took place, respectively the 6th and 7th of July. When the grass was fully grown, two measuring day were carried on the 9th and 10th of August, to follow 5 days of consequently measurements from the 14th to the 18th of August, in which 2 boxes were continuously watered and the 2 others were left dry. In the 13th of September was lead one last measurement, in which this time the vegetation cover was removed in 2 of the boxes, in order to have the bare soil.

(37)

Figure 3.6: Grass cover for the boxes of sand and loam in the month of August, the

layer of organic soil is well visible.

• 6th-7th July: sparse vegetation and all soils watered • 9th-10th August: dense vegetation and all soils watered

• 14th-18th August: dense vegetation, 2 soils dried and 2 watered. • 13th September: 2 bare soils and 2 with vegetation, all watered.

The measurements were performed every time proceeding first to the reset of the scales, in order to remove the off-set and convey the shifted value to zero during the calibration. Every scale needed to be lifted in order to reset them, so this procedure was done for each scale at the time, using a pulleys system, to overcome the weight of each box between 120 and 150 kg. The thermal infrared camera was than fixed every time before measuring, centering the interested area of objective. The camera was left recording, taking a snapshot every 5 minutes of the soils surface temperature. The soils were than watered with 3 liter of water, watering homogeneously through all the surface, and left than drying for the rest of the measurement hours.

The recording hours were from 9 a.m. to 12 a.m., this choice was due to the presence of the building in the sud-west part of the terrace, which was producing shadow on the boxes, starting from noon. Otherwise this effect was then reflected as an elbow with minor slope in the surface temperature time series.

Another external factor that was controlling the goodness of the measurements was the cloudy cover. The days chosen for measuring were those with most clear sky conditions as possible, in order to have the direct short-wave radiation from the sun, leading to the energy processes and have less disturbance and diffusion of the

(38)

beam by the clouds.

After noon a video at the high frequencies of 25 Hz was recorded for 15 minutes. Every time was following the dis-installation of the camera, and the data from the soil sensors and meteorological station were read out from the data-logger.

3.3

Temporal and spatial analysis of thermal

im-ages

3.3.1

Reading out the thermal images

As the outputs of the thermal camera were set as text files, .csv files were stored in the memory of the associated device to the camera, for every snapshot taken. Csv file is a comma-separated values file, which stores tabular data in a plain text, for which each line of the file is a data record.

All the analysis were conducted through MATLAB (MATrix LABoratory), a multi-paradigm numerical computing environment, using its version R2015b.

Starting from the reading of the .csv files in Matlab environment, the values of temperature reported as comma-separated values for every file, were imported as a matrix. Those matrix had 382 x 288 size, which corresponds to the pixels resolution 382 x 288.

Every numeric value is the recorded value of the surface temperature taken from the corresponding pixel, expressed in °C, and every matrix corresponds to a thermal image taken at a certain time step, as it can be seen in Fig.3.7.

Figura

Figure 2.1: Electromagnetic spectrum, detail of visible and infrared ranges.
Figure 2.3: Representation of Planck’s law: spectral specific radiations of the black body depending on the wavelength.
Figure 2.4: Scheme of the energy balance at the land surface.
Figure 3.1: Photography and model of the instrumental setting for the experiment.
+7

Riferimenti

Documenti correlati

In questo periodo di emergenza idrica nazionale è doveroso sottolineare che la più consistente risorsa idrica del territorio italiano è costituita dalle acque sotterranee

Since a huge number of microorganisms are involved in food production, foodstuffs should be considered as complex matrices where any microbial component has a precise role and

Equivalent ߠ כ for sand, loam and silt loam are also given, as found by [17] using the SDM for artificially hydrophobised (again via silanisation) material tested at different

In addition to the amount of air bubbles, the value of the excess pore water pressure depends mainly on the ratio between the velocity of the water level change and

The calculated bioaccumulation coefficients of copper and zinc (WB) in maize (Zea mays L.) and English plantain (Plantagolanceolata L.) are shown graphically in Figure 4..

The ambient air temperature (Ta), the brine temperature at the borehole number 1 and 3 outlet (Tbrine,1 and Tbrine,3) and the average power specific extraction rate of the

The analysis of the problem of ensuring the WSS reliability shown that the problem of analysis and ensuring reliability requires time-consuming, multivariate calculations of the

In order to improve the temperature distribution in the corner and to avoid the condensation of water vapor, it is reasonable to use an increased layer of thermal insulation of