Reconstruction of GRACE Total Water Storage Through Automated Machine Learning

2020 ◽  
Author(s):  
Alex Sun ◽  
Bridget Scanlon ◽  
Himanshu Save ◽  
Ashraf Rateb
Keyword(s):  
Machine Learning ◽  
Spatial Variability ◽  
Large Scale ◽  
Water Storage ◽  
Total Water ◽  
New Paradigm ◽  
Satellite Mission ◽  
Automated Machine Learning ◽  
The Mean ◽  
The Impact

<p>The GRACE satellite mission and its follow-on, GRACE-FO, have provided unprecedented opportunities to quantify the impact of climate extremes and human activities on total water storage at large scales. The approximately one-year data gap between the two GRACE missions needs to be filled to maintain data continuity and maximize mission benefits. There is strong interest in using machine learning (ML) algorithms to reconstruct GRACE-like data to fill this gap. So far, most studies attempted to train and select a single ML algorithm to work for global basins. However, hydrometeorological predictors may exhibit strong spatial variability which, in turn, may affect the performance of ML models. Existing studies have already shown that no single algorithm consistently outperformed others over all global basins. In this study, we applied an automated machine learning (AutoML) workflow to perform GRACE data reconstruction. AutoML represents a new paradigm for optimal model structure selection, hyperparameter tuning, and model ensemble stacking, addressing some of the most challenging issues related to ML applications. We demonstrated the AutoML workflow over the conterminous U.S. (CONUS) using six types of ML algorithms and multiple groups of meteorological and climatic variables as predictors. Results indicate that the AutoML-assisted gap filling achieved satisfactory performance over the CONUS. For the testing period (2014/06–2017/06), the mean gridwise Nash-Sutcliffe efficiency is around 0.85, the mean correlation coefficient is around 0.95, and the mean normalized root-mean square error is about 0.09. Trained models maintain good performance when extrapolating to the mission gap and to GRACE-FO periods (after 2017/06). Results further suggest that no single algorithm provides the best predictive performance over the entire CONUS, stressing the importance of using an end-to-end workflow to train, optimize, and combine multiple machine learning models to deliver robust performance, especially when building large-scale hydrological prediction systems and when predictor importance exhibits strong spatial variability.</p>

Reconstruction of GRACE Total Water Storage Through Automated Machine Learning

2020 ◽  
Author(s):  
Alexander Y. Sun ◽  
Bridget R. Scanlon ◽  
Himanshu Save ◽  
Ashraf Rateb
Keyword(s):  
Machine Learning ◽  
Water Storage ◽  
Total Water ◽  
Automated Machine Learning

scholarly journals RECOG RL01: Correcting GRACE total water storage estimates for global lakes/reservoirs and earthquakes

2020 ◽  
Author(s):  
Simon Deggim ◽  
Annette Eicker ◽  
Lennart Schawohl ◽  
Helena Gerdener ◽  
Kerstin Schulze ◽  
...  
Keyword(s):  
Time Series ◽  
Surface Water ◽  
Regional Scale ◽  
Water Storage ◽  
Water Volume ◽  
Optical Remote Sensing ◽  
Hydrological Models ◽  
Total Water ◽  
Satellite Mission ◽  
The Impact

Abstract. Observations of changes in terrestrial water storage obtained from the satellite mission GRACE (Gravity Recovery and Climate Experiment) have frequently been used for water cycle studies and for the improvement of hydrological models by means of calibration and data assimilation. However, due to a low spatial resolution of the gravity field models spatially localized water storage changes, such as those occurring in lakes and reservoirs, cannot properly be represented in the GRACE estimates. As surface storage changes can represent a large part of total water storage, this leads to leakage effects and results in surface water signals becoming erroneously assimilated into other water storage compartments of neighboring model grid cells. As a consequence, a simple mass balance at grid/regional scale is not sufficient to deconvolve the impact of surface water on TWS. Furthermore, non-hydrology related phenomena contained in the GRACE time series, such as the mass redistribution caused by major earthquakes, hamper the use of GRACE for hydrological studies in affected regions. In this paper, we present the first release (RL01) of the global correction product RECOG (REgional COrrections for GRACE), which accounts for both the surface water (lakes & reservoirs, RECOG-LR) and earthquake effects (RECOG-EQ). RECOG-LR is computed from forward-modelling surface water volume estimates derived from satellite altimetry and (optical) remote sensing and allows both a removal of these signals from GRACE and a re-location of the mass change to its origin within the outline of the lakes/reservoirs. The earthquake correction RECOG-EQ includes both the co-seismic and post-seismic signals of two major earthquakes with magnitudes above 9 Mw. We can show that applying the correction dataset (1) reduces the GRACE signal variability by up to 75 % around major lakes and explains a large part of GRACE seasonal variations and trends, (2) avoids the introduction of spurious trends caused by leakage signals of nearby lakes when calibrating/assimilating hydrological models with GRACE, even in neighboring river basins, and (3) enables a clearer detection of hydrological droughts in areas affected by earthquakes. A first validation of the corrected GRACE time series using GPS-derived vertical station displacements shows a consistent improvement of the fit between GRACE and GNSS after applying the correction. Data are made available as open access via the Pangea database (RECOG-LR: Deggim et al. (2020a) https://doi.org/10.1594/PANGAEA.921851; RECOG-EQ: Gerdener et al. (2020b, under revision), https://doi.pangaea.de/10.1594/PANGAEA.921923).

scholarly journals Water Storage Variations in Tibet from GRACE, ICESat, and Hydrological Data

2019 ◽  
Vol 11 (9) ◽  
pp. 1103 ◽  
Author(s):  
Fang Zou ◽  
Robert Tenzer ◽  
Shuanggen Jin
Keyword(s):  
Tibetan Plateau ◽  
Water Budget ◽  
Hydrological Cycle ◽  
Water Storage ◽  
The Tibetan Plateau ◽  
Total Water ◽  
Satellite Mission ◽  
Mountain Glaciers ◽  
Grace Data ◽  
The Impact

The monitoring of water storage variations is essential not only for the management of water resources, but also for a better understanding of the impact of climate change on hydrological cycle, particularly in Tibet. In this study, we estimated and analyzed changes of the total water budget on the Tibetan Plateau from the Gravity Recovery And Climate Experiment (GRACE) satellite mission over 15 years prior to 2017. To suppress overall leakage effect of GRACE monthly solutions in Tibet, we applied a forward modeling technique to reconstruct hydrological signals from GRACE data. The results reveal a considerable decrease in the total water budget at an average annual rate of −6.22 ± 1.74 Gt during the period from August 2002 to December 2016. In addition to the secular trend, seasonal variations controlled mainly by annual changes in precipitation were detected, with maxima in September and minima in December. A rising temperature on the plateau is likely a principal factor causing a continuous decline of the total water budget attributed to increase melting of mountain glaciers, permafrost, and snow cover. We also demonstrate that a substantial decrease in the total water budget due to melting of mountain glaciers was partially moderated by the increasing water storage of lakes. This is evident from results of ICESat data for selected major lakes and glaciers. The ICESat results confirm a substantial retreat of mountain glaciers and an increasing trend of major lakes. An increasing volume of lakes is mainly due to an inflow of the meltwater from glaciers and precipitation. Our estimates of the total water budget on the Tibetan Plateau are affected by a hydrological signal from neighboring regions. Probably the most significant are aliasing signals due to ground water depletion in Northwest India and decreasing precipitation in the Eastern Himalayas. Nevertheless, an integral downtrend in the total water budget on the Tibetan Plateau caused by melting of glaciers prevails over the investigated period.

scholarly journals GRACE water storage estimates for the Middle East and other regions with significant reservoir and lake storage

2012 ◽  
Vol 9 (10) ◽  
pp. 11131-11159 ◽  
Author(s):  
L. Longuevergne ◽  
C. R. Wilson ◽  
B. R. Scanlon ◽  
J. F. Crétaux
Keyword(s):  
Spatial Distribution ◽  
Surface Water ◽  
Numerical Experiments ◽  
Water Storage ◽  
Accurate Estimation ◽  
Total Water ◽  
Satellite Mission ◽  
Reservoir Storage ◽  
Aquifer Systems ◽  
The Impact

Abstract. While GRACE (Gravity Recovery and Climate Experiment) satellites are increasingly being used to monitor water storage changes globally, the impact of spatial distribution of water storage within a basin is generally ignored but may be substantial. In many basins, water may be stored in reservoirs, lakes, flooded areas, small aquifer systems, and other localized regions with sizes typically below GRACE resolution. The objective of this study was to assess the impact of non-uniform water storage distribution on GRACE estimates as basin-wide averages, focusing on surface water reservoirs. Analysis included numerical experiments testing the effect of mass size and position within a basin, and application to the Lower Nile (Lake Nasser) and Tigri–Euphrates (TE) basins as examples. Numerical experiments show that by assuming uniform mass distribution, GRACE estimates may under- or over-estimate basin-average water storage by up to a factor of two, depending on reservoir location and extent. Although their spatial extent may be unresolved by GRACE, reservoir storage may dominate in some basins. For example, it accounts for 95% of seasonal variations in the Lower Nile and 10% in the TE basins. Because reservoirs are used to mitigate droughts and buffer against climate extremes, their influence on interannual time scales can be large, for example accounting for 50% of total water storage decline during the 2007–2009 drought in the TE basin. Effects on GRACE estimates are not easily accounted for via simple multiplicative scaling, but in many cases independent information may be available to improve estimates. Accurate estimation of the reservoir contribution is critical, especially when separating groundwater from GRACE total water storage changes. Because the influence of spatially concentrated water storage – and more generally water distribution – is significant, GRACE estimates will be improved when it is possible to combine independent spatial distribution information with GRACE observations, even when reservoir storage is not a major factor. In this regard, data from the upcoming Surface Water Ocean Topography (SWOT) satellite mission should be an especially important companion to GRACE-FO observations.

Forming Canonical 8D Data Clouds to Explore the Transient Dynamics of Physical Stiffened Plates: A Machine Learning Approach for Complex Mechanical Structures

2020 ◽  
Author(s):  
Ioannis T. Georgiou
Keyword(s):  
Machine Learning ◽  
Large Scale ◽  
Complex Structure ◽  
Spatial Dimension ◽  
Physical Structure ◽  
Transient Dynamics ◽  
Alloy Plate ◽  
Considerable Uncertainty ◽  
Acceleration Signal ◽  
The Impact

Abstract This work presents a data-driven explorative study of the physics of the dynamics of a physical structure of complicated geometry. The geometric complexity of the physical system renders the typical single sensor acceleration signal quite complicated for a physics interpretation. We need the spatial dimension to resolve the single sensory signal over its entire time horizon. Thus we are introducing the spatial dimension by the canonical eight-dimensional data cloud (Canonical 8D-Data Cloud) concept to build methods to explore the impact-induced free dynamics of physical complex mechanical structures. The complex structure in this study is a large scale aluminum alloy plate stiffened by a frame made of T-section beams. The Canonical 8D-Data Cloud is identified with the simultaneous acceleration measurements by eight piezoelectric sensors equally spaced and attached on the periphery of a circular material curve drawn on the uniform surface of the stiffened plate. The Data Cloud approach leads to a systematic exploration-discovery-quantification of uncertainty in this physical complex structure. It is found that considerable uncertainty is stemming from the sensitivity of transient dynamics on the parameters of space-time localized force pulses, the latter being used as a means to diagnose the presence of structural anomalies. The Data Cloud approach leads to aspects of machine learning such as reduced dynamics analytics of big sensory data by means of heavenly machine-assisted computations to carry out the unparalleled data reduction analysis enabled by the Advanced Proper Orthogonal Decomposition Transform. Emphasized is the connection between the characteristic geometric features of high-dimensional datasets as a whole, the Data Cloud, and the modal physics of the dynamics.

scholarly journals Observing Snow Cover and Water Resource Changes in the High Mountain Asia Region in Comparison with Global Mountain Trends over 2000–2018

2020 ◽  
Vol 12 (23) ◽  
pp. 3913
Author(s):  
Claudia Notarnicola
Keyword(s):  
Water Resources ◽  
Confidence Interval ◽  
Snow Cover ◽  
Water Storage ◽  
High Mountain ◽  
Global Level ◽  
Total Water ◽  
Mountain Areas ◽  
Snow Cover Extent ◽  
The Impact

The quantification of snow cover changes and of the related water resources in mountain areas has a key role for understanding the impact on several sectors such as ecosystem services, tourism and energy production. By using NASA-Moderate Resolution Imaging Spectroradiometer (MODIS) images from 2000 to 2018, this study analyzes changes in snow cover in the High Mountain Asia region and compares them with global mountain areas. Globally, snow cover extent and duration are declining with significant trends in around 78% of mountain areas, and the High Mountain Asia region follows similar trends in around 86% of the areas. As an example, Shaluli Shan area in China shows significant negative trends for both snow cover extent and duration, with −11.4% (confidence interval: −17.7%, −5.5%) and −47.3 days (confidence interval: −70.4 days, −24.4 days) at elevations >5500 m a.s.l. respectively. In spring, an earlier snowmelt of −13.5 days (confidence interval: −24.3 days, −2.0 days) in 4000–5500 m a.s.l. is detected. On the other side, Tien Shan area shows an earlier snow onset of −28.8 days (confidence interval: −44.3 days, −8.2 days) between 2500 and 4000 m a.s.l., governed by decreasing temperature and increasing snowfall. In the current analysis, the Tibetan Plateau shows no significant changes. Regarding water resources, by using Gravity Recovery and Climate Experiment (GRACE) data it was found that around 50% of areas in the High Mountain Asia region and 30% at global level are suffering from significant negative temporal trends of total water storage (including groundwater, soil moisture, surface water, snow, and ice) in the period 2002–2015. In the High Mountain Asia region, this negative trend involves around 54% of the areas during spring period, while at a global level this percentage lies between 25% and 30% for all seasons. Positive trends for water storage are detected in a maximum 10% of the areas in High Mountain Asia region and in around 20% of the areas at global level. Overall snow mass changes determine a significant contribution to the total water storage changes up to 30% of the areas in winter and spring time over 2002–2015.

scholarly journals What Is the Impact of Additional Tropical Observations on a Modern Data Assimilation System?

2019 ◽  
Vol 147 (7) ◽  
pp. 2433-2449
Author(s):  
Laura C. Slivinski ◽  
Gilbert P. Compo ◽  
Jeffrey S. Whitaker ◽  
Prashant D. Sardeshmukh ◽  
Jih-Wang A. Wang ◽  
...  
Keyword(s):  
El Niño ◽  
Large Scale ◽  
El Nino ◽  
Mean Square ◽  
Observational Network ◽  
In Situ Observations ◽  
The Mean ◽  
The Impact ◽  
Assimilation System

Abstract Given the network of satellite and aircraft observations around the globe, do additional in situ observations impact analyses within a global forecast system? Despite the dense observational network at many levels in the tropical troposphere, assimilating additional sounding observations taken in the eastern tropical Pacific Ocean during the 2016 El Niño Rapid Response (ENRR) locally improves wind, temperature, and humidity 6-h forecasts using a modern assimilation system. Fields from a 50-km reanalysis that assimilates all available observations, including those taken during the ENRR, are compared with those from an otherwise-identical reanalysis that denies all ENRR observations. These observations reveal a bias in the 200-hPa divergence of the assimilating model during a strong El Niño. While the existing observational network partially corrects this bias, the ENRR observations provide a stronger mean correction in the analysis. Significant improvements in the mean-square fit of the first-guess fields to the assimilated ENRR observations demonstrate that they are valuable within the existing network. The effects of the ENRR observations are pronounced in levels of the troposphere that are sparsely observed, particularly 500–800 hPa. Assimilating ENRR observations has mixed effects on the mean-square difference with nearby non-ENRR observations. Using a similar system but with a higher-resolution forecast model yields comparable results to the lower-resolution system. These findings imply a limited improvement in large-scale forecast variability from additional in situ observations, but significant improvements in local 6-h forecasts.

Large-scale fuel deposition patterns on northern Spanish shores following the ‘Prestige’ oil spill

2008 ◽  
Vol 88 (3) ◽  
pp. 463-468 ◽  
Author(s):  
José Luis Acuña ◽  
Araceli Puente ◽  
Ricardo Anadón ◽  
Consolación Fernández ◽  
María Luisa Vera ◽  
...  
Keyword(s):  
Large Scale ◽  
Northern Spain ◽  
The West ◽  
Deposition Patterns ◽  
Prestige Oil Spill ◽  
Cantabrian Coast ◽  
The Mean ◽  
The Impact ◽  
Oil Tanker ◽  
Fuel Deposition

Following the accident of the oil tanker ‘Prestige’, we surveyed the large scale fuel deposition patterns on the Cantabrian shore (northern Spain) covering three regions (from west to east): (i) Asturias, west of Cape Peñas (24 segments surveyed); (ii) Asturias, east of Cape Peñas (33 segments surveyed); and (iii) Cantabria (also east of Cape Peñas, 256 segments surveyed). Fuel arrived to the Cantabrian Coast as a single oil wave which was more intense to the east than to the west of Cape Peñas. The mean percentage of coast length affected was 25, 41 and 15% in western Asturias, eastern Asturias and Cantabria, respectively. However, less than 10% of the substrate was covered by fuel in oiled patches, thus the impact was moderate. We conclude that these patterns are consistent with fuel transport by the Iberian Poleward Current, a hydrographic feature typical of this region during winter.

Drivers of biases in the extratropical storm tracks in CMIP6

2020 ◽  
Author(s):  
Matthew Priestley ◽  
Duncan Ackerley ◽  
Jennifer Catto ◽  
Kevin Hodges ◽  
Ruth McDonald ◽  
...  
Keyword(s):  
Large Scale ◽  
Climate Models ◽  
Storm Track ◽  
Storm Tracks ◽  
Extratropical Cyclones ◽  
Ocean Coupling ◽  
Coupled Climate Models ◽  
The Mean ◽  
The Impact

<p>Extratropical cyclones are the leading driver of the day-to-day weather variability and wintertime losses for Europe. In the latest generation of coupled climate models, CMIP6, it is hoped that with improved modelling capabilities come improvements in the structure of the storm track and the associated cyclones. Using an objective cyclone identification and tracking algorithm the mean state of the storm tracks in the CMIP6 models is assessed as well as the representation of explosively deepening cyclones. Any developments and improvements since the previous generation of models in CMIP5 are discussed, with focus on the impact of model resolution on storm track representation. Furthermore, large-scale drivers of any biases are investigated, with particular focus on the role of atmosphere-ocean coupling via associated AMIP simulations and also the influence of large-scale dynamical and thermodynamical features.</p>

scholarly journals GRACE water storage estimates for the Middle East and other regions with significant reservoir and lake storage

2013 ◽  
Vol 17 (12) ◽  
pp. 4817-4830 ◽  
Author(s):  
L. Longuevergne ◽  
C. R. Wilson ◽  
B. R. Scanlon ◽  
J. F. Crétaux
Keyword(s):  
Spatial Distribution ◽  
Surface Water ◽  
Numerical Experiments ◽  
Water Storage ◽  
Accurate Estimation ◽  
Areal Extent ◽  
Total Water ◽  
Reservoir Storage ◽  
Average Water ◽  
The Impact

Abstract. While GRACE (Gravity Recovery and Climate Experiment) satellites are increasingly being used to monitor total water storage (TWS) changes globally, the impact of spatial distribution of water storage within a basin is generally ignored but may be substantial. In many basins, water is often stored in reservoirs or lakes, flooded areas, small aquifer systems, and other localized regions with areas typically below GRACE resolution (~200 000 km2). The objective of this study was to assess the impact of nonuniform water storage distribution on GRACE estimates of TWS changes as basin-wide averages, focusing on surface water reservoirs and using a priori information on reservoir storage from radar altimetry. Analysis included numerical experiments testing effects of location and areal extent of the localized mass (reservoirs) within a basin on basin-wide average water storage changes, and application to the lower Nile (Lake Nasser) and Tigris–Euphrates basins as examples. Numerical experiments show that by assuming uniform mass distribution, GRACE estimates may under- or overestimate basin-wide average water storage by up to a factor of ~2, depending on reservoir location and areal extent. Although reservoirs generally cover less than 1% of the basin area, and their spatial extent may be unresolved by GRACE, reservoir storage may dominate water storage changes in some basins. For example, reservoir storage accounts for ~95% of seasonal water storage changes in the lower Nile and 10% in the Tigris–Euphrates. Because reservoirs are used to mitigate droughts and buffer against climate extremes, their influence on interannual timescales can be large. For example, TWS decline during the 2007–2009 drought in the Tigris–Euphrates basin measured by GRACE was ~93 km3. Actual reservoir storage from satellite altimetry was limited to 27 km3, but their apparent impact on GRACE reached 45 km3, i.e., 50% of GRACE trend. Therefore, the actual impact of reservoirs would have been greatly underestimated (27 km3) if reservoir storage changes were assumed uniform in the basin. Consequently, estimated groundwater contribution from GRACE would have been largely overestimated in this region if the actual distribution of water was not explicitly taken into account. Effects of point masses on GRACE estimates are not easily accounted for via simple multiplicative scaling, but in many cases independent information may be available to improve estimates. Accurate estimation of the reservoir contribution is critical, especially when separating estimating groundwater storage changes from GRACE total water storage (TWS) changes. Because the influence of spatially concentrated water storage – and more generally water distribution – is significant, GRACE estimates will be improved by combining independent water mass spatial distribution information with GRACE observations, even when reservoir storage is not the dominant mechanism. In this regard, data from the upcoming Surface Water Ocean Topography (SWOT) satellite mission should be an especially important companion to GRACE-FO (Follow-On) observations.

Sign in / Sign up

Export Citation Format

Share Document