1. Introduction
Soils are essential elements for life, due to their contribution to Ecosystem Services as providers of food production, water and climate regulation, energy provision and biodiversity [
1]. Among many parameters considered as soil quality indicators [
2], knowing the thermal conditions of the soil is important for several reasons: (i) soil temperature has an effect on soil biota for plant growth [
3], (ii) thermal conductivity is critical in different fields, such as the automotive and aerospace industries, ceramics, glass and building materials sectors and the energy one, especially in geothermal applications, in which this parameter conditions the performance of the underground thermal exchange [
4], (iii) the thermal properties of the soil are directly related with soil moisture [
5] and (iv) soil thermal properties are the main factors for mass and energy exchange processes on Earth [
6]. In the geotechnical context, controlling the thermal conductivity of the ground could act as a support for defining changes in composition, humidity and other influential properties such as porosity, resistance to freezing, etc. When trying to measure the thermal conductivity of the soil, practice problems are mainly linked to difficulties for covering a representative volume of the sample, in which the variations in the composition and properties are reflected.
In this sense, there are many different models for the determination of thermal conductivity of soil, both theoretical and experimental [
7]. Regarding theoretical models, the authors of Reference [
8] present an empirical model for estimation of soil thermal conductivity, which is based on the model in Reference [
9] but including the influence of organic matter content and particle composition. This model is validated for 19 soils, in contrast to other models such as in References [
10] and [
11], that were specifically developed for sandy and peat soils, and for sand-kaolin clay mixtures, respectively.
Among experimental measurements, the authors of Reference [
12] prove the capacity of a thermo-TDR (Time-Domain Reflectometry) probe to measure thermal conductivity of sand, while the authors of Reference [
13] measure the thermal conductivity of different samples with a device based on the Guarded Hot Plate method.
However, all the models are based on the performance of measurements of certain parameters on punctual positions of the area under study. This procedure makes the determination of thermal conductivity time-consuming and restricted to certain areas. In order to perform larger measurements, satellite data has shown to be a valid input for the determination of ground thermal conditions at large-scale, as well as the determination of the spatial distribution of the parameter under study [
14].
All models and measuring methodologies mentioned so far focus on the determination of the thermal conductivity of the soil, omitting the other heat transfer methods (convection, radiation). Conduction–convection algorithms are used for the estimation of soil apparent thermal diffusivity, since these result in more accurate thermal diffusivity values because of their consideration of the vertical heterogeneity of the soil [
15]. Heat transfer through convection is also considered for the estimation of thermal diffusion, which is a key parameter for the design of Ground Source Heat Pump Systems [
16]. Heat transfer through radiation has also been considered in the case of studies for horizontal ground-coupled heat pumps [
17,
18], but radiation was only considered from a theoretical point of view, calculating the input from the sun in the position under study. Thus, the a priori analysis of soil materials would benefit from a laboratory test for the determination of all the heat fluxes involved in the thermal imbalance. In this sense, infrared thermography can be an adequate technique, in its nature of producing thermal (temperature) images of the area under study [
19].
The performance of a valid measurement with infrared thermography requires the consideration of certain requirements, since the thermal infrared technique is based on the measurement of the incoming radiation to the camera, and its conversion to temperature values [
20], which presents many parameters of influence, mainly: radiation from surrounding objects, the emissivity of the object under study, the distance from camera to object and ambient temperature and humidity [
21]. In addition to the consideration of the parameters of interest, the conversion from incoming thermal infrared radiation to temperature values requires the thermal or radiometric calibration of the camera, that is, the equation that inputs radiation and outputs temperature values.
Traditionally, thermal infrared cameras are calibrated using a temperature modulated black body target, under laboratory conditions [
22]. This type of calibration reaches an accuracy of 0.06–0.2 K [
22] for static measurements, going up to 0.32 K for measurements performed from Unmanned Aerial Vehicles, UAV [
23]. It is a time-consuming and complicated process, with a high cost regarding the final price of the camera [
24]. The radiometric calibration is usually based on Plank curves for the determination of the correlation between infrared radiation and temperature [
25], although other mathematical relations, such as linear and polynomial models, have been tested [
26]. Recent advances have focused on image processing techniques and neural networks for the radiometric calibration, in an image-by-image calibration [
27] or with specifically trained artificial neural networks [
26].
Provided the complexity of the radiometric calibration procedures, efforts are being made towards the simplification of the process, starting from the design of plate radiators in contrast to the more expensive cavity blackbodies [
28]. The authors of Reference [
29] prove the validity of a three-point calibration for measurements performed from UAV platforms, while the authors of Reference [
30] perform a radiometric calibration using images acquired by the thermographic camera of an object at constant temperature.
The radiometric calibration can be completed with a geometric calibration that allows knowing the interior geometry of the camera: dimensions of the pixels, focal length of the camera lens and principal or central point of the image. In the case of thermal infrared cameras, the geometric calibration requires calibration boards specifically designed, either based on a temperature difference [
31] or on an emissivity difference [
32] between the targets and the background.
This paper presents a novel approach for the thermal characterization of geothermal environments. This approach consists of the monitoring of the soil surface with a thermographic camera, which has been radiometrically calibrated with a novel, low-cost approach. The thermal imbalances within the soil depth and in the soil surface allow for the determination of the thermal conductivity of the soil material, which is complemented with the thermographic monitoring for the computation of the convection effect. In addition, the comparison of thermal conductivity values computed in-depth and at the surface works as a validation of the radiometric calibration of the thermographic camera, together with its comparison with manufacturer values. Thus, the paper is structured as follows:
Section 2 describes the devices used for the study, the methodology for the radiometric calibration of the camera and the modeling of the thermal behavior of the system (underground and at the surface).
Section 3 shows the results obtained from the thermal imbalances in-depth (thermal conductivity) and at the surface (convection effect).
Section 4 includes a discussion of the results, while
Section 5 presents the conclusions reached.