Comparison of Daily GRACE Gravity Field and Numerical Water Storage Models for De-aliasing of Satellite Gravimetry Observations

Reducing aliasing effects of insufficiently modelled high-frequent, non-tidal mass variations of the atmosphere, the oceans and the hydrosphere in gravity field models derived from the Gravity Recovery and Climate Experiment (GRACE) satellite mission is the topic of this study. The signal content of the daily GRACE gravity field model series (ITG-Kalman) is compared to high-frequency bottom pressure variability and terrestrially stored water variations obtained from recent numerical simulations from an ocean circulation model (OMCT) and two hydrological models (WaterGAP Global Hydrology Model, Land Surface Discharge Model). Our results show that daily estimates of ocean bottom pressure from the most recent OMCT simulations and the daily ITG-Kalman solutions are able to explain up to 40 % of extra-tropical sea-level variability in the Southern Ocean. In contrast to this, the daily ITG-Kalman series and simulated continental total water storage variability largely disagree at periods below 30 days. Therefore, as long as no adequate hydrological model will become available, the daily ITG-Kalman series can be regarded as a good initial proxy for high-frequency mass variations at a global scale. As a second result of this study, based on monthly solutions as well as daily observation residuals, it is shown that applying this GRACE-derived de-aliasing model supports the determination of the time-variable gravity field from GRACE data and the subsequent geophysical interpretation. This leads us to the recommendation that future satellite concepts for determining mass variations in the Earth system should be capable of observing higher frequeny signals with sufficient spatial resolution.     

Significance of secular trends of mass variations determined from GRACE solutions

Since 2002 the Earth’s gravity field is globally observed by the Gravity Recovery and Climate Experiment (GRACE) satellite mission. The GRACE monthly gravity field solutions, available from several analysis centres, reflect mass variations in the atmosphere, hydrosphere and geosphere. Due to correlated noise contained in these solutions, it is, however, first necessary to apply an appropriate filtering technique. The resulting, smoothed time series are applied not only to determine variations with different periodic signatures (e.g., seasonal, short and medium-term), but to derive long-periodic mass variations and secular trends as well. As the GRACE monthly solutions always show the integral effect of all mass variations, for separation of single processes, like the GIA (Glacial isostatic adjustment)-related mass increase in Fennoscandia, appropriate reduction models (e.g. from hydrology) are necessary.In this study we show for the example of the Fennoscandian uplift area that GRACE solutions from different analysis centres yield considerably different secular trends. Furthermore, it turns out that the inevitable filtering of the monthly gravity field models affects not only the amplitudes of the signals, but also their spatial resolution and distribution such as the spatial form of the detected signals. It also becomes evident that the determination of trends has to be performed together with the determination of periodic components. All periodic terms which are really contained in the data, and only such, have to be included. The restricted time span of the available GRACE measurements, however, limits the separation of long-periodic and secular signals. It is shown that varying the analysis time span affects the results considerably. Finally, a reduction of hydrological signals from the detected integral secular trends using global hydrological models (WGHM, LaDWorld, GLDAS) is attempted. The differences among the trends resulting from different models illustrate that the state-of-the-art hydrology models are not suitable for this purpose as yet. Consequently, taking the GRACE monthly gravity field solutions from one centre, choosing a single filter and applying an insufficiently reliable reduction model leads sometimes to a misinterpretation of considered geophysical processes. Therefore, one has to be cautious with the final interpretation of the results.     

联合星载GPS和KBRR星间测速数据反演GRACE时变重力场模型

高精度GRACE卫星时变重力场反演一直是卫星重力测量中的难题.为了恢复高精度的时变地球重力场模型,本文联合GRACE卫星的星载GPS和KBR星间测速观测数据,在对GRACE卫星进行精密定轨的同时,解算出60阶月平均地球重力场模型.通过对GRACE卫星的定轨精度、星载GPS相位和KBR星间测速数据的拟合残差以及时变地球重力场模型解算精度等分析,表明:(1)与美国宇航局喷气推进实验室(JPL)发布的约化动力学精密轨道相比,本文确定GRACE卫星轨道三维位置误差小于5 cm.(2)星载GPS相位数据拟合残差为5~8 mm,KBR星间测速数据拟合残差为0.18~0.30μm·s~(-1).(3)解算的月平均重力场模型与美国德克萨斯大学空间研究中心(CSR)、德国地学研究中心(GFZ)和JPL发布的RL05模型精度接近,时变信号在全球范围内具有很好的空间分布一致性.通过计算亚马逊流域和长江流域的水储量变化,本文与上述三个机构的计算结果无明显差异,且相关系数均达0.9以上.可见,本文建立的卫星轨道与重力场同解算法具有反演高精度GRACE时变重力场能力,为我国卫星重力场反演提供了重要的技术支持.     

Calibration/Data Assimilation Approach for Integrating GRACE Data into the WaterGAP Global Hydrology Model (WGHM) Using an Ensemble Kalman Filter: First Results

We introduce a new ensemble-based Kalman filter approach to assimilate GRACE satellite gravity data into the WaterGAP Global Hydrology Model. The approach (1) enables the use of the spatial resolution provided by GRACE by including the satellite observations as a gridded data product, (2) accounts for the complex spatial GRACE error correlation pattern by rigorous error propagation from the monthly GRACE solutions, and (3) allows us to integrate model parameter calibration and data assimilation within a unified framework. We investigate the formal contribution of GRACE observations to the Kalman filter update by analysis of the Kalman gain matrix. We then present first model runs, calibrated via data assimilation, for two different experiments: the first one assimilates GRACE basin averages of total water storage and the second one introduces gridded GRACE data at \(5^\circ\) resolution into the assimilation. We finally validate the assimilated model by running it in free mode (i.e., without adding any further GRACE information) for a period of 3 years following the assimilation phase and comparing the results to the GRACE observations available for this period.     

Tackling mass redistribution phenomena by time-dependent GRACE- and terrestrial gravity observations

Time variable gravity field models derived from the satellite mission GRACE have been demonstrated to be consistent with water mass variations in the global hydrological cycle. Independent observations are provided by terrestrial measurements. In order to achieve a maximum of reliability and information gain, ground-based gravity observations may be deployed for comparison with the gravity field variations derived from the GRACE satellite mission. In this context, the data of the network of superconducting gravimeters (SG) of the ‘Global Geodynamics Project’ (GGP) are of particular interest. This study is focused on the dense SG network in Central Europe with its long-term gravity observations. It is shown that after the separation and reduction of local hydrological effects in the SG observations especially for subsurface stations, the time-variable gravity signals from GRACE agree well with the terrestrial observations from the SG station cluster.Station stability of the SG sites with respect to vertical deformations was checked by GNSS based observations. Most of the variability can be explained by loading effects due to changes in continental water storage, and, in general, the stability of all stations has been confirmed.From comparisons based on correlation and coherence analyses in combination with the root mean square (RMS) variability of the time series emerges, that the maximum correspondence between the SG and GRACE time series is achieved when filtering the GRACE data with Gaussian filters of about 1000 km filter length, which is in accordance with previous publications.Empirical Orthogonal Functions (EOF) analysis was applied to the gravity time series in order to identify common characteristic spatial and temporal patterns. The high correspondence of the first modes for GRACE and SG data implies that the first EOF mode represents a large-scale (Central European) time-variable gravity signal seen by both the GRACE satellites and the SG cluster.     

Retrieval of Large-Scale Hydrological Signals in Africa from GRACE Time-Variable Gravity Fields

Since its launch in April 2002, the Gravity Recovery and Climate Experiment (GRACE) mission is recording the Earth’s time-variable gravity field with temporal and spatial resolutions of typically 7–30?days and a few hundreds of kilometers, allowing the monitoring of continental water storage variations from both continental and river-basin scales. We investigate here large scale hydrological variations in Africa using different GRACE spherical harmonic solutions, using different processing strategies (constrained and unconstrained solutions). We compare our GRACE estimates to different global hydrology models, with different land-surface schemes and also precipitation forcing. We validate GRACE observations through two different techniques: first by studying desert areas, providing an estimate of the precision. Then we compare GRACE recovered mass variations of main lakes to volume changes derived from radar altimetry measurements. We also study the differences between different publicly available precipitation datasets from both space measurements and ground rain gauges, and their impact on soil-moisture estimates.     

基于变差函数的插值方法计算华北地区重力场变化

定点重复重力测量是获取区域重力场变化的主要手段之一, 重力场特征与地形起伏、 构造走向等因素相关。 以华北地区为例, 考虑区域内地形和构造的北东向分布规律, 从EGM2008重力模型中拟合各向异性变差函数参数, 利用变差函数网格化插值, 对华北地区2009—2013年期间重力场观测数据进行网格化重建, 获取华北地区重力场时空变化结果。 研究结果表明基于区域地形、 构造特征的各向异性变差函数插值方法, 获得的空间重力场变化在重力异常梯级带上更加明显, 重力变化与活动构造分布具有更好的一致性。 本文研究方法对于恢复区域时空重力场异常具有重要意义, 有助于提高应用重力资料划分潜在地震危险区空间位置的精度, 为华北地区震情研判和构建地震预报定量指标体系提供可靠的地球物理场数据。     

Gravity field variations from superconducting gravimeters for GRACE validation

The network of superconducting gravimeters (SG) of the ‘Global Geodynamics Project’ (GGP) offers the unique opportunity to supplement and validate the gravity field variations derived from the GRACE satellite mission. Because of the different spatial and temporal resolutions of the gravity data a combination of all datasets can be used to retrieve a maximum of information regarding mass transfers especially related to hydrology which is deployable as constraint for hydrological modelling.For a consistent combination of the datasets the gap between terrestrial data of superconducting and absolute gravimeters (AG) and from satellite data has to be bridged. A successful combination of SG and AG data could be realized for several stations which resulted in time series of the highest accuracy and long-term stability.In principle, the same reductions applied to GRACE data have to be taken into account for the terrestrial data. The separation of local hydrological effects in SG observations is crucial for the comparison with satellite-derived gravity data. It is shown that even for stations with a hydrological challenging situation such as Moxa/Germany local hydrology-induced effects can be successfully modelled.Currently, the study focuses on Europe with its dense and long-term observation network. Regarding the consistency of the SG gravity variations they are representative for a larger region. From a comparison with GRACE-derived gravity field changes, and the variations due to hydrological models a principle good agreement emerges.     

利用动力学方法解算GRACE时变重力场研究

本文利用动力学方法建立GRACE(Gravity Recovery And Climate Experiment)K波段距离变率(KBRR)观测、轨道观测与重力场系数的观测方程,通过GRACE Level 1B观测数据,成功解算出全球月时变重力场模型——IGG时变重力场模型,并将2008—2009年的解算结果与GRACE三大数据处理机构美国德克萨斯大学空间中心CSR(Center for Space Research)、美国宇航局喷气推进实验室JPL(Jet Propulsion Laboratory)和德国地学研究中心GFZ(GeoForschungs Zentrum)发布的最新全球时变重力场模型进行详细对比分析.结果表明:IGG结果在全球质量异常、中国及周边地区质量异常的趋势变化、全球质量异常均方差、2~60每阶位系数差值以及亚马逊流域和撒哈拉沙漠等典型区域平均质量异常等方面与CSR、JPL和GFZ解算的RL05结果较为一致.其中,IGG解算结果在2~20阶与CSR、GFZ和JPL最新解算结果基本一致,20~40阶IGG解算结果与GFZ、JPL单位最新解算结果较为接近,大于40阶IGG结果介于CSR与GFZ、JPL之间;亚马逊流域平均质量异常周年振幅IGG、CSR、GFZ和JPL获取到的结果分别为17.6±1.1cm、18.9±1.2cm、17.8±0.9cm和18.9±1.0cm等效水柱高.利用撒哈拉沙漠地区的平均质量异常做反演精度评定,IGG、CSR、GFZ和JPL的时变重力场获取到的平均质量异常均方差分别为1.1cm、0.9cm、0.8cm和1.2cm,表明IGG解算结果与CSR、GFZ和JPL最新发布的RL05结果在同一精度水平.     

Separation of mass signals within GRACE monthly gravity field models by means of empirical orthogonal functions

Since 2002 the two GRACE satellites observe the time varying gravity signal mainly caused by the sum of mass variations within the Earth subsystems ocean, atmosphere, and continental hydrosphere. It is a challenging problem to separate the integral GRACE signal and to identify and quantify the mass variations of the individual subsystems. This work proves first by a closed loop simulation that such a decomposition is successful by means of empirical orthogonal functions (EOF) derived from geophysical models and a least-squares adjustment with a multivariate Gauss–Markov model with time coefficients parameterized. The geophysical models are used to synthesize GRACE observations which are subsequently separated leading to time coefficients coinciding with those of the predefined models. In a second step the separation is performed with real, unfiltered time series of 5 years of monthly GRACE gravity field models (with atmospheric and oceanic background models reconstructed) and a limited number of EOFs. The reconstructed time coefficients are in good agreement with the original ones and exhibit high correlations (0.70 for ocean, 0.91 for atmosphere and 0.93 for continental hydrosphere). Analysis of GRACE residuals and the correlation among the time coefficients substantiate a successful identification.     

利用径向基函数RBF解算GRACE全球时变重力场

本文利用GRACE(Gravity Recovery And Climate Experiment)level 1b数据和径向基函数RBF(radial basis function)方法解算了全球时变地球重力场.RBF基函数相比传统球谐(spherical harmonic)基函数,其高度的空域局部特性使得正则化过程易于添加先验协方差信息,从而可能揭示更加准确的重力场信号.本文研究表明,RBF基函数算法在精化现有的GRACE全球时变重力场模型,如提升部分区域信号幅度等方面具有一定优势.本文通过将RBF的尺度因子作为待解参数,基于GRACE卫星的Level 1b数据和变分方程法,成功获取了2009-2010年90阶无约束全球时变重力场RBF模型Hust-IGG03,以及正则化全球时变重力场RBF模型Hust-IGG04.通过与GRACE官方数据处理中心GFZ发布的最新90阶球谐基时变模型RL05a进行对比,结果表明:(1)无约束RBF模型Hust-IGG03和GFZ RL05a在空域和频域表现基本一致;(2)正则化RBF模型Hust-IGG04无需进行后处理滤波已经显示较高信噪比,噪音水平接近于球谐基模型GFZ RL05a经400 km高斯滤波后的效果;(3)HustIGG04相比400 km高斯滤波GFZ RL05a在周年振幅图和趋势图上显示出更多的细节信息,并且呈现出更强的信号幅度,如在格陵兰冰川融化趋势估计上Hust-IGG04比GFZ RL05a提高了24.2%.以上结果均显示RBF方法有助于进一步挖掘GRACE观测值所包含的时变重力场信息.     

Seismologic applications of GRACE time-variable gravity measurements

The Gravity Recovery and Climate Experiment (GRACE) has been measuring temporal and spatial variations of mass redistribution within the Earth system since 2002. As large earthquakes cause significant mass changes on and under the Earth’s surface, GRACE provides a new means from space to observe mass redistribution due to earthquake deformations. GRACE serves as a good complement to other earthquake measurements because of its extensive spatial coverage and being free from terrestrial restriction. During its over 10 years mission, GRACE has successfully detected seismic gravitational changes of several giant earthquakes, which include the 2004 Sumatra–Andaman earthquake, 2010 Maule (Chile) earthquake, and 2011 Tohoku-Oki (Japan) earthquake. In this review, we describe by examples how to process GRACE time-variable gravity data to retrieve seismic signals, and summarize the results of recent studies that apply GRACE observations to detect co- and post-seismic signals and constrain fault slip models and viscous lithospheric structures. We also discuss major problems and give an outlook in this field of GRACE application.     

基于GRACE的空间约束方法监测华北平原地下水储量变化

华北平原作为我国重要的工农业基地和政治经济中心,面临着严重的水资源危机.因此,开展对华北平原地下水储量变化的监测工作具有重要现实意义与科学价值.本文基于GRACE重力卫星的空间约束方法,研究了华北平原地下水储量变化的时空分布规律,并与地面水井实测与地下水模型结果进行了综合比较和分析.结果表明:2002-2014年,华北平原地下水存在明显的长期亏损,GRACE估计的亏损速率为-7.4±0.9 km~3·a~(-1),而地面水井资料估计的浅层地下水亏损速率为-1.2 km~3·a~1,对比两者之间的差异可以发现,华北平原的地下水亏损以深层地下水为主.2002-2008年,GRACE估计的华北平原地下水亏损速率为-5.3±2.2 km~3·a~(-1),这与华北平原两个地下水模型得到的平均亏损速率-5.4 km~3·a~(-1)十分吻合.通过华北平原区域地下水模型的独立验证,说明GRACE可以有效评估华北平原的地下水储量变化趋势.除了长期亏损的趋势项之外,华北平原地下水还存在明显的年际变化特征,并与该地区年降雨量变化特征一致.在降雨偏少的2002年、2005-2009年和2014年,华北平原地下水储量显著减少.在空间分布上,GRACE结果表明,华北平原的地下水储量减少主要发生在山前平原和中部平原区,这也与水井实测资料和区域地下水模型结果较为吻合.与GRACE和区域地下水模型相比,目前的全球水文模型仍无法准确估计华北平原地下水变化的空间分布和亏损速率.上述研究表明,GRACE提供了评估华北平原地下水储量变化的重要监测手段.     

Seasonal Water Storage Variations as Impacted by Water Abstractions: Comparing the Output of a Global Hydrological Model with GRACE and GPS Observations

Better quantification of continental water storage variations is expected to improve our understanding of water flows, including evapotranspiration, runoff and river discharge as well as human water abstractions. For the first time, total water storage (TWS) on the land area of the globe as computed by the global water model WaterGAP (Water Global Assessment and Prognosis) was compared to both gravity recovery and climate experiment (GRACE) and global positioning system (GPS) observations. The GRACE satellites sense the effect of TWS on the dynamic gravity field of the Earth. GPS reference points are displaced due to crustal deformation caused by time-varying TWS. Unfortunately, the worldwide coverage of the GPS tracking network is irregular, while GRACE provides global coverage albeit with low spatial resolution. Detrended TWS time series were analyzed by determining scaling factors for mean annual amplitude (f _GRACE) and time series of monthly TWS (f _GPS). Both GRACE and GPS indicate that WaterGAP underestimates seasonal variations of TWS on most of the land area of the globe. In addition, seasonal maximum TWS occurs 1 month earlier according to WaterGAP than according to GRACE on most land areas. While WaterGAP TWS is sensitive to the applied climate input data, none of the two data sets result in a clearly better fit to the observations. Due to the low number of GPS sites, GPS observations are less useful for validating global hydrological models than GRACE observations, but they serve to support the validity of GRACE TWS as observational target for hydrological modeling. For unknown reasons, WaterGAP appears to fit better to GPS than to GRACE. Both GPS and GRACE data, however, are rather uncertain due to a number of reasons, in particular in dry regions. It is not possible to benefit from either GPS or GRACE observations to monitor and quantify human water abstractions if only detrended (seasonal) TWS variations are considered. Regarding GRACE, this is mainly caused by the attenuation of the TWS differences between water abstraction variants due to the filtering required for GRACE TWS. Regarding GPS, station density is too low. Only if water abstractions lead to long-term changes in TWS by depletion or restoration of water storage in groundwater or large surface water bodies, GRACE may be used to support the quantification of human water abstractions.     

Wiener optimal filtering of GRACE data

We present a spatial averaging method for Gravity Recovery and Climate Experiment (GRACE) gravity-field solutions based on the Wiener optimal filtering. The optimal filter is designed from the least-square minimization of the difference between the desired and filtered signals. It requires information about the power spectra of the desired gravitational signal and the contaminating noise, which is inferred from the average GRACE degree-power spectrum. We show that the signal decreases with increasing spherical harmonic degree j with approximately j^−b, where b = 1.5 for GRACE data investigations. This is termed the Second Kaula rule of thumb for temporal variations of the Earth’s gravity field. The degree power of the noise increases, in the logarithmic scale, linearly with increasing j. The Wiener optimal filter obtained for the signal model with b = 1.5 closely corresponds to a Gaussian filter with a spatial half width of 4° (∼440 km). We find that the filtered GRACE gravity signal is relatively insensitive to the exponent b of the signal model, which indicates the robustness of Wiener optimal filtering. This is demonstrated using the GFZ-GRACE gravity-field solution for April 2004.     

GOCE, Satellite Gravimetry and Antarctic Mass Transports

In 2009 the European Space Agency satellite mission GOCE (Gravity Field and Steady-State Ocean Circulation Explorer) was launched. Its objectives are the precise and detailed determination of the Earth’s gravity field and geoid. Its core instrument, a three axis gravitational gradiometer, measures the gravity gradient components V _xx , V _yy , V _zz and V _xz (second-order derivatives of the gravity potential V) with high precision and V _xy , V _yz with low precision, all in the instrument reference frame. The long wavelength gravity field is recovered from the orbit, measured by GPS (Global Positioning System). Characteristic elements of the mission are precise star tracking, a Sun-synchronous and very low (260 km) orbit, angular control by magnetic torquing and an extremely stiff and thermally stable instrument environment. GOCE is complementary to GRACE (Gravity Recovery and Climate Experiment), another satellite gravity mission, launched in 2002. While GRACE is designed to measure temporal gravity variations, albeit with limited spatial resolution, GOCE is aiming at maximum spatial resolution, at the expense of accuracy at large spatial scales. Thus, GOCE will not provide temporal variations but is tailored to the recovery of the fine scales of the stationary field. GRACE is very successful in delivering time series of large-scale mass changes of the Antarctic ice sheet, among other things. Currently, emphasis of respective GRACE analyses is on regional refinement and on changes of temporal trends. One of the challenges is the separation of ice mass changes from glacial isostatic adjustment. Already from a few months of GOCE data, detailed gravity gradients can be recovered. They are presented here for the area of Antarctica. As one application, GOCE gravity gradients are an important addition to the sparse gravity data of Antarctica. They will help studies of the crustal and lithospheric field. A second area of application is ocean circulation. The geoid surface from the gravity field model GOCO01S allows us now to generate rather detailed maps of the mean dynamic ocean topography and of geostrophic flow velocities in the region of the Antarctic Circumpolar Current.     

基于GRACEKBRR数据的动力积分法反演时变重力场模型

基于动力积分法恢复了一组60阶的时变重力场模型WHU-Grace01s,且在位系数解算过程中仅使用KBRR数据.通过与CSR、GFZ和JPL发布的Release 05模型的阶方差和位系数误差谱对比可知,WHU-Grace01s模型在高阶次部分的阶方差较小,且对轨道共振现象不敏感.将WHU-Grace01s时变重力场模型与CSR、GFZ、JPL、DEOS、Tongji、ITG、AIUB和GRGS等8家机构发布模型通过相同的滤波处理,获得了全球地表质量变化的时空分布,从结果可以看出:各个模型计算的时变信号在空域上分布十分接近,且WHU-Grace01s模型计算的太平洋中心和撒哈拉沙漠区域的质量变化较小;对比几个典型质量变化区域,WHU-Grace01s模型和JPL模型计算的长江流域和珠江流域时变信号呈强相关,其相关系数分别为0.948和0.976,且与上述8个模型计算的两个流域时变信号的相关系数均达到0.9以上;在南极区域和格陵兰岛,WHU-Grace01s模型和其他各个模型均能反映区域冰川质量的积累或消融,且各模型计算获得的长期趋势变化结果相当.研究结果表明,WHU-Grace01s模型和国内外已发布机构模型具有很好的一致性,且受到轨道共振影响较小.     

利用GPS与GRACE监测陆地水负荷导致的季节性水平形变：以喜马拉雅山地区为例

地表陆地水负荷变化是引起重力场和地壳形变呈现季节性特征的主要因素,并且能够利用地表及空间大地测量技术对其进行有效的监测.本文通过对质量负荷形变效应的理论模拟,描述了水平分量的形变指向以及垂直与水平分量的幅值比可以提高对负荷区域的辨别程度,并且联合GPS坐标时间序列及GRACE模型对喜马拉雅山地区的季节性负荷形变进行了详细对比分析,研究结果显示两者垂直分量的季节性变化具有较好的一致性,且GPS周年项幅值要大于GRACE.而由GRACE解算得到的水平分量结果表明该地区季节性形变主要受东南亚及印度东北部地区的陆地水负荷控制,位于喜马拉雅山地区多数GPS台站的垂直分量及北向分量的初相位与GRACE模型解算结果相近,而部分GPS台站的东向分量与GRACE模型存在明显不同,由此导致GPS与GRACE监测到的形变指向存在差异.通过对GRACE估算精度以及GPS垂直与水平分量幅值比的深入分析,发现GPS对局部周边地区的河流、谷地及农田灌溉等负荷变化造成的形变效应较为敏感,而GRACE由于截断阶次及平滑滤波等影响因素,不仅造成在水平分量上的分辨率远低于垂直分量,而且整体估算精度要低于GPS观测得到的形变信息.     

Multivariate Prediction of Total Water Storage Changes Over West Africa from Multi-Satellite Data

West African countries have been exposed to changes in rainfall patterns over the last decades, including a significant negative trend. This causes adverse effects on water resources of the region, for instance, reduced freshwater availability. Assessing and predicting large-scale total water storage (TWS) variations are necessary for West Africa, due to its environmental, social, and economical impacts. Hydrological models, however, may perform poorly over West Africa due to data scarcity. This study describes a new statistical, data-driven approach for predicting West African TWS changes from (past) gravity data obtained from the gravity recovery and climate experiment (GRACE), and (concurrent) rainfall data from the tropical rainfall measuring mission (TRMM) and sea surface temperature (SST) data over the Atlantic, Pacific, and Indian Oceans. The proposed method, therefore, capitalizes on the availability of remotely sensed observations for predicting monthly TWS, a quantity which is hard to observe in the field but important for measuring regional energy balance, as well as for agricultural, and water resource management. Major teleconnections within these data sets were identified using independent component analysis and linked via low-degree autoregressive models to build a predictive framework. After a learning phase of 72 months, our approach predicted TWS from rainfall and SST data alone that fitted to the observed GRACE-TWS better than that from a global hydrological model. Our results indicated a fit of 79 % and 67 % for the first-year prediction of the two dominant annual and inter-annual modes of TWS variations. This fit reduces to 62 % and 57 % for the second year of projection. The proposed approach, therefore, represents strong potential to predict the TWS over West Africa up to 2 years. It also has the potential to bridge the present GRACE data gaps of 1 month about each 162 days as well as a—hopefully—limited gap between GRACE and the GRACE follow-on mission over West Africa. The method presented could also be used to generate a near-real-time GRACE forecast over the regions that exhibit strong teleconnections.