Subsidence, the downward vertical land motion (VLM), plays a pivotal role in contributing to the risk of coastal flooding. Accurately estimating VLM and identifying its potential features related to subsidence can provide crucial information for stakeholders to make better informed decisions. This study aimed to estimate large-scale subsidence at the Texas Gulf Coast and identify potential subsidence features using explainable models. Nine potential features were considered for modeling the VLM, ranging from natural terrain variations to anthropogenic activities. These features were used to train a random forest (RF) machine learning (ML) model. Explainable artificial intelligence (XAI) techniques including SHapley Additive exPlanations (SHAP) and impurity- and permutation-based feature importance were used to identify the contributions to subsidence. The results demonstrated favorable performance of the RF model, achieving an $R^{2}$ value of 0.56 during validation. XAI results underscored the significance of the digital elevation model (DEM) on explaining subsidence at the Texas Coast. Additionally, XAI analysis highlighted the overall contribution of subsidence from anthropogenic activities, such as hydrocarbon extraction and groundwater withdrawal. Furthermore, the sample-level SHAP map provided detailed and reasonable subsidence-attribution results across the study area, showing potentials for automatic and data-driven explanations of the VLM.

A double-difference (DD) method was used to estimate vertical land motion (VLM) at 26 tide gauge (TG) sites with record lengths of at least ten years across the Texas Gulf Coast, USA, between 1993 and 2020. In the method, the first difference was conducted by coupling nearby correlated TG stations to remove sea-level variability for both TG and satellite radar altimetry (SRA) data. Upon completion of the first difference, a second difference was performed by subtracting between TG and SRA data. The results obtained from the DD method were compared against that of: 1) a single-difference (SD) method through subtraction between measurements from TG and SRA and 2) a global navigation satellite system (GNSS) precise point positioning (PPP) method. The results showed that the DD method improved the performance of VLM estimation with an uncertainty below 1.0 mm/yr at most TG stations. Meanwhile, the estimated VLM trends acquired from the DD method correlated better to that of the ground-truth GNSS PPP solutions than the SD method. The DD method possesses great potential to discover VLM knowledge, particularly along coastal regions where other techniques such as GNSS and interferometric synthetic aperture radar (InSAR) are of impaired estimation capability.

Global sea-level is rising at an unprecedented rate in recent decades largely due to glacier/ice-sheet melting and ocean warming expansion, which is referred to as the absolute sea-level rise (ASLR). The combination of vertical land motion (VLM) and ASLR leads to a faster sea-level rise relative to the land, which is referred to as the relative sea-level rise (RSLR). The Texas Gulf Coast is one of the leading hotspots subject to VLM and/or RSLR issues in the United States. GNSS has long been used to monitor long-term and accurate VLM with a continuous GNSS (cGNSS) observation network. However, the observation network density offered by cGNSS stations is often limited. This study investigated the feasibility of estimating VLM using the sea-level change differences between the ASLR and RSLR data observed by tide gauges and satellite radar altimetry using 5-day, 15-day, and 30-day time windows. The results were validated by comparing against the GNSS measurements processed by GipsyX software. Independent time-series of VLM processed by using sea-level change differences and GNSS were obtained at Galveston and Rockport, two areas of the Gulf Coast experiencing substantial subsidence. The trends of the VLM time-series were estimated with Hector software. At Galveston, Texas, the VLM trend from GNSS (-3.9 ± 0.4 mm/yr) matched well with that from sea-level change difference estimated at the tide gauge (-3.8 ± 0.7 mm/yr with a 15-day mean window). At Rockport, Texas, an approximately 1.0 mm/yr VLM difference was observed between GNSS (-4.2 ± 0.3 mm/yr) and sea-level change difference (-5.2 ± 1.1 mm/yr with a 15-day mean window) within the confidence intervals of the two methods. This preliminary study demonstrated the feasibility of VLM estimates with sea-level change differences in coastal areas where longterm static cGNSS observations are absent. However, further investigation is needed to model error sources and improve noise mitigation for reliable VLM trend estimate.