Global Navigation Satellite System Reflectometry (GNSS-R) is a bistatic technique that reads navigation signals reflected from Earth to infer surface properties. Its core observable, the delay-Doppler map (DDM), changes shape with the scattering regime: smooth surfaces yield compact, high-amplitude coherent returns, while rough terrain and dense vegetation produce broader, diffuse, “horseshoe-shaped” returns with lower peak power. Over land, the coherent–noncoherent transition depends on vegetation, topography, inland water and possibly shallow subsurface soil properties. Misclassifying this transition can bias retrievals and add uncertainty. Earlier work often focused on one surface type or region and used heuristic thresholds, leaving a gap in joint, nonlinear, pan-tropical assessments. Given these challenges, in-depth research is needed on how multiple land-surface factors jointly modulate GNSS-R coherence across the pan-tropics.
Published (DOI: 10.1186/s43020-026-00221-z) online on 21 October 2026 in Satellite Navigation, the study was led by Qingyun Yan of Nanjing University of Information Science and Technology, with collaborators from the Henan Polytechnic University and Memorial University of Newfoundland. The team combined Cyclone Global Navigation Satellite System (CYGNSS) observations with reference datasets for topography, surface water, soil moisture and vegetation, then used an explainable machine-learning model to quantify how these properties modulate DDM coherence across the pan-tropics. It also tested model generalization and sensitivity to incidence angle, threshold choice and spatial resolution.
The framework used a PN proxy inversely related to coherence from CYGNSS Level-1 DDM power and trained a Light Gradient Boosting Machine (LightGBM) model with topographic, hydrological, soil-moisture and vegetation predictors. Under spatial cross-validation, the model explained about 60% of spatial variance in the proxy and generalized reasonably in regional holdout tests. Shapley additive explanations (SHAP) and accumulated local effects (ALE) showed coherence rises over smooth, wet and moderately vegetated surfaces but falls over rough, steep and densely vegetated terrain. Terrain roughness was the strongest individual predictor; aggregated by process group, topography dominated about 68% of land, surface water and vegetation each about 14%, and soil moisture about 4%. ALE curves highlighted approximate empirical transitions near 15 m roughness, 1.1° slope, 0.5% water occurrence, 0.35 m³ m⁻³ soil moisture, 107 Mg ha⁻¹ biomass and 1.25 leaf area index. Ablation tests in four regions supported these groups as leading controls, while showing that multiple processes shape DDM coherence.
The authors said the value of the work lies in turning a black-box model into a map of physical controls. They said the results do not propose a new scattering mechanism; rather, they show where topography, water, soil moisture and vegetation are likely to matter most, and where coherence-based products should be treated with caution. They added that the transition ranges are empirical guides, not operational thresholds, and should help users interpret GNSS-R land observations and design better scene-selection and quality-control steps.
For GNSS-R land applications, the findings offer a practical way to anticipate signal quality before retrieval. Coherence-sensitive products such as inland-water mapping, soil-moisture estimation and vegetation studies can use the dominant-control maps to flag mountainous, steep or densely forested scenes where diffuse scattering and phase decorrelation are more likely. The empirical transition ranges can guide scene selection and quality control, while the process-group maps help explain regional differences in performance. The authors caution that the ranges reflect annually averaged, pan-tropical conditions and should be tested with time-resolved data and raw intermediate-frequency observations before being used as operational thresholds.
###
References
DOI
10.1186/s43020-026-00221-z
Original Source URL
https://doi.org/10.1186/s43020-026-00221-z
Funding information
National Natural Science Foundation of China, Grant 42001362; Henan International Science and Technology Cooperation Key Program, Grant 241111520700.
About Satellite Navigation
Satellite Navigation (ISSN: 2662-1363; ISSN: 2662-9291) Satellite Navigation is the official journal of the Aerospace Information Research Institute. The journal aims to report innovative ideas, new results, and progress in the theories, techniques, and applications of satellite navigation. The journal welcomes original articles, reviews and commentaries.