Spectral frequency-increment measurement noise

Spectral frequency-increment measurement noise is the contribution of frequency-estimation errors to a statistic comparing a transition frequency at two times. It can create an apparent nonzero spectral-diffusion frequency-increment variance even when the underlying frequency has not changed.

For an explicit measurement model, write $Y_t={\omega }_t+{ϵ}_t$ and define $\Delta \omega ={\omega }_T-{\omega }_0$, $\Delta ϵ={ϵ}_T-{ϵ}_0$. With zero-mean endpoint errors,The last term vanishes only under an appropriate lack-of-correlation assumption between the physical increment and its estimation errors. Merely naming the errors noise does not establish that assumption.

If that cross term vanishes, equal endpoint standard uncertainties $u$ and error correlation coefficient ${\rho }_{ϵ}$ giveIndependent errors give $2u^2$. A perfectly shared offset cancels from the difference, although the absolute frequencies can remain uncertain. This correlated spectral endpoint error is an ordinary application of correlated-input uncertainty propagation.^[ref]

An original example has true root-mean-square increment $0.1\mathrm{rad}\mathrm{p}{\mathrm{s}}^{-1}$ and $u=0.2\mathrm{rad}\mathrm{p}{\mathrm{s}}^{-1}$. Independent endpoint errors give $D_{\mathrm{obs}}=0.01+0.08=0.09$ and apparent root-mean-square increment $0.3\mathrm{rad}\mathrm{p}{\mathrm{s}}^{-1}$. At ${\rho }_{ϵ}=0.75$, the added term is only $0.02$, giving an apparent increment about $0.173\mathrm{rad}\mathrm{p}{\mathrm{s}}^{-1}$.

A noise subtraction can also be uncertain. Observed second moment $0.03$ minus estimated noise contribution $0.08$ gives $-0.05$ in the same squared-frequency units. That negative estimate does not establish a negative physical variance; it flags uncertainty or a mismatch in the assumptions used to separate the contributions.

Repeated observations of a stable reference can test the estimator's noise and temporal correlation. Filtering, common references and reuse of the same data can correlate errors at nearby times. Primary two-dimensional infrared spectroscopy analysis connects spectral geometry to a frequency-correlation model rather than directly recording every molecule's frequency trajectory.^[ref] The additive-error equations apply only when the stated frequency estimator and error model describe the actual measurement.

Correlated spectral endpoint error means that the errors in two estimated frequencies have nonzero covariance. Their effect on a frequency difference depends on that covariance as well as on the uncertainty of each endpoint. Shared errors can cancel in a difference without making the absolute frequencies accurately known.

Let endpoint errors be ${ϵ}_0,{ϵ}_T$, with standard uncertainties $u_0,u_T$ and correlation coefficient $\rho$. The error in the increment is ${ϵ}_T-{ϵ}_0$, soFor equal uncertainties $u$, the result is $2u^2(1-\rho )$. Independent errors give $2u^2$, perfectly shared errors give zero, and perfect anticorrelation gives $4u^2$. This is correlated-input uncertainty propagation applied to a difference.^[ref]

An original common-offset model writes ${ϵ}_0=b+{\eta }_0$ and ${ϵ}_T=b+{\eta }_T$. Suppose $b$ has variance $s_b^2$ and the two independent residual errors each have variance $s^2$, all mutually independent. Each endpoint then has variance $s_b^2+s^2$ and their covariance is $s_b^2$, butWith $s_b=0.3$ and $s=0.1$ in compatible frequency units, each endpoint has standard uncertainty about $0.316$, while the increment error has standard uncertainty about $0.141$. This spectral common-mode offset rejection follows from subtraction, rather than from knowing the common offset's value.

A shared gain error behaves differently. If $Y_t=(1+g){\omega }_t+b$, the constant offset cancels but $Y_T-Y_0=(1+g)({\omega }_T-{\omega }_0)$. A shared $1\%$ scale error therefore remains a $1\%$ increment error. Calling an error common does not establish which part cancels.

To estimate covariance, repeated measurements should preserve the reference and data-processing relationships relevant to the two endpoints. Filtering or overlapping fitting windows can introduce correlations. The spectral frequency-increment measurement noise contribution also depends on covariance between physical fluctuations and estimation errors; the endpoint formula alone does not justify dropping that separate term. A two-dimensional infrared spectroscopy statistic should be interpreted through its actual estimator and response model, not by automatically adding independent noise to two abstract frequency labels.

Spectral common-mode offset rejection is cancellation of an additive frequency-axis shift when the same shift enters two spectral estimates with equal sensitivity and the estimates are subtracted. It improves a frequency difference without establishing accurate absolute frequencies. The relevant measurement model must identify what is shared, and over which observations.

For true frequencies ${\omega }_1,{\omega }_2$, write measured estimates asThe unknown common offset $b$ disappears exactly fromFor independent residual errors of standard uncertainty $s$, the difference still has standard uncertainty $\sqrt{2}s$. Cancellation of $b$ is not cancellation of all measurement noise. With correlated residuals, the covariance term must be retained in uncertainty propagation.^[ref]

An original example uses two spectral positions at $1600$ and $1650{\mathrm{cm}}^{-1}$. An additive shift of $3{\mathrm{cm}}^{-1}$ gives measured positions $1603$ and $1653{\mathrm{cm}}^{-1}$, with the same $50{\mathrm{cm}}^{-1}$ separation. A shared gain error is different: $Y_i=1.01{\omega }_i+3$ gives separation $50.5{\mathrm{cm}}^{-1}$. Subtraction rejects the offset but retains the scale error.

The algebra generalizes to a contrast of several spectral positions. If $Z={\sum }_i{w}_i{Y}_i$ and each $Y_i$ contains the same additive $b$, its contribution is $b{\sum }_i{w}_i$. Thus weights satisfying ${\sum }_i{w}_i=0$ reject that offset. Averaging positions with weights summing to one does not. This is an original linear-model derivation of the rejection condition, not a claim that every spectral estimator has this form.

Two failures of the ideal assumption are especially informative. With unequal spectral offset sensitivity, $Y_i={\omega }_i+a_{i}b+{\eta }_i$, subtraction leaves $(a_2-a_1)b$. If the offset drifts between observations, it instead leaves $b(t_2)-b(t_1)$. For a stationary drift with variance $s_b^2$ and correlation ${\rho }_b(\tau )$ at lag $\tau$, this residual has varianceA common instrument or reference does not alone imply perfect cancellation at different times.

An apparatus-level check measures two stable reference features using the same acquisition and fitting pipeline, then deliberately shifts the assigned axis origin. Their separation should stay fixed under an additive reassignment; a scale change should alter it. Repeated reference observations test temporal drift and residual covariance. In two-dimensional infrared spectroscopy, this test concerns frequency-estimation errors; spectral diffusion also describes physical fluctuations and cannot be identified or removed merely by subtracting an instrumental offset.

Unequal spectral offset sensitivity occurs when a shared instrumental or reference offset influences two spectral estimates by different amounts. It limits spectral common-mode offset rejection: subtraction cancels a common disturbance only to the extent that its two response coefficients match.

For frequencies ${\omega }_1,{\omega }_2$, write an explicit measurement modelwhere $b$ is the shared offset, $a_i$ are dimensionless sensitivities and ${\eta }_i$ are residual errors. ThenThus even perfectly shared $b$ leaves a residual when $a_1\ne a_2$.

An original uncertainty example uses $a_1=0.98$, $a_2=1.02$ and standard uncertainty $u_b=0.50{\mathrm{cm}}^{-1}$. The offset contribution to the difference is $0.04u_b=0.020{\mathrm{cm}}^{-1}$. With independent residual standard uncertainties of $0.010{\mathrm{cm}}^{-1}$ each and no correlation with $b$, the combined uncertainty isA general calculation retains the relevant covariances; the independent-error simplification is an assumption, not a consequence of a shared instrument.^[ref]

Reweighting also changes what is measured. Subtracting $Y_1/a_1$ from $Y_2/a_2$ cancels $b$ when both sensitivities are nonzero, but estimates ${\omega }_2/a_2-{\omega }_1/a_1$, not generally the original frequency difference. Achieving cancellation is insufficient unless the resulting estimator still answers the intended physical question.

A controlled reference perturbation can estimate the slopes $a_1$ and $a_2$ while the true spectral features are held stable. Their difference predicts leakage. Spectral offset identifiability asks whether the available observations can distinguish a real change of frequency separation from this leakage; one observed difference alone cannot determine both an unknown physical change and an unknown offset. Reference features, controlled perturbations or independently established response coefficients supply additional information.

Spectral offset identifiability asks whether the available spectral observations can distinguish a physical frequency change from an instrumental offset. Identifiability is a property of the measurement model and its observations. Reducing random noise cannot by itself separate quantities that always enter the model in the same combination.

For an original local model, let an observed change bewhere $\theta$ is the physical change, $b$ an unknown reference offset and $a$ a known sensitivity. If $a\ne 0$, every replacementgives the same $z$. The single observation therefore does not identify both unknowns. With $z=0.10{\mathrm{cm}}^{-1}$ and $a=0.04$, $(\theta ,b)=(0.10,0)$ and $(0.02,2.0)$ in compatible frequency units are observationally indistinguishable, even without measurement noise.

Reference-assisted spectral offset separation adds an observation $r=cb$ from a stable reference with known nonzero sensitivity $c$. The combined linear system isIts determinant is $c$, so the two ideal unknowns are identifiable when $c\ne 0$. The reference must genuinely exclude the physical change represented by $\theta$; otherwise the assumed second equation is wrong.

Identifiability also differs from precision. For fixed known $a/c$, correlated-input uncertainty propagation givesA very weak reference response can make the identified estimate noisy. Uncertainty in the sensitivities requires additional terms. The primary GUM supplies the measurement-model and covariance framework; the rank and invariance arguments above are explicit derivations for this stated spectral model.^[ref]

An experiment can test the separation by measuring stable reference features and the sample while deliberately changing the assigned axis offset. Repeated spectra characterize noise; reference perturbations characterize sensitivity. Neither exercise establishes spectral offset identifiability unless the observations distinguish the competing physical and instrumental changes.

Reference-assisted spectral offset separation uses an additional reference observation to distinguish a physical spectral change from an instrumental offset. The reference must respond to the offset in a known way while excluding, or separately modelling, the sample’s physical change. Merely collecting another signal does not guarantee spectral offset identifiability.

For a local linear measurement model, writewhere $\theta$ is the desired change, $b$ the shared offset and $c\ne 0$. With known sensitivities and zero-mean residual errors,removes the offset contribution. This is a subtraction of modelled responses, not a claim that the uncorrected sample and reference frequencies are physically identical.

An original example uses $z=0.10{\mathrm{cm}}^{-1}$, $r=2.0{\mathrm{cm}}^{-1}$ and $q=0.04$, giving $\hat{\theta }=0.02{\mathrm{cm}}^{-1}$. A reference may have a much larger offset response than the sample; the response ratio supplies the necessary conversion.

The uncertainty of a reference-corrected spectral change depends on joint errors. For fixed $q$,With $u_z=0.010{\mathrm{cm}}^{-1}$, $u_r=0.20{\mathrm{cm}}^{-1}$ and correlation $0.5$, this gives approximately $0.00917{\mathrm{cm}}^{-1}$; assuming independence instead gives $0.0128{\mathrm{cm}}^{-1}$. The primary GUM supplies the correlated-input propagation rule used in this calculation.^[ref]

If $q$ is itself estimated, its uncertainty also matters. Under an additional independent-input first-order approximation, the contribution is $r^2{u}_q^2$. Large uncertainties, correlations with the fitted ratio or nonlinear responses require a more complete analysis.

A controlled experiment records stable reference features through the same acquisition and processing chain while deliberately perturbing the instrumental offset. This can estimate $a/c$ and test whether the response is linear. Sample/reference timing must also match the assumed offset: a drifting offset at different acquisition times can leave an uncorrected residual. Repeated paired data, rather than unrelated error estimates, are needed to characterize the relevant covariance.

For reference-assisted spectral offset separation, the corrected estimate $\hat{\theta }=z-qr$ depends on the measured sample change $z$, reference change $r$ and sensitivity ratio $q$. Its measurement uncertainty must include their joint variation, including uncertainty in a fitted sensitivity ratio. Subtracting a reference can reduce shared noise while adding independent noise.

The gradient of the measurement model $z-qr$ with respect to $(z,r,q)$ is $g=(1,-q,-r)$. If $\Sigma$ is their covariance matrix, first-order uncertainty propagation gives $u_{\theta }^2\approx g\Sigma g^{\mathsf{T}}$, orThe signs follow from the sensitivity coefficients, rather than a rule that all correlation reduces uncertainty. This is an application of the correlated-input propagation law in the primary GUM.^[ref]

For an original independent-input example, take $z=0.12\mathrm{c}{\mathrm{m}}^{-1}$, $r=2.0\mathrm{c}{\mathrm{m}}^{-1}$, $q=0.04$, $u_z=0.010\mathrm{c}{\mathrm{m}}^{-1}$, $u_r=0.20\mathrm{c}{\mathrm{m}}^{-1}$ and $u_q=0.010$. Then $\hat{\theta }=0.040\mathrm{c}{\mathrm{m}}^{-1}$ andTreating $q$ as exact would give only $0.01281\mathrm{c}{\mathrm{m}}^{-1}$. A large reference response amplifies uncertainty in its conversion ratio, even when that response is measured precisely.

For mutually independent random inputs, the exact product variance also includes $u_q^2{u}_r^2$:where $q,r$ on the right denote the input means. It follows by expanding $\mathbb{E}(q^2{r}^2)=(q^2+u_q^2)(r^2+u_r^2)$. This term is omitted by linearization. In the example it changes the standard uncertainty to about $0.02383\mathrm{c}{\mathrm{m}}^{-1}$.

Experimentally, use paired sample/reference spectra to estimate shared covariance, and independent sensitivity measurements to test the proposed value of $q$. If the same spectra estimate both $q$ and the correction, independence cannot simply be assumed. Drift between acquisitions or a nonlinear reference response is a measurement model problem as well as a precision problem.