Peter Zaspel
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Peter Zaspel has a total of 28 co-authors. Here are the top co-authors by number of shared publications:
- Vivin Vinod
University of Wuppertal
10 shared publications - Kshema Shaju
Alfred Wegener Institute, Potsdam, Germany
8 shared publications - Thomas Laepple
Alfred-Wegener-Institut Helmholtz-Zentrum für Polar- und Meeresforschung, Research Unit Potsdam, Potsdam, Germany
8 shared publications - Nora Hirsch
Alfred Wegener Institute Helmholtz Centre for Polar and Marine Research Potsdam Germany
7 shared publications - Ulrich Kleinekathöfer
Constructor University
6 shared publications - Dongyu Lyu
Constructor University
3 shared publications
Latest Publications
- 21 May 2026
- Published by American Chemical Society (ACS) in Journal of Chemical Theory and Computation
- Vol. 22 (11) , 5637-5648
LFaB: Low Fidelity as Bias for Active Learning in the Chemical Configuration Space
Active learning promises to provide an optimal training sample selection procedure in the construction of machine learning models. It often relies on minimizing the model's variance, which is assumed to decrease the prediction error. Still, it is frequently even less efficient than pure random sampling. Motivated by the bias-variance decomposition, we propose to minimize the model's bias instead of its variance. By doing so, we are able to almost exactly match the best-case error over all possible greedy sample selection procedures for a relevant application. Our bias approximation is based on using cheap to calculate low fidelity data as known from Δ-ML or multifidelity machine learning. We exemplify our approach for a wider class of applications in quantum chemistry including predicting excitation energies and ab initio potential energy surfaces. Here, the proposed method reduces training data consumption by up to an order of magnitude compared to standard active learning.
- 21 May 2026
- Published by American Chemical Society (ACS) in Journal of Chemical Theory and Computation
- Vol. 22 (11) , 5637-5648
LFaB: Low Fidelity as Bias for Active Learning in the Chemical Configuration Space
Active learning promises to provide an optimal training sample selection procedure in the construction of machine learning models. It often relies on minimizing the model's variance, which is assumed to decrease the prediction error. Still, it is frequently even less efficient than pure random sampling. Motivated by the bias-variance decomposition, we propose to minimize the model's bias instead of its variance. By doing so, we are able to almost exactly match the best-case error over all possible greedy sample selection procedures for a relevant application. Our bias approximation is based on using cheap to calculate low fidelity data as known from Δ-ML or multifidelity machine learning. We exemplify our approach for a wider class of applications in quantum chemistry including predicting excitation energies and ab initio potential energy surfaces. Here, the proposed method reduces training data consumption by up to an order of magnitude compared to standard active learning.
- Open Access
- 10 December 2025
- Published by Copernicus GmbH in Geoscientific Instrumentation, Methods and Data Systems
- Vol. 14 (2) , 459-474
Ice borehole thermometry: sensor placement using greedy optimal sampling
Borehole thermometry is an important tool for reconstructing past climate conditions, assessing changes in land energy storage, and understanding subsurface thermal regimes such as permafrost and glacial dynamics. Optimizing the temperature sensor placement within boreholes allows us to maximize the informativeness of temperature measurements, particularly in polar regions where operational constraints necessitate cost-effective solutions. Traditional sensor placement methods such as linear or exponential spacing, often overlook site-specific subsurface ice heat distribution characteristics, potentially limiting the accuracy of the measured temperature profile. In this paper, we propose a greedy optimal sampling technique for strategically placing temperature sensors in ice boreholes. Utilizing heat transfer model simulations, this method selects sensor locations that minimize interpolation errors in reconstructed temperature profiles. We apply our approach to two distinct ice borehole sites: EPICA Dronning Maud Land site in East Antarctica and the Greenland Ice Core Project site, each with unique surface conditions. Our results demonstrate that the greedy optimal sensor placement significantly outperforms conventional linear and exponential spacing methods, reducing sampling errors by up to a factor of ten and thus achieving similar informativeness with fewer sensors. This strategy offers a cost–effective means to maximize the information obtained from ice borehole temperature measurements, thereby potentially enhancing the precision of climate reconstructions.
- 10 December 2025
- Published by Copernicus GmbH in Geoscientific Instrumentation, Methods and Data Systems
- Vol. 14 (2) , 459-474
Open AccessIce borehole thermometry: sensor placement using greedy optimal sampling
Borehole thermometry is an important tool for reconstructing past climate conditions, assessing changes in land energy storage, and understanding subsurface thermal regimes such as permafrost and glacial dynamics. Optimizing the temperature sensor placement within boreholes allows us to maximize the informativeness of temperature measurements, particularly in polar regions where operational constraints necessitate cost-effective solutions. Traditional sensor placement methods such as linear or exponential spacing, often overlook site-specific subsurface ice heat distribution characteristics, potentially limiting the accuracy of the measured temperature profile. In this paper, we propose a greedy optimal sampling technique for strategically placing temperature sensors in ice boreholes. Utilizing heat transfer model simulations, this method selects sensor locations that minimize interpolation errors in reconstructed temperature profiles. We apply our approach to two distinct ice borehole sites: EPICA Dronning Maud Land site in East Antarctica and the Greenland Ice Core Project site, each with unique surface conditions. Our results demonstrate that the greedy optimal sensor placement significantly outperforms conventional linear and exponential spacing methods, reducing sampling errors by up to a factor of ten and thus achieving similar informativeness with fewer sensors. This strategy offers a cost–effective means to maximize the information obtained from ice borehole temperature measurements, thereby potentially enhancing the precision of climate reconstructions.
- Open Access
- 27 October 2025
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 6 (4) , 045019
Evaluation of uncertainty estimations for Gaussian process regression based machine learning interatomic potentials
Uncertainty estimations for machine learning interatomic potentials (MLIPs) are crucial for quantifying model error and identifying informative training samples in active learning (AL) strategies. In this study, we evaluate uncertainty estimations of Gaussian process regression (GPR)-based MLIPs, including the predictive GPR standard deviation and ensemble-based uncertainties. We do this in terms of calibration and in terms of impact on model performance in an AL scheme. We consider GPR models with Coulomb and smooth overlap of atomic positions representations as inputs to predict potential energy surfaces and excitation energies of molecules. Regarding calibration, we find that ensemble-based uncertainty estimations show already poor global calibration (e.g. averaged over the whole test set). In contrast, the GPR standard deviation shows good global calibration, but when grouping predictions by their uncertainty, we observe a systematical bias for predictions with high uncertainty. Although an increasing uncertainty correlates with an increasing bias, the bias is not captured quantitatively by the uncertainty. Therefore, the GPR standard deviation can be useful to identify predictions with a high bias and error but, without further knowledge, should not be interpreted as a quantitative measure for a potential error range. Selecting the samples with the highest GPR standard deviation from a fixed configuration space leads to a model that overemphasizes the borders of the configuration space represented in the fixed dataset. This may result in worse performance in more densely sampled areas but better generalization for extrapolation tasks.
- 27 October 2025
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 6 (4) , 045019
Open AccessEvaluation of uncertainty estimations for Gaussian process regression based machine learning interatomic potentials
Uncertainty estimations for machine learning interatomic potentials (MLIPs) are crucial for quantifying model error and identifying informative training samples in active learning (AL) strategies. In this study, we evaluate uncertainty estimations of Gaussian process regression (GPR)-based MLIPs, including the predictive GPR standard deviation and ensemble-based uncertainties. We do this in terms of calibration and in terms of impact on model performance in an AL scheme. We consider GPR models with Coulomb and smooth overlap of atomic positions representations as inputs to predict potential energy surfaces and excitation energies of molecules. Regarding calibration, we find that ensemble-based uncertainty estimations show already poor global calibration (e.g. averaged over the whole test set). In contrast, the GPR standard deviation shows good global calibration, but when grouping predictions by their uncertainty, we observe a systematical bias for predictions with high uncertainty. Although an increasing uncertainty correlates with an increasing bias, the bias is not captured quantitatively by the uncertainty. Therefore, the GPR standard deviation can be useful to identify predictions with a high bias and error but, without further knowledge, should not be interpreted as a quantitative measure for a potential error range. Selecting the samples with the highest GPR standard deviation from a fixed configuration space leads to a model that overemphasizes the borders of the configuration space represented in the fixed dataset. This may result in worse performance in more densely sampled areas but better generalization for extrapolation tasks.
- 11 July 2025
- Published by AIP Publishing in The Journal of Chemical Physics
- Vol. 163 (2)
Benchmarking data efficiency in Δ-ML and multifidelity models for quantum chemistry
The development of machine learning (ML) methods has made quantum chemistry (QC) calculations more accessible by reducing the computational cost incurred in conventional QC methods. This has since been translated into the overhead cost of generating training data. Increased work in reducing the cost of generating training data resulted in the development of Δ-ML and multifidelity machine learning methods, which use data at more than one QC level of accuracy, or fidelity. This work compares the data costs associated with Δ-ML, multifidelity machine learning (MFML), and optimized MFML in contrast with a newly introduced MultifidelityΔ-Machine Learning (MFΔML) method for the prediction of ground state energies, vertical excitation energies, and the magnitude of the electronic contribution of molecular dipole moments from the multifidelity benchmark dataset QeMFi. This assessment is made on the basis of the training data generation cost associated with each model and is compared with the single fidelity kernel ridge regression case. The results indicate that the use of multifidelity methods surpasses the standard Δ-ML approaches in cases of a large number of predictions. In applications where only a few numbers of predictions/evaluations are to be made using ML models, the herein developed MFΔML method is shown to provide an added advantage over conventional Δ-ML.
- 11 July 2025
- Published by AIP Publishing in The Journal of Chemical Physics
- Vol. 163 (2)
Benchmarking data efficiency in Δ-ML and multifidelity models for quantum chemistry
The development of machine learning (ML) methods has made quantum chemistry (QC) calculations more accessible by reducing the computational cost incurred in conventional QC methods. This has since been translated into the overhead cost of generating training data. Increased work in reducing the cost of generating training data resulted in the development of Δ-ML and multifidelity machine learning methods, which use data at more than one QC level of accuracy, or fidelity. This work compares the data costs associated with Δ-ML, multifidelity machine learning (MFML), and optimized MFML in contrast with a newly introduced MultifidelityΔ-Machine Learning (MFΔML) method for the prediction of ground state energies, vertical excitation energies, and the magnitude of the electronic contribution of molecular dipole moments from the multifidelity benchmark dataset QeMFi. This assessment is made on the basis of the training data generation cost associated with each model and is compared with the single fidelity kernel ridge regression case. The results indicate that the use of multifidelity methods surpasses the standard Δ-ML approaches in cases of a large number of predictions. In applications where only a few numbers of predictions/evaluations are to be made using ML models, the herein developed MFΔML method is shown to provide an added advantage over conventional Δ-ML.
- Open Access
- 30 June 2025
- Published by Copernicus GmbH
- Vol. 2025, 1-25
Reply on RC2
Borehole thermometry is an important tool for reconstructing past climate conditions, assessing changes in land energy storage, and understanding subsurface thermal regimes such as permafrost and glacial dynamics. Optimizing the temperature sensor placement within boreholes allows us to maximize the informativeness of temperature measurements, particularly in polar regions where operational constraints necessitate cost-effective solutions. Traditional sensor placement methods such as linear or exponential spacing, often overlook site-specific subsurface heat distribution characteristics, potentially limiting the accuracy of the measured temperature profile. In this paper, we propose a greedy optimal sampling technique for strategically placing temperature sensors in ice boreholes. Utilizing heat transfer model simulations, this method selects sensor locations that minimize interpolation errors in reconstructed temperature profiles. We apply our approach to two distinct borehole sites: EPICA Dronning Maud Land site in East Antarctica and the Greenland Ice Core Project site, each with unique surface conditions. Our results demonstrate that the greedy optimal sensor placement significantly outperforms conventional linear and exponential spacing methods, reducing sampling errors by up to a factor of ten and thus achieving similar informativeness with fewer sensors. This strategy offers a cost-effective means to maximize the information obtained from borehole temperature measurements, thereby potentially enhancing the precision of climate reconstructions.
- 30 June 2025
- Published by Copernicus GmbH
- Vol. 2025, 1-25
Open AccessReply on RC2
Borehole thermometry is an important tool for reconstructing past climate conditions, assessing changes in land energy storage, and understanding subsurface thermal regimes such as permafrost and glacial dynamics. Optimizing the temperature sensor placement within boreholes allows us to maximize the informativeness of temperature measurements, particularly in polar regions where operational constraints necessitate cost-effective solutions. Traditional sensor placement methods such as linear or exponential spacing, often overlook site-specific subsurface heat distribution characteristics, potentially limiting the accuracy of the measured temperature profile. In this paper, we propose a greedy optimal sampling technique for strategically placing temperature sensors in ice boreholes. Utilizing heat transfer model simulations, this method selects sensor locations that minimize interpolation errors in reconstructed temperature profiles. We apply our approach to two distinct borehole sites: EPICA Dronning Maud Land site in East Antarctica and the Greenland Ice Core Project site, each with unique surface conditions. Our results demonstrate that the greedy optimal sensor placement significantly outperforms conventional linear and exponential spacing methods, reducing sampling errors by up to a factor of ten and thus achieving similar informativeness with fewer sensors. This strategy offers a cost-effective means to maximize the information obtained from borehole temperature measurements, thereby potentially enhancing the precision of climate reconstructions.
- Open Access
- 30 June 2025
- Published by Copernicus GmbH
- Vol. 2025, 1-25
Reply on RC1
Borehole thermometry is an important tool for reconstructing past climate conditions, assessing changes in land energy storage, and understanding subsurface thermal regimes such as permafrost and glacial dynamics. Optimizing the temperature sensor placement within boreholes allows us to maximize the informativeness of temperature measurements, particularly in polar regions where operational constraints necessitate cost-effective solutions. Traditional sensor placement methods such as linear or exponential spacing, often overlook site-specific subsurface heat distribution characteristics, potentially limiting the accuracy of the measured temperature profile. In this paper, we propose a greedy optimal sampling technique for strategically placing temperature sensors in ice boreholes. Utilizing heat transfer model simulations, this method selects sensor locations that minimize interpolation errors in reconstructed temperature profiles. We apply our approach to two distinct borehole sites: EPICA Dronning Maud Land site in East Antarctica and the Greenland Ice Core Project site, each with unique surface conditions. Our results demonstrate that the greedy optimal sensor placement significantly outperforms conventional linear and exponential spacing methods, reducing sampling errors by up to a factor of ten and thus achieving similar informativeness with fewer sensors. This strategy offers a cost-effective means to maximize the information obtained from borehole temperature measurements, thereby potentially enhancing the precision of climate reconstructions.
- 30 June 2025
- Published by Copernicus GmbH
- Vol. 2025, 1-25
Open AccessReply on RC1
Borehole thermometry is an important tool for reconstructing past climate conditions, assessing changes in land energy storage, and understanding subsurface thermal regimes such as permafrost and glacial dynamics. Optimizing the temperature sensor placement within boreholes allows us to maximize the informativeness of temperature measurements, particularly in polar regions where operational constraints necessitate cost-effective solutions. Traditional sensor placement methods such as linear or exponential spacing, often overlook site-specific subsurface heat distribution characteristics, potentially limiting the accuracy of the measured temperature profile. In this paper, we propose a greedy optimal sampling technique for strategically placing temperature sensors in ice boreholes. Utilizing heat transfer model simulations, this method selects sensor locations that minimize interpolation errors in reconstructed temperature profiles. We apply our approach to two distinct borehole sites: EPICA Dronning Maud Land site in East Antarctica and the Greenland Ice Core Project site, each with unique surface conditions. Our results demonstrate that the greedy optimal sensor placement significantly outperforms conventional linear and exponential spacing methods, reducing sampling errors by up to a factor of ten and thus achieving similar informativeness with fewer sensors. This strategy offers a cost-effective means to maximize the information obtained from borehole temperature measurements, thereby potentially enhancing the precision of climate reconstructions.
- Open Access
- 19 June 2025
- Published by Wiley in Advanced Theory and Simulations
- Vol. 8 (11)
Excitation Energy Transfer between Porphyrin Dyes on a Clay Surface: A Study Employing Multifidelity Machine Learning
Natural light‐harvesting antenna complexes efficiently capture solar energy mostly using chlorophyll molecules, i.e., magnesium porphyrin pigments, embedded in a protein matrix. Inspired by this natural configuration, artificial clay‐porphyrin antenna structures have experimentally been synthesized and demonstrated to exhibit remarkable excitation energy transfer properties. The current study presents a computational design and simulations of a synthetic light‐harvesting system that emulates natural mechanisms by arranging cationic free‐base porphyrin molecules on an anionic clay surface. The transfer of excitation energy among the porphyrin dyes is investigated using a multiscale quantum mechanics/molecular mechanics (QM/MM) approach based on the semi‐empirical density functional‐based tight‐binding theory for the ground state dynamics. To improve the accuracy of the results, an innovative multifidelity machine learning approach is incorporated, which allows the prediction of excitation energies at the numerically demanding time‐dependent density functional theory level together with the def2‐SVP basis set. This approach is applied to an extensive dataset of 640 k geometries for the 90‐atom porphyrin structures, facilitating a thorough analysis of the excitation energy diffusion among the porphyrin molecules adsorbed to the clay surface. The insights gained from this study, inspired by natural light‐harvesting complexes, demonstrate the potential of porphyrin‐clay systems as effective energy transfer systems.
- 19 June 2025
- Published by Wiley in Advanced Theory and Simulations
- Vol. 8 (11)
Open AccessExcitation Energy Transfer between Porphyrin Dyes on a Clay Surface: A Study Employing Multifidelity Machine Learning
Natural light‐harvesting antenna complexes efficiently capture solar energy mostly using chlorophyll molecules, i.e., magnesium porphyrin pigments, embedded in a protein matrix. Inspired by this natural configuration, artificial clay‐porphyrin antenna structures have experimentally been synthesized and demonstrated to exhibit remarkable excitation energy transfer properties. The current study presents a computational design and simulations of a synthetic light‐harvesting system that emulates natural mechanisms by arranging cationic free‐base porphyrin molecules on an anionic clay surface. The transfer of excitation energy among the porphyrin dyes is investigated using a multiscale quantum mechanics/molecular mechanics (QM/MM) approach based on the semi‐empirical density functional‐based tight‐binding theory for the ground state dynamics. To improve the accuracy of the results, an innovative multifidelity machine learning approach is incorporated, which allows the prediction of excitation energies at the numerically demanding time‐dependent density functional theory level together with the def2‐SVP basis set. This approach is applied to an extensive dataset of 640 k geometries for the 90‐atom porphyrin structures, facilitating a thorough analysis of the excitation energy diffusion among the porphyrin molecules adsorbed to the clay surface. The insights gained from this study, inspired by natural light‐harvesting complexes, demonstrate the potential of porphyrin‐clay systems as effective energy transfer systems.
- Open Access
- 2 June 2025
- Published by Copernicus GmbH
- Vol. 2025, 1-25
Comment on egusphere-2024-3755
Borehole thermometry is an important tool for reconstructing past climate conditions, assessing changes in land energy storage, and understanding subsurface thermal regimes such as permafrost and glacial dynamics. Optimizing the temperature sensor placement within boreholes allows us to maximize the informativeness of temperature measurements, particularly in polar regions where operational constraints necessitate cost-effective solutions. Traditional sensor placement methods such as linear or exponential spacing, often overlook site-specific subsurface heat distribution characteristics, potentially limiting the accuracy of the measured temperature profile. In this paper, we propose a greedy optimal sampling technique for strategically placing temperature sensors in ice boreholes. Utilizing heat transfer model simulations, this method selects sensor locations that minimize interpolation errors in reconstructed temperature profiles. We apply our approach to two distinct borehole sites: EPICA Dronning Maud Land site in East Antarctica and the Greenland Ice Core Project site, each with unique surface conditions. Our results demonstrate that the greedy optimal sensor placement significantly outperforms conventional linear and exponential spacing methods, reducing sampling errors by up to a factor of ten and thus achieving similar informativeness with fewer sensors. This strategy offers a cost-effective means to maximize the information obtained from borehole temperature measurements, thereby potentially enhancing the precision of climate reconstructions.
- 2 June 2025
- Published by Copernicus GmbH
- Vol. 2025, 1-25
Open AccessComment on egusphere-2024-3755
Borehole thermometry is an important tool for reconstructing past climate conditions, assessing changes in land energy storage, and understanding subsurface thermal regimes such as permafrost and glacial dynamics. Optimizing the temperature sensor placement within boreholes allows us to maximize the informativeness of temperature measurements, particularly in polar regions where operational constraints necessitate cost-effective solutions. Traditional sensor placement methods such as linear or exponential spacing, often overlook site-specific subsurface heat distribution characteristics, potentially limiting the accuracy of the measured temperature profile. In this paper, we propose a greedy optimal sampling technique for strategically placing temperature sensors in ice boreholes. Utilizing heat transfer model simulations, this method selects sensor locations that minimize interpolation errors in reconstructed temperature profiles. We apply our approach to two distinct borehole sites: EPICA Dronning Maud Land site in East Antarctica and the Greenland Ice Core Project site, each with unique surface conditions. Our results demonstrate that the greedy optimal sensor placement significantly outperforms conventional linear and exponential spacing methods, reducing sampling errors by up to a factor of ten and thus achieving similar informativeness with fewer sensors. This strategy offers a cost-effective means to maximize the information obtained from borehole temperature measurements, thereby potentially enhancing the precision of climate reconstructions.
- Open Access
- 18 March 2025
- Published by Copernicus GmbH
Shallow Ice Borehole Thermometry: Bayesian inversion techniques for reconstructing the Antarctic temperature evolution of the last centuries
Reconstructing paleo-climate information using subsurface temperature data from ice borehole thermometry has proven to be a promising approach in climate reconstruction. Using Bayesian inversion, we aim to reconstruct the temperature evolution of Antarctica over the last centuries. A forward heat transfer model is implemented that simulates borehole temperature profiles for time-dependent surface temperatures. We invert this forward model using Bayesian inference by modeling known uncertainties as priors and obtain the reconstructed surface temperatures with associated uncertainty information. Through synthetic examples, we highlight the extent to which signals can be retrieved from a 200-meter borehole along with its uncertainty information. We further apply this to measured data for surface temperature reconstruction from EPICA Dronning Maud Land and analyze the results.
- 18 March 2025
- Published by Copernicus GmbH
Open AccessShallow Ice Borehole Thermometry: Bayesian inversion techniques for reconstructing the Antarctic temperature evolution of the last centuries
Reconstructing paleo-climate information using subsurface temperature data from ice borehole thermometry has proven to be a promising approach in climate reconstruction. Using Bayesian inversion, we aim to reconstruct the temperature evolution of Antarctica over the last centuries. A forward heat transfer model is implemented that simulates borehole temperature profiles for time-dependent surface temperatures. We invert this forward model using Bayesian inference by modeling known uncertainties as priors and obtain the reconstructed surface temperatures with associated uncertainty information. Through synthetic examples, we highlight the extent to which signals can be retrieved from a 200-meter borehole along with its uncertainty information. We further apply this to measured data for surface temperature reconstruction from EPICA Dronning Maud Land and analyze the results.
- Open Access
- 16 March 2025
- Published by Copernicus GmbH
- Vol. 2025, 1-25
Comment on egusphere-2024-3755
Borehole thermometry is an important tool for reconstructing past climate conditions, assessing changes in land energy storage, and understanding subsurface thermal regimes such as permafrost and glacial dynamics. Optimizing the temperature sensor placement within boreholes allows us to maximize the informativeness of temperature measurements, particularly in polar regions where operational constraints necessitate cost-effective solutions. Traditional sensor placement methods such as linear or exponential spacing, often overlook site-specific subsurface heat distribution characteristics, potentially limiting the accuracy of the measured temperature profile. In this paper, we propose a greedy optimal sampling technique for strategically placing temperature sensors in ice boreholes. Utilizing heat transfer model simulations, this method selects sensor locations that minimize interpolation errors in reconstructed temperature profiles. We apply our approach to two distinct borehole sites: EPICA Dronning Maud Land site in East Antarctica and the Greenland Ice Core Project site, each with unique surface conditions. Our results demonstrate that the greedy optimal sensor placement significantly outperforms conventional linear and exponential spacing methods, reducing sampling errors by up to a factor of ten and thus achieving similar informativeness with fewer sensors. This strategy offers a cost-effective means to maximize the information obtained from borehole temperature measurements, thereby potentially enhancing the precision of climate reconstructions.
- 16 March 2025
- Published by Copernicus GmbH
- Vol. 2025, 1-25
Open AccessComment on egusphere-2024-3755
Borehole thermometry is an important tool for reconstructing past climate conditions, assessing changes in land energy storage, and understanding subsurface thermal regimes such as permafrost and glacial dynamics. Optimizing the temperature sensor placement within boreholes allows us to maximize the informativeness of temperature measurements, particularly in polar regions where operational constraints necessitate cost-effective solutions. Traditional sensor placement methods such as linear or exponential spacing, often overlook site-specific subsurface heat distribution characteristics, potentially limiting the accuracy of the measured temperature profile. In this paper, we propose a greedy optimal sampling technique for strategically placing temperature sensors in ice boreholes. Utilizing heat transfer model simulations, this method selects sensor locations that minimize interpolation errors in reconstructed temperature profiles. We apply our approach to two distinct borehole sites: EPICA Dronning Maud Land site in East Antarctica and the Greenland Ice Core Project site, each with unique surface conditions. Our results demonstrate that the greedy optimal sensor placement significantly outperforms conventional linear and exponential spacing methods, reducing sampling errors by up to a factor of ten and thus achieving similar informativeness with fewer sensors. This strategy offers a cost-effective means to maximize the information obtained from borehole temperature measurements, thereby potentially enhancing the precision of climate reconstructions.