Probabilistic geothermal resource assessment method and apparatus, system, storage medium

Abstract

The application discloses a kind of probabilistic geothermal resource evaluation method and device, system, storage medium, comprising: step S1, by three-factor adaptive weight function fusion temperature field and measured data, obtain synthetic temperature field data;Step S2, based on geothermal parameter physical coupling constraint parameter inter-coupling relationship is established;Step S3, training sample is generated;Step S4, according to training sample, by heat balance constraint training proxy model;Step S5, based on the proxy model of training based on ternary risk index Monte Carlo prediction is carried out and Sobol sensitivity analysis is carried out.The technical scheme of the application is used, and efficient, reliable, dynamic updating resource quantity uncertainty quantification evaluation is realized.

Description

Technical Field

[0001] This invention belongs to the field of geothermal resource exploration and evaluation technology, specifically relating to a probabilistic geothermal resource evaluation method, device, system, and storage medium, applicable to sedimentary basin hydrothermal geothermal systems where heat conduction is the dominant heat transfer mechanism. Background Technology

[0002] Geothermal resource assessment is crucial for project investment decisions. In the early stages of exploration, direct data (such as drilling data) is scarce, and the assessment faces two major challenges: 1) difficulties in model building and parameter determination due to small sample sizes; 2) extremely high computational costs in probabilistic analysis due to the use of complex numerical models in Monte Carlo simulations.

[0003] Sedimentary basin-type hydrothermal geothermal systems have unique physical characteristics that distinguish them from volcanic and fractured geothermal systems: (1) The reservoir is a porous medium (sandstone, siltstone, etc.) with relatively low permeability (typically 10-100 mD), and heat transfer is dominated by thermal conduction, with Peckley numbers usually less than 1; (2) The strata are horizontally layered, and the temperature field is mainly controlled by vertical heat flow, which can be described by a one-dimensional layered model; (3) The reservoir parameters (porosity, permeability, thermal conductivity) have a clear physical coupling relationship with the sedimentary lithology. These characteristics provide a theoretical basis for constructing a physical constraint proxy model, and also require that the evaluation method must specifically embed the physical constraints dominated by thermal conduction, rather than simply applying general methods applicable to convection-dominated systems.

[0004] Currently, the probabilistic volume method combined with Monte Carlo simulation is the standard approach for handling parameter uncertainty. However, existing practices often rely on human experience to determine parameter distribution and use full-order numerical simulators (such as TOUGH2) for capacity forecasting. This results in problems such as high subjectivity and excessive computation time (ten thousand simulations are usually not feasible), making it difficult to achieve fast and dynamic engineering decision-making.

[0005] The specific shortcomings of existing technologies include: 1) Crude data fusion methods. Existing physical enhancement methods mostly use fixed weights to fuse synthetic data with measured data, without considering the effects of depth variations, spatial distance, and measurement errors, leading to systematic biases in sparse data regions. 2) Lack of physical constraints in surrogate models. Traditional surrogate models such as Gaussian process regression are based solely on data fitting and do not embed the physical laws of thermal storage, which may produce non-physical predictions that violate the laws of heat conduction, especially in the parameter space boundary region. 3) Limited extrapolation risk assessment. Existing methods mostly rely on prediction variance or input boundaries for extrapolation judgments, lacking a systematic risk assessment mechanism that comprehensively considers statistical uncertainty and physical rationality. 4) Neglect of physical coupling in parameter sampling. In Monte Carlo sampling, each parameter is extracted independently or only statistical correlation is considered, without incorporating physical causal constraints such as permeability and porosity, resulting in a large number of invalid samples and reduced computational efficiency. Summary of the Invention

[0006] To address the problems existing in the prior art, this invention provides a probabilistic geothermal resource evaluation method, device, system, and storage medium specifically for heat conduction-dominated sedimentary basin geothermal systems. Through deep adaptive physical-data fusion, thermal balance constraint proxy model, and ternary risk index extrapolation protection, it achieves efficient, reliable, and dynamically updatable quantitative evaluation of resource uncertainty.

[0007] To achieve the above objectives, the present invention provides the following solution: A probabilistic geothermal resource assessment method, comprising: Step S1: The temperature field and measured data are fused using a three-factor adaptive weighting function to obtain synthetic temperature field data; Step S2: Establish the physical coupling relationship between parameters based on the physical coupling constraints of geothermal parameters; Step S3: Generate training samples; Step S4: Train the surrogate model using thermal balance constraints based on the training samples; Step S5: Perform Monte Carlo prediction based on the ternary risk index based on the trained agent model and conduct Sobol sensitivity analysis.

[0008] As a preferred embodiment, it also includes: step S6, automatically triggering agent model revision, distribution update and agent retraining when new data is input, forming a closed-loop iteration.

[0009] As a preferred approach, the physical coupling constraints of geothermal parameters include: permeability-porosity coupling, temperature-depth coupling, and thermal conductivity-lithology coupling.

[0010] The present invention also provides a probabilistic geothermal resource assessment device, comprising: The first processing module is used to fuse the temperature field and measured data through a three-factor adaptive weighting function to obtain synthetic temperature field data. The second processing module is used to establish the physical coupling relationship between parameters based on the physical coupling constraints of geothermal parameters; The third processing module is used to generate training samples; The fourth processing module is used to train the surrogate model based on the training samples and through thermal balance constraints. The fifth processing module is used to perform Monte Carlo prediction based on the ternary risk index based on the trained agent model and to conduct Sobol sensitivity analysis.

[0011] As a preferred embodiment, it also includes a sixth processing module, which is used to automatically trigger agent model revision, distribution update and agent retraining when new data is input, forming a closed-loop iteration.

[0012] As a preferred approach, the physical coupling constraints of geothermal parameters include: permeability-porosity coupling, temperature-depth coupling, and thermal conductivity-lithology coupling.

[0013] The present invention also provides a probabilistic geothermal resource evaluation system, comprising a memory and a processor, wherein the memory stores a computer program executed by the processor, and the computer program executes a probabilistic geothermal resource evaluation method when executed by the processor.

[0014] The present invention also provides a storage medium storing a computer program, which executes a probabilistic geothermal resource evaluation method when running.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention addresses the spatial bias problem in small-sample modeling through a deep adaptive physical-data fusion algorithm, ensures the physical rationality of predictions through a thermal balance-constrained Gaussian process surrogate model, achieves robust extrapolation protection through a ternary risk index, improves the effective sample rate through physical coupling sampling of geothermal parameters, and constructs a closed-loop iterative workflow to achieve dynamically updated resource evaluation as exploration progresses. Attached Figure Description

[0016] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of the probabilistic geothermal resource evaluation method according to an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] Example 1 like Figure 1 As shown, the present invention provides a probabilistic geothermal resource evaluation method, including: S1, Synthetic Temperature Field Data It allows users to set prior ranges for substrate heat flux, thermal conductivity, etc., and uses Latin hypercube / optimal Latin hypercube sampling to generate parameter combinations. It then calls the finite difference / finite element solver to solve the steady-state heat conduction equation and generate synthetic temperature field data.

[0021] The physical model used in this step is for sedimentary basin-type geothermal reservoir systems dominated by thermal conduction. In such systems, formation permeability is low (typically 10-100 mD), groundwater flow is slow, and Peckley numbers are high. (in The volumetric heat capacity of the fluid. For Darcy velocity, For characteristic length, Since the thermal conductivity is negligible, convective heat transfer is negligible, and the temperature field is mainly controlled by vertical heat conduction.

[0022] The steady-state heat conduction equation is ; in Thermal conductivity (unit: W / (m·K)) The radioactive heat generation rate (unit: μW / m³, typical value for sedimentary rocks: 0.5-2.0 μW / m³). For temperature.

[0023] Since the strata in the sedimentary basin are horizontally layered, one-dimensional vertical layering discretization is used. ; This represents the top boundary heat flow (unit: mW / m², typical value for sedimentary basins: 40-80 mW / m²).

[0024] A three-factor adaptive weighting function is used to fuse synthetic and measured data. ; in, For depth The variance of the synthesized temperature field (in °C²) reflects the uncertainty of the physical model at that depth; For depth Measurement error variance of the measured data (unit: °C²); For depth Spatial distance to the nearest measured point (in meters); It is the distance attenuation coefficient (unit: ℃² / m), which physically represents the equivalent variance increment per unit distance.

[0025] The enhanced temperature field after fusion is ; Near the actual measurement point ( Small and high measurement accuracy When the actual measurement point is far away or the measurement error is large, the weight of the physical model synthesized data increases, thus achieving optimal fusion that is spatially adaptive.

[0026] Distance attenuation coefficient Determined by one of the following methods: Method 1 (applicable to scenarios with ≥3 measured points): Perform leave-one-out cross-validation at known measured points, aiming to minimize the root mean square error of the fused temperature field prediction at the left-out points, and optimize accordingly. value; Method 2 (suitable for insufficient measurement points or as the initial value for Method 1): Based on the accuracy of the temperature measuring instrument and regional geothermal gradient variability Using empirical formulas It is confirmed that, among them, This is a calibration constant, with a typical value range of 0.5-2.0.

[0027] The enhanced temperature field data output in this step This will serve as the base temperature field for building the physical model and setting the initial conditions for the simulation in subsequent steps.

[0028] S2. Establish physical coupling relationships between parameters based on physical coupling constraints of geothermal parameters. The physical coupling constraints established in this step are specifically designed for the porosity characteristics of sedimentary basin-type hydrothermal reservoirs. Sedimentary basin-type hydrothermal reservoirs are mainly composed of porous media such as sandstone, siltstone, and carbonate rocks. Their permeability-porosity relationship follows the classical granular media flow theory, which is different from the discrete fracture network model of fracture-type hydrothermal reservoirs.

[0029] The physical coupling constraints of geothermal parameters include the following: (1) Permeability-porosity coupling (Kozeny; Carman constraint) is ; in, Permeability (unit: m) 2 1 mD = 9.87 × 10 -16 m 2 ), Porosity (dimensionless, typical value for sedimentary sandstone: 0.05-0.30). Equivalent particle diameter (unit: m, 0.1-0.25 mm for fine sand, 0.25-0.5 mm for medium sand). During sampling, the following parameters are given. and back, The parameters must fall within the range allowed by the equation. This constraint ensures that the sampling parameters satisfy the physical laws of Darcy flow.

[0030] (2) Temperature-depth coupling (geothermal gradient constraint), based on the temperature field characteristics of the heat conduction-dominated system. ; in, , This represents the physically reasonable range of the regional geothermal gradient (typically 15-45℃ / km in sedimentary basins, and up to 60℃ / km in high heat flow areas). This constraint excludes abnormal temperature distributions caused by improper parameter combinations.

[0031] (3) Thermal conductivity-lithology coupling is based on the physical relationship between the mineral composition and thermal conductivity of sedimentary rocks. A preset lithology-thermal conductivity mapping database is used: sandstone 2.5-4.5 W / (m·K) (higher thermal conductivity is associated with higher quartz content); mudstone / shale 1.5-2.5 W / (m·K); carbonate rocks 2.0-3.5 W / (m·K); evaporite 3.5-6.0 W / (m·K). During sampling, the thermal conductivity a priori range is automatically matched according to the lithology of the strata.

[0032] Sampling strategy: When performing Monte Carlo sampling, perform the above three types of physical coupling checks on each set of parameters and reject combinations that violate any of the constraints, rather than removing them afterward, which can significantly improve the effective sample rate.

[0033] S3. Generating training samples based on physical constraints and enhanced temperature field This step generates samples for training the surrogate model through experimental design. Its execution directly depends on the output of steps S1 and S2. The specific process is as follows: (1) Determine the parameter space and sampling rules (depending on S2): Based on the physical coupling constraints of geothermal parameters established in step S2, the reasonable value range and joint distribution of each input parameter are determined. Optimal Latin hypercube sampling (LHS) is used to extract N sets of parameter combinations within this physically constrained parameter space. The sampling here incorporates physical coupling verification to ensure that each set of parameters is physically reasonable.

[0034] (2) Construct a numerical model and set initial conditions (depending on S1): For each set of parameters Construct a corresponding high-fidelity numerical model (such as the TOUGH2 model). Then, use the spatially adaptive enhanced temperature field generated in step S1. This serves as the initial temperature field for the model, ensuring that the starting point for the numerical simulation is a spatially optimal temperature distribution that incorporates measured information.

[0035] (3) Perform the simulation and obtain the output sample: For each set of parameters and its corresponding model, run a numerical simulation under the specified scenario. Upon successful completion, extract the target output variables (such as cumulative heat generation). ).Will( , () as a training sample.

[0036] (4) Quality control: Check the convergence of all simulations, remove simulations that fail or whose results clearly violate physical laws (such as energy non-conservation), and finally obtain M valid (input, output) training sample pairs.

[0037] S4. Based on the training samples, train the surrogate model using thermal balance constraints. The surrogate model constructed in this step is specifically designed for heat conduction-dominated systems, applying the steady-state heat conduction law. By directly embedding the product term of the kernel function, a "thermal balance constraint kernel function" is formed. This design allows parameter combinations that violate the law of heat conduction to automatically receive extremely low correlation weights in the kernel function calculation, thus effectively suppressing them in the prediction stage and exhibiting stronger rigidity than the mean function constraint.

[0038] The surrogate model is trained using a thermal equilibrium-constrained Gaussian process regression, with the kernel function being: ; in, Standard radial basis kernel function; thermal equilibrium constraint penalty term A function defined as an input parameter vector; Standard RBF core is ; in, The variance of the signal is the overall variation of the output of the control function. For the first The length scale parameter of the input dimension controls the smoothness of the function's variation along that dimension; To observe the noise variance and characterize the measurement error in the training data; For the Kronecker delta function, when The value is 1 if it is true, and 0 otherwise.

[0039] The thermal equilibrium constraint penalty term (in the functional form of the input parameters) is: ; The physical residual function is ; in, For the heat flux component in the input parameter vector, For thermal conductivity components, The temperature gradient is derived from the input parameters using the heat conduction equation.

[0040] The positive definiteness proof is that the penalty term can be decomposed into ,in Always positive. Since the product of positive definite kernels is still a positive definite kernel, therefore... Maintain positive definiteness.

[0041] The physical meaning is required by the law of heat conduction. When the parameter combination makes Penalty item when deviating from zero Approaching zero, the weight of this combination in Gaussian process prediction is reduced, thereby automatically suppressing non-physical understanding.

[0042] The constraint strength coefficient The initial setting is determined adaptively in the following manner: A smaller value (e.g., 0.1); during the training of the surrogate model, monitor the physical residuals on the validation set. The distribution of; defining the physical violation rate. Exceeding the physical reasonable threshold The sample proportion; if the physical violation rate Then increase (e.g., multiply by 1.5); iteratively adjust until the physical violation rate drops below 5%, while maintaining the surrogate model's prediction accuracy. ;like If it drops below 0.90, it will revert. Up to the previous valid value.

[0043] S5. Perform Monte Carlo prediction based on ternary risk index protection based on the trained agent model and conduct Sobol sensitivity analysis. The Monte Carlo prediction uses a ternary risk index for extrapolation protection, and the ternary risk index is defined as follows: ; in, The Mahalanobis distance from the input point to the boundary of the convex hull of the training data (e.g., calculated using the mean and covariance of the training set) is used to measure the degree of statistical extrapolation of the input parameter space. Normalized prediction variance ( (The mean of the target variable in the training set) measures the uncertainty of the surrogate model in the output function space; The thermal equilibrium residual is used to assess physical plausibility. It should be noted that... and Although there is some correlation in Gaussian processes, the former captures the statistical extrapolation of parameter combinations from the perspective of the input space, while the latter reflects the uncertainty of the predicted function from the perspective of the output space. The two perspectives are complementary, and their combined use can more comprehensively assess the extrapolation risk.

[0044] To eliminate dimensional differences, the risk index is calculated using a standardized summation form: ; in, , , These are the standard deviations of the three component indicators on the training set; As the importance weight, the default value is 100%. The formula achieves dimensionlessness by dividing by the standard deviation.

[0045] Critical threshold The method for determining this is to perform leave-one-out cross-validation on the training data and calculate the leave-out sample's... Combine prediction error with a plot of the error-risk index relationship curve, selecting the inflection point where the error rises sharply (which can be determined through curvature maximization or piecewise linear fitting) as the inflection point. .

[0046] Protection strategy for when When this occurs, the sample triggers full-order simulation rollback or is marked as low confidence, and is downweighted or removed in sensitivity analysis.

[0047] The Monte Carlo simulation engine employs a joint convergence criterion combining quantiles, variance, and Sobol metrics. Quantile convergence: ,in , , ; Variance convergence: , ; Sobol convergence: , ; Convergence is determined when all three of the above indicators are met simultaneously.

[0048] For high The sensitivity results of the samples are marked as low confidence or removed from the Sobol calculation.

[0049] S6. Information Gain-Driven Iterative Updates and Version Management After new data is input, the information gain metric is calculated and a corresponding update strategy is triggered, with version records supporting backtracking; specifically: Calculate the information gain metric for new data, including the shift in the distribution of key parameters. and the proportion of new samples ; when or At that time, the proxy model is fully retrained. When validation set performance degrades within a threshold (NLL increases) , decline When this is the case, only fine-tune the hyperparameters; Records the version number, hyperparameters, verification metrics, sensitivity sorting, and data source for each update, and supports version rollback and rollback.

[0050] The embodiments of the present invention have the following technical effects: 1. Spatial adaptive fusion is a three-factor adaptive weight function that dynamically adjusts the fusion weights based on depth, distance and measurement error, overcoming the systematic bias of fixed weight methods in sparse data areas and improving the modeling reliability in small sample scenarios.

[0051] 2. The physical constraint surrogate uses a thermal equilibrium constraint kernel function to enable the surrogate model to automatically suppress non-physical interpretations that violate the law of heat conduction during training and prediction, thereby improving the physical rationality of the prediction and ensuring the positive definiteness of the kernel function.

[0052] 3. Robust extrapolation protection: The three-element risk index comprehensively considers the degree of statistical extrapolation, forecast uncertainty and physical rationality. The weight adaptive normalization ensures that the contribution of each indicator is balanced, and identifies high-risk predictions more comprehensively than a single indicator.

[0053] 4. Efficient physical sampling: Physically coupled sampling of geothermal parameters rejects non-physical combinations during the sampling stage, which significantly improves the effective sample rate (from 65% to 92%) compared to post-sampling and reduces computational waste.

[0054] 5. Computational acceleration reduces the time required for 10,000 Monte Carlo simulations from "monthly" to "hourly" for surrogate models; supports parallel / batch prediction.

[0055] 6. Quantify uncertainty by outputting P10 / P50 / P90, intervals, and sensitivity, and provide Sobol confidence intervals to facilitate risk decision-making.

[0056] 7. The dynamic closed-loop retraining strategy, driven by information gain, intelligently switches between incremental fine-tuning and full retraining, balancing computational cost and model timeliness.

[0057] Example 2 The present invention also provides a probabilistic geothermal resource assessment device, comprising: The first processing module is used to fuse the temperature field and measured data through a three-factor adaptive weighting function to obtain synthetic temperature field data. The second processing module is used to establish the physical coupling relationship between parameters based on the physical coupling constraints of geothermal parameters; The third processing module is used to generate training samples; The fourth processing module is used to train the surrogate model based on the training samples and through thermal balance constraints. The fifth processing module is used to perform Monte Carlo prediction based on the ternary risk index based on the trained agent model and to conduct Sobol sensitivity analysis.

[0058] As a preferred embodiment, it also includes a sixth processing module, which is used to automatically trigger agent model revision, distribution update and agent retraining when new data is input, forming a closed-loop iteration.

[0059] As a preferred approach, the physical coupling constraints of geothermal parameters include: permeability-porosity coupling, temperature-depth coupling, and thermal conductivity-lithology coupling.

[0060] Example 3 The present invention also provides a probabilistic geothermal resource evaluation system, comprising a memory and a processor, wherein the memory stores a computer program executed by the processor, and the computer program executes a probabilistic geothermal resource evaluation method when executed by the processor.

[0061] Example 4 The present invention also provides a storage medium storing a computer program, which executes a probabilistic geothermal resource evaluation method when running.

[0062] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A probabilistic geothermal resource evaluation method, characterized in that, include: Step S1: The temperature field and measured data are fused using a three-factor adaptive weighting function to obtain synthetic temperature field data; Step S2: Establish the physical coupling relationship between parameters based on the physical coupling constraints of geothermal parameters; Step S3: Generate training samples; Step S4: Train the surrogate model using thermal balance constraints based on the training samples; Step S5: Perform Monte Carlo prediction based on the ternary risk index based on the trained agent model and conduct Sobol sensitivity analysis; Among them, the physical coupling constraint of geothermal parameters in S2 is aimed at the porosity characteristics of sedimentary basin-type geothermal reservoirs; sedimentary basin-type geothermal reservoirs are composed of sandstone, siltstone, and carbonate rock porosity. The physical coupling constraints of geothermal parameters include the following: (1) The permeability-porosity coupling constraint is: ； in, For penetration rate, Porosity The equivalent particle diameter; the permeability-porosity coupling constraint ensures that the sampling parameters satisfy the physical laws of Darcy flow; (2) The temperature-depth coupling constraint is: ； in, , The physically reasonable range of the regional geothermal gradient; temperature-depth coupling constraints eliminate abnormal temperature distributions caused by improper parameter combinations; (3) Thermal conductivity-lithology coupling is based on the physical relationship between the mineral composition and thermal conductivity of sedimentary rocks. A preset lithology-thermal conductivity mapping database is used: the thermal conductivity of sandstone is 2.5-4.5; the thermal conductivity of mudstone / shale is 1.5-2.5; the thermal conductivity of carbonate rocks is 2.0-3.5; and the thermal conductivity of evaporites is 3.5-6.

0. During sampling, the thermal conductivity prior range is automatically matched according to the lithology of the strata. During Monte Carlo sampling, each set of parameters is subjected to the three types of physical coupling checks mentioned above, and combinations that violate any constraint are rejected, rather than being eliminated afterward.

2. The probabilistic geothermal resource evaluation method as described in claim 1, characterized in that, It also includes: step S6, automatically triggering agent model revision, distribution update and agent retraining when new data is input, forming a closed loop iteration.

3. The probabilistic geothermal resource evaluation method as described in claim 1, characterized in that, The physical coupling constraints of geothermal parameters include: permeability-porosity coupling, temperature-depth coupling, and thermal conductivity-lithology coupling.

4. A probabilistic geothermal resource evaluation device for implementing the probabilistic geothermal resource evaluation method of claim 1, characterized in that, include: The first processing module is used to fuse the temperature field and measured data through a three-factor adaptive weighting function to obtain synthetic temperature field data. The second processing module is used to establish the physical coupling relationship between parameters based on the physical coupling constraints of geothermal parameters; The third processing module is used to generate training samples; The fourth processing module is used to train the surrogate model based on the training samples and through thermal balance constraints. The fifth processing module is used to perform Monte Carlo prediction based on the ternary risk index based on the trained agent model and to conduct Sobol sensitivity analysis.

5. The probabilistic geothermal resource evaluation device as described in claim 4, characterized in that, Also includes: The sixth processing module is used to automatically trigger agent model revision, distribution update and agent retraining when new data is input, forming a closed loop iteration.

6. The probabilistic geothermal resource evaluation device as described in claim 5, characterized in that, The physical coupling constraints of geothermal parameters include: permeability-porosity coupling, temperature-depth coupling, and thermal conductivity-lithology coupling.

7. A probabilistic geothermal resource evaluation system, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that is executed by the processor, and the computer program, when executed by the processor, performs the probabilistic geothermal resource evaluation method as described in any one of claims 1-3.

8. A storage medium, characterized in that, The storage medium stores a computer program, which executes the probabilistic geothermal resource evaluation method as described in any one of claims 1-3 when it is run.

Images