FWHM: A Thorough Guide to the Full Width at Half Maximum

The term FWHM, standing for Full Width at Half Maximum, is a cornerstone concept across optics, spectroscopy, astronomy, and image analysis. It provides a single, interpretable measure of how broad a peak or a point-spread function appears in an observed signal. In practical terms, the FWHM tells you how wide a feature is when its peak intensity is reduced to half of its maximum value. This simple idea unlocks insights into resolution, instrument performance, and the fundamental properties of the observed system. Here we explore FWHM in depth—from the mathematics behind it to the hands-on methods you can use in real-world data analysis, with clear guidance for researchers, engineers and students alike.

FWHM explained: What does FWHM measure and why it matters

At its core, FWHM is a width metric that is easy to interpret. For a peak in a 1D signal, such as a spectral line, the FWHM marks the distance between the two points on the curve where the signal equals half of the peak value. In two dimensions or three dimensions, the concept generalises to the width of a peak or blob at half of its maximum intensity, often applied to point-spread functions (PSFs) in imaging systems. The FWHM is intimately connected with resolution: narrower FWHM implies sharper features and higher resolving power, while broader FWHM indicates more blurring or instrument-imposed limitations.

Because FWHM is defined relative to a peak, it is robust to moderate background variation and is widely used when the exact shape of the distribution may deviate from a perfect Gaussian. Nevertheless, for idealised cases, such as a purely Gaussian peak, the FWHM links directly to the standard deviation, providing a bridge between two common characterisations of width.

Mathematical foundations of FWHM

FWHM for a Gaussian distribution

In many practical contexts the peak is well described by a Gaussian. If the signal follows a Gaussian profile with standard deviation σ, the Full Width at Half Maximum is given by:

FWHM = 2√(2 ln 2) · σ ≈ 2.355 · σ

This relationship is central when interpreting instrument response or PSF widths in microscopy and astronomy. It also provides a convenient way to convert between FWHM and σ when comparing results across studies or simulations.

Relation to sigma and units

The sigma parameter represents the spread of the distribution in the same units as the axis of the data. Consequently, FWHM inherits those same units. If you measure a PSF in arcseconds, the FWHM will be expressed in arcseconds; if in pixels, the FWHM will be in pixels. When reporting FWHM, it is important to include the unit and, where relevant, the sampling interval of the data to avoid misinterpretation.

Other shapes and general definitions

Not all peaks are Gaussian. For other shapes—Lorentzians, Voigt profiles, or asymmetrical features—the FWHM is still defined as the width between the two half-maximum points, but the numerical relationship to σ is no longer simple. In these cases, the FWHM is a practical descriptor of width, while other moments or curvature-based measures may provide complementary information about the peak shape. For non-Gaussian features, reporting the FWHM alongside a description of the peak shape is good practice.

FWHM in practice across disciplines

In imaging and microscopy

In imaging systems, the FWHM of the PSF characterises the system’s resolving power. A typical scenario is laboratory fluorescence microscopy where the measured PSF width represents how a point source (or a sub-resolution object) is spread by the optics and detector. A smaller FWHM indicates better localisation of tiny features, sharper edges, and higher detail. When you compare instruments or objective lenses, the FWHM provides a straightforward metric to rank optical quality. In practice, researchers estimate FWHM by capturing images of sub-resolution beads, extracting intensity profiles, and determining the half-maximum positions along principal axes.

In spectroscopy and spectral lines

For spectral data, the FWHM of a line profile conveys the spectral resolution and the kinetics of the emitting or absorbing species. Instrumental broadening can contribute to an observed line with an FWHM that exceeds the intrinsic width of the transition. Techniques such as fitting Gaussian or Voigt profiles to emission or absorption lines yield FWHM values that assist in identifying physical conditions, such as temperature, turbulence, and velocity dispersion. When reporting spectral FWHM, it is common to specify the fitting model and include uncertainties arising from noise and continuum placement.

In astronomy and telescope optics

Astronomical images are subject to atmospheric seeing, telescope optics, and detector effects. The net FWHM of stellar images encodes this combination of factors. Astronomers often quote the FWHM of the PSF at a given wavelength as a measure of image sharpness and as input to deconvolution or photometric measurements. Understanding the FWHM across the field of view helps account for spatial variation in resolution, which is especially important in crowded fields or when performing precise aperture photometry.

Measuring FWHM from data

Step-by-step procedures for 1D profiles

1D profiles are common in spectroscopy and line-scans. A practical approach is as follows:

• Identify the peak value and locate its maximum intensity.
• Determine the half-maximum level: half of the peak intensity.
• Find the two points on either side of the peak where the profile crosses the half-maximum level. If the data are noisy, apply a smoothing step or interpolate between data points.
• Compute the distance between these two points along the independent axis to obtain the FWHM.

Interpolation is key for accuracy. Linear interpolation can be sufficient, but cubic or spline interpolation often yields more precise half-maximum positions, particularly when sampling is coarse relative to the expected FWHM.

2D and 3D estimates: PSFs and imaging data

For a 2D PSF, several strategies exist:

• Extract 1D cross-sections along the major and minor axes through the peak and compute FWHM for each. Averaging the results provides a robust estimate of the PSF width in different directions.
• Fit a 2D Gaussian (or another suitable model) to the PSF, and derive the FWHM from the fitted parameters using FWHM = 2√(2 ln 2)σx for the x-axis and FWHM = 2√(2 ln 2)σy for the y-axis.
• In 3D data, extend the approach to the three principal axes or fit a 3D Gaussian. Report the FWHM in each dimension to capture anisotropy in the optical system.

When the peak is not isolated or the background is varying, carefully model the background and consider fitting a profile rather than relying on raw half-maximum crossing. This reduces bias in the FWHM estimate.

Handling noise and sampling

Noise broadens an observed peak and can bias the FWHM if not accounted for. Smoothing can mitigate high-frequency noise, but excessive smoothing may artificially widen the peak. A balanced approach is to use low-pass filtering or locally weighted regression (LOESS) prior to half-maximum estimation, followed by interpolation for sub-sample accuracy. Additionally, ensure your sampling rate satisfies the Nyquist criterion for the features of interest to avoid aliasing of the FWHM.

Interpolation methods to locate half-maximum

The accuracy of FWHM measurements improves with robust interpolation. Common methods include:

• Linear interpolation between adjacent data points around the half-maximum crossing.
• Cubic spline interpolation to locate a more precise crossing point in smoother data.
• Polynomial fits to a local window around the peak, then solving for the half-maximum crossing analytically.

Deconvolution and FWHM

Observed features are the convolution of the true object with the instrument’s PSF. In many cases you want to infer the intrinsic width of the object or separate the instrument width from the observed FWHM. The general principle is:

Observed FWHM^2 ≈ Intrinsic FWHM^2 + Instrumental FWHM^2

For Gaussian profiles, this relation holds approximately because widths add in quadrature. If either the intrinsic profile or the PSF deviates from a Gaussian, the relationship becomes more complex and deconvolution methods may be required. Deconvolution can recover finer structure but is sensitive to noise; regularisation and careful validation are essential.

Uncertainty and reporting FWHM

Reporting FWHM with an uncertainty helps readers assess the reliability of the measurement. Common approaches include:

• Estimating the standard error of the fitted FWHM from the covariance matrix in a least-squares fit.
• Using bootstrapping or Monte Carlo simulations to propagate noise and background variations into an FWHM estimate.
• Providing confidence intervals for the half-maximum crossing points obtained via interpolation.

When publishing FWHM values, include:

• The method used to estimate FWHM (direct half-maximum, fitting, or deconvolution).
• The model or profile assumed (Gaussian, Voigt, etc.).
• Units of measurement and the sampling interval.
• Uncertainties or confidence bounds.

Practical considerations and pitfalls

Be mindful of several common issues that can affect FWHM accuracy:

• Background variations and slope: misplacing the half-maximum level can bias the estimate.
• Asymmetry: non-symmetric peaks yield different FWHM values on each side; report directional FWHMs or fit an asymmetric model.
• Blending: when multiple peaks are close, deblending is necessary before measuring an individual FWHM.
• Sampling and discretisation: coarse sampling makes half-maximum detection less precise without interpolation.

Software and tools

A range of software supports FWHM calculations across disciplines. In microscopy and astronomy, practitioners frequently use Python with libraries such as NumPy, SciPy, and Astropy for profile extraction and fitting; MATLAB is another popular option for curve fitting and 2D/3D Gaussian models. Dedicated image processing packages may provide built-in PSF analysis tools, while spectroscopy software often includes peak fitting routines. Regardless of the tool, ensure you document the fitting model, interpolation method, and any smoothing steps used to derive the FWHM.

FWHM and reportable practice: best-practice guidelines

For robust, reproducible science, adopt a consistent approach to FWHM reporting:

• State the target profile type and the fitting model if applicable (e.g., Gaussian FWHM, Voigt FWHM).
• Specify the dimension(s) in which FWHM is measured (1D profile, 2D PSF axes, etc.).
• Include the estimation method and any data processing steps (smoothing, background subtraction, interpolation).
• Offer the numerical value, units, and the associated uncertainty or confidence interval.

FWHM as a bridge between theory and observation

The concept of FWHM connects theoretical line or PSF models with observed data. In laboratory optics, FWHM translates a theoretical point-spread function into a measurable width. In astronomy, FWHM helps researchers compare atmospheric conditions with telescope performance. In spectroscopy, FWHM provides a quantitative handle on instrumental resolution and physical conditions of the source. Across all these contexts, FWHM remains a practical, interpretable, and widely understood width metric.

Frequently asked questions about FWHM

What is the difference between FWHM and FWTM?

FWHM refers to the width of a peak at half its maximum value. FWTM (full width at tenth maximum) or other fractions describe widths at additional reference levels, but these are less commonly used as standard measures of resolution. FWHM remains the most widely recognised descriptor for optical and spectroscopic sharpness.

Can I use FWHM to compare different instruments?

Yes. When comparing instruments, ensure you measure FWHM under the same sampling, wavelength or energy, and same data processing conditions. Differences in detector pixel size, sampling rate, and background can influence the measured FWHM. Where possible, measurements should be normalised or conducted under equivalent experimental settings.

Is FWHM sufficient to characterise a PSF?

FWHM provides a succinct width measure but does not capture all aspects of a PSF, such as asymmetry, skew, or extended wings. For comprehensive PSF characterisation, report additional metrics (e.g., asymmetry index, kurtosis, central peak sharpness) and consider full profile modelling rather than relying on a single width value.

How does noise affect FWHM?

Noise can blur the half-maximum threshold and bias FWHM estimates, especially for faint peaks. Applying careful smoothing or smoothing filters, along with interpolation for sub-pixel accuracy, helps mitigate noise effects. Always report how noise was handled and provide uncertainty estimates.

A concluding note on FWHM in modern analysis

The Full Width at Half Maximum remains a fundamental, intuitive, and versatile descriptor across scientific disciplines. Whether you are diagnosing instrument performance, interpreting astronomical images, or extracting physical properties from spectral lines, FWHM offers a consistent language for width and resolution. By combining rigorous measurement techniques with transparent reporting, researchers can use FWHM to illuminate subtle features, compare results across studies, and advance our understanding of the systems we observe. As technologies evolve, the FWHM continues to adapt—serving as a reliable yardstick for sharpness, clarity, and the precision of scientific measurement.