Pith. sign in

REVIEW 4 major objections 3 minor 36 references

This paper proposes an observation-specific bias correction for back-transforming log-rainfall in enhancement trials; the corrected estimator is unbiased, forces zero attribution for controls, and puts the Oman ionizer effect at 6.68% (95%

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-01 06:07 UTC pith:GUQNVBKI

load-bearing objection A clean and useful bias correction for back-transformed LMM predictions, but the applied 'significant positive effect' claim rests on an unexamined positive-rainfall selection assumption. the 4 major comments →

arxiv 2607.21991 v1 pith:GUQNVBKI submitted 2026-07-24 stat.ME stat.AP

Bias-Adjusted Attribution Estimation for Rainfall Enhancement Trials

Zhi Yang Tho , Raymond Chambers , A.H. Welsh This is my paper
classification stat.ME stat.AP MSC 62F4062J0562P12
keywords weather modificationcloud ionizationattribution estimationtransformation biaslog-normal datalinear mixed modelbootstrap inferencerainfall enhancement
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper tackles a statistical artefact that appears whenever rainfall enhancement is evaluated on the log scale and then transformed back. Because exponentiating a predicted log value is not unbiased in expectation, the usual estimate of 'natural rainfall' (what would have fallen without ionizers) is systematically wrong, and so is the amount of rain attributed to the intervention. The paper derives a per-observation correction factor from the estimated covariance of the treatment coefficients and shows that the corrected estimator is unbiased, makes control observations receive exactly zero attributed rain, and yields reliable bootstrap confidence intervals. Applied to the Oman 2013–2018 ionization trial, the corrected estimate is a 6.68% increase in downwind rainfall, with a 95% bootstrap interval from 2.21% to 12.88%; simulations indicate the earlier smearing-type estimator overstates the effect (12.52%) and its intervals badly undercover.

Core claim

Under the linear mixed model log(y_it) = x'_it α + z'_it β + u_t + e_it, the paper defines latent natural rainfall as R_it = y_it exp(−z'_it β), so that attributed rain is A_it = y_it − R_it. The problem is that replacing β by its estimate produces E[exp(−z'_it β̂)] = exp(−z'_it β) exp(½ z'_it Σ_β z_it), which is biased. The paper's central discovery is that multiplying the naive counterfactual by the observation-specific term λ̂_it = exp(−½ z'_it Σ̂_β z_it) removes this bias in expectation, giving R̂_it = y_it λ̂_it exp(−z'_it β̂) and hence Â_it = y_it − R̂_it. This estimator is coherent: when z_it = 0 the correction factor is 1, so control observations have zero attributed rain. For the O

What carries the argument

The load-bearing object is the observation-specific bias-adjustment term λ̂_it = exp(−½ z'_it Σ̂_β z_it), where z_it is the gauge-day exposure vector and Σ̂_β is the estimated covariance matrix of the fixed-effect treatment coefficients β̂. It arises from the exact expectation E[exp(−z' β̂)] = exp(−z' β) exp(½ z' Σ_β z), so multiplying the naive counterfactual y_it exp(−z'_it β̂) by this factor corrects the retransformation bias. The overall target is θ = (Σ A_it)/(Σ R_it) × 100%, estimated by the ratio of sums of Â_it and R̂_it. The paper couples this correction with a proportional random effect block bootstrap that resamples day-level random effects and unit-level residuals, with probabil

Load-bearing premise

The estimate is valid only if, among the gauge-days with rain, the log-linear model correctly captures what natural rainfall would have been without ionization, and if whether any rain falls is unaffected by the intervention — conditioning on positive rainfall, which the paper does, makes this second condition silently load-bearing.

What would settle it

Refit the same model on the full gauge-day record with zero-rainfall days included through a hurdle or two-part model; if the estimated percentage changes materially, loses significance, or changes sign, then the paper's decision to condition on positive rainfall is the decisive flaw rather than a harmless simplification.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Reported effect sizes from log-linear retransformation analyses of rainfall enhancement trials are likely overstated; the corrected estimator lowers the Oman estimate from 12.52% to 6.68%.
  • Control observations receive zero attributed rainfall by construction, so future trial estimates align directly with the randomized cross-over design.
  • Bootstrap confidence intervals around the corrected estimate attain near-nominal coverage (96% in simulations), while the existing estimator's intervals cover only 22.6% of true values.
  • The estimator supplies the uncertainty quantification demanded by international weather-modification reporting guidance.
  • The correction transfers directly to other linear mixed models with exponentiated predictions, including spatial covariance structures.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Because the analysis keeps only gauge-days with positive rainfall, the 6.68% is conditional on rain having occurred; if ionization changes the probability of any rainfall, part of the treatment effect is invisible to this estimator and selection bias may enter.
  • The correction shrinks attribution toward zero for low-exposure observations, which is the likely mechanism behind the smaller magnitude relative to the older constant adjustment; this shrinkage pattern is testable on other trials.
  • The same retransformation identity applies any time a log-scale model is exponentiated — cost, health expenditure, or count outcomes — so the bias adjustment could alter effect sizes in many applied fields, not just weather modification.
  • A direct stress test would be to simulate a data-generating process with zero-inflation or an intervention effect on rainfall occurrence; if the estimator becomes biased there, a hurdle/two-part extension is needed.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 3 minor

Summary. The paper proposes a new estimator for the raw-scale rainfall amount attributable to a rainfall-enhancement intervention when log-rainfall is modelled by a linear mixed model. The estimator multiplies the usual back-transformed counterfactual by an observation-specific term exp(-0.5 z' \hat\Sigma_\beta z), derived from the asymptotic covariance matrix of the fixed-effect estimator, and is shown to satisfy a coherence property: observations with no intervention have exactly zero estimated attribution. The authors apply the method to the Oman 2013-2018 ionization trial, obtaining an overall attribution estimate of 6.68% with a bootstrap 95% confidence interval of 2.21% to 12.88%, and compare it with the Chambers et al. (2022) estimator. A simulation study reports lower bias and MSE and better bootstrap coverage for the proposed estimator.

Significance. The paper addresses a real and underappreciated problem: when estimating a multiplicative back-transformed quantity in a log-scale mixed model, the uncertainty in the fixed-effect coefficients induces a transformation bias that standard smearing adjustments handle only arbitrarily. The proposed observation-specific correction is a principled idea, and the exact zero-attribution property for unexposed observations is a clear improvement. The PREB bootstrap is a useful extension to unbalanced day-level clusters. If the model and the positive-rainfall selection assumption hold, the estimator and its bootstrap inference perform well in the reported simulations. However, the theoretical justification is incomplete, and the applied conclusion rests on an untested assumption about the effect of ionization on rainfall occurrence.

major comments (4)
  1. [Section 4, Eq. (4)] The derivation of E[exp(-z_it' \hat\beta)] is correct as a statement about the lognormal moment of the estimator, but the proposed estimator is \hat R_it = y_it \hat\lambda_it exp(-z_it' \hat\beta). Since \hat\beta is estimated from the same data that determine y_it, y_it and \hat\beta are not independent; the expectation of the product is not the product of the expectations. The displayed calculation therefore does not imply E[\hat R_it] = R_it or E[\hat A_it] = A_it. If the intended argument is conditional on the observed y_it or relies on a higher-order asymptotic expansion, it must be stated explicitly. As written, the paper's central unbiasedness claim is not established.
  2. [Section 2 and Section 5] The Oman analysis uses only the 4168 downwind gauge-day observations with positive rainfall; all zero-rainfall observations are excluded, with the justification that enhancement methods are designed to enhance rather than create rainfall. This is an untested structural assumption. If ionization can change the probability of any rainfall, then the condition y_it>0 is a post-treatment selection event, and the estimated target-versus-control contrast is biased for the effect on total rainfall (including occurrence). The randomized on/off schedule does not remove this bias because the filtering occurs after treatment assignment. The authors should either test this assumption, model occurrence explicitly (e.g., a two-part model), or provide a sensitivity analysis. As it stands, the statistically significant applied conclusion is conditional on this assumption.
  3. [Section 6] The simulation is not independent validation of the applied conclusion. It generates data from model (1) using the same covariates x_it, z_it and the same parameter estimates from Table 2, then defines the true A_it and theta using equations (2)-(3). Within this data-generating process, the proposed estimator - which was derived from model (1) - naturally outperforms the alternative. The simulation is useful for comparing the two bias-correction formulas under the assumed model, but it cannot speak to the validity of the positive-rainfall exclusion or to model misspecification. The statement in Section 7 that the simulation provides 'empirical support for the conclusions drawn from the proposed method' overstates the evidential value.
  4. [Section 4, after Eq. (4)] The plug-in of \hat\Sigma_\beta into \hat\lambda_it is not addressed. The unbiasedness factor exp(-0.5 z_it' \Sigma_\beta z_it) is derived under the assumption that \Sigma_\beta is known; replacing it with a function of \hat\sigma^2_u and \hat\sigma^2_e changes the expectation in a way that is not quantified. A first-order justification for the plug-in version, or a simulation-based check, should be given. The bootstrap accounts for estimation uncertainty in interval construction but not for a systematic bias in the point estimator.
minor comments (3)
  1. [Throughout] There are several typos: 'underlying' for 'underlying' and 'coaslescence' for 'coalescence' in Section 2; 'occured' in Section 1; 'constrast' in Section 4; 'reflated' appears twice in Algorithm 1 (likely 'inflated' or 'reflated' should be defined). Please proofread.
  2. [Section 5] In the description of the H1 target indicator, 'on day i' should presumably be 'on day t'. Also, Table 2 would benefit from standard errors or confidence intervals for the fixed-effect estimates, especially because the bias correction depends on their covariance.
  3. [Figure 4] The caption states yellow and purple, but the text referring to the figure should be checked for colour accessibility; consider also reporting the range and quantiles of \hat\lambda_it in the text.

Circularity Check

0 steps flagged

No significant circularity: the bias-adjustment derivation is self-contained; simulation is a standard model-based self-consistency check.

full rationale

The paper's central methodological claim is the construction of lambda_hat_it = exp(-0.5 z_it' Sigma_hat_beta z_it) to correct back-transformation bias. Section 4 derives this directly from the asymptotic normality of beta_hat: E exp(-z_it' beta_hat) = exp(-z_it' beta) exp(0.5 z_it' Sigma_beta z_it), so the proposed estimator targets the model-defined counterfactual R_it = y_it exp(-z_it' beta). This is a self-contained mathematical derivation, not a fitted parameter relabeled as a prediction. The coherence property (zero attribution for z=0) is an algebraic consequence of the definition, not a hidden input. The applied conclusion inherits the usual assumption that the posited counterfactual R_it is valid; that is a correctness/identification concern, not circularity. The simulation sets true parameter values equal to the Table 2 estimates and generates data from model (1), with 'true' A_it computed from equations (2)-(3). This is a standard Monte Carlo check of the estimator against the model's own estimand; it does not provide external validation of the causal model, but it is not a circular step because the estimator does not use the true values and the simulation openly states the generating mechanism. Self-citations to Chambers et al. (2012/2016/2022) and Tho et al. (2025) supply the model, covariate selection, and bootstrap procedure; these are prior methods imported as building blocks rather than a self-referential uniqueness argument, and the new estimator's derivation does not reduce to them. Therefore no circular step meeting the quoting-and-reduction standard is present.

Axiom & Free-Parameter Ledger

1 free parameters · 6 axioms · 1 invented entities

The proposed estimator has essentially no new arbitrary free parameters, which is a strength. However, it is built on a strong parametric model, a model-based counterfactual, and a positive-rainfall selection rule. The simulation uses the same fitted parameters as truth, so the ledger's key burden is the domain assumptions, not the tuning constants.

free parameters (1)
  • REML model parameters (alpha, beta, sigma2_u, sigma2_e) = Intercept 0.280, Elevation -0.125, Expected log-rainfall 0.856; beta values in Table 2; sigma2_u = 0.274, sigma2_e = 1.8
    Estimated from the Oman data and used both in the applied analysis and as the true parameter values for the simulation. The simulation truth therefore inherits the fitted model rather than testing it.
axioms (6)
  • domain assumption Gaussian linear mixed model on log rainfall: log(y) = X*alpha + Z*beta + u + e with i.i.d. day effects and errors
    Equation (1), Section 3. The entire attribution framework and the bias correction rely on this model being correctly specified for positive rainfall.
  • domain assumption Randomization / exogeneity of treatment exposure z with respect to u and e
    Section 5 uses target indicators based on randomized on/off status and dynamic downwind corridors. Unmeasured confounding or exposure misclassification would bias beta and hence attribution.
  • domain assumption Multiplicative enhancement decomposition: R_it = y_it * exp(-z_it' beta)
    Equation (3), Section 3. This defines natural rainfall as observed rainfall with the fitted treatment effect removed; it is a model-based counterfactual, not an independent measurement.
  • standard math Asymptotic normality of the REML fixed-effect estimator (Laird and Ware 1982)
    Equation (4), Section 4. Needed for E[exp(-z' beta_hat)] = exp(-z' beta + z' Sigma_beta z / 2).
  • standard math Approximate independence of beta_hat from random effects and errors for unconditional unbiasedness of R_hat
    Section 4 uses E[y * lambda_hat * exp(-z' beta_hat)] = R; this assumes the fixed-effect estimator does not depend on the realized u and e beyond the asymptotic approximation.
  • domain assumption Enhancement does not create rain, so restricting to positive-rainfall gauge-days is harmless
    Section 2: 'the analysis focuses on downwind gauge-day observations with positive rainfall amounts.' This selection is not modeled and is a potential source of collider bias.
invented entities (1)
  • Latent natural rainfall R_it no independent evidence
    purpose: Counterfactual rainfall in the absence of enhancement; defines attribution A_it = y_it - R_it and the overall attribution theta
    R_it is unobservable and defined by the fitted model as y_it * exp(-z_it' beta). The entire estimation target depends on this latent construct, which is not validated by any external measurement.

pith-pipeline@v1.3.0-alltime-deepseek · 3965 in / 4147 out tokens · 158420 ms · 2026-08-01T06:07:13.045963+00:00 · methodology

0 comments
read the original abstract

Model-based analyses of rainfall enhancement trial data typically involve modelling log-transformed rainfall using linear mixed models to assess the effectiveness of enhancement methods under real-world conditions. This approach improves on traditional average-based analyses by allowing explicit control for the effects of meteorological and topographical covariates that may affect precipitation amounts. However, a key issue with such analyses is the bias that arises when back-transforming the log-rainfall to the original scale for estimating attribution, defined as the additional raw-scale rainfall attributable to the enhancement method. To address this issue, we propose a new attribution estimator that incorporates theoretically justified, observation-specific bias-adjustment terms. The proposed estimator improves upon existing estimators that rely on arbitrary adjustments, and satisfies a coherence property that ensures zero estimated attribution for observations without enhancement intervention. A proportional random effect block bootstrap is further used to conduct inference on the attribution quantities. Applying both the proposed estimator and an existing estimator to the Oman rainfall enhancement trial from 2013 to 2018, we find statistically significant positive effect of the ground-based ionization technology on downwind rainfall at the 5% significance level, with our proposed estimator indicating a smaller effect than the existing method. A simulation study further support the findings based on the proposed estimator, demonstrating its superior estimation accuracy and improved inferential performance of the associated bootstrap confidence intervals.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

36 extracted references

  1. [1]

    Manton and Loredana Warren and Suzanne L

    Michael J. Manton and Loredana Warren and Suzanne L. Kenyon and Andrew D. Peace and Shane P. Bilish and Karen Kemsley , title =. Journal of Applied Meteorology and Climatology , year =

  2. [2]

    Manton and Loredana Warren , journal =

    Michael J. Manton and Loredana Warren , journal =

  3. [3]

    Manton and Andrew D

    Michael J. Manton and Andrew D. Peace and Karen Kemsley and Suzanne Kenyon and Johanna C. Speirs and Loredana Warren and John Denholm , keywords =. Atmospheric Research , volume =. 2017 , issn =

    2017

  4. [4]

    Breed, Daniel and Rasmussen, Roy and Weeks, Courtney and Boe, Bruce and Deshler, Terry , year =

  5. [5]

    Rasmussen, Roy and Tessendorf, Sarah and Xue, Lulin and Weeks, Courtney and Ikeda, Kyoko and Landolt, Scott and Breed, Dan and Deshler, Terry and Lawrence, Barry , year =

  6. [6]

    French and Katja Friedrich and Sarah A

    Jeffrey R. French and Katja Friedrich and Sarah A. Tessendorf and Robert M. Rauber and Bart Geerts and Roy M. Rasmussen and Lulin Xue and Melvin L. Kunkel and Derek R. Blestrud , title =. Proceedings of the National Academy of Sciences , volume =

  7. [7]

    Tessendorf, Sarah and French, Jeffrey and Friedrich, Katja and Geerts, Bart and Rauber, Robert and Rasmussen, Roy and Xue, Lulin and Ikeda, Kyoko and Blestrud, Derek and Kunkel, Melvin and Parkinson, Shaun and Snider, Jefferson and Aikins, Joshua and Faber, Spencer and Majewski, Adam and Grasmick, Coltin and Bergmaier, Philip and Janiszeski, Andrew and Sp...

  8. [8]

    DeFelice and D

    T.P. DeFelice and D. Axisa and John J. Bird and C. Alexander Hirst and Eric W. Frew and R.P. Burger and D. Baumgardner and Gerhard Botha and Henno Havenga and Dan Breed and S. Bornstein and C. Choate and Ceu Gomez-Faulk and Michael Rhodes , keywords =. Atmospheric Research , volume =. 2023 , issn =

    2023

  9. [9]

    Miller, Anna and Ramelli, Fabiola and Fuchs, Christopher and Omanovic, Nadja and Spirig, Robert and Zhang, Huiying and Lohmann, Ulrike and Kanji, Zamin and Henneberger, Jan , year =

  10. [10]

    Advances in Meteorology , volume =

    Jung, Woonseon and Cha, Joo Wan and Ko, A.-Reum and Chae, Sanghee and Ro, Yonghun and Hwang, Hyun Jun and Kim, Bu-Yo and Ku, Jung Mo and Chang, Ki-Ho and Lee, Chulkyu , title =. Advances in Meteorology , volume =

  11. [11]

    Harrison, R. and Nicoll, Keri and Tilley, Douglas and Marlton, Graeme and Stefan, Chindea and Dingley, Gavin and Iravani, Pejman and Cleaver, David and Du Bois, Jonathan and Brus, David , year =. Demonstration of a Remotely Piloted Atmospheric Measurement and Charge Release Platform for Geoengineering , volume =

  12. [12]

    Charged Particle (Negative Ion)-Based Cloud Seeding and Rain Enhancement Trial Design and Implementation , volume =

    Zheng, Wei and Fengming, Xue and Zhang, Ming and Wu, Qiqi and Yang, Zhou and Ma, Shaoxiang and Liang, Haotian and Wang, Chuliang and Wang, Yuxing and Ai, Xinkun and Yang, Yong and Yu, Kexun , year =. Charged Particle (Negative Ion)-Based Cloud Seeding and Rain Enhancement Trial Design and Implementation , volume =

  13. [13]

    The Journal of Weather Modification , volume=

    Statistical modeling of rainfall enhancement , author=. The Journal of Weather Modification , volume=. 2010 , publisher=

    2010

  14. [14]

    The Journal of Weather Modification , volume=

    Accounting for spatiotemporal variation of rainfall measurements when evaluating ground-based methods of weather modification , author=. The Journal of Weather Modification , volume=. 2011 , publisher=

    2011

  15. [15]

    Fajardo and G

    C. Fajardo and G. Costa and L.T. Ortiz and M. Nande and M.L. Rodríguez-Membibre and M. Martín and S. Sánchez-Fortún , keywords =. Ecotoxicology and Environmental Safety , volume =. 2016 , issn =

    2016

  16. [16]

    D. A. Klein and E. M. Molise , journal =. Ecological Ramifications of Silver Iodide Nucleating Agent Accumulation in a Semi-Arid Grassland Environment , urldate =

  17. [17]

    and Jolly, William C

    Cooper, Charles F. and Jolly, William C. , title =. Water Resources Research , volume =

  18. [18]

    Vonnegut, Bernard and Moore, C. B. , title =. Recent Advances in Atmospheric Electricity , editor =

  19. [19]

    and Cheng, H

    Ma, S. and Cheng, H. and Li, J. and Xu, M. and Liu, D. and Ostrikov, K. , TITLE =. Atmospheric Chemistry and Physics , VOLUME =. 2020 , PAGES =

    2020

  20. [20]

    Droplet charging by high voltage discharges and its influence on precipitation enhancement , journal =

    A.A. Droplet charging by high voltage discharges and its influence on precipitation enhancement , journal =. 2012 , issn =

    2012

  21. [21]

    1995 , issn =

    Quantitative descriptions of ice formation mechanisms of silver iodide-type aerosols , journal =. 1995 , issn =

    1995

  22. [22]

    Water , VOLUME =

    Zheng, Wei and Ma, Hengben and Zhang, Ming and Xue, Fengming and Yu, Kexun and Yang, Yong and Ma, Shaoxiang and Wang, Chuliang and Pan, Yuan and Shu, Zhiliang and Mu, Jianhua and Yang, Wenqing and Yin, Xianzhi , TITLE =. Water , VOLUME =. 2021 , pages =

    2021

  23. [23]

    Morrison, Anthony and Siems, Steven and Manton, Michael and Nazarov, Alex , year =

  24. [24]

    Atmospheric Research , volume =

    Zev Levin and Noam Halfon and Pinhas Alpert , keywords =. Atmospheric Research , volume =. 2010 , issn =

    2010

  25. [25]

    Journal of Weather Modification , volume=

    Using Dynamically Defined Controls to Evaluate the Impact of An Ionization Technology , author=. Journal of Weather Modification , volume=

  26. [26]

    Chambers, Ray and Beare, Stephen and Peak, Scott and Al-Kalbani, Mohammed , year =

  27. [27]

    International Statistical Review , volume =

    Chambers, Ray and Beare, Stephen and Peak, Scott and Al-Kalbani, Mohammed , title =. International Statistical Review , volume =

  28. [28]

    Journal of the American Statistical Association , volume =

    Naihua Duan , title =. Journal of the American Statistical Association , volume =. 1983 , publisher =

    1983

  29. [29]

    2025 , eprint=

    A Proportional Random Effect Block Bootstrap for General Clustered Data , author=. 2025 , eprint=

    2025

  30. [30]

    2000 , publisher=

    Mixed-Effects Models in S and S-PLUS , author=. 2000 , publisher=

    2000

  31. [31]

    Journal of Statistical Software , author=

    Fitting Linear Mixed-Effects Models Using lme4 , volume=. Journal of Statistical Software , author=. 2015 , pages=

    2015

  32. [32]

    Harville , journal =

    David A. Harville , journal =. Maximum Likelihood Approaches to Variance Component Estimation and to Related Problems , urldate =

  33. [33]

    H. D. Patterson and R. Thompson , journal =. Recovery of Inter-Block Information when Block Sizes are Unequal , urldate =

  34. [34]

    Laird and James H

    Nan M. Laird and James H. Ware , journal =. Random-Effects Models for Longitudinal Data , urldate =

  35. [35]

    Spatial variation : Stochastic models and their application to some problems in forest surveys and other sampling investigations , booktitle =

    Bertil Mat. Spatial variation : Stochastic models and their application to some problems in forest surveys and other sampling investigations , booktitle =. 1960 , volume =

    1960

  36. [36]

    1997 , publisher=

    Bootstrap Methods and Their Application , author=. 1997 , publisher=

    1997