Floods are among the most destructive hydrometeorological disasters worldwide, highlighting the need for reliable large-scale monitoring of flood-prone conditions. This study constructed the Optimal Flood Potential Index (OFPI) for higher-accuracy and broader-scale monitoring of storage-based flood-prone regions, based on a monthly terrestrial water storage anomaly (TWSA) dataset with improved spatial resolution (0.5°). OFPI was then used to assess the spatiotemporal evolution of flood-prone regions across China from 2003 to 2022. Spatiotemporal analysis shows that flood-prone conditions in China are mainly concentrated from June to August, with July and August representing the peak period. High-risk areas are primarily located in southern China in June and shift northward in July and August. Long-term trend analysis reveals clear regional differences: northern China generally exhibits a decreasing tendency in flood risk, whereas southern China shows an increasing trend. Seasonal analyses indicate that winter flood risk remains low with limited variation, spring and autumn changes are relatively weak and scattered, and summer presents the most pronounced and spatially coherent trends. Performance evaluation demonstrates that OFPI improves flood-prone region identification relative to Flood Potential Index (FPI). Validation based on the Dartmouth Flood Observatory (DFO) flood records indicates that OFPI shows better detection performance than FPI. Furthermore, quantitative assessment based on the reference sample set shows that recall increases from 0.609 to 0.715, and F1-score rises from 0.633 to 0.659, indicating a stronger capability to detect high-risk conditions while reducing missed detections. Overall, OFPI provides a more effective framework for monitoring the spatiotemporal evolution of flood-prone regions.

The Gravity Recovery and Climate Experiment (GRACE) Level-2 products, i.e., monthly gravity field models expressed in spheric harmonic (SH) coefficients, have been released in several versions since 2002. In this study, we analyze the extraction of co-seismic signals associated with the 2004 Sumatra-Andaman MW9.1 earthquake, using GRACE data from four versions, namely the RL01, RL04, RL05, and RL06, provided by the Center for Space Research (CSR), University of Texas at Austin. Results indicate that the co-seismic signals extracted from later versions of GRACE data are less affected by noise. Moreover, the earliest RL01 vision can also reflect the co-seismic gravity-change signals of the 2004 Sumatra-Andaman earthquake. In particular, we find that the spatial pattern of the co-seismic signals extracted from the RL04 version of GRACE data exhibits significant distortion, which is probably due to the large errors introduced in the processing of the atmosphere and ocean models in the gravity field inversion. Comparison between the dislocation model prediction and GRACE observation suggests that the GRACE results from later versions are more consistent with the model prediction. In addition, the dislocation models with stronger constraints from near-field measurements during the fault-slip inversion provide a more consistent prediction with the GRACE observation. The uncertainty estimation of GRACE data over the global oceanic region reveals that the noise levels gradually decrease from the earlier to later versions, with the RMSs of 7.1, 4.9, 2.4, and 2.1 μGal (with 300 km spatial smoothing) for the RL01, RL04, RL05, and RL06 versions, respectively.

As a leading indicator of global climate change, contemporary global mean sea level (GMSL) change is mainly driven by thermosteric (thermal expansion) and barystatic (ocean mass increase) contributions. GMSL change has been continuously measured by satellite altimetry since 1993, while thermosteric sea level change can be inferred from in-situ hydrographic measurements dating back to the 1970s. However, direct observations of barystatic sea level change were generally lacking until the launch of the Gravity Recovery and Climate Experiment (GRACE) in 2002. In the absence of GRACE, barystatic sea level estimation relies primarily on the so-called mass budget approach by summing individual surface mass change estimates (e.g. ice sheets, glaciers, and terrestrial water storage) obtained from different remote sensing or geophysical modelling techniques, providing an indirect observation due to the lack of global constraints. In this study, we use low-degree gravity fields obtained from satellite laser ranging (SLR), a traditional space geodetic technique over decades, to directly estimate barystatic sea level changes since 1993. To this end, we effectively address the issues of signal leakage and missing geocenter motion for SLR gravity fields using the forward modelling technique. Our SLR-based barystatic sea level estimates allow the direct observation-based assessment of the GMSL budget over the satellite altimetry era, and also provide an independent dataset for cross-validation and gap-filling between GRACE and its successor GRACE-FO. Using reprocessed altimetry data from NASA's Goddard Space Flight Center and updated thermosteric sea level ensembles, we reconcile the GMSL rise budget from 1993 to 2022. Our results show that the sum of thermosteric and SLR-based barystatic contributions (3.16 ± 0.64 mm/yr) agrees well with the altimetry-observed GMSL rate (3.22 ± 0.28 mm/yr), suggesting that the GMSL budget can be closed within uncertainties over the last three decades. Nevertheless, we observe increased budget residuals when using different altimetry datasets, especially in recent years, highlighting the ongoing challenges in accurately observing GMSL change and robustly closing the GMSL budget.

The sea level anomaly (SLA) has been accurately tracked by satellite altimetry, yet its barotropic and depth-integrated baroclinic components are routinely interpreted using theoretical or modeled vertical structures. In this study, we utilized manometric SLA from the Gravity Recovery and Climate Experiment (GRACE) and steric SLA from observational and reanalysis temperature/salinity databases to evaluate their roles in seasonal variations of geostrophic velocity anomalies and eddy kinetic energy (EKE) in the South China Sea (SCS). Through the empirical orthogonal function analysis, we found that the manometric component of geostrophic velocity anomalies is closely associated with the western boundary current, reflecting a barotropic response to seasonally reversed wind stress in summer and winter. The steric component, primarily driven by baroclinic instability, shapes two large cyclonic (anticyclonic) gyres (Luzon and Nansha Gyres) in the northeastern and southern SCS during summer (winter), as well as small mesoscale anomalies in the northwestern SCS during spring and autumn. The cross-correlation analysis demonstrates considerable influence of wind stress on the surface dynamics throughout SCS, while wind stress curl predominantly contributes to the gyres and dipole system off the Vietnamese coasts. Opposing covariances between manometric and steric EKE along the eastern deep and western shelf sides of the southwestern continental slope were investigated via vertical density, temperature, and salinity anomalies along three transects. These patterns arise from seasonally distinct horizontal and vertical mixing structure in upper-layer and near-bottom cross-shelf currents, providing observational evidence for significant interactions between baroclinic and barotropic instabilities in coastal regions.

Abstract Assessments of the global mean sea level (GMSL) budget over the satellite altimetry era (since the early 1990s) have concluded that the GMSL budget is closed within data uncertainties until 2016. However, studies have shown that since then, the sea level budget based on Argo data down to 2,000 m for the thermosteric contribution is no longer closed. Using an ocean reanalysis with no altimetry data assimilation, we show that accounting for deep ocean thermosteric contribution (below 2,000 m, not sampled by Argo) allows the GMSL budget to be almost closed since 2016. The deep ocean contribution over 2005–2022 is estimated to be 0.4 ± 0.15 mm/yr, that is, about 10% of the observed GMSL rise over that period. This represents a substantial increase of the deep ocean contribution to sea level rise, previously estimated on the order of 0.1 mm/yr only over 1980–2010. This finding reveals that deep ocean warming is gaining importance and that ocean heat uptake has now reached several regions below 2,000 m depth, notably the Northwestern Atlantic Ocean and Southern Ocean.

The interactions between vegetation and hydro-climatic factors play a critical role in key geophysical processes, including carbon and hydrological cycles. However, the interaction between vegetation and hydro-climatic factors in China remains unclear. Here, nonlinear Granger causality tests were employed to analyze the bidirectional relationships between vegetation and key hydro-climatic variables from 2002 to 2021. The results indicate that temperature was the dominant Granger-causal factor influencing vegetation growth (43.49 % of grid cells), followed by terrestrial water storage (16.49 %), soil moisture (11.44 %), precipitation (10.92 %), and solar radiation (3.11 %). In the reverse direction, vegetation exerted the strongest feedback on terrestrial water storage (31.18 %) and precipitation (28.77 %), with weaker effects on soil moisture (20.68 %), solar radiation (9.67 %), and temperature (3.79 %). Bidirectional Granger causality was observed in 33.57 %-46.77 % of the assessed areas, with grasslands showing the highest proportion of significant causal relationships. These findings improve our understanding between vegetation and hydro-climatic factors across China, providing valuable insights that can guide the refinement and optimization of these processes in ecological and hydrological models.

The large-scale global mass distribution affects Earth's oblateness J2. Since the 21st century, the previously observed secular decrease in J2 attributed to glacial isostatic adjustment has been reversed, with J2 now increasing under climate change. Revealing the spatial roles of mass redistribution contributing to the J2 trend is essential for understanding Earth's dynamic processes. In this paper, we quantify the cryospheric and hydrological contributions to the J2 trend utilizing the Gravity Recovery and Climate Experiment (GRACE) and GRACE Follow-On (GFO) measurements. Unlike previous efforts that drew merely general conclusions about J2 contribution sources, we conduct a detailed analysis to identify the underlying spatial patterns of mass migration associated with these sources. The Greenland and Antarctic ice sheets (GrIS and AIS) remain the dominant contributors to the increasing J2 trend; however, the contribution from AIS melting has slowed and recently tended toward stabilization, due to mass gains in East Antarctica. Mass changes in mountain glaciers, independently determined through GRACE/GFO and results from satellite radiometry measurements of changes in glacier size, provide a substantially large contribution to the J2 trend, mainly from mountain glaciers in mid-to high-latitude North America and Eurasia, even surpassing the effect of AIS melting. The contributions of land hydrology to the increasing J2 trend are ascertained to be driven by increased terrestrial water storage (TWS) in Africa and by TWS depletion together with potential ice loss in Eurasia. These findings provide an improved understanding of how global ice and TWS changes influence geodynamic processes. (c) 2026 COSPAR. Published by Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

Droughts are intensifying under global warming, causing severe impacts on water resources, agriculture, and ecosystems. Different types of droughts are interconnected through propagation within the hydrological cycle, yet their dynamics and nonlinear responses to climate change remain poorly understood. Here we apply the PCR-GLOBWB 2 model driven by bias-corrected CMIP6 climate projections to investigate the evolution and propagation of meteorological, hydrological, and groundwater droughts in the Pearl River Basin (PRB), China, across historical (1979-2014) and future period (2015-2099) under four Shared Socioeconomic Pathways (SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5). Historical simulations show that meteorological drought is the shortest (similar to 1.6 months) and most frequent, whereas groundwater drought is the longest (similar to 8.28 months) but least frequent. Future projections show enhanced drought frequency and shortened duration, with rapid propagation from the atmosphere to surface hydrology (1-4 months) but delayed transmission into groundwater (6-24 months). This indicates that surface hydrology responds quickly and recovers easily, whereas groundwater exhibits delayed transmission, accumulation, and persistent signals. Nonlinear propagation behaviors exist under different emission scenarios, with extremes occurring primarily under moderate forcing rather than the highest-emission scenario, which is associated with opposing effects of precipitation and temperature, with higher temperatures enhancing evapotranspiration and variable precipitation offsetting drought intensity. As scenarios escalate, the differences between the near- and far-future periods become more pronounced. These findings advance understanding of cross-sphere drought connectivity and highlight the need to incorporate subsurface delay responses into future drought projections.

Abstract Limitations in the temporal resolution of contemporary gravity satellite missions hinder the precise monitoring of rapid Earth surface mass changes. By the early 2030s, unprecedented high‐temporal monitoring of Earth's dynamic mass redistribution will be available using the temporal gravity field derived from the Hybrid Gravity Satellite Ensemble (referred to as the “HGSE” in this study), which contains GRACE‐FO, ChiGaM, TIANQIN‐2, GRACE‐C, and NGGM. This paper proposes a Hybrid‐Augmented Resolution Dealiasing (HARD) algorithm that utilizes a sliding window technique to co‐estimate 3‐day low‐degree and daily high‐degree spherical harmonic coefficients. The HARD algorithm reduces temporal aliasing errors by 18.4%–30.7% compared to conventional processing strategies. Based on predefined noise levels for each satellite, closed‐loop simulations demonstrate that the HGSE yields daily gravity field solutions (with a maximum degree and order of 60) that can effectively reduce noise by approximately 76.2% in long‐term trends and 39.3% in annual amplitudes compared to classical monthly solutions. Applications in terrestrial water storage (TWS) change, glacier mass change, and co‐seismic deformation reveal significant improvements: 39.4% enhanced TWS signal recovery in large river basins, 21.2% higher accuracy in monitoring Tibetan Plateau glacier mass variation, and 69.4% superior co‐seismic signal recovery for megathrust earthquakes. These findings underscore the potential of HGSE to advance high‐frequency gravity field monitoring, offering critical references for the performance analysis of future gravity satellite missions monitoring the Earth's dynamic system processes on a daily scale.

Active volcanoes are often deformed by magmatic activity occurring at depth. Here, we report the deformation of Mount Fuji, central Japan, in relation to hydrological activity induced by rain. Through an analysis of the daily coordinates of global navigation satellite system stations deployed around Mount Fuji, we detected transient surface uplift of 1-2 cm correlated with heavy rains and deduced that this is caused by the expansion of shallow aquifers within Shin-Fuji lava layers. Such hydrological inflation of the volcano, lasting for a day or two, occurs within similar to 25 km of the summit. The uplift gradually decays with distance and is replaced by large-area subsidence through rainwater loading beyond the end of these lava layers. Understanding such "cold" volcanic inflation assists in the correct interpretation of "hot" changes associated with magmatic forcing mechanisms.

The pole tide has been defined as the oceanic response to polar motion, with a component due to the Chandler Wobble (CW), at a period near 14 months. The magnitude of the CW was greatly reduced after about 2015, and a corresponding global attenuation of the ocean pole tide amplitude is expected. We analyzed global tide gauge measurements using empirical orthogonal functions (EOF) methods to separate temporal and spatial patterns of regional pole tide signals. Our results show that tide gauge arrays in Japan and along the east coast of North America exhibit 14-month pole tide signals whose temporal variability is broadly consistent with the CW amplitude reduction after 2015. At stations in both regions, EOF-based mean pole tide amplitudes since 1980 fall in the range of approximately 0.5–1 cm, representing a more coherent signal than direct spectral estimates from individual stations. These amplitudes exceed the expected equilibrium response (less than 0.5 cm), suggesting that they reflect a mixture of equilibrium and moderate non-equilibrium contributions. Tide gauges in the North Sea and Baltic Sea show no detectable amplitude decrease after 2015, consistent with previous studies documenting strong non-equilibrium pole tide variability in these regions. Together with the partial agreement with equilibrium expectations in Japan and along the North American east coast, these results suggest that regional pole tide signals likely reflect a combination of equilibrium response and additional forcing, potentially including long-period atmospheric variability.

We derive Greenland ice mass changes from the gravity recovery and climate experiment (GRACE) and GRACE follow-on mascon solutions and 30 input-output method (IOM) combinations, and conduct a comprehensive comparison for the period 2002-2018. Using mascon solutions as constraints, we identify an optimal IOM combination to estimate the linear and quadratic trends for Greenland ice mass change, which enables us to extend the analysis to 2023. The IOM-estimated ice mass loss rate of -257.7 +/- 19.8 Gt & centerdot;yr(-1) closely matches estimates from the three mascon solutions (-258.9 +/- 13.2, -264.1 +/- 12.0, and -271.8 +/- 12.0 Gt & centerdot;yr(-1), respectively), provided by the Center for Space Research (CSR), Goddard Space Flight Center (GSFC), and Jet Propulsion Laboratory (JPL). A deceleration in ice mass losses is observed across all datasets, with rates of 3.2 +/- 3.8, 4.5 +/- 2.2, 3.8 +/- 1.9, and 6.5 +/- 2.2 Gt & centerdot;yr(-2) for the IOM combination, CSR, GSFC, and JPL mascon solutions, respectively. Partitioning results suggest that surface mass balance (SMB), solid ice discharge, and peripheral glaciers' mass balance contribute about 41%, 50%, and 9%, respectively, to the total loss, with the deceleration mainly originating from SMB. Further decomposition of SMB reveals that runoff anomaly accounts for 136% of SMB-related ice mass loss, with 37% offset by positive precipitation anomaly. The observed deceleration in Greenland ice mass loss that has only been evident in recent years is temporally coherent with the quadratic increase in cumulative snowfall anomaly (4.1 +/- 1.6 Gt & centerdot;yr(-2)).

Active volcanoes often deform by magmatic activities at depth. Here we report that they deform also by hydrological activities induced by rains. By analyzing the daily coordinates of global navigation satellite system stations deployed around the Fuji volcano, the highest mountain of the country in central Japan, we detected transient surface uplift of 1-2 centimeters correlated with heavy rains. We consider they were caused by the expansion of shallow aquifers within Shin-Fuji lava layers. Such hydrological inflation of the volcano, lasting for a day or two, occurs within ~25 km from the summit. The uplift gradually decays with distance and is replaced with large-area subsidence by rainwater loading beyond the end of these lava layers. Subsidence is proportional to daily rains, rather than cumulative rains, suggesting dynamic equilibrium of precipitation and run-off. Understanding such ‘cold’ deformation of active volcanoes would help us correctly interpret ‘hot’ ones by magmatic activities.

This study focuses on the feasibility of detecting coseismic gravity changes from inland large earthquakes using Gravity Recovery and Climate Experiment Follow-On (GRACE-FO) observations. The 2023 Turkey double earthquakes (MW7.8 and MW7.5) serve as the specific case study. Dislocation model predictions indicate that the peak-to-peak coseismic gravity changes range from −0.10 to +0.06 μGal at the spatial resolution of GRACE-FO monthly gravity field solutions. An RMS analysis of CSR RL06 data (2018–2025) demonstrates that observation noise reaches up to 1.26 μGal over the near-field region of the Turkey double earthquakes. This indicates that the noise amplitude exceeds the target signal by an order of magnitude, which suggests that the coseismic gravity-change signal is difficult to detect. Moreover, uncertainty estimation using the discrepancies between the GLDAS and WGHM land surface models shows that the hydrological model uncertainty reaches up to about 2.59 μGal, which further hinders the feasibility of coseismic signal detection. Furthermore, simulative tests indicate that improved precision in both GRACE-FO observations and hydrological models is necessary to detect the coseismic signals of earthquakes comparable to the Turkey double earthquakes (about MW7.8). For marine earthquakes, GRACE-FO noise needs to be reduced to 60% of its current level. For inland earthquakes, besides the reduction of GRACE-FO noise, the hydrological model errors should be reduced to 5%. Consequently, successful GRACE/GRACE-FO detection of the effect of inland earthquakes with a magnitude of around MW8.0 or lower requires advancements in satellite gravimetry coupled with enhanced hydrological models.

Study region: Tianshan regionStudy focus: The Tianshan region, a typical mountain–desert transition zone in Central Asia, supports downstream oasis ecosystems and agriculture heavily reliant on groundwater. While GRACE satellites reveal significant mass loss, the relative contributions of glacier melt and groundwater depletion remain unclear. This study combines GRACE, reanalysis diagnostics, and independent glacier estimates to identify freshwater decline drivers from 2003 to 2024.New hydrological insights for the region: We identify a prominent ~7.3-year interannual signal strongly coupled with El Niño–Southern Oscillation (ENSO) variability. Incorporating this low-frequency oscillation into the model improves the fit (R^2 from 0.88 to 0.91) and prevents a ~4% overestimation of the long-term trend. Furthermore, stochastic analysis shows that Total Water Storage (TWS) residuals exhibit colored noise. A Generalized Gauss–Markov (GGM) noise model expands trend uncertainty from ±0.02 to ±0.08 cm/yr. Comparing 2006–2016 and 2012–2022, surface water persistently increased while TWS declined, indicating that lake storage partially offset overall losses. Additionally, groundwater exhibits a persistent downward trend, dominating the regional water deficit (~66%). These findings demonstrate that properly accounting for low-frequency oscillations and noise characteristics is crucial for accurate freshwater assessments and sustainable water resource management in vulnerable mountain regions.

Climate change is intensifying the hydrologic cycle, and terrestrial water storage (TWS) provides an integrated indicator of these shifts. In this study, we analyzed long-term and interannual TWS variations across Africa over April 2002-December 2024 using GRACE/GRACE Follow-On (GRACE/-FO) satellite gravimetry, together with land surface models (LSMs) and satellite altimetry derived lake storage. GRACE/-FO revealed a continental TWS increase of 96.1 +/- 13.4 Gt yr-1 (2 sigma), with an acceleration of 11.6 +/- 3.83 Gt yr-2. Basin-scale attribution exhibited that large lakes/reservoirs storage increases largely contributed to the observed TWS gains, whereas model predicted soil moisture showed a general declining trend, likely indicating the limitations of LSMs in this region. The long-term TWS increase was further quantitatively supported by a rising water balance. Interannual variability was structured by large-scale climate modes: El Nino-Southern Oscillation (ENSO) and the Indian Ocean Dipole (IOD) paced continental and tripole-like responses, respectively, as identified by empirical orthogonal function (EOF) analysis. Our results elucidated new observational constraints on the drivers of African TWS dynamics and their responses to a changing climate.

Seasonal mass changes along shorelines have been largely overlooked due to the coarse spatial resolution of the Gravity Recovery and Climate Experiment (GRACE) measurements. Here, we examine the monthly mean ocean mass along the east coast of China using GRACE data and identify a pronounced seasonal cycle and its southward propagation, with peak-to-peak amplitudes of similar to 150 mm, reaching the minimum in April and maximum in October, opposite in phase to seasonal mass changes over adjacent land. Independent estimates from satellite altimetry sea level anomalies corrected for thermosteric effects and global Navigation Satellite Systems displacements confirm the GRACE results, all revealing significant seasonal ocean mass changes in this region. Furthermore, wind-stress forced model simulations effectively replicates both the phase and magnitude of the observed variations, demonstrating that wind stress is the primary driver of the seasonal ocean mass changes and southward propagating transports.

Large-scale ecological restoration has enhanced ecosystem services, but its impacts on terrestrial water storage (TWS) remain debated. Our study challenges the prevailing assumption that restoration activities inevitably reduce TWS. Here we construct a long-term TWS record (1987-2020) to analyze hydrological dynamics in the Mu Us Sandyland in China, a region undergoing intensified ecological restoration since 1999. Our findings delineate three distinct hydrological phases: stable pre-restoration conditions (1987-1998), rapid TWS depletion (2003-2010), and subsequent unintended recovery (2011-2020). Satellite and modeling data reveal that ecological restoration and agricultural expansion initially reduce TWS through enhanced plant transpiration, while vegetation-mediated atmospheric feedback later appears to boost regional precipitation. Climate projections suggest this region may experience 51

The Gravity Recovery and Climate Experiment (GRACE) mission and its successor, GRACE Follow‐On (GFO), effectively monitor terrestrial water storage anomaly (TWSA). However, their constrained spatial resolution imposes limitations, with leakage and attenuation potentially impacting the accuracy of regional TWSA. While GRACE/GFO observations capture the ongoing TWSA decline in the Middle East due to excessive groundwater extraction, the nearby Caspian Sea's long‐term water loss, combined with the seasonal signals from the coastal sea, complicate accurate TWSA estimation through signal attenuation and leakage. To address these issues, we propose a combined approach, that is, independent component analysis (ICA)‐based forward modeling (IFM), to discern and isolate the leakage effect and improve the recovery of TWSA signal. We demonstrate the impact of signal attenuation and leakage through simulation, and validate the effectiveness of IFM. This method is also confirmed through steric‐corrected altimetry estimates in the Caspian Sea, Red Sea, and Persian Gulf, and further validated in Greenland and Victoria Lake. Our results show considerable leakage in GRACE/GFO TWSA estimates for Saudi Arabia and Iran. Leakage from the Red Sea and Persian Gulf introduces a 28.6% bias in Saudi Arabia's TWSA trend, while leakage from the Caspian Sea results in a 36.4% bias in Iran. After IFM recovery, the TWSA decline rates for Saudi Arabia, Iraq, and Iran are 11.48 ± 0.32, 3.56 ± 0.44, and 7.75 ± 0.45 km 3 /yr, respectively. This study demonstrates the effectiveness of IFM in deriving refined TWSA signal, providing valuable insights for water resource management in arid regions.

We introduce an improved Kalman filtering (i.e. the Affine transformation based Kalman filtering, named Aff-Kalman filtering) into the regional Mascon inversion model to recover reliable weekly surface mass variation (SMV) over the Amazon River basin from the Gravity Recovery and Climate Experiment (GRACE)-based geopotential difference (GPD) data. The performance of Aff-Kalman filtering was validated through closed-loop simulation and comparative analyses against official spherical harmonic coefficient (SHC) and Mascon solutions and independent hydro-meteorological datasets. The simulation results demonstrate that the weekly SMVs from Aff-Kalman filtering have better consistency with input 'real' signals than those from traditional Kalman filtering and GPD Mascon estimates, and the corresponding determination coefficient (R2) increased by 3.2% and 2.9%, root mean square error (RMSE) decreased by 24.97 mm and 13.81 mm, and mean absolute percentage error (MAPE) decreased by 2.82% and 0.39%, respectively. Additionally, the weekly and monthly SMVs derived from Aff-Kalman filtering exhibit strong consistency with official weekly GFZ SHC and monthly CSR Mascon solutions in the spatio-temporal domains. Meanwhile, the weekly RMSE between the Aff-Kalman filtering and GFZ SHC solutions (29.73 mm) is smaller than that between the GPD Mascon and GFZ SHC solutions (34.85 mm). Furthermore, the first-order difference of terrestrial water storage changes (i.e. dS/dt) derived from Aff-Kalman filtering present better consistency (i.e. higher correlation and lower RMSE) with P-ET-R time series (derived from precipitation-P, evapotranspiration-ET and runoff-R based on the water budget closure) than those of GFZ SHC, CSR Mascon, and GPD Mascon solutions from weekly to monthly scales. For the monthly solutions, the corresponding R2 and RMSE values are 0.843 and 29.577 mm/month (Aff-Kalman filtering), 0.839 and 30.154 mm/month (CSR Mascon), and 0.842 and 30.216 mm/month (GPD Mascon) on the temporal scale, respectively. Our inversion method provides an alternative tool for estimating reliable regional SMVs with higher temporal resolution from GRACE data.