A retrospective research series

My Papers,
Re-reviewed.

Published work should survive more than peer review. In this series, I return to my own papers, keep the contribution that still holds, and name the questions I did not yet know how to ask.

Ying-Fan Lin, published as Ye-Chen Lin, et al. (2016) · Journal of Hydrology

A general solution to a very specific aquifer.

The paper derived an analytical pumping-test model for a radial, finite, two-zone confined aquifer with a Robin-type outer boundary. Ten years later, the derivation is not the part I would challenge.

The visual annotations are retrospective commentary. The paper image and technical statements refer to the published article.

What still holds

The analytical contribution was real.

The work produced a general solution within its stated conceptual model and examined how the Robin-type outer boundary, the finite radial extent, and the two-zone structure affect drawdown and wellbore response. That is the paper's legitimate contribution.

What I would change

Solvability is not field readiness.

The concentric outer boundary is difficult to defend for most field settings. The paper did not fit field data, test parameter identifiability, or turn its root-search procedure into a reproducible inversion workflow. Its one-at-a-time sensitivity analysis also left parameter interactions largely unexamined.

“The mathematics was real. So was the later lesson.”

Reviewer #3: Future Me

Why publish the criticism?

Because a research trajectory should show correction, not just accumulation.

This is not a retraction and it does not claim that the derivation is wrong. It is a record of how my standard changed: from asking whether a groundwater problem admits an analytical solution to asking whether the model can be identified, reproduced, compared with alternatives, and used without hiding consequential uncertainty.

Read the published paper Open this paper in the publication record

Ying-Fan Lin, published as Ye-Chen Lin, et al. (2017) · Advances in Water Resources

The forward problem had a solution. Did the inverse problem have one answer?

The paper coupled saturated and unsaturated flow in a leaky unconfined aquifer under constant-head pumping. It also marked a practical change in how I worked: a reviewer request led me onto the campus bus, into FEMWATER, and toward a broader standard of evidence.

Original paper pages and Figure 1 appear in the film. The pumping-test photograph is a public-domain USGS image used illustratively; it does not depict the Iowa field site.

What still holds

The model crossed several difficult boundaries.

The study joined Richards' equation with saturated groundwater flow, treated the screened and unscreened parts of a partially penetrating well as a mixed boundary, and obtained the solution through Laplace and Weber transforms. It then compared the result with FEMWATER and fitted records from four observation wells.

What I would change

A fitted table is not an identifiability analysis.

Five parameters were estimated with Mathematica FindFit, but the paper did not report confidence intervals, parameter correlations, or a joint identifiability assessment. A new analysis should combine head and wellbore-flow information, compare defensible conceptual models, and propagate uncertainty beyond the fitted curves.

“The forward problem had a solution. The inverse problem may not have had one answer.”

Reviewer #3: Future Me

What the paper changed

An aquifer-test model should eventually answer an aquifer-test question.

The derivation mattered, but the field purpose matters more: what can the observations actually teach us about the aquifer? That question now makes parameter estimation, uncertainty, and model comparison part of the theory rather than optional work after the analytical solution is complete.

Read the published paper Open this paper in the publication record

Ying-Fan Lin, published as Ye-Chen Lin, et al. (2019) · Water Resources Research

The paper I put in a freezer.

My first master's topic spent years in cold storage. It returned as a corrected aquifer thermal energy storage model with a Robin-type injection-well boundary, global sensitivity analysis, and a field application in Mobile, Alabama.

Original paper pages and reconstructed technical diagrams appear in the film. Paper evidence, author recollection, and retrospective questions are labeled separately.

What still holds

The injection-period model earned its place.

The paper coupled conduction and dispersion, compared the Robin-type injection-well boundary with alternative analytical solutions, screened parameter effects with the Morris method, and reproduced the reported injection-period temperature response reasonably well.

What I would change

An injection solution is not an operating cycle.

A complete injection-storage-recovery analysis cannot be obtained by casually superposing states governed by different conditions. Today I would carry the terminal state forward as the next initial condition and design the operating cycle from the outset.

“Accept the injection paper. Write the sequel.”

Reviewer #3: Future Me

What the paper changed

A boundary condition can be correct without completing the engineering problem.

The paper repaired the injection model and clarified when its boundary representation matters. The later lesson is operational: a useful thermal-storage analysis must preserve the system state as the forcing changes, rather than treating each phase as an interchangeable copy of the first.

Read the published paper Open this paper in the publication record