-
Initial validation of a soil-based mass-balance approach for empirical monitoring of enhanced rock weathering rates
Authors:
Tom Reershemius,
Mike E. Kelland,
Jacob S. Jordan,
Isabelle R. Davis,
Rocco D'Ascanio,
Boriana Kalderon-Asael,
Dan Asael,
T. Jesper Suhrhoff,
Dimitar Z. Epihov,
David J. Beerling,
Christopher T. Reinhard,
Noah J. Planavsky
Abstract:
Enhanced Rock Weathering (ERW) is a promising scalable and cost-effective Carbon Dioxide Removal (CDR) strategy with significant environmental and agronomic co-benefits. A major barrier to large-scale implementation of ERW is a robust Monitoring, Reporting, and Verification (MRV) framework. To successfully quantify the amount of carbon dioxide removed by ERW, MRV must be accurate, precise, and cos…
▽ More
Enhanced Rock Weathering (ERW) is a promising scalable and cost-effective Carbon Dioxide Removal (CDR) strategy with significant environmental and agronomic co-benefits. A major barrier to large-scale implementation of ERW is a robust Monitoring, Reporting, and Verification (MRV) framework. To successfully quantify the amount of carbon dioxide removed by ERW, MRV must be accurate, precise, and cost-effective. Here, we outline a mass-balance-based method where analysis of the chemical composition of soil samples is used to track in-situ silicate rock weathering. We show that signal-to-noise issues of in-situ soil analysis can be mitigated by using isotope-dilution mass spectrometry to reduce analytical error. We implement a proof-of-concept experiment demonstrating the method in controlled mesocosms. In our experiment, basalt rock feedstock is added to soil columns containing the cereal crop Sorghum bicolor at a rate equivalent to 50 t ha$^{-1}$. Using our approach, we calculate rock weathering corresponding to an average initial CDR value of 1.44 +/- 0.27 tCO$_2$eq ha$^{-1}$ from our experiments after 235 days, within error of an independent estimate calculated using conventional elemental budgeting of reaction products. Our method provides a robust time-integrated estimate of initial CDR, to feed into models that track and validate large-scale carbon removal through ERW.
△ Less
Submitted 22 October, 2023; v1 submitted 9 February, 2023;
originally announced February 2023.
-
An investigation of factors affecting high-precision Sr isotope analyses (87Sr/86Sr and δ88/86Sr) by MC-ICP-MS
Authors:
Jiuyuan Wang,
Dan Asael,
Noah J. Planavsky,
Lidya G. Tarhan
Abstract:
The abundance and fractionation of the stable strontium (Sr) isotope system are being increasingly utilized to move forward our understanding in geological and cosmological processes. Two analytical techniques are commonly used to measure stable Sr isotopes: 1) double-spike thermal ionization mass spectrometry (DS-TIMS) and 2) Zr-doped sample-standard bracketing multi-collector inductively coupled…
▽ More
The abundance and fractionation of the stable strontium (Sr) isotope system are being increasingly utilized to move forward our understanding in geological and cosmological processes. Two analytical techniques are commonly used to measure stable Sr isotopes: 1) double-spike thermal ionization mass spectrometry (DS-TIMS) and 2) Zr-doped sample-standard bracketing multi-collector inductively coupled plasma mass spectrometry (Zr-doped SSB via MC-ICP-MS). Relative to DS-TIMS, Zr-doped SSB via MC-ICP-MS allows simultaneous determinations of both 87Sr/86Sr and 88Sr/86Sr ratios, increasing measurement efficiency and sample throughput. However, this technique is currently associated with greater uncertainties in measurement precision and accuracy. In this study, we evaluated potential factors that can affect the quality of Sr isotope measurements during Zr-doped SSB. Our tests show that incomplete Sr recovery during chromatographic separation, mismatches of Sr and Zr concentrations and acid molarity between sample and bracketing standard, and cation contamination could all affect the precision and accuracy of Sr isotope measurements. We present evidence that, with updated preparation procedures and diligent concentration checks, a long-term reproducibility (2σSD: 87Sr/86Sr = +/-0.000015 and δ88/86Sr = +/-0.03 permil) comparable to that of DS-TIMS is achievable when using the Zr-doped SSB method via MC-ICP-MC.
△ Less
Submitted 4 November, 2021;
originally announced November 2021.
-
A Generative Approach for Mitigating Structural Biases in Natural Language Inference
Authors:
Dimion Asael,
Zachary Ziegler,
Yonatan Belinkov
Abstract:
Many natural language inference (NLI) datasets contain biases that allow models to perform well by only using a biased subset of the input, without considering the remainder features. For instance, models are able to make a classification decision by only using the hypothesis, without learning the true relationship between it and the premise. These structural biases lead discriminative models to l…
▽ More
Many natural language inference (NLI) datasets contain biases that allow models to perform well by only using a biased subset of the input, without considering the remainder features. For instance, models are able to make a classification decision by only using the hypothesis, without learning the true relationship between it and the premise. These structural biases lead discriminative models to learn unintended superficial features and to generalize poorly out of the training distribution. In this work, we reformulate the NLI task as a generative task, where a model is conditioned on the biased subset of the input and the label and generates the remaining subset of the input. We show that by imposing a uniform prior, we obtain a provably unbiased model. Through synthetic experiments, we find that this approach is highly robust to large amounts of bias. We then demonstrate empirically on two types of natural bias that this approach leads to fully unbiased models in practice. However, we find that generative models are difficult to train and they generally perform worse than discriminative baselines. We highlight the difficulty of the generative modeling task in the context of NLI as a cause for this worse performance. Finally, by fine-tuning the generative model with a discriminative objective, we reduce the performance gap between the generative model and the discriminative baseline, while allowing for a small amount of bias.
△ Less
Submitted 31 August, 2021;
originally announced August 2021.
