Continental waterstorage plays a major role in Earth's climate system. However, temporal and spatial variations of continental water are poorly known, particularly in Africa. Gravity Recovery and Climate Experiment (GRACE) satellite mission provides an opportunity to estimate terrestrial waterstorage (TWS) variations at both continental and river-basin scales. In this paper, seasonal and secular variations of TWS within Africa for the period from January 2003 to July 2013 are assessed using monthly GRACE coefficients from three processing centers (Centre for Space Research, the German Research Centre for Geo- sciences, and NASA's Jet Propulsion Laboratory). Monthly grids from Global Land Data Assimilation System (GLDAS)-1 and from the Tropical Rainfall Measuring Mission (TRMM)- 3B43 models are also used in order to understand the reasons of increasing or decreasing waterstorage. Results from GRACE processing centers show similar TWS estimates at seasonal timescales with some differences concerning inter-annual trend variations. The largest annual signals of GRACE TWS are observed in Zambezi and Okavango River basins and in Volta River Basin. An increasing trend of 11.60 mm/a is found in Zambezi River Basin and of 9 mm/a in Volta River Basin. A phase shift is found between rainfall and GRACE TWS (GRACE TWS is preceded by rainfall) by 2e3 months in parts of south central Africa. Comparing GLDAS rainfall with TRMM model, it is found that GLDAS has a dry bias from TRMM model.
A priori corrections of GRCTellus ensemble mean GRACE signals using a set of LSM-derived scaling factors (i.e. amplitude gain) can lead to substantial uncertainty in 1TWS (Long et al., 2015). We show that the amplitude of simulated terrestrial water mass over the Upper Nile Basin varies substantially among various LSMs (see Fig. S15). Most of these LSMs (GLDAS models: CLM, NOAH, VIC) do not include surface water or groundwater storage (Scanlon et al., 2012). Although CLM (v.4.0 and 4.5) includes a simple representation (i.e. shallow unconfined aquifer) of ground- water (Niu et al., 2007; Oleson et al., 2008), it does not con- sider recharge from irrigation return flows. In addition, many of these LSMs do not consider lakes and reservoirs and, most critically, LSMs are not reconciled with in situ observations. The combined measurement and leakage errors, √ (bias 2 + leak 2 ) (Swenson and Wahr, 2006) for GRCTellus 1TWS based on CLM4.0 model for the LVB and LKB are 7.2 and 6.6 cm respectively. These values, however, do not represent mass leakage from the lake to the surrounding area within the basin itself. A sensitivity analysis of GRCTellus and GRGS signals reveal that signal leakage occurs from lake to its sur- rounding basin area as well as between basins. For instance, GRACE signal leakage into the LKB from the LVB, which is 3 times larger in area than the LKB, is 3.4 times bigger for both GRCTellus GRACE and GRGS products. Further- more, the analysis shows that leakage from Lake Victoria to the LVB for GRCTellus is substantially greater than GRGS product by a factor of ∼ 2.6. In other words, 1 mm change in the level of Lake Victoria represents an equivalent change of 0.12 mm in 1TWS in the LVB for GRCTellus compared to 0.32 mm for GRGS. Consequently, changes in the amplitude of GRGS 1TWS are much greater ( ∼ 38 %) than GRCTel- lus. During the observed reduction in 1TWS (83 km 3 ) from 2003 to 2006, the computed volumetric reduction for GRGS is found to be 69 km 3 whereas it is 31 km 3 for GRCTellus.
Abstract. We investigate the interannual and interdecadal hydrological changes in the Amazon River basin and its sub- basins during the 1980–2015 period using GRACE satellite data and a physically based, 2 km grid continental-scale hy- drological model (LEAF-Hydro-Flood) that includes a prog- nostic groundwater scheme and accounts for the effects of land use–land cover (LULC) change. The analyses focus on the dominant mechanisms that modulate terrestrial waterstorage (TWS) variations and droughts. We find that (1) the model simulates the basin-averaged TWS variations remark- ably well; however, disagreements are observed in spatial patterns of temporal trends, especially for the post-2008 pe- riod. (2) The 2010s is the driest period since 1980, charac- terized by a major shift in the decadal mean compared to the 2000s caused by increased drought frequency. (3) Long-term trends in TWS suggest that the Amazon overall is getting wetter (1.13 mm yr −1 ), but its southern and southeastern sub- basins are undergoing significant negative TWS changes, caused primarily by intensified LULC changes. (4) Increas- ing divergence between dry-season total water deficit and TWS release suggests a strengthening dry season, especially in the southern and southeastern sub-basins. (5) The sub- surface storage regulates the propagation of meteorological droughts into hydrological droughts by strongly modulating TWS release with respect to its storage preceding the drought condition. Our simulations provide crucial insight into the importance of sub-surface storage in alleviating surface wa- ter deficit across Amazon and open pathways for improving prediction and mitigation of extreme droughts under chang-
Abstract. The amount of water stored on continents is an important constraint for water mass and energy ex- changes in the Earth system and exhibits large inter-annual variability at both local and continental scales. From 2002 to 2017, the satellites of the Gravity Recovery and Climate Experiment (GRACE) mission have observedchanges in terrestrial waterstorage (TWS) with an unprecedented level of accuracy. In this paper, we use a statistical model trained with GRACE observations to reconstruct past climate-driven changes in TWS from his- torical and near-real-time meteorological datasets at daily and monthly scales. Unlike most hydrological models which represent water reservoirs individually (e.g., snow, soil moisture) and usually provide a single model run, the presented approach directly reconstructs total TWS changes and includes hundreds of ensemble members which can be used to quantify predictive uncertainty. We compare these data-driven TWS estimates with other independent evaluation datasets such as the sea level budget, large-scale water balance from atmospheric re- analysis, and in situ streamflow measurements. We find that the presented approach performs overall as well or better than a set of state-of-the-art global hydrological models (Water Resources Reanalysis version 2). We provide reconstructed TWS anomalies at a spatial resolution of 0.5 ◦ , at both daily and monthly scales over the period 1901 to present, based on two different GRACE products and three different meteorological forcing datasets, resulting in six reconstructed TWS datasets of 100 ensemble members each. Possible user groups and applications include hydrological modeling and model benchmarking, sea level budget studies, assessments of long-term changes in the frequency of droughts, the analysis of climate signals in geodetic time series, and the interpretation of the data gap between the GRACE and GRACE Follow-On missions. The presented dataset is published at https://doi.org/10.6084/m9.figshare.7670849 (Humphrey and Gudmundsson, 2019) and updates will be published regularly.
The dynamic approach is used by the three official processing centers and by the Graz Univer- sity of Technology (ITSG-Grace2014, ITSG-Grace2016, ITSG-Grace2018; Mayer-Gürr et al., 2016, 2018), the GNSS Research Center of Wuhan University (WHU-Grace01s; Zhou et al., 2015), the Huazhong University of Science and Technology (HUST-Grace2016; Zhou et al., 2016), the Institute of Geodesy and Geophysics at the Chinese Academy of Sciences (IGG- RL01; Wang et al., 2015), the Leibniz University Hannover (LUH-GRACE2018; Naeimi et al., 2018), the Faculty of Geoscience and Environment Engineering at Southwest Jiaotong Univer- sity (SWJTU-Grace-RL01), and the Space Geodesy Research Group (CNES/GRGS; Lemoine et al., 2007; Bruinsma et al., 2010). ITSG and WHU-Grace2016 provide monthly solutions for d/o 60, 90, and 120. ITSG solutions also include daily solutions, which are estimated within an Kalman smoother framework using temporal correlation patterns derived from geophyical models (Kurtenbach et al., 2012). Additionally, ITSG-Grace2016 is provided with full covari- ance information. Besides the three official processing centers, ITSG-Grace2018 is currently the only publicly available solution that uses the AOD1B-RL06 product instead of AOD1B- RL05. Unlike other solutions, the CNES/GRGS-RL03 gravity fields are constrained towards a mean field, which stabilizes the solutions and makes filtering unnecessary (indicated by the ** in Table 2.1).
Global climate models (GCMs) still plays a vital role for the assessment of climate variability and change, more so in regions where in situ data are not well documented. However, at a scale of 100-250 km, current GCMs only have the potential to simulate the main characteristics of general circulation at the range of that scale (Shongwe et al. 2011). To that extent, GCMs are not necessarily capable of capturing the detailed processes associated with regional to local climate variability and changes that are required for regional and national climate change assessments (Denis et al. 2002; Wang et al. 2004; Knutti et al. 2005; Giorgi et al. 2009). This is particularly so for heterogeneous regions such as the GHA region, where subgrid scale variations in topography, vegetation, soils, and coastlines have a significant effect on the climate (Mutail et al. 2000). In addition, at coarse grid resolutions, the magnitude and intensity of subgrid-scale extreme events such as heavy rainfall are often not captured, nor realistically reproduced. Generally, GCM data have been used to describe the climate processes of many African regions and to produce the climate information for applications in different socioeconomic sectors including agriculture, water, and health (e.g Alley et al. 2007; Anyah and Qiu 2012; Otieno and Anyah 2013). However, in order to formulate adaptation policies in response to climate change impacts, reliable climate change information is usually required at finer spatial scales than that of a typical GCM.
Infiltration can be measured using an infiltrometer, which is a wide diameter tube surrounding an area of soil (Fig.2.3). It is generally provided with an outer ring also. The rings are filled to a certain depth and continually refilled to maintain the depth and the inflow measured, which gives the infiltration. The purpose of the outer ring is to eliminate to some extent the edge effect of the surrounding dryer soil. Another way to measure infiltration is to simulate rainfall by a sprinkler and collect the runoff from the plot. The difference between the water supplied to the sprinkler and the runoff collected is assumed to have infiltrated. In practice, approximation of infiltration losses can be made using infiltration indices. Of the various indices, the Φ-index is the simplest to use. The Φ-index is defined as the average rainfall intensity above which the volume of rainfall equals the volume of runoff 6. Although the Φ-index does not account for the change in infiltration rate with time
Abstract. We present a new conceptual scheme of the inter- action between unsaturated and saturated zones of the MO- BIDIC (MOdello Bilancio Idrologico DIstributo e Continuo) hydrological model which is applicable to shallow water ta- ble conditions. First, MODFLOW was coupled to MOBIDIC as the physically based alternative to the conceptual ground- water component of the MOBIDIC–MODFLOW. Then, as- suming a hydrostatic equilibrium moisture profile in the un- saturated zone, a dynamic specific yield that is dependent on the water table level was added to MOBIDIC–MODFLOW, and calculation of the groundwater recharge in MOBIDIC was revisited using a power-type equation based on the infil- tration rate, soil moisture deficit, and a calibration parameter linked to the initial water table depth, soil type, and rainfall intensity. Using the water table fluctuation (WTF) method for a homogeneous soil column, the parameter of the proposed groundwater recharge equation was determined for four soil types, i.e. sand, loamy sand, sandy loam, and loam under a pulse of rain with different intensities. The fidelity of the in- troduced modifications in MOBIDIC–MODFLOW was as- sessed by comparison of the simulated water tables against those of MIKE SHE, a physically based integrated hydro- logical modelling system simulating surface and groundwa- ter flow, in two numerical experiments: a two-dimensional case of a hypothetical watershed in a vertical plane (constant slope) under a 1 cm d −1 uniform rainfall rate and a quasi- real three-dimensional watershed under 1 month of a mea- sured daily rainfall hyetograph. The comparative analysis confirmed that the simplified approach can mimic simple and complex groundwater systems with an acceptable level of ac- curacy. In addition, the computational efficiency of the pro- posed approach (MIKE SHE took 180 times longer to solve
All storage options are potentially vulnerable to the impacts of CC. By modifying both water availability and water demand, CC will affect the performance, cost and adverse impacts of different types of waterstorage option. In some situations, certain storage options will be rendered completely impracticable whilst the viability of others may be increased. For example, CC may have significant impacts on soil moisture. In arid regions, the proportional change in soil moisture can be much greater than the proportional change in rainfall (Chiew et al. 1995; de Wit and Stankiewicz 2006). Hence, less rainfall and longer dry periods mean that SWC measures may fail to increase and maintain soil moisture sufficiently, leading to increased frequency of crop failure. Groundwater recharge may be reduced if rainfall decreases or its temporal distribution changes in such a way that infiltration declines. Many aquifers near the coast will be at risk from saltwater intrusion as a result of sea-level rise. Ponds and tanks may not fill to capacity or the frequency of filling may be reduced, so that they are unable to provide sufficient water for supplemental irrigation. Changes in river flows may mean that reservoir yields, and hence assurance of water supplies, decline. Storage in ponds, tanks and reservoirs may also be reduced more rapidly as a consequence of increased evaporation and/ or greater sediment inflows. Furthermore, both large and small dams as well as ponds and tanks and local initiatives, with minimal planning. In some
One of the challenges in using GRACE data is their temporal resolu- tion, which is limited to one month, as well as their coarse spatial reso- lution (typically N300 km). Unconstrained GRACE products require the application of some form of spatial ﬁltering to reduce the effects of high- frequency errors inherent to the publicly available GRACE ﬁelds. This spatial ﬁltering redistributes the signal over the ﬁlter radius, commonly referred to as signal leakage, requiring additional processing to restore this leaked signal if accurate TWS results over a speci ﬁc target area are desired. Several signal restoration methods have been described in the literature for this purpose. Landerer and Swenson (2012) applied a scal- ing factor computed as the ratio between the true TWS and ﬁltered TWS, based on a hydrological model. The procedure is simple but may introduce a bias caused by the dependency on a particular hydrological model. Baur, Kuhn, and Featherstone (2009) applied a correction based on known signal geometry. Their method was developed to restore the signal along the coastal zone of Greenland. The method does not rely on external data and can be very effective, but requires a controlled environment, where the surrounding signal is smaller than the target one, and the signal location is known. More recently, Chen et al. (2013, 2014) proposed a strategy similar to that of Baur et al. (2009) but without the known signal geometry requirement. The main idea is to mitigate the leakage out signal (from land to ocean) using GRACE data directly, so that the signal damping effect near the coast is effec- tively reduced ( Chen et al., 2013 ). This strategy is straightforward, easy to implement, and has been proven effective for inland applica- tions ( Chen, Li, Zhang, & Ni, 2014 ). As will be shown later, the results produced compared well with independent validation data, suggesting the approach is suitable for this study as well.
A significant challenge in water resource planning is to ensure a smooth inte- gration of the provision of water supplies for domestic use and water for other purposes leading to economic production, particularly in rural areas. Water for domestic supplies in rural areas is used for various household purposes such as cooking, washing, food gardens, stock watering and small businesses. If water is provided mainly for irrigation, it can also be used for domestic purposes, and if water is provided for domestic purposes, it will be used for other purposes too. In South Africa, water is stored for multiple uses, which include: irrigation for food production; fisheries; energy production; industrial water supply; mining; drinking water and sanitation; stock water; flood management and drought mitigation; recreation; pollution control; game watering; and environmental ser- vices (Figure 9). At national level, much of water is stored for irrigation, mu- nicipal and industrial use, domestic use and hydropower generation.
The first goal of this study was to investigate the impact of as- similating GRACE into the OpenStreams wflow_hbv model on the estimated TWS, GW storage and streamflow in the Rhine River basin. GRACE observations were assimilated into each grid cell of the model with an EnKF to update the soil moisture and UZ and LZ storage terms of the model. In general, assimilation drew the EnOL estimated TWS closer to the GRACE observations. In the absence of independent TWS observations, a qualitative analysis of the increments in TWS indicated that GRACE assimilation could partially correct the TWS estimate for the influence of errors in the meteorological forcing data and model parameters. As a re- sult, an improvement in the GW estimate after assimilating GRACE data was noticeable. In the best case, correlation co- efficient increased from 0.31 to 0.53 and RMSE decreased by 35 % with respect to the EnOL case. However, it is found that the improvement in TWS estimates did not always trans- late to an improved agreement between the estimated and ob- served GW storage variation at certain well locations. The differences may be due to the OpenStreams wflow_hbv pa- rameters: if the upper limit on soil moisture storage is too high (low), then the GW variations could be under (over)- estimated. This is particularly relevant in the type of model where the calibration is per sub-basin. This does not allow for local differences on the order of single or a few grid cells. The issue of scale is also significant because GRACE ob- serves monthly variations on the order of hundreds of kilo- metres. Groundwater variations can be influenced by local features at finer scales. When the basin average is consid- ered, validation against a denser network of well data or an independent GW model could be used to determine if an im- provement occurs at the scale of the entire basin.
In many semi-parametric models, `regular' parameters can be estimated by (semi-parametric) maximum likelihood estimators. The asymptotic theory for such estimators has been developed for a number of models of practical interest, and is similar to the asymptotic theory for maximum likelihood estimators in classical parametric models. In particular, the maximum likelihood estimators are asymptotically normal, where the inverse of the `ef®cient Fisher information matrix' gives the asymptotic covariance matrix. The latter matrix is the Fisher information matrix corrected for the presence of an in®nite-dimensional nuisance parameter. See, for example, Bickel et al. (1993) for an extensive review of information bounds. See Gill (1989), Chang (1990), Gu and Zhang (1993), Qin (1993), van der Laan (1993), Qin and Lawless (1994), van der Vaart (1994a; 1994b; 1994c; 1996), Murphy (1995), Gill et al. (1995), Huang (1996), Parner (1998), Qin and Wong (1996) and Mammen and van de Geer (1997) for results on the asymptotics of particular maximum likelihood estimators. It is natural to use the asymptotic normality of the estimator in order to form con®dence intervals and test statistics. This requires an estimator of the standard error or equivalently of the Fisher information matrix. In some speci®c cases the ef®cient Fisher information matrix is of closed form. For example, under the assumption that the observation time is independent of the covariates, Huang (1996) gives an explicit estimator of the asymptotic variance in a proportional hazards model applied to current status data. Sometimes the `ef®cient score' or `ef®cient in¯uence function' is explicit. Then since the ef®cient Fisher information matrix is the covariance of the ef®cient score function, one may estimate the
Compared to classical reconstructions applied to atmo- spheric or oceanic data sets which combine grids and sparse records of the same physical quantity (e.g., surface atmo- spheric pressure or sea level), here this is not the case as we combine gridded TWS from GRACE with in situ river level records. River level is a component of the total TWS, related in a nonlinear way to inundation extent and thus sur- face water volumes. However, in the Amazon Basin, previ- ous studies (e.g., Xavier et al., 2010; Vaz de Almeida, 2009) have shown that at seasonal and interannual time scales, river water level fluctuations can locally be correlated to TWS (as observed by GRACE). Such a correlation suggests that, at these time scales, TWS (including underground waters) and surface waters co-vary in a similar way. In the present study, we take advantage of this correlation at the interannual time scale and compute a scaling factor between river level and GRACE-based TWS over the GRACE time span (since 2002). This allows us to construct virtual multi-decade long TWS time series at the gauging sites for further combination of 2-D TWS grids from GRACE (of limited time duration) with sparse but long virtual TWS time series (based on the re-scaled river level time series). The final products are grid- ded (i.e., 2-D) time series of past TWS.
We used 108 months of GRACE and streamflow data over nine water years (WY; October–September; 2004–2012). These data comprises positive, neutral, and negative phases of the El Niño-Southern Oscillation and negative and pos- itive phases of the Pacific Decadal Oscillation (Feng et al., 2014; Iizumi et al., 2014). As a result, the data provide years of above- and below-average precipitation, snowpack, and streamflow for the region. The three watersheds were delineated upstream from United States Geological Survey (USGS) stream gages at 1 ◦ resolution, which is the reso- lution of GRACE data. In the CRB, these grid cells repre- sent a dimension of approximately 80 by 120 km. The Up- per Columbia consists of the area upstream of the Columbia River at the International Boundary gage (USGS 12399500), just downstream of the confluence of the Columbia and Pend- Oreille Rivers. The Pend-Oreille is a major watershed in the upper portions of the CRB. The Snake River gage at Weiser (USGS 13269000) provides gauged streamflow data above Hell’s Canyon Reservoir, the largest impoundment in the Snake River basin. The USGS gage at The Dalles (USGS 14105700) provides the most downstream streamflow data for the CRB. Monthly mean runoff (R; mm) was calculated for each of the three gages using the USGS streamflow data. Measurements of TWSA were obtained from the GRACE RL-05 (Swenson and Wahr, 2006; Landerer and Swenson, 2012) data set from NASA’s Tellus website (http://grace.jpl. nasa.gov). The errors present in the gridded GRACE data ex- ist primarily as a result of truncation (i.e., a low number of harmonics) in the spherical harmonic solution, and smooth- ing and systematic noise removal (called “de-striping”) that is applied after GRACE level-2 processing to remove spa- tially correlated noise (called “stripes”; Swenson and Wahr, 2006). This smoothing tends to smear adjacent signals to- gether (within the radius of the filtering function), resulting in smaller signals being lost, and larger signals having a coarser footprint and a loss of spatial information.
Average air temperatures and precipitation sums in the period of 2002–2016 hydrological years indicate changeability of weather condi- tions (tab. 1). Very cold and very warm as well as very dry and extremely moist half-years were observed. The results of Mann-Kendal statistical calculation for the air temperature suggest a slight increase in the average half-year temperatures, however the S value is not statistically significant. The S describing the amount of precipitation sug- gests a slight decrease in precipitation sums in the winter half-year and their slight increase in the summer half-year. However, both S values are not statistically significant.
The municipal sewage sample was collected from singanallur (open drainage) of Coimbatore. Plastic containers were used for collection and the samples were immediately brought to the laboratory and refrigerated at Cyprinus carpio were selected as the test animal and healthy specimens of fishes were procured from a local fresh water pond in Coimbatore. The fishes were acclimatized to laboratory conditions for a period of fifteen During the period of acclimatization fishes were fed regularly with conventional diet (rice bran and Feeding was stopped one day prior to the
It should be remarked that in the com- mand areas of the Upper and Lower Gugera Branch Canal, water tables are now 3 to 8 meters deep and appear to change less over time than would be expected from the above calculations of the water balance of the irri- gated areas. The primary source of ground- water recharge in IIMI’s research sites in the Upper Gugera command area is seepage from the Upper Gugera Branch Canal itself, which carries about 180 m 3 /s in this reach, and the Qadirabad-Balloki Link Canal, which carries around 540 m 3 /s (Greenman, Swarzenski, and Bennett 1967). Both flow parallel and close to the head reach of the distributaries (Mananwala and Lagar) in which the data were collected. This flow pat- tern accounts for the reported gradient in water tables and groundwater quality from the head reach to the tails in the command areas of the distributaries that flow more or less perpendicular to these two canals (Kijne and Vander Velde 1992). Recharge to groundwater in the area of the research sites in the command of the Lower Gugera Branch Canal is in large measure from the Ravi River, which flows as close as 2 kilome- ters from parts of the downstream half of one of the sample distributary canals in this area (Pir Mahal Distributary). The hydraulic gradient for a water table depth of 8 meters over a distance of 2 kilometers is small, and therefore the impact of local recharge can only be limited in spite of the fairly large horizontal transmissibility of the aquifer (Greenman, Swar-zenski, and Bennett 1967). The water balance calculations presented here do not take into account lateral ground- water flow from sources of good quality water that affect the hydrological conditions