Enhancing Spatiotemporal InSAR Retrieval of Snow Water Equivalent Using Multichannel and MAP Estimators

Abstract

Differential repeat-pass interferometric synthetic aperture radar (d-InSAR) is an emerging method for retrieving seasonal snow water equivalent (SWE). The technique presents several limitations and sources of uncertainty related to decorrelation and phase ambiguities. For this reason, low-frequency bands are commonly preferred as they offer greater unambiguous retrieval ranges and coherence conservation. However, they lack sensitivity to small changes in snow mass due to long wavelengths. High-frequency bands provide higher sensitivity but suffer from decorrelation and phase wrapping under heavy snow accumulation. Future SAR satellite missions may enable simultaneous acquisitions. MultiChannel (MCh) techniques leverage the combination of statistically independent measurements to reduce noise and ambiguities. This study formulates MCh techniques for retrieving changes in SWE ( $\Delta \mathrm {SWE}$ ) using d-InSAR. Three MCh configurations were evaluated: L- and S-, L- and C-, and C- and X-bands. The theoretical performance was analyzed using the Cramér–Rao lower bound (CRLB). Experimental validation employed SodSAR, a 1–10 GHz tower-based SAR system, and compared single-band and MCh retrievals of $\Delta \mathrm {SWE}$ against in situ snow scale measurements. Additionally, semisynthetic simulations based on ALOS-2 data and a numerical snow model are provided for assessing distributed retrievals employing spatial filtering and Maximum a Posteriori (MAP) estimates. Results demonstrate that MCh configurations have the potential to improve retrieval accuracy, reduce noise, and resolve phase ambiguities.