GRACE Data Sets Research Articles

Expert-based prior uncertainty analysis of gridded water balance components: Application to the irrigated Hindon River Basin, India

Study regionHindon River Basin, North India. Study focusAccurate estimation of water balance components is crucial for water management applications yet challenging due to errors in monthly gridded water balance data products. Error and uncertainty quantification is especially important in the absence of extensive in-situ data. This paper presents a prior uncertainty analysis for such situations consisting of two components: (i) quantification of prior uncertainties using metrics that quantify errors in individual products and variability and consistency between products, and (ii) reduction of prior uncertainties by eliminating unrealistic water balance estimates. New hydrological insights for the regionGrid-scale inter-product uncertainty or variability, computed as the coefficient of variation (CV, %) at various temporal scales, reveals discrepancies between water balance products due to a combination of factors, including methodological differences, inherent spatial variability, data sources, and resolution disparities. At the mean annual scale, P fluxes display a lower grid-scale inter-product uncertainty (5–9 %) than ET (20–55 %), while the ∆TWS from GRACE solutions show a moderate mean annual grid-scale inter-product uncertainty (15–19 %). Grid-scale inter-product uncertainties of ∆SMS for July – representing the onset of the monsoon season – are high (CV = 54–122 %), indicating that the uncertainty in estimates of this component may have a large impact on water balance analyses. P fluxes exhibited fewer spatio-temporal uncertainties (R2 above 0.8) than ET fluxes (R2 less than 0.75). The exclusion of the unreliable data sets resulted in (a) reducing uncertainties in input water balance components with triple collocation range shifting from 15–38 to 17–23 mm/month for ET and from 16–52 to 11–23 mm/month for P, (b) obtaining updated prior estimates of seasonal water balance. The updated priors of water balance variables per season suggest a net basin outflow (from −318 to −57 mm/season) during the monsoon (rainy) season and net basin inflow (from −38 to 330 mm/season) during the non-monsoon (dry) season, the latter related to surface-water imports from outside the basin. All GRACE data sets exhibit a regional long-term decreasing trend in total water storage (ranging from −31 to −61 mm/year), qualitatively confirming previously documented unsustainable groundwater depletion in the basin. Prior ranges and uncertainties for all water balance variables reported here can be used as input into a posterior analysis that uses in-situ data for locally calibrating (bias-correcting, noise-filtering) and further updating the prior estimates.

Advancing Hydrology through Machine Learning: Insights, Challenges, and Future Directions Using the CAMELS, Caravan, GRDC, CHIRPS, PERSIANN, NLDAS, GLDAS, and GRACE Datasets

Machine learning (ML) applications in hydrology are revolutionizing our understanding and prediction of hydrological processes, driven by advancements in artificial intelligence and the availability of large, high-quality datasets. This review explores the current state of ML applications in hydrology, emphasizing the utilization of extensive datasets such as CAMELS, Caravan, GRDC, CHIRPS, NLDAS, GLDAS, PERSIANN, and GRACE. These datasets provide critical data for modeling various hydrological parameters, including streamflow, precipitation, groundwater levels, and flood frequency, particularly in data-scarce regions. We discuss the type of ML methods used in hydrology and significant successes achieved through those ML models, highlighting their enhanced predictive accuracy and the integration of diverse data sources. The review also addresses the challenges inherent in hydrological ML applications, such as data heterogeneity, spatial and temporal inconsistencies, issues regarding downscaling the LSH, and the need for incorporating human activities. In addition to discussing the limitations, this article highlights the benefits of utilizing high-resolution datasets compared to traditional ones. Additionally, we examine the emerging trends and future directions, including the integration of real-time data and the quantification of uncertainties to improve model reliability. We also place a strong emphasis on incorporating citizen science and the IoT for data collection in hydrology. By synthesizing the latest research, this paper aims to guide future efforts in leveraging large datasets and ML techniques to advance hydrological science and enhance water resource management practices.

The importance of local observation wells, reanalysis, and satellite data on gravity anomaly, climate, and land use to improve groundwater management in the Urucuia aquifer system

The Urucuia Aquifer System (UAS) is an important water source for multiple users and the perennity of the San Francisco River. However, it has been subject to intensive land use transition (LUT), which provoked concerns due to continuous groundwater depletion. Therefore, understanding this depletion and potential drivers is essential for improved management towards water security. In this study, satellite data of the GRACE-project and in-situ data of the RIMAS-project were used to quantify the groundwater depletion between 2003 and 2021. Two important drivers were assessed for their impact on the UAS: (i) climate variability and (ii) LUT. The impact of climate variability was evaluated by using the vertical water balance (VWB; rainfall - evapotranspiration) on basis of validated rainfall and air temperature data from reanalysis (1979–2020). LUT was retrieved from the MAPBIOMAS database (1985–2022). The UAS was subdivided into quadrants of 0.75 × 0.75° (6700 km2) to assess the spatial variation of groundwater depletion and its drivers. Groundwater depletion was characterized by trend analysis and contrasted with the temporal behavior of VWB and LUT across the UAS. The results show a continuous decrease of the groundwater table in the study period (−22 mm/y) with a steeper decline between 2012 and 2018 (−52 mm/y). A direct cause-effect-relation can be observed with the negative trend of the VWB, pushed by decreasing rainfall and increasing air temperature over time. On the other hand, groundwater depletion correlates with LUT and is more accentuated where native vegetation is replaced by irrigated crops. Moreover, the GRACE dataset could be used for regional studies, but in-situ monitoring is recommended for local management as validation (R2 = 56%) indicates an underestimation of groundwater depletion by satellite data. In conclusion, extending the in-situ monitoring network, limiting LUT, and irrigation are crucial to long-term sustainability of the UAS.

Satellite-based drought assessment in the endorheic basin of Lake Balkhash

Introduction: This study investigates into the drought-prone region of the Lake Balkhash basin, conducting a thorough analysis spanning 2 decades, emphasizing its significant impact on agriculture and water challenges in Central Asia. Beyond the specific focus on this region, the research aims to contribute valuable insights that extend our understanding of the broader implications of drought in the area.Methods: Utilizing MODIS satellite imagery, the study employs the Vegetation Health Index a comprehensive indicator combining the Vegetation Condition Index and Temperature Condition Index ranging from 0 (extreme aridity) to 100 (optimal moisture). The assessment of drought conditions from 2000 to 2020 includes probabilistic mapping, trend analysis, and Pearson’s correlation coefficients. Connections with hydrological factors, such as river water levels and the Balkhash Lake area, are explored, along with in-depth analyses of land use data and the GRACE dataset on water equivalent thickness, enhancing the study’s depth and reliability.Results: Drought affected approximately 44% of the Balkhash Lake Basin during the study period, impacting both Kazakhstan and China. Severe drought episodes occurred in 2000, 2008, 2014, and 2015, highlighting the region’s vulnerability. Analysis of drought trends revealed diverse patterns: 23% exhibited an increase, 17% showed a decrease, and 60% remained stable. Correlations between drought and hydrological parameters varied among stations, with positive correlations at Kapshagay and Shelek Stations, a weak correlation at Ayagoz Station, and a significant positive correlation at Lepsy despite the elevation.Discussion: This research underscores the intricate link between drought and hydrological factors in the Balkhash Lake Basin, emphasizing the need for precise water resource management and climate adaptation. Crucial strategies include proactive monitoring, tailored interventions, and the application of probabilistic drought mapping to enhance water supply management, contributing actionable insights for sustainable practices in the region.Conclusion: This study significantly advances our understanding of drought dynamics in the Balkhash Lake Basin, recommending adaptive strategies, site-specific interventions, and sustainable water management. The findings provide a crucial foundation for informed water resource decisions in Central Asia, emphasizing the importance of region-specific approaches to address diverse challenges posed by drought.

Thermosphere model assessment for geomagnetic storms from 2001 to 2023

We present an updated study for thermosphere model assessment under geomagnetic storm conditions, defined when the geomagnetic index ap = 80 or larger. Comparisons between five empirical models, NRLMSISE-00, JB2008, and three versions of DTM2020, and CHAMP, GRACE, GOCE, Swarm-A, and GRACE-FO neutral density data sets for 152 storms are presented. The storms are categorized according to ap, as single peak or multiple peaks. After applying a model debiasing procedure using the density data just before the onset of a storm, the models are on average only slightly biased, often only a few percent. This is an unexpected and reassuring result for these relatively simple models, which were fitted to different observations. The standard deviations of these averages are however up to 12% (1-sigma), which places the small biases into perspective. The smallest biases are achieved at the lowest altitude when comparing with GOCE data, and the highest for GRACE. The best results, i.e. smallest bias and standard deviation on average over all single-peak storms, over the entire 4-Phase storm period are obtained with DTM2020_Intermediate and DTM2020_Research models, while the oldest model, NRLMSISE-00, is the least precise. However, NRLMSISE-00 is the least biased when compared to multiple-peak storms. As could be expected, multiple-peak storms are reproduced with less precision than single-peak storms. The assessment reveals that model precision decreases with altitude, but that bias is independent of altitude, at least in the range covered by the data, 250–550 km.

Insight into land cover dynamics and water challenges under anthropogenic and climatic changes in the eastern Nile Delta: Inference from remote sensing and GIS data

The destabilization of delta's worldwide due to climate change and human activities presents challenges in meeting the growing demands for freshwater and food. The Nile Delta in Egypt is a prime example of a vulnerable region facing various stressors. In order to preserve land and water resources, it is crucial to monitor the spatial and temporal changes in Land Use/Land Cover (LULC), shoreline, and Terrestrial Water Storage (TWS) in these vulnerable regions This study comprehensively investigates the dynamic changes in LULC and their associated water and soil responses in the Eastern Nile Delta under these combined impacts. To achieve this goal, a combination of remote sensing techniques utilizing Landsat (5, 8, and 9), and GRACE datasets, along with field observations and Geographic Information System (GIS) tools, was employed. Accordingly, shoreline changes show coastal erosion rates ranging from 5.28 to 34.92 m/year due to climate change-induced SLR, with continued inland movement predicted for the next 20 years. Moreover, the dynamic changes in urbanization and alterations in agricultural cover have considerable penalties for water demand. Analysis of GRACE data indicates a notable reduction in average TWS by 77.89 mm between 2002 and 2017, with an annual rate, estimated at −5.821 mm/year. Soil sampling in highly vulnerable areas confirms agricultural degradation attributed to elevated salinity levels, with EC values ranging from 3.60 to 190 ds/m. These finds provide valuable insights for stakeholders and policymakers, to make reliable strategies regarding water allocation, land use regulations, and climate change adaptation in the worldwide vulnerable deltas.

PREDICTING THE IMPACT OF RAPID URBANISATION ON URBAN GROUNDWATER LEVEL VARIATIONS AND ANALYZING CLIMATE DYNAMICS IN PUNJAB'S CITIES, INDIA: AN APPROACH TOWARDS SUSTAINABLE CITIES

Abstract. In the past few decades, India has witnessed an unparallel population growth across its urban centres. The significant rise in urban population has led to acute demands for resources to sustain them, contributing to anthropogenic activities that impact urban resources and climate. This study was carried out to analyse the impact of rapid urbanisation on the groundwater level variation and climate dynamics. Focus was laid on Ludhiana and Patiala urban centres of Punjab state, India. Multi-temporal analysis of LULC was computed using Landsat-4, 5, 7 & 8 datasets to analyse urban sprawl during 1990–2023 for both the cities. A significant increase of 30% urban area was observed during 1990–2023. Further, to analyse the impact of urbanisation on ground water levels, the groundwater water storage variation was analysed using the GRACE and GRACE-FO mission datasets. It was found that the groundwater withdrawal was greater than the amount of total groundwater recharge during 1990–2023. Both these cities Ludhiana and Patiala has experienced an extreme decline in groundwater storage levels. Further, the time-series statistical relationships were explored to understand the relation between urban sprawl and annual temperature, rainfall patterns and total water storage for these cities during the years 2000-2022 for analysing the effect of rapid urbanization.

Quantifying basin-scale changes in groundwater storage using GRACE and one-way coupled hydrological and groundwater flow model in the data-scarce Bandung groundwater Basin, Indonesia

Assessing basin-scale groundwater storage changes is often difficult when groundwater table data are scarce. In this study, we quantify the groundwater storage changes by using the Wflow_sbm hydrological model coupled with the MODLOW groundwater model in the data-scarce area of the Bandung groundwater basin, Indonesia. The soil moisture storage change calculated by Wflow_sbm plus the groundwater storage change calculated by MODFLOW is compared to the water storage change estimated by the GRACE satellite between 2005 and 2015. The calculated cross-correlation coefficient is 0.502, and 62.1% of the simulated water storage change falls within GRACE's estimated uncertainty bounds. The important context in the water storage comparison are GRACE temporally local time-lags, data gaps in GRACE datasets, and differences in model seasonal performance and analyzed domain characteristics. The two latter factors highlight the importance of considering local groundwater-related information over large-scale global datasets in basin-scale groundwater storage change assessment. Based on the groundwater flow model, the current predicament of groundwater abstraction in the Bandung groundwater basin is highly unsustainable for future groundwater uses. On average, the groundwater storage in the study area is dwindling at the rate of 87 million m3/year between 2005 and 2018, dominantly a consequence of groundwater abstraction, whose effect is rippled to further impact the whole groundwater flow regime. Agreed by the situation shown by the limited data, the simulated groundwater table drawdown, spatially, is found to be locally and non-uniformly distributed. The capability of a one-way coupled hydrological and groundwater model to investigate basin-scale groundwater storage change, with comparable estimates to the GRACE dataset, unravel the opportunity of using such methods to estimate the behavior of future groundwater storage dynamics under the changing anthropogenic and climatic factors in catchment-scale studies.

Spatio-temporal dynamics of groundwater storage changes in the Yellow River Basin

Groundwater is an important source of water supply and ecosystem resilience. However, limited information on spatio-temporal dynamics makes a complete assessment of available groundwater resources difficult, impairing sustainable water management. The Gravity Recovery and Climate Experiment mission (GRACE) has recently made this possible. In this study, we used the Yellow River Basin (YRB) as a model system to explore the use of spatio-temporal dynamics information about groundwater change derived from the GRACE datasets for regional groundwater management. While there was an overall decreasing trend (R2 = 0.57) during the last 14 year, the groundwater storage over the whole basin decreased significantly (p < 0.0001, slope changed from −0.0137 cm/month to −0.0684 cm/month) since 2010 (2010–2016) and showed stronger fluctuations than the time before (2003–2009). The range and the standard deviation of groundwater storage change also increased in recent years especially after 2010. At the basin scale, locations which exhibited higher variabilities (large standard deviation) over time generally showed radical decrease of groundwater storage. The results indicated that groundwater depletion may reduce the aquifers’ function for ecosystem resilience, thus posing risks to the ecosystem of the YRB and threatening its people to climate change and extreme events. Despite the overall trend, the changes were heterogeneous if looking at finer scales: spatially, there was a gradual decline of storage from west to east (e.g. the change in December 2016 was −3.6, −9.1 and −25.8 cm for the upper, middle and lower reach, respectively); and temporally, the timeseries among the reaches were significantly different (p = 0.023). Our hotspot analysis also indicated the heterogeneity in groundwater decline across the basin and through the time. Additionally, it showed that human factors (e.g. groundwater consumption) become dominant in determining the groundwater change pattern over climatic variations. We therefore call for more attention to groundwater in developing sustainable water management strategies and suggest a closer cooperation of neighboring provinces in the YRB to have a reciprocal strategic plan for water regulation, protection, and management.

Groundwater Withdrawals Under Drought: Reconciling GRACE and Land Surface Models in the United States High Plains Aquifer

AbstractAdvanced Land Surface Models (LSM) offer a powerful tool for studying hydrological variability. Highly managed systems, however, present a challenge for these models, which typically have simplified or incomplete representations of human water use. Here we examine recent groundwater declines in the US High Plains Aquifer (HPA), a region that is heavily utilized for irrigation and that is also affected by episodic drought. To understand observed decline in groundwater and terrestrial water storage during a recent multiyear drought, we modify the Noah‐MP LSM to include a groundwater irrigation scheme. To account for seasonal and interannual variability in active irrigated area, we apply a monthly time‐varying greenness vegetation fraction (GVF) data set within the model. A set of five experiments were performed to study the impact of groundwater irrigation on the simulated hydrological cycle of the HPA and to assess the importance of time‐varying GVF when simulating drought conditions. The results show that including the groundwater irrigation scheme improves model agreement with ALEXI ET data, mascon‐based GRACE TWS data, and depth‐to‐groundwater measurements in the southern HPA, including Texas and Kansas, and that accounting for time‐varying GVF is important for model realism under drought. Results for the HPA in Nebraska are mixed, likely due to the model's weaknesses in representing subsurface hydrology in this region. This study highlights the value of GRACE data sets for model evaluation and development and the potential to advance the dynamic representations of the interactions between human water use and the hydrological cycle.

High-fidelity geometry models for improving the consistency of CHAMP, GRACE, GOCE and Swarm thermospheric density data sets

During the last two decades, accelerometers on board of the CHAMP, GRACE, GOCE and Swarm satellites have provided high-resolution thermosphere density data to improve our knowledge on atmospheric dynamics and coupling processes in the thermosphere-ionosphere region. Most users of the data have focused on relative density variations. Scale differences between datasets and models have been largely neglected or removed using ad hoc scale factors. The origin of these scale differences arises from errors in the aerodynamic modelling, specifically in the modelling of the satellite outer surface geometry and of the gas-surface interactions. Therefore, the first step to remove the scale differences is to enhance the geometry modelling. This work forms the foundation for the future improvement of characterization of satellite aerodynamics and gas-surface interactions models at TU Delft, as well as for extending the use of sideways and angular accelerations in the aerodynamic analysis of accelerations and derivation of thermosphere datasets. Although work to improve geometry and aerodynamic force models by other authors has focused on CHAMP and GRACE, this paper includes the GOCE and Swarm satellites as well. In addition, it uses a density determination algorithm that is valid for arbitrary attitude orientations, enabling a validation making use of attitude manoeuvres. The results show an improvement in the consistency of density data between these four missions, and of data obtained before, during and after attitude manoeuvres of CHAMP and Swarm. The new models result in larger densities, compared to the previously used panel method. The largest average rescaling of density, by switching to the new geometry models is reached for Swarm at 32%, the smallest for GRACE at 5%. For CHAMP and GOCE, mean differences of 11% and 9% are obtained respectively. In this paper, an overview of the improvements and comparisons of data sets is provided together with an introduction to the next research phase on the gas-surface interactions.

The Stalled Recovery of the Iraqi Marshes

The Iraqi (Mesopotamian) Marshes, an extensive wetlands system in Iraq, has been heavily impacted by both human and climate forces over the past decades. In the period leading up to the Second Gulf War in 2002, the marshlands were shrinking due to both a policy of draining and water diversion in Iraq and construction of dams upstream on the Tigris and Euphrates rivers. Following the war through 2006, this trend was reversed as the diversions were removed and active draining stopped. A combination of MODIS and GRACE datasets were used to determine the change in surface water area (SWA) in the marshes, marshland extent and change in mass both upriver in the Tigris and Euphrates watersheds and in the marshlands. Results suggest that the post war dam removal and decreased pumping in 2003 provided only temporary respite for the marshlands (2003–2006 SWA: 1,477 km2 increase (600%), water equivalent depth (WED): +2.0 cm/yr.; 2006–2009: −860 km2 (−41%) WED: −3.9 cm/yr.). Unlike in the period 2003–2006, from 2006 forward the mass variations in the marshes are highly correlated with those in the upper and middle watershed (R = 0.86 and 0.92 respectively), suggesting that any recovery due to that removal is complete, and that all future changes are tied more strongly to any climate changes that will affect recharge in the upper Tigris-Euphrates system. Precipitation changes in the watershed show a reduction of an average of 15% below the 15 yr mean in 2007–2011 This corresponds with published ensemble predictions for the 2071–2099 time period, that suggested similar marshland shrinkage should be expected in that time period.

PRM119 Mapping Between Composite Measures in Psoriatic Arthritis and the SF-6D: Analysis From the Grace Dataset

PRM119 Mapping Between Composite Measures in Psoriatic Arthritis and the SF-6D: Analysis From the Grace Dataset

Evaluation of high-frequency oceanographic signal in GRACE data: Implications for de-aliasing

[1] The sub-monthly portion of a recent daily-resolution GRACE data set (ITG-Grace2010) is evaluated over the ocean by comparing with the high-frequency component of sea level variability measured by satellite altimetry. The current Atmosphere and Ocean De-aliasing (AOD) model used to remove the high-frequency non-tidal ocean mass variations in GRACE data processing is also assessed. We demonstrate that the OMCT-based AOD model does not adequately represent the true sub-monthly variability in non-tidal ocean mass variability, but that ITG-Grace2010 does. The differences are not small, and indicate that a new model, or a data set derived from sub-monthly GRACE information, should be considered for de-aliasing in future GRACE processing.

Vertical deformations from homogeneously processed GRACE and global GPS long-term series

Temporal variations in the geographic distribution of surface mass cause surface displacements. Surface displacements derived from GRACE gravity field coefficient time series also should be observed in GPS coordinate time series, if both time series are sufficiently free of systematic errors. A successful validation can be an important contribution to climate change research, as the biggest contributors to mass variability in the system Earth include the movement of oceanic, atmospheric, and continental water and ice. In our analysis, we find that if the signals are larger than their precision, both geodetic sensor systems see common signals for almost all the 115 stations surveyed. Almost 80% of the stations have their signal WRMS decreased, when we subtract monthly GRACE surface displacements from those observed by GPS data. Almost all other stations are on ocean islands or small peninsulas, where the physically expected loading signals are very small. For a fair comparison, the data (79 months from September 2002 to April 2009) had to be treated appropriately: the GPS data were completely reprocessed with state-of-the-art models. We used an objective cluster analysis to identify and eliminate stations, where local effects or technical artifacts dominated the signals. In addition, it was necessary for both sets of results to be expressed in equivalent reference frames, meaning that net translations between the GPS and GRACE data sets had to be treated adequately. These data sets are then compared and statistically analyzed: we determine the stability (precision) of GRACE-derived, monthly vertical deformation data to be ~1.2 mm, using the data from three GRACE processing centers. We statistically analyze the mean annual signals, computed from the GPS and GRACE series. There is a detailed discussion of the results for five overall representative stations, in order to help the reader to link the displayed criteria of similarity to real data. A series of tests were performed with the goal of explaining the remaining GPS–GRACE residuals.

Analysis of gravity field variations derived from Superconducting Gravimeter recordings, the GRACE satellite and hydrological models at selected European sites

Abstract If we restrict the spatial resolution to a half-wavelength of about 1500 km and the temporal resolution to 1 month, GRACE-derived temporal gravity variations can be resolved within the μgal (10-8 m/s2) range. A comparison with ground gravity measurements from selected Superconducting Gravimeter (SG) stations forming the Global Geodynamics Project (GGP) provides an independent validation. For this study, five European SGstations were selected that both cover a large test field and allow closely located SG-stations to be studied. Prior to this comparison, GRACE and SG data sets have to be reduced for the same known gravity effects due to Earth and ocean tides, pole tide, and atmosphere. After these reductions, the remaining part can be mainly attributed to mass changes in terrestrial water storage. For this reason, gravity variations derived from global hydrological models are included in the comparison of SG and GRACE results. Conversely, the hydrology models can be checked by gravity variations determined from GRACE and SG observations. For most of the SG locations investigated here, the comparison based primarily on computed correlations shows quite a good agreement among the gravity variation derived from the three different kinds of data sets: SG, GRACE, and hydrology models. The variations in SG gravity (point measurements) prove to be representative for a large area within the μgal accuracy range, if local gravity effects are removed correctly. Additionally, a methodology for an analysis of dominant common features based on the EOF-technique is proposed and illustrated. The first principal component shows strong periodicity, and the search for arbitrary periods confirms a strong common annual component, which reduces the total signal content considerably. The first eigenvector reveals common features and differences between distinct SG stations. Discrepancies between SG, GRACE, and hydrology models at individual SG stations, detected by both methods, may provide valuable hints for further investigations of respective data series.

Estimation of steric sea level variations from combined GRACE and Jason-1 data

We estimate the mean steric sea level variations over the 60°S–60°N oceanic domain for the recent period (from August 2002 to April 2006), by combining sea level data from Jason-1 altimetry with time-variable gravity data from GRACE. The observed global mean sea level change from satellite altimetry results in total from steric plus ocean mass change. As GRACE measurements averaged over the ocean represents the ocean mass change component only, the difference between GRACE and altimetry observations provides an estimate of the mean steric sea level. Two different sets of GRACE geoid solutions (the GRGS EIGEN-GL04 and the GFZ EIGEN-GRACE04S products) have been used. Each GRACE data set ranges over approximately 3 yr or more (August 2002–April 2006 for the GRGS geoids and February 2003–February 2006 for the GFZ geoids).We first focus on the seasonal variations. The two GRACE data sets agree very well in terms of mean annual mass variation, both in amplitude (approximately 7.3 mm equivalent sea level) and phase (maximum on October 14). For both time spans, Jason-1 sea level minus GRACE ocean mass has an annual amplitude of approximately 5.8 mm with a maximum on March 11. The latter signal compares well with the steric annual sea level estimated from the World Ocean 2004 climatology and the Ishii et al. [M. Ishii, M. Kimoto, K. Sakamoto, and S.I. Iwasaki, Steric sea level changes estimated from historical ocean subsurface temperature and salinity analyses, Journal of Oceanography, 62 (2) (2006) 155–170.] ocean temperature data. We also examine the interannual fluctuations of the Jason-1 minus GRACE sea level. The two resulting steric sea level time series (based on the two GRACE data sets) agree well. The inferred steric sea level curve exhibits an increasing trend during the last 3.5 yr (August 2002–April 2006), of the same order of magnitude as the 1993–2003 steric sea level trend computed with in situ hydrographic data. However, over the last 3.5 yr, we note a strong discrepancy between altimetry minus GRACE and in situ-based steric sea level trend, the latter exhibiting a negative slope. The cause for such a discrepancy is yet unknown but may be related to inadequate sampling of in situ ocean temperature measurements.

Land water storage within the Congo Basin inferred from GRACE satellite gravity data

GRACE satellite gravity data is used to estimate terrestrial (surface plus ground) water storage within the Congo Basin in Africa for the period of April, 2002–May, 2006. These estimates exhibit significant seasonal (30 ± 6 mm of equivalent water thickness) and long‐term trends, the latter yielding a total loss of ∼280 km3 of water over the 50‐month span of data. We also combine GRACE and precipitation data sets (CMAP, TRMM) to explore the relative contributions of the source term to the seasonal hydrological balance within the Congo Basin. We find that the seasonal water storage tends to saturate for anomalies greater than 30–40 mm of equivalent water thickness. Furthermore, precipitation contributed roughly three times the peak water storage after anomalously rainy seasons, in early 2003 and 2005, implying a ∼60–70% loss from runoff and evapotranspiration. Finally, a comparison of residual land water storage (monthly estimates minus best‐fitting trends) in the Congo and Amazon Basins shows an anti‐correlation, in agreement with the “see‐saw” variability inferred by others from runoff data.