Advances in Thermal Infrared Remote Sensing Technology for Geothermal Resource Detection

Abstract

This paper discusses thermal infrared (TIR) remote sensing technology applied to the delineation of geothermal resources, a significant renewable energy source. The technical characteristics and current status of TIR remote sensing is discussed and related to the integration of geological structure, geophysical data, and geochemical analyses. Also discussed are surface temperature inversion algorithms used to delineate anomalous ground-surface temperatures. Unlike traditional geophysical and geochemical exploration methods, remote sensing technology exhibits considerable advantages in terms of convenience and coverage extent. The paper addresses the major challenges and issues associated with using TIR remote sensing technology in geothermal prospecting.

1. Introduction

Amidst the escalating depletion of traditional fossil fuel resources and the hastened pace of global energy transition, renewable energy sources are increasingly assuming a prominent position within the global energy regime. Geothermal, solar, wind, tidal, and hydroelectric energies, as core constituents of green new energy forms, have been witnessing a steady rise in their share within the global energy structure, thereby prompting extensive and profound scholarly research and practical exploration [1]. By the 1990s, geothermal resources had emerged as a widely recognized and accessible clean and renewable energy source, garnering substantial international interest and fostering intensified development endeavors [2,3,4]. Since 2010, the global direct utilization of geothermal energy has experienced rapid growth, with the installed capacities and annual heat utilization reaching approximately 108 GWt and 283,580 GWh, respectively [5,6,7]. China possesses abundant geothermal resources, accounting for approximately one-sixth of the world’s total geothermal reserves, indicating substantial potential for further development and utilization [8,9]. However, the formation and spatial distribution of geothermal resources are significantly influenced by specific geological structural features [8,10]. Given the highly uneven distribution of global geothermal resources, this inherently poses a major challenge to their extensive and intensive exploitation worldwide. Consequently, it is imperative to conduct systematic geothermal surveys, scientifically evaluate advantageous geothermal regions, and prioritize the selection of appropriate target exploration areas.

The identification of geothermal anomaly zones, particularly those characterized by high temperatures, constitutes a pivotal prerequisite for mapping the distribution pattern of geothermal resources and advancing the related research and practical endeavors [10]. This process is an intricate, interdisciplinary system that integrates various methodologies. Among numerous methods employed in identifying geothermal anomalies are TIR remote sensing technology [11], geophysical exploration techniques [12], geochemical analytical methods [13], mathematical statistical models [14], and spatial analysis techniques [15]. TIR remote sensing technology, due to its extensive information content, high detection precision, and capability for rapid large-scale identification with minimal constraints from ground conditions, offers significant advantages in terms of efficiency and cost-effectiveness. It effectively demarcates areas of surface temperature anomalies, providing a fresh perspective and technological approach for the exploration and development of geothermal resources [1].

This paper endeavors to provide a comprehensive overview of the research advancements in the application of TIR remote sensing technology for geothermal resource exploration, critically examining its multifaceted utility in addressing challenges pertinent to geothermal resource investigation. The paper presents an analysis of the current limitations inherent in the remote sensing exploration for geothermal resources. The paper describes a forward-looking perspective on the future developmental trajectory of remote sensing technology in this domain and aims to offer strategic and guiding recommendations for harnessing this technology to facilitate the precise detection and assessment of geothermal resources.

2. A Concise Introduction to TIR Remote Sensing Technology

Infrared remote sensing technology refers to a detection method employing sensors that operate within the infrared band spectrum ranging from 0.76 to 1000 μm (

Table 1. Information sources in TIR bands for remote sensing data acquisition.

Infrared Remote Sensing	Band Range	The Sensor Receives Information from the Source
Near infrared/Short-wave infrared	0.76~2.5 μm	It is mainly the reflected energy of the surface to the solar radiation, and the contribution of the Earth’s own radiation is very small
Middle infrared	2.5~6.0 μm	It contains both the target’s own thermal radiation and the target’s reflected radiation to the sun’s mid-infrared radiation, both of which are of the same order of magnitude
Thermal infrared	6.0~15.0 μm	The thermal radiation of the ground object is the main part, and the reflected solar radiation can be ignored
Ultra-far infrared	15.0~1000 μm	This band is rarely used in Earth remote sensing observation

). This technology discerns and interprets the infrared energy reflected or radiated by target objects, thereby acquiring pertinent information about the nature, condition, and patterns of change for these ground features [16]. Diverging from visible light and near-infrared remote sensing, TIR remote sensing predominantly relies on the intrinsic emitted energy signals from Earth’s surface materials. Specifically, it harnesses TIR detectors to capture and document the TIR radiation data emitted by ground objects, which are imperceptible to the human eye, and, subsequently, utilizes this TIR information for object identification and the inversion of key surface parameters such as temperature, emissivity, humidity, and heat flux, making it a critically important technique within the realm of remote sensing [17,18].

In its narrow definition, TIR remote sensing specifically pertains to the utilization of satellite-borne sensors to capture spectral information from objects within the TIR band. However, in a broader sense, it involves obtaining the TIR spectral data of targets using various platforms such as ground-based, spaceborne, and airborne instruments (Figure 1). In geothermal resource exploration, surface temperature constitutes one of the key parameters. A multitude of satellites equipped with sensors like Thematic Mapper (TM), Enhanced Thematic Mapper Plus (ETM+), Topocentric Intermediate Reference System (TIRS), Topocentric Intermediate Reference System-2 (TIRS-2), Moderate Resolution Imaging Spectroradiometer (MODIS), Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), and Advanced Very High Resolution Radiometer (AVHRR) are capable of providing data that include the TIR band, thus rendering TIR remote sensing technology instrumental in the inversion of surface temperatures and widely applicable across diverse fields. These applications span a broad array of geological and environmental research areas, including but not limited to mineral mapping [19], urban climatology studies [20], urban heat island effect analyses [21], drought monitoring [22], soil moisture assessments [23], geothermal resource prospecting [10], geothermal anomaly detection [24], real-time volcanic activity monitoring [25], and dormant volcano research [26]. Moreover, this technology is also employed in mapping heat flux and heat loss [27], as well as characterizing hydrothermal alteration features [28], further highlighting its extensive potential for application in geological research and environmental monitoring endeavors.

Since the early 1960s, the United States Army Cold Region Research and Engineering Laboratory collaborated with the University of Michigan to conduct groundbreaking geothermal surveys in Yellowstone National Park using infrared scanning imaging technology. This effort successfully identified hot springs and other geothermal anomalies in the region, establishing a robust foundation for subsequent geothermal exploration endeavors [29]. By the 1970s, geothermal remote sensing technology had made significant strides on a global scale. Especially in 1978, Lee’s use of TIR remote sensing led to the discovery of important geothermal anomalies in the Lodsburg area of New Mexico, USA, indicating the maturation of TIR remote sensing applications within the field of geothermal exploration [30]. Concurrently, researchers from New Zealand [31] and Mexico [32] also carried out geothermal remote sensing surveys, and these pioneering studies contributed valuable practical experience to the worldwide exploration and development of geothermal resources. In the 1980s, Japanese scholar Sekioka utilized TIR sensors mounted on helicopters to conduct mapping studies in geothermal regions, which significantly propelled the widespread application and development of geothermal remote sensing technology in Japan [33]. By contrast, China’s research into TIR remote sensing for geothermal exploration commenced at a relatively later stage; however, since the early 1980s, several geothermal anomalies were successfully identified during geothermal survey experiments conducted in southern Liaoning Province, accumulating valuable experience [34]. This groundwork laid a robust foundation for subsequent geothermal remote sensing investigations in China.

Despite extensive international research efforts in applying TIR remote sensing technology to geothermal resource exploration that have led to remarkable advancements, this technology currently does not completely supplant traditional geophysical and geochemical exploration methods. Nevertheless, with the ongoing rapid evolution and development of TIR remote sensing technology, its applications in geothermal resource exploration are poised for an increasingly broad expansion.

3. Fundamental Principles of TIR Remote Sensing for Geothermal Resource Detection

3.1. The Inversion Methodology for Retrieving LST

Surface temperature serves as a critical physical parameter indicating the thermal state of Earth’s surface, which is governed by the intricate interplay between solar radiation inputs and internal energy exchanges within the planet [35,36]. Given the rapid spatiotemporal variability of surface temperatures, the reliance solely on ground-based point measurements is inadequate for capturing the comprehensive spatial–temporal distribution patterns at regional scales. Consequently, satellite remote sensing technology has emerged as a focal point of research for many scholars due to its ability to provide extensive, multi-temporal data on surface temperatures across large areas [16,37,38]. According to their varying spatial resolutions, TIR remote sensing data can be classified into several categories: low-spatial-resolution data, exemplified by MODIS imagery with a 1 km resolution and AVHRR data at a 1.1 km spatial resolution; intermediate-spatial-resolution data, such as the 120 m resolution images from the TM sensor, and TIRS/TIRS-2 data offering a 100 m spatial resolution; and medium- to high-spatial-resolution data, which include ASTER imagery featuring a 90 m resolution and ETM+ data with a 60 m spatial resolution. The availability of data with various spatial resolutions enables researchers to select appropriate remote sensing materials tailored to their specific research needs, thereby facilitating a more precise inversion of LST and better understanding of the spatial distribution patterns and temporal variations thereof. In the process of LST inversion, the preliminary step involves a judicious screening of the remote sensing data, which must undergo a series of preprocessing stages including georeferencing, radiometric calibration, atmospheric correction, and terrain normalization. Following this, an appropriate inversion method is applied to estimate the LST within the study area. Lastly, the inverted LST results should be subjected to an error analysis and accuracy assessment, ensuring the validity and reliability of the derived outcomes (Figure 2).

3.2. LST Inversion Algorithms

With the growing demand for large-scale LST data in research areas such as geothermal resource exploration, geothermal anomaly detection, the mapping of heat flux and thermal loss, and hydrothermal alteration mapping, the use of satellite TIR remote sensing data to retrieve LST has seen a rapid advancement over the past few decades. During this period, a variety of inversion algorithms have been developed, tailored to different satellite platforms, among which some have become relatively mature and widely used, including but not limited to the single-channel algorithm, multi-channel algorithm, multi-angle algorithm, multi-temporal algorithm, and hyperspectral algorithm [39,40,41,42,43,44]. In the realm of geothermal resource exploration, the Landsat series of remote sensing satellites, jointly managed by National Aeronautics and Space Administration (NASA) and United States Geological Survey (USGS), have become the most widely utilized data source for TIR remote sensing applications both domestically and internationally due to their continuous time-series of TIR data and relatively high-spatial-resolution capabilities [45,46,47]. Based on the Landsat TIR remote sensing data, the LST inversion algorithms mainly include the single-channel algorithm based on the radiation transmission equation [48], single-window algorithm [39], universal single-channel algorithm [41], practical single-channel algorithm [49], and split-window algorithm [50]. It is particularly worth noting that, although the Landsat-8 image has two TIR bands and the split-window algorithm can be used to invert the LST in theory, the absolute calibration accuracy of the TIRS 11th band has a large error due to the interference of stray light outside the field of view, and its accuracy also has a large uncertainty. However, Landsat-9 has solved the above problems and can theoretically apply the split-window algorithm for LST inversion [51].

3.2.1. Radiative Transfer Equation (RTE) Method

The surface temperature inversion method, grounded in the RTE, is commonly referred to as the atmospheric correction method. It fundamentally relies on the TIR radiation transfer equation and represents one of the pioneering techniques developed for LST retrieval. This method serves as a foundational element within the theoretical framework of LST inversion algorithms. With its extensive applicability, it enables the inversion of LST from a majority of TIR remote sensing datasets through atmospheric correction procedures. The atmospheric correction method boasts several advantages, such as a streamlined algorithmic process, a relatively small number of required input parameters, and an easily comprehensible set of underlying principles [48]. In the process of LST retrieval, TIR sensors detect radiance values that encompass not only the Earth’s emitted thermal radiation but also contributions from upward atmospheric radiation and downward atmospheric radiation after reflection at the ground surface. Therefore, the RTE method utilizes relevant atmospheric modeling software to simulate real-time atmospheric conditions, thereby acquiring a variety of atmospheric profile parameters, including atmospheric transmittance, upward atmospheric irradiance, and downward atmospheric irradiance. Subsequently, by subtracting the energy component influenced by the atmosphere from the total received radiation energy as measured by remote sensing satellite sensors, one can obtain an estimate of the true surface radiation value. Once the surface emissivity is determined, the LST can be inversely computed using the inverse Planck function. The expression of the atmospheric correction algorithm formula can be represented as follows:

$$L_{\lambda }=\left[\epsilon \cdot B\left(T_S\right)+\left(1-\epsilon \right)L\downarrow \right]\cdot \tau +L\uparrow$$

(1)

In the formula, L_λ represents the brightness temperature of TIR radiation at a given wavelength λ. ε denotes the surface emissivity, T_S signifies the true LST in Kelvin (K), B(T_S) stands for the blackbody radiation brightness corresponding to T_S, L↓ indicates the downwelling atmospheric radiance, τ is the atmospheric transmittance, and L↑ represents the upwelling atmospheric radiance.

For atmospheric upwelling radiation brightness, downwelling radiation brightness, and atmospheric transmittance, among other profile parameters, these can be obtained from the official website provided by NASA. By inputting relevant parameters such as image acquisition time and central latitude, one can directly acquire data suitable for inverting the LST in a TIR band. For Landsat-8 remote sensing imagery, Band 10 can be utilized for LST inversion. According to Equation (1) representing the atmospheric correction algorithm, the conversion relationship between the true LST T_S and the radiant brightness value B(T_S) of a blackbody at the same temperature in the TIR band is given by Equation (2). Subsequently, according to Planck’s Law, the calculation formula for inversely determining the true LST T_S is expressed as Equation (3).

$$B\left(T_S\right)=\left[L_{\lambda }-L\uparrow -\tau \cdot (1-\epsilon )L\downarrow \right]/\tau \cdot \epsilon$$

(2)

$$T_S=\frac{K_2}{\ln \left(\frac{{\mathrm{K}}_1}{B\left(T_S\right)}+1\right)}$$

(3)

In the formula, K₁ and K₂ are constants that can be obtained from the header file. For TIR Band 10, K₁ = 774.8 W/(m²·μm·sr), and K₂ = 1321.08 K; for TIR Band 11, K₁ = 480.89 W/(m²·μm·sr), and K₂ = 1201.14 K.

3.2.2. Single-Window Algorithm

The single-window algorithm, as introduced by Qin et al. in 2001 [39], is a method developed for LST retrieval based on the principles of terrestrial thermal radiation transfer equations. This algorithm presents a simplified and relatively high-precision solution, particularly tailored to Landsat-5 TM TIR imagery. It notably diminishes the dependence on detailed atmospheric profiles while holistically addressing both atmospheric and surface influences. In the context of this method, the inversion of LST necessitates only three key parameters: the atmospheric transmittance, surface emissivity, and mean atmospheric temperature [49]. Through the establishment of interrelations among these parameters, the mathematical formulation of the single-window algorithm can be articulated as represented in Equation (4).

$$T_S=\raisebox{1ex}{$\left\{\mathrm{a}\left(1-C-D\right)+\left(\mathrm{b}\left(1-C-D\right)+C+D\right)T_{\mathrm{sensor}}-D{T}_{\mathrm{a}}\right\}$}\!\left/ \!\raisebox{-1ex}{$C$}\right.$$

(4)

In the equation, T_S represents the LST (K), T_sensor denotes the sensor brightness temperature (K), and T_a signifies the average atmospheric temperature (K). The constants a and b are set to −67.355351 and 0.458606, respectively, within the temperature range of 0 °C to 70 °C. C and D are intermediate variables. The expressions for T_sensor, T_a, C, and D are presented in Formulas (5), (6), and (7), respectively.

$$T_{\mathrm{sensor}}=\frac{K_2}{\ln \left(\frac{{\mathrm{K}}_1}{L_{\lambda }}+1\right)}$$

(5)

$$T_{\mathrm{a}}=\left\{\begin{array}{l}25.9396+0.88045\ast T_0,\mathrm{United} \mathrm{States} 1967 \mathrm{standard} \mathrm{atmosphere}\\ 17.9796+0.91715\ast T_0,\mathrm{Tropical} \mathrm{atmosphere}\\ 16.0110+0.92621\ast T_0,\mathrm{Mid}-\mathrm{latitude} \mathrm{summer} \mathrm{atmosphere}\\ 19.2704+0.91118\ast T_0,\mathrm{Mid}-\mathrm{latitude} \mathrm{winter} \mathrm{atmosphere}\end{array}\right.$$

(6)

$$\left\{\begin{array}{l}C=\epsilon \cdot \tau \\ D=\left(1-\tau \right)\left[1+\left(1-\epsilon \right)\tau \right]\end{array}\right.$$

(7)

In the formula, for TIR Band 10 within Landsat-8, K₁ equals 774.89 W/(m²·μm·sr), and K₂ equals 1321.08 K; L_λ represents the brightness temperature of TIR radiation; T₀ signifies the near-surface air temperature (K); ε denotes the surface emissivity; while τ refers to the atmospheric transmittance.

3.2.3. Generalized Single-Channel Algorithm

The generalized single-channel algorithm, as introduced by Jiménez-Muñoz and Sobrino in 2003 [41], is an LST retrieval technique. This methodology was formulated through the application of a first-order Taylor series expansion to the Planck function around a specific temperature value, enabling its broad applicability across various types of TIR data, encompassing those from Landsat-5 TM and Landsat-8 thermal sensors alike. The algorithm’s core simplicity and reduced parameter requirement are notable, demanding only three essential inputs: the effective wavelength, atmospheric water vapor content, and sensor radiance. It is particularly well-suited for regions with low atmospheric moisture levels due to this minimal input necessity. A significant strength of the generalized single-channel algorithm lies in its versatility and adaptability, allowing it to employ uniform equations and coefficients across a diverse array of thermal sensors. Consequently, it showcases a high universality and flexibility in LST retrievals. The formula embodying the single-channel algorithm’s inversion process for LST is expressed in Equation (8).

$$T_S=\gamma \left[{\phi }_1{L}_{\lambda }+{\phi }_2/\epsilon +{\phi }_3\right]+\delta$$

(8)

In the formula, γ and δ are two intermediate variables obtained through linear expansion based on Planck’s law; L_λ represents the TIR radiance; ε is the surface emissivity; and φ₁, φ₂, and φ₃ are atmospheric parameters. The calculation methods for γ and δ are, respectively, given by Equations (9) and (10).

$$\gamma \approx \raisebox{1ex}{$T_{\mathrm{sensor}}^2$}\!\left/ \!\raisebox{-1ex}{${\mathrm{b}}_{\lambda }{L}_{\lambda }$}\right.$$

(9)

$$\delta \approx T_{\mathrm{sensor}}-\raisebox{1ex}{$T_{\mathrm{sensor}}^2$}\!\left/ \!\raisebox{-1ex}{${\mathrm{b}}_{\lambda }$}\right.$$

(10)

$$T_{\mathrm{sensor}}=\frac{{\mathrm{c}}_2}{\lambda \ln \left(\raisebox{1ex}{${\mathrm{c}}_1$}\!\left/ \!\raisebox{-1ex}{${\lambda }^5{L}_{\lambda }$}\right.+1\right)}$$

(11)

In the formula, λ represents the effective wavelength (in micrometers); T_sensor denotes the sensor brightness temperature (K); c₁ and c₂ are Planck radiation constants, where c₁ = 1.19104 × 10⁸ W·um⁴·m⁻²·sr⁻¹, and c₂ = 14,387.685 um·K; b_λ is a constant, with values of 1324 K for Landsat-8 TIR Band 10 and 1199 K for Band 11.

The calculation formulae for atmospheric parameters φ₁, φ₂, and φ₃ are presented in Equation (12).

$$\left[\begin{array}{c}{\phi }_1\\ {\phi }_2\\ {\phi }_3\end{array}\right]=\left[\begin{array}{ccc}{\mathrm{c}}_{11}& {\mathrm{c}}_{12}& {\mathrm{c}}_{13}\\ {\mathrm{c}}_{21}& {\mathrm{c}}_{22}& {\mathrm{c}}_{23}\\ {\mathrm{c}}_{31}& {\mathrm{c}}_{32}& {\mathrm{c}}_{33}\end{array}\right]\left[\begin{array}{c}{\omega }^2\\ \omega \\ 1\end{array}\right]$$

(12)

In the formula, ω signifies the atmospheric water vapor content (g·cm⁻²); and c_ij (where i = 1, 2, 3 and j = 1, 2, 3) are atmospheric parameters that are related to ω.

The mathematical relationships between the three atmospheric functions for Landsat-8 data and the atmospheric water vapor content are given by Equation (13).

$$\left\{\begin{array}{l}{\phi }_1=0.04019{\omega }^2+0.02916\omega +1.01523\\ {\phi }_2=-0.38333{\omega }^2-1.50294\omega +0.20324\\ {\phi }_3=0.00918{\omega }^2+1.36072\omega -0.27514\end{array}\right.$$

(13)

3.2.4. Practical Single-Channel Algorithm

To address the inherent inaccuracies introduced by linearizing the Planck function and solving atmospheric parameter systems in conventional single-channel algorithms, Wang et al. (2019) have developed the practical single-channel algorithm (PSC) [49]. This innovative approach bypasses the errors stemming from the linear approximation of the Planck function by directly establishing a robust correlation between the on-board sensor radiance and the terrestrial blackbody radiation brightness. Furthermore, the PSC algorithm employs an advanced atmospheric parameter analysis to formulate an optimized model for estimating these parameters, utilizing global fitting methodologies to calibrate its coefficients, thus effectively mitigating the compounded effects of errors related to multiple atmospheric variables. The practical single-channel (PSC) algorithm is divided into two versions: (1) the practical single-channel algorithm based on atmospheric water vapor content (PSC_w), and (2) the practical single-channel algorithm based on both atmospheric water vapor content and near-surface air temperature (PSC_w&Ta), where the value of parameter "a" can refer to the research of Wang et al. [49].The algorithm representation is as follows:

$$T_S=\frac{K_2}{\ln \left(\frac{{\mathrm{K}}_1}{B{\left(T_S\right)}_{\mathrm{x}}}+1\right)}$$

(14)

where B (T_S)_x is B (T_S)_w or B (T_S) _w&Ta

$$\begin{array}{l}B{\left(T_S\right)}_{\mathrm{w}}={\mathrm{a}}_0+{\mathrm{a}}_1\mathrm{w}+\left({\mathrm{a}}_2+{\mathrm{a}}_3\mathrm{w}+{\mathrm{a}}_4{\mathrm{w}}^2\right)\frac{1}{\epsilon }\\ +\left(a_5+a_{6}w+a_7{w}^2\right)\frac{1}{\epsilon }{L}_{sen}\end{array}$$

(15)

$$\begin{array}{l}B{\left(T_S\right)}_{\mathrm{w}\&T\mathrm{a}}={\mathrm{a}}_0+{\mathrm{a}}_1\mathrm{w}+\left({\mathrm{a}}_2+{\mathrm{a}}_3\mathrm{w}\right)T_{\mathrm{a}}+\left(\left(a_4+a_{5}w+a_6{w}^2\right)+\left(a_7+a_{8}w+a_9{w}^2\right)T_a\right)\\ +\left(a_{10}+a_{11}w+a_{12}{w}^2\right)\frac{1}{\epsilon }{L}_{sen}\end{array}$$

(16)

3.2.5. Split-Window Algorithm

In the realm of LST retrieval studies, where the precise and real-time acquisition of atmospheric parameters often proves challenging, researchers have devised a methodology that diminishes the dependency on these variables. This strategy capitalizes on the inherent information within the remote sensing data to facilitate the atmospheric correction. The fundamental principle of the split-window algorithm lies in exploiting the differential sensitivity to atmospheric absorption properties, particularly with respect to the water vapor content, between two neighboring spectral bands situated within an atmospheric transparency window, having central wavelengths around 11 μm and 12 μm. By judiciously utilizing various combinations of measurements from these two bands, the algorithm successfully mitigates atmospheric interference, thereby accomplishing atmospheric correction. Despite the considerable absolute radiometric calibration inaccuracies affecting Band 11 of Landsat-8′s TIRS, which conventionally preclude its use in surface temperature retrieval via the split-window algorithm, researchers, including Jiménez-Muñoz et al. (2014) and Rozenstein et al. (2014), have devised alternative split-window methodologies that incorporate Bands 10 and 11 of TIRS [50,52].

According to the research by Rozenstein et al. (2014) [52], the calculation formula for the split-window algorithm used in LST retrieval from the Landsat-8 data is as follows:

$$T_S=A_0+A_1{T}_{10}-A_2{T}_{11}$$

(17)

$$A_0=\frac{{\mathrm{a}}_{10}{D}_{11}\left(1-C_{10}-D_{10}\right)}{D_{11}{C}_{10}-D_{10}{C}_{11}}-\frac{a_{11}{D}_{10}\left(1-C_{11}-D_{11}\right)}{D_{11}{C}_{10}-D_{10}{C}_{11}}$$

(18)

$$A_1=1+\frac{D_{10}}{D_{11}{C}_{10}-D_{10}{C}_{11}}+\frac{{\mathrm{b}}_{10}{D}_{11}\left(1-C_{11}-D_{11}\right)}{D_{11}{C}_{10}-D_{10}{C}_{11}}$$

(19)

$$A_2=\frac{D_{10}}{D_{11}{C}_{10}-D_{10}{C}_{11}}+\frac{{\mathrm{b}}_{11}{D}_{10}\left(1-C_{11}-D_{11}\right)}{D_{11}{C}_{10}-D_{10}{C}_{11}}$$

(20)

$$C_{\mathrm{i}}={\epsilon }_i{\tau }_i$$

(21)

$$D_{\mathrm{i}}=\left[1-{\tau }_{\mathrm{i}}\right]\left[1+\left(1-{\epsilon }_{\mathrm{i}}\right){\tau }_{\mathrm{i}}\right]$$

(22)

T_s represents the LST (K), and T₁₀ and T₁₁ are the on-board brightness temperatures in Bands 10 and 11 (K), where i refers to either Band 10 or Band 11. ε_i denotes the surface emissivity for the i-th band, while τ_i represents the atmospheric transmittance for the same i-th band. The coefficients a_i and b_i are regression coefficients determined according to different temperature ranges for the TIRS Bands 10 and 11; these regression coefficients can be referred to the research of Qin et al. [52].

4. The Application of TIR Remote Sensing Technology in Geothermal Resource Exploration

In earth science, geothermal anomalies usually refer to places where the flow of heat in the earth’s crust is significantly enhanced, or where the earth’s temperature is abnormally elevated. Specifically, geothermal anomalies are not only manifested as temperature differences between geothermal anomaly zones and their surrounding environments but also deeply rooted in the mechanisms of geothermal fluid flow and storage, both of which are critically governed by geological conditions. Consequently, when identifying geothermal anomaly regions, it is essential to integrate a variety of key features across multiple Earth science domains, encompassing regional temperature anomalies, topographic and geomorphological patterns, geological structures, geophysical data, geochemical indicators, and rock alteration characteristics, among others (Figure 3). The employment of mathematical models and spatial analysis techniques can assist in locating potential geothermal resource areas. This interdisciplinary, integrated analytical approach compensates for the shortcomings of a single detection method and improves the accuracy of geothermal resource exploration, thereby providing a more robust theoretical foundation for the development and utilization of geothermal energy. The synergistic integration of remote sensing technology with disciplines such as geology, geophysics, and geochemistry enables a comprehensive and profound understanding of the spatial distribution and energy potential of geothermal resources [10,24].

4.1. The Integration of TIR Remote Sensing Technology with Geological Structure Interpretation

TIR remote sensing technology plays an important role in geothermal resource exploration by integrating multi-source high-resolution TIR data to inversely derive surface temperatures, thereby effectively uncovering potential geothermal anomaly zones. Simultaneously, when delving into the geological conditions that are intimately tied to the generation and storage of geothermal energy, remote sensing techniques are employed to precisely identify key determinants such as tectonic landforms, lithological distribution patterns, and subsurface rock bodies [53,54,55]. Its strengths lie in its ability to comprehensively analyze the spatial structure and dynamic processes of geothermal systems across macroscopic to microscopic scales, thus providing scientific underpinnings for the assessment of the geothermal resource potential and the development of strategic exploitation plans. Scholars frequently integrate thermal anomaly information derived from remote sensing technology with geological condition data, conducting a thorough analysis by examining existing geothermal datasets and delving into the geological structural context. They systematically identify favorable tectonic settings conducive to geothermal activity and further investigate the spatial correlation between the results of surface temperature inversions and known geothermal resources, as well as fault structures [56]. This integrated approach enables researchers to establish rigorous criteria for delineating geothermal anomaly zones, effectively eliminating pseudo-anomalies unrelated to geothermal processes. Moreover, through qualitative or semi-quantitative methods, they target areas for geothermal exploration, thereby optimizing the outcomes in geothermal resource assessment and potential forecasting. The synergistic application of remote sensing technology and geological analysis not only significantly boosts the efficiency in geothermal resource detection but also enhances the understanding of the complex dynamics within geothermal systems, providing a robust scientific foundation for the sustainable development and rational utilization of geothermal energy.

To eliminate spurious anomalies induced by non-geothermal resource factors such as topography, urban heat island effects, and vegetation coverage, various scholars have proposed distinct methods [53,57,58]. Among these strategies, mitigating the impact of terrain is particularly paramount. Besides utilizing nighttime TIR remote sensing data from ASTER satellites to minimize obstructions due to topographic relief and radiation errors, digital elevation model (DEM)-based terrain correction is also a widely adopted method in the international community [49]. This technique adjusts surface temperature measurements to reflect genuine geothermal conditions rather than artifacts caused by variations in topography. Li et al. (2023) introduced an innovative approach for efficiently calculating the multi-temporal LST in the Damxung–Yangbajing basin of the Qinghai–Tibet Plateau using Google Earth Engine (GEE) [59]. This methodology integrates terrain correction, elevation adjustment, and multi-temporal sequence analysis to discern geothermal anomaly signals. In regions with a complex topography, the study partitions the research area based on varying degrees of terrain undulation and slope aspects, performing separate inversions of LST, thereby effectively mitigating the “same object, different spectra” and “same spectrum, different objects” phenomena that arise from disparate solar radiation intensities received by various ground features, thus enhancing the precision of LST retrieval [60,61,62]. To eliminate temperature anomalies caused by urban heat island effects and other human activities, as well as natural factors, scholars often conduct comprehensive analyses that integrate a variety of data sources including human geography data, hydrogeological survey findings, groundwater dynamics monitoring records, deep-source gas emission characteristics, meteorological observations, borehole temperature measurements from fieldwork, and the spatial distribution patterns of known hot springs [63,64,65]. Due to the influence of vegetation cover on the extraction of temperature anomaly information, employing multi-source, multi-temporal, and long-term winter LST datasets in conjunction with threshold-based methods constitutes an effective and cost-efficient approach for detecting geothermal resources by distinguishing and filtering out non-geothermal-derived anomalies [62,66].

In the Tengchong area of Yunnan Province, China, a series of extensive research projects on geothermal resource exploration have been carried out by scholars [53,60,67,68]. These studies ingeniously employ remote sensing techniques to conduct a detailed interpretation of geological structures within the Tengchong region and utilize surface temperature inversion methodologies to reveal subsurface thermal conditions. By systematically extracting anomalous information from TIR remote sensing data, coupled with meticulous structural interpretations and overlay analyses of geothermal anomaly distribution zones, researchers not only expose the interplay between geological structures and geothermal anomalies in this region but also delve into the mechanisms governing the generation and transfer of geothermal energy. Consequently, they scientifically elucidate the spatial patterns of geothermal resource enrichment and their controlling factors in the Tengchong area. Building upon previous work, researchers including Xiong et al. (2016) advanced a statistically based approach to identify areas of LST anomalies [69]. This strategy specifically accounts for the influence of elevation on the distribution pattern of surface temperatures in the Tengchong region and innovatively employs a zoning method based on elevation data to accurately differentiate and pinpoint potential temperature anomaly zones. In this study, researchers employed remote sensing techniques for the geological structure interpretation and the mapping of altered rock bodies to accurately delineate the spatial distribution pattern of geothermal anomalies in the Tengchong region, juxtaposing these findings with existing data on geothermal fields, hot springs, and volcanic occurrences. The research findings indicate a strong consistency between the distribution of geothermal zones in the Tengchong area and the development of fault structures, highlighting the critical role played by the fault system. Furthermore, deep-seated magma activities serve as a substantial heat source for the geothermal energy within the region. Meanwhile, the interconnected network of faults acts as an efficient conduit for the upward transfer of heat from the Earth’s interior, significantly contributing to the formation and concentration of surface geothermal anomalies. Darge et al. (2019) also reached a similar conclusion in their study of the Tulu Moye geothermal area in Ethiopia [63], demonstrating the significant value of integrating multiple detection methods to elucidate geothermal resource distribution patterns. Specifically, they combined TIR remote sensing techniques for identifying geothermal anomalies with remote sensing geological interpretations and detailed field survey data on geological structures, topography, and geomorphology. Using these integrated datasets and theoretical knowledge of geothermal mechanisms, the researchers developed an informative model that serves as a critical scientific framework for guiding the exploration and assessment of potential geothermal resources in uncharted territories. The research has indeed proven that such a comprehensive analytical approach constitutes an efficacious means of probing and delineating prospective geothermal regions.

4.2. The Integration of TIR Remote Sensing Technology and Geophysical Methods

Geophysical techniques play an essential and indispensable role in the exploration and development of geothermal resources [70]. This domain encompasses a comprehensive set of exploration technologies from deep subsurface geological structure investigation to surface feature analysis, coupled with a sophisticated suite of data processing and interpretation methodologies. For instance, geophysical methods assist with the understanding of dynamic changes in deep magma systems and the spatial distribution of specific heat-producing rock bodies (strata), and conduct an in-depth analysis of large-scale fault structures closely associated with heat transfer processes. Moreover, they effectively determine the rock formations that serve as reservoir media and their accompanying fracture systems, as well as accurately assess the position of insulating caprock layers controlling heat loss, along with the burial depths of their top and bottom interfaces [70,71].

Currently prevalent geophysical techniques in geothermal resource exploration include active seismic methods, a variety of electromagnetic methods (such as magnetotellurics, audio-magnetotellurics, controlled-source audio-magnetotellurics, high-density electrical resistivity imaging, and wide-angle electromagnetic sounding), magnetic methods, and gravity methods [72,73]. TIR remote sensing data effectively detect thermal anomaly patterns in surface spatial distribution. Integrating diverse sets of geophysical data allows for a more comprehensive understanding of the subsurface geological structures and their inherent relationship with geothermal systems. This multi-dimensional analytical approach that combines both surface and subsurface perspectives provides a more effective comparative analysis for the potential distribution of geothermal resources within the study area, thereby significantly enhancing the efficiency and accuracy of geothermal resource exploration. These techniques may be effectively used to reveal complex geological factors tied to the generation, conduction, and storage of geothermal energy.

Tian et al. (2015) conducted a pioneering study in the Hokkaido region of northern Japan where they employed 28,476 temperature data points collected from 433 boreholes to develop a subsurface temperature distribution model ranging from a 100 to 1500 m depth using the kriging estimation method (KED) [74]. By integrating detailed well log data with high-resolution TIR imagery (from Landsat-8’s thermal infrared sensor), researchers successfully assembled a comprehensive dataset that captures both subsurface and surface temperature distributions. This integrated approach was instrumental in accurately identifying and delineating geothermal prospect zones, leading to significant advancements in their research findings. Chan et al. (2018) deployed deep electromagnetic sounding profiles in the Wuerhe region, Xinjiang Province, China, and, through a comprehensive analysis of geoelectric information and surface temperature anomaly data, elucidated the causative factors behind thermal anomalies pertaining to geothermal resources [24]. This study demonstrates that the synergistic application of TIR remote sensing and magnetotelluric techniques constitutes a promising approach for geothermal exploration.

Moreover, the integration of residual gravity anomalies, aeromagnetic data, seismic data, and TIR remote sensing information in the spatial modeling for the geothermal potential assessment is a prevalent methodological approach in geothermal resource exploration [72,75,76,77]. Abdel et al. (2018) used various tools such as remote sensing, seismic events, aeromagnetic, and gravity data to evaluate the geothermal potential of the Ferrara Oasis and its applicability for geothermal development in the western desert [78]. Similar, Li et al. (2017) conducted a case study in the Tengchong area of Yunnan Province, China, where various datasets were utilized, including earthquake epicenter distributions, fault patterns, Bouguer gravity anomalies, magnetic anomalies, and Landsat 7 ETM+ imagery [68]. These datasets were systematically processed into five critical influencing factor maps: the Gutenberg–Richter b-value map, distance to the nearest fault lines, proximity to major graben structures, magnetic anomaly distribution, and LST maps. Ultimately, through the application of an improved certainty factor method that combined these factors, the study successfully predicted unexplored and undeveloped prospective geothermal zones within the Nujiang Basin. Abuzied et al. (2020) combined remote sensing techniques, geophysical methods, and geographic information systems (GISs) in an innovative manner to derive factor layers indicative of the geothermal energy distribution using multi-source datasets [64]. These factor layers included topographic and geological maps, Bouguer gravity anomalies, aeromagnetic anomalies, and satellite imagery, as well as seismic activity records and bottom hole temperatures (BHT) from oil and gas wells. Through Bayesian statistical modeling, the study modeled potential areas along the Suez Gulf coastline for geothermal potential, culminating in the creation of a geothermal potential map. The synergistic application of these diverse methods significantly enhances the accuracy and efficiency of geothermal resource exploration, providing robust scientific grounding for the development and utilization of geothermal energy resources.

4.3. The Integration of TIR Remote Sensing Technology with Geochemical Methods

Geochemical methods serve as a cost-effective and efficient approach widely utilized in geothermal resource exploration. By systematically collecting representative samples, such as groundwater, surface water, gas emissions, minerals, and sediments, and subjecting them to a meticulous chemical composition and isotope analysis, these methods enable an in-depth understanding of the underlying mechanisms driving the formation and evolution of geothermal fluids, and help accurately depict the spatial distribution range of geothermal anomalies based on derived geochemical characteristics [79,80,81]. Consequently, this provides a scientific foundation for the further investigation and exploitation of geothermal resources. In the exploration of geothermal resources, geochemical techniques play an integral role and are implemented through diverse media analysis methods, notably encompassing four main branches: rock geochemistry, hydrogeochemistry, gas geochemistry, and soil geochemistry [82,83,84]. Given the unique geological attributes of geothermal resources and the advanced state of both the theoretical research and practical exploration in geochemical technologies, currently recognized mature exploration strategies primarily focus on these four areas. Firstly, rock geochemistry investigates deep geothermal activities by precisely measuring the elemental distributions and isotopic compositions within subsurface rock formations. Secondly, hydrogeochemistry employs chemical changes in groundwater and surface hot water systems to infer the thermal fluid circulation paths and reservoir characteristics. Thirdly, gas geochemistry indirectly infers the heat source properties and generation mechanisms of geothermal systems by analyzing the composition and isotopic features of gases emitted from geothermal regions. Lastly, soil geochemistry utilizes soils as a reactive medium at the Earth’s near-surface to identify and assess surficial effects associated with geothermal activity [85]. The integration of geochemical methods with TIR remote sensing technology presents several key advantages in geothermal resource exploration, such as rapid large-scale surveying, the efficient identification of anomalous areas, and the mitigation of false anomalies. A common practice currently employed is to combine soil radon gas concentration measurements, water chemistry analysis, and TIR remote sensing data to delineate potential geothermal prospect zones. This synergistic approach significantly enhances the accuracy and cost-effectiveness of geothermal resource recognition and evaluation, providing a critical scientific underpinning for the development and utilization of geothermal energy [65].

In geothermal systems, particularly those governed by faulting activities, surface temperature anomalies are typically associated with the development of faults [24]. In geothermal research, the precise location of faults can be determined through measurements of soil gas radon concentrations. By integrating remote sensing interpretations with soil radon measurements, the position of potential geothermal fault zones can be identified with a relatively high degree of accuracy. Generally, areas with higher soil radon concentrations correspond to elevated surface temperatures, which serve as a valuable indicator for assessing the shallow geothermal energy potential. Sun et al. (2022) successfully employed soil radon concentration data along with LST inversion derived from TIR remote sensing data of the Landsat-8 satellite to guide the drilling of geothermal wells in the Tangjiashan area, Sichuan Province, China [58]. Moreover, hydrochemical analysis is a commonly used method in geothermal exploration. Zhang et al. (2022), leveraging Landsat-8 remote sensing data, conducted an extensive interpretation of atmospheric-corrected surface temperatures and hydrothermal alteration anomalies in the Xian’ning area, Hubei Province, China [86]. By integrating favorable geothermal segments, geological settings, and detailed results from hydrochemical analyses, as well as radon anomaly data, she effectively delineated the target geothermal prospect zone and achieved significant outcomes in this process.

5. The Main Problems with Detecting Geothermal Resources Using TIR Remote Sensing Technology

The integration of TIR remote sensing technology with geological, geophysical, and geochemical methods effectively enhances the efficiency and accuracy of geothermal resource prospecting. This multidisciplinary approach provides a crucial scientific foundation for the development and utilization of geothermal energy. However, there is currently a series of challenges that TIR remote sensing technology faces in its application to geothermal exploration:

(1)

The long-term series of ground temperature anomaly observations play a significant role in enhancing the accuracy of geothermal resource detection using TIR remote sensing [57,59,67]. Initially, these observations contribute to the compilation of a comprehensive ground temperature database, thereby furnishing a critical benchmark against which remotely sensed surface temperature retrievals can be validated and refined. Secondly, the meticulous tracking and interpretation of temporal–spatial variations in ground temperature data, with an emphasis on identifying persistent or cyclic temperature anomalies, play a pivotal role in unearthing likely zones of geothermal activity. Moreover, sustained geothermal monitoring plays a crucial role in assessing the stability and dynamic characteristics of geothermal systems, an understanding that is vital for devising strategies conducive to the sustainable exploitation of geothermal energy. Additionally, integrating empirical ground temperature measurements with remotely sensed data significantly enhances the precision of surface temperature inversions. This integrated approach allows for comparative analyses that can pinpoint regions exhibiting geothermal anomalies, thus facilitating the accurate localization of geothermal resource distributions. Furthermore, leveraging such data enables researchers to delve deeper into the actual potential and size of geothermal reserves through the examination of parameters such as geothermal gradients and associated variables [59,87].

(2)

The current research commonly employs global or local threshold methods to extract high-temperature anomalies from remote sensing data and combines these with regional geological structural characteristics for delineating geothermal anomaly zones [49,53,62]. Furthermore, some studies have used nighttime remotely sensed data to minimize the influence of solar radiation or adopted terrain correction techniques to reduce interference caused by factors such as elevation and slope [88,89]. However, relying solely on single remote sensing techniques in geothermal exploration may lead to numerous false positives, complicating the accurate identification of genuine geothermal anomalies and increasing the workload for subsequent field verification. Currently, scholars are experimenting with more advanced methods like discriminant functions, principal component analysis (PCA), logistic regression equations, Dempster–Shafer evidence theory, and logical operator algorithms to demarcate geothermal anomaly areas [66,67,88,89]. Nonetheless, these approaches heavily rely on indirect data, possess a degree of subjectivity, require substantial professional expertise to guide their application, and often provide insufficient explanations for the formation mechanisms of geothermal anomaly regions, thereby limiting their reliability as a dependable reference in practical resource development and utilization.

(3)

While satellite remote sensing data can effectively be used for the large-scale detection of geothermal anomalies and provide insights into the current state and dynamic changes of geothermal activity within a region, this approach typically yields only an approximate estimation of the spatial extent of the geothermal presence. A prevalent issue in the current research is the overreliance on TIR remote sensing bands, often overlooking the importance of incorporating other spectral bands that could significantly contribute to characterizing geothermal resources. The integrated application of multi-band and multi-source remote sensing data has the potential to reveal a more comprehensive understanding of the properties and characteristics of geothermal resources [57,66,90].

(4)

The integration of remote sensing technology with GIS presents substantial practical utility in geothermal resource exploration [91,92,93]. This integrated approach provides extensive and multidimensional data that aid in identifying potential geothermal zones and analyzing their geothermal energy potential. However, given the diverse mechanisms driving the geothermal formation across different regions, the factors to consider when constructing predictive models can vary significantly [65,93,94,95]. A critical issue during modeling is the careful selection and quantification of relevant factors, which requires a thorough analysis and investigation [95,96]. Parameters such as geological structures, surface temperatures, geochemical indicators, and geomagnetic data may all be important variables for model development. Furthermore, the temporal and spatial resolution of the data, data quality, and errors introduced during processing can all impact the accuracy of the models. The choice of model assumptions and methods is equally crucial because these can lead to inherent biases and influence results [93]. If an inappropriate model or faulty assumptions are adopted, misleading predictions may arise, thereby potentially leading decision-makers astray with adverse effects on their decisions.

6. Summary and Prospect

TIR remote sensing holds vast potential for development in the field of geothermal resource exploration. With advancements in sensor technology and the advent of novel high-precision sensors, TIR remote sensing is poised to achieve increased accuracy and sensitivity in detecting the distribution and activity patterns of geothermal resources. The integration with other types of remote sensing techniques, such as combining with optical remote sensing for high-accuracy measurements of temperature gradients, or coupling with hyperspectral remote sensing for the precise identification of minerals indicative of geothermal processes, will further propel the advancement of TIR sensing in geothermal exploration.

(1)

In geothermal resource exploration, the pivotal task lies in intensifying the study of multi-source remote sensing data fusion. Due to the ill-posed nature of surface temperature inversion equations, where the spectral information is limited and assumptions about atmospheric properties or emissivity must be made, this inherently restricts the accuracy of temperature retrieval. To address this issue, it is essential to integrate various types of remote sensing data, such as high-spectral TIR imagery, and combine multiple remote sensing techniques including optical, infrared, and radar for comprehensive data fusion analysis. This approach significantly improves the accuracy and reliability of detecting geothermal resources. By merging multi-source data and fully utilizing the information gathered from different sensors, more informative insights can be obtained to mitigate the effects of uneven surface temperatures and atmospheric disturbances. Moreover, harnessing a variety of remote sensing devices to acquire high-quality data and increasing observation frequencies whenever possible allows for the capturing of more nuanced details related to geothermal resources.

(2)

Space analysis technology is an important component of geothermal resource exploration. Through GIS, researchers can integrate remotely sensed data with geological and geophysical parameters such as terrestrial heat flow, granite distribution, fault locations, Bouguer gravity anomalies, aeromagnetic anomalies, and depths to basement rocks. Utilizing models like information content models, evidential weight models, fuzzy logic models, and the analytic hierarchy process (AHP) for spatial modeling analysis allows for a more holistic understanding of the complexity inherent in geothermal systems. This integrated spatial analysis approach aids in the comprehensive assessment of potential geothermal zones from multiple perspectives and scales, thereby significantly enhancing the accuracy and reliability of geothermal resource detection. Moreover, conducting predictive accuracy assessments helps validate the model effectiveness and ensures the scientific validity and practical utility of the forecasting outcomes.

(3)

Artificial intelligence and mechanical learning technologies are driving revolutionary progress in the field of geothermal resource exploration. These technologies, which have already demonstrated significant potential in information retrieval and data mining, are now being leveraged to enhance the accuracy and efficiency of geothermal resource detection. By analyzing vast amounts of historical data through advanced learning algorithms, predictive models and classifiers can be developed to achieve precise predictions and quantitative evaluations of geothermal resource distributions. The integration of AI and machine learning into geothermal prospecting not only significantly boosts the speed of exploration but also delivers more accurate prediction outcomes, providing robust support for the development and utilization of geothermal energy, driving the field towards higher levels of sophistication and effectiveness.

(4)

Interdisciplinary comprehensive research is becoming increasingly important in the exploration and evaluation of geothermal resources. By strengthening interdisciplinary collaboration among fields such as geology, geophysics, geochemistry, geography, and hydrology, researchers can delve more deeply into the formation mechanisms and storage characteristics of geothermal resources, thereby significantly enhancing the accuracy of geothermal resource detection. Through the integration of theories and methods from multiple disciplines, scientists can provide a more comprehensive assessment of the potential of geothermal resources, offering more scientifically robust and reliable foundations for the development and utilization of geothermal energy.

Looking ahead, the synergistic application of high-resolution remote sensing technologies, multi-source data fusion and spatial analysis, machine learning, and artificial intelligence applications, alongside the integration of subsurface detection methods with geophysical and geochemical exploration promises a real-time and dynamically updated high-precision and efficient system for exploring and assessing geothermal resources. This comprehensive approach aims to provide accurate, reliable, and sustainable scientific underpinnings for the development and utilization of geothermal energy, thereby significantly driving the growth of renewable energy sectors.