Workshop 1 (December 11, 2008):

Output and results

IGCP 565 Project Workshops

IGCP 565 Workshop 1: Science of geodetic monitoring of the hydrological cycle

Notes from the First Workshop (2008)

Hans-Peter Plag opened the workshop and welcomed the participants. He briefly introduce the IGCP 565 Project and provided some background on the IGCP program. He emphasized the importance of the Project and the Workshop by referring to the discussions at the International Conference on Water Scarcity, Global Change and Groundwater Management Responses, which had taken place a week before in Irvine, California (see At this conference, Peter Gleick, President and co-founder of the Pacific Institute, stated that there was no excuse for the fact that one billion people do not have access to sufficient drinking water; in his view it was unacceptable and a failure of humanity as a whole. Although many of the problems related to water supply are caused by a crisis of governance and not a lack of abundance (see e.g. the press release announcing the Second United Nations Water Development Report), it is also obvious that improved regional water management needs scientific support based on Earth observations and modelling.

In the first presentation, Norman Miller illustrated the severity of water scarcity by showing the projected situation in 2025, in which large areas of the global land areas are impacted by scarcity. One of the most difficult component in the hydrological cycle to quantify is the Total Groundwater, and the large uncertainties hamper water management extremely. He then demonstrated that drought will be the big challenge of the future and showed that droughts have meteorological (insufficient net precipitation); hydrological (insufficient surface/ground water); agricultural (insufficient soil moisture); and socio-economic (insufficient water resource supply to meet growing demands) causes and impacts. Drought-related indices derived from different hydrological and remote sensing data underline the current problems, and he used the example of North America to illustrate the severity of the situation. Droughts are cause by a number of processes, including sea surface temperature patterns (e.g., Australia, Sub-Sahara, Atlanta, Georgia), large-scale circulation changes, shifts in jets and storm tracks (e.g., LLJ shift), atmospheric patterns leading to blocking, and, not least, a warming planet leading to decreasing snowpack, snow/rain ratio and a associates increase in winter run-off. Most of these causes have in common that they are associated with large-scale features, in particular sea surface temperature. The geodetic observing system already contributes to the monitoring of these large-scale features and support the detection of droughts.

Miller also mentioned that climate variability or change reduced snowpack in the Sierra Nevada by up to 60%. He reported on a drought study for the Central Valley with simulations of the 'relative' response of the water table to surface flow reductions. These studies show that a huge deficit is building up fast in the southern part of the Central Valley. This impacts water delivery as well as delivery from pumping, and it necessitate more infrastructure.

Miller also showed that satellite gravity observations help to monitor the droughts. However, monitoring is not the issue. The key issue is forecasting, which is worked on by a different community. The main uncertainties for GRACE are the limitation in spatial and temporal scales. An important step forward would be the integration of different observations into models.

Rick Lawford in his presentation introduced the Integrated Global Water Cycle Observation (IGWCO), which is developed in support of helping to solve the world's water problems with observations and information. In the frame of IGWCO, several components of the water cycle are addressed in separate activities. There are many relation between these activities and the IGCP 565 Project, and the project could support capacity building.

Lawford informed that for precipitation, there are many data sets, with no single set sticking out as the best one. For soil moisture, the main goal is higher spatial and temporal resolution. With respect to water quality, he remarked that several factors determine quality, and many groups are concerned with water quality. Most monitoring was and still is in situ and site-specific. Remote sensing can guide where to place in situ observations. For run-off, he mentioned the HARON project. This project has three phases: (1) upgrade of in situ observations; (2) combination with satellite observations; and (3) implementation of integrated water cycle models for user applications. He speculated that geodesy might come in for groundwater, and emphasized that all water cycle-related data could help to resolve lack of consistency between measurements in different regions. He stated that currently WMO standards do not have a big impact on how water-related measurements are being carried out.

With respect to capacity building, he mentioned the TIGER project in Africa (TIGER), and the Water-Cycle initiative led by Japan in South-East Asia. A new initiative for South America is under development. End to End (E2E) initiatives will look at the full suite of products and the complete chain from observations to applications.

Lawford concluded his talk with an example of the cost-benefit ratio of climate observations, and a brief reviewed of the GEO Water tasks. He stated that the IGCP 565 Project could contribute to these tasks. With respect to the way forward, he mentioned that applications and demonstration of the impact of geodetic observations have not be developed sufficiently.

Markus Rothacher focused on GRACE. He mentioned that the predicted lifetime of GRACE is up to 2013, and informed that the USA, China and Germany are major users of GRACE observations among a total of some 20 countries. The outputs based on GRACE observations include the gravity field. He reported on a reprocessing of CHAMP observations, which resulted in good agreement between GRACE and CHAMP results. The plans at GFZ include improvements of the background models (e.g., tides), and he reported that GFZ is carrying out a study for a follow-on mission.

He emphasized that satellite gravity missions are unique in measuring the subsurface water, but he also stated that there are other geodetic techniques which allow deduction of subsurface water contents. Despite the fact that GRACE and superconducting gravimeters provide very different observations with very different characteristics in resolution, accuracy, and drift, a careful comparison of superconducting gravimeters, GRACE and hydrological models show a very good agreement at some stations, but larger differences at others, where hydrology is difficult to model.

With respect to global models, he stated that more complicated Earth System models are needed. He reported that GFZ is working on a 4-D coupled model with data assimilation. For separation of the different effects, one needs models to remove all the other effects. He concluded by stating that there are still many disagreements between inversions and hydrological models.

Xiaoping Wu reported on inversions of geodetic observations for land water storage. He pointed out that there are different temporal and spatial scales in water storage changes, which lead to multiple geodetic signatures. This compares to the objectives of global coverage, complete spatio-teomporal spectum, high resolution and accuracy. There are different ways for the inversion, using, for example, spherical harmonics, mascons, grids, etc. In his work, he focuses on spherical harmonics as the mathematical tool.

Inversions using GPS observations and models have errors at the level of a few centimeters at a few hundred kilometers wavelength, while GRACE at 200 km has an error larger than 20 cm. He speculated that combination of the observation with models may be the key to fully exploit the geodetic observations. Results of combined GRACE/SLR/VLBI/GPS/ECCO inversion show a dramatic melting of Greenland. Model predictions for geoid changes associated with post-glacial rebound and GRACE results show a good agreement except for the poles, where not enough data is available. The velocity of the Center of Figure (CF) of the solid Earth and the Center of Mass of the Earth system (CM) determined in the inversion is on the order of -0.25 mm/yr compared to predictions of -0.6 mm/yr for post-glacial rebound models, which does not seem to be consistent. For the mass contribution to mean global sea level changes, he reported 0.55+-0.17 mm/yr. However, uncertainties are still rather large in many areas.

In his second presentation, Richard Lawford introduced the GEWEX program. GEWEX is around for about 20 years. There are many projects associated with it in various fields of the global water and energy cycle. One project looks at monsoons and the forcing, and this could be supported by geodesy. Questions are whether changes in land use and aerosols have impacted monsoons. Extremes is another area of study for GEWEX. Forecasts of the drought conditions in 2005 in Canada failed completely. There is a need to do something about seasonal prediction. There are periods of relatively quiet variations and then there are periods of extremes. He considered that maybe the geodetic observations could be used to identify these phases. At least, geodetic observation could support the validation of models. He also presented the GEWEX road map, which assumes an end of the current program by 2012.

Matthew Rodell presented land surface hydrological models, which are needed to get high spatial resolution. Assimilation of meteorological and hydrological data in these models is state-of-the-art. Geodetic data may be very valuable since these data have no limitation in terms of penetration depth. Challenges arise from the fact that the observed and modeled variables differ, that the spatial resolution of geodetic observations is coarse, the temporal resolution is too low, and the latency too large.

Incorporation of geodetic observation into hydrological models can be done at different levels. Currently, terrestrial water storage as derived from GRACE is incorporated into the models. At a higher level, the gravity field and rotation observations would be incorporated directly into a model that does forward modeling of these quantities.

GRACE data assimilation uses 10-year spin-up prior to 2002. The results have higher resolution that GRACE alone, and the accuracy is better than a model alone would provide. Current assimilation uses gridded fields with 500 km resolution. The match between model and observations is not perfect, but in areas where no hydrological observation are available, GRACE gives at least a reasonable estimate of the changes in land water storage.

Jakob Flury focused on error sources in the GRACE mission. He emphasized the need for higher accuracy as well as higher spatial resolution. Comparison of GRACE results for ice changes in Antarctica compare on large scales well with ICESat results but in details there are also significant differences. He also indicated the need for higher temporal resolution to better capture those processes that have significant energy at shorter time scales. As limitations hampering today progress towards higher accuracy he listed:

  • sensor accuracy;
  • disturbances;
  • spatial and temporal sampling;
  • parameterization, modeling, representations;
Jean Dickey wanted to added to this list also the height of the satellite, but Jacob Flury responded that he only considered parameters of the current mission, for which the height is fixed.

Satellite induced disturbances are very small but still much larger than the sensor accuracy, and these disturbances are partly due to heater switching. Several types of vibrations occur, and the 'Twangs' are one of them. These vibrations are not well understood, and some happen when sun light hits the satellite bottom, while others happen at other times. An important objective of all these studies is to learn more for future satellite missions. Thus, GRACE can be considered as a precision test laboratory. It could be and can be expected that the high accuracy of the GRACE sensor would lead to unexpected effects. He stated that calibration and validation is limited since there is no better global data source than GRACE. However, it was remarked that GPS can by now be considered to be global and used for cross-calibration/validation with GRACE, particularly if the relationships of the Load Love numbers are fully exploited.

In his conclusions, Jacob Flury emphasized the need to continue GRACE-like missions ASAP. He also requested the development and exploration of new sensor technologies, improved 'laboratory' conditions by eliminating or reducing some of the known disturbances, and envisaged sensor combination and satellite constellations as potential future developments.

Markus Rothacher reported on a project led by GFZ, which focuses on periodic variations in water storage as determined from GRACE and global hydrology models. For annual signals, a very good agreement between GRACE and model predictions was found, and several annual eigenmodes could be identified, which are associated with different spatial patterns. However, the hydrological model (WGHM) appears to be one months earlier than GRACE. Reconstruction of the GRACE data from main modes helps to reduce the noise.

Richard Gross showed the intercomparison of gravity (SLR and GRACE), deformation (various), and rotation results for degree 2. Some results agree well but others do not agree at all. He concluded that comparison with models could help to answer the question which of the techniques is doing best in measuring which coefficient.

Hans-Peter Plag discussed ways of improving spatial and temporal resolution through combined geodetic observations. He concluded that a key improvement has to happen in terms of an integrated Earth system model, which would allow to make full use of the integrated geodetic observations without having to separate all the different effects.

Eric Calais presented the Hermanus Project, which is a study of local hydrological effects in geodetic observations of surface displacements.

In the subsequent discussion, the following questions were considered:

  • What are the key monitoring gaps? There was considerable discussion. Although Richard Lawford identified groundwater, evaporation and evapotranspiration as key components where additional observations would be very valuable, he could not give a clear prioritization. A main conclusion was that monitoring surface mass variations in general would support hydrological applications. Richard Lawford also suggested to rephrase the question and preferred to ask "What can we do?" and then to offer the best products. Hans-Peter Plag asked whether we could give a prioritization, if the objective was to support the Millennium Development Goal of reducing the number of people without access to sufficient drinking water by 50%, but no such prioritization emerged. Hans-Peter Plag pointed out that there are two possible approaches for the exploitation of the geodetic observations, with one approach starting from the local (point) hydrological observations and aiming to extend these to scales of several tens of kilometers, and the other approach coming from the current global to regional scale techniques and trying to increase spatial and temporal resolution. Norman Miller emphasized a hybrid approach of a localized implementation combined with the global observations. The use of models could guide the selection of the best location for the local implementations. There was agreement that for closing the hydrological budget across all scales, deep groundwater was most important. Even more so, agreement was on the importance of an immediate follow-on mission for GRACE.
  • How does groundwater pumping related to earthquakes? Jean Dickey posed this question, and she motivating it with earthquakes induced near Denver by replacing extracted oil by water. Therefore, pumping in some of the areas in California might have undesired and potentially disastrous consequences. It was mentioned that induced seismicity associated with changes in pore pressure is know from some reservoirs, and in a few cases this has led to large earthquakes. However, pumping is likely to reduce pore pressure and it is not clear how this would interact with the stress field and potential earthquakes.
  • What about removing salt from salt water to satisfy demands? This question was also brought up by Jean Dickey who wanted to know to what extent desalination could be used to reduce the pressure on non-renewable groundwater resources. It was reported that desalination is still relatively expensive and energy demanding and mainly a solution in coastal areas where no other alternatives exist. However, Richard Lawford pointed out that in the future, when water could become a commodity with all the costs integrated in the price and the current subsidies removed, the price of desalinated water might become competitive to water from other resources.
  • How can the operational availability of the geodetic products for application be improved, e.g. by getting operational agencies involved in the project? It was commented that a potential agency to take responsibility for the products could be the U.S. Department of Agriculture, which is already involved in some related activities (e.g. drought monitoring).

Other questions noted during the presentations were not further discussed due to lack of time. These questions include:

  • What are the advantages and disadvantages of inversion/separation versus assimilation/simulation?
  • What are the long-term perspective for gravity satellite missions?
  • What is the fingerprint of present-day melting in vertical displacement, sea level and gravity?
  • How can geodetic observations best help in improving predictability, up to seasonal time scales?
  • How can we best facilitate the full assimilation of data in models and what improvements do we expect?
  • How can cross-calibration between the geodetic techniques help to improve GRACE validation?
  • What can we learn from low-degree cross-correlations in terms of technique problems and models?

In case of problems, mail to