Diffraction-Driven Photometry: A Novel Method for Stellar Temperature Determination

Context Estimating stellar temperatures is a fundamental task in astrophysics, typically achieved using color indices (CI) derived from multi-filter photometry. However, this method requires sequential observations through different filters, making it vulnerable to brightness variations between exposures—such as those caused by stellar variability or transits—and sensitive to inter-filter calibration errors, both of which can compromise temperature estimation. Aims We aim to introduce a new technique—diffraction-driven photometry—that enables stellar temperature estimation from a single broadband image by extracting spectral information directly from the Polychromatic Point Spread Function (P-PSF). Methods. We analytically derive the relationship between the P-PSF morphology and stellar temperature using Planck’s law and diffraction theory. A new metric, the P-PSF Intensity Ratio (PIR), is defined and shown to correlate monotonically with temperature. We perform extensive numerical simulations to verify this behavior and quantify performance under realistic instrumental noise using synthetic detector models. Results Our simulations show that the PIR is a reliable estimator of stellar temperature and remains monotonic within the Airy disk across a broad range of stellar types. The method is intrinsically robust against global flux variations and can be adapted to a wide range of optical systems through calibration. It performs best at high signal-to-noise ratios, but remains effective for spectral classification at lower SNRs. Conclusions Diffraction-driven photometry provides an efficient alternative to classical CI-based temperature estimation, particularly suited to the constraints and design efficiencies required in space-based instrumentation. Its simplicity makes it ideal for small satellites and enables retroactive application to archival data, offering a practical path toward robust temperature retrieval from single-band observations.