Global Water Storage: A New Measurement Model

Scientists Junyang Gou and Professor Benedikt Soja from ETH Zurich have developed a new model for measuring global water storage using a novel deep learning approach. Their research, published in Nature Water, integrates satellite observations with hydrological models to achieve remarkable accuracy even in smaller basins.

What is global water storage?

Global water storage (GWS) is the total amount of water stored on Earth’s surface and subsurface, including soil moisture, groundwater, surface water, snow and ice. GWS is an important indicator of the hydrological cycle and the water balance of the planet. It also affects various aspects of human and natural systems, such as agriculture, water resources management, flood and drought prediction, and climate change adaptation.

How is global water storage measured?

Traditionally, GWS is measured by ground-based instruments, such as river gauges, wells, and soil moisture sensors. However, these methods have limitations in terms of spatial coverage, temporal resolution, and data availability. In recent years, satellite-based methods have emerged as a promising alternative for measuring GWS. One of the most widely used satellite missions is the Gravity Recovery and Climate Experiment (GRACE), which measures the changes in Earth’s gravity field caused by the mass variations of GWS. However, GRACE has a coarse spatial resolution of about 300 km, which makes it difficult to capture the local and regional variations of GWS.

What is the new model for measuring global water storage?

The new model proposed by Gou and Soja is based on a deep learning algorithm that combines GRACE data with hydrological models to produce high-resolution estimates of GWS anomalies (the deviations from the long-term mean). The algorithm uses a self-supervised data assimilation approach, which means that it learns from its own predictions and adjusts them to match the observations. The algorithm also uses a residual block structure, which allows it to capture complex nonlinear relationships between the input and output features.

The new model can produce GWS anomaly maps with a spatial resolution of 0.25 degrees (about 28 km), which is an order of magnitude higher than GRACE. The model also shows high accuracy and consistency with independent validation data sets, such as in situ measurements and remote sensing products. The model can capture the seasonal and interannual variations of GWS in different regions of the world, as well as the impacts of extreme events, such as floods and droughts.

What are the benefits of the new model

The new model offers several benefits for various domains that rely on GWS information, such as:

  • Hydrology: The new model can provide more detailed and reliable information on the water cycle and the water balance of different regions and basins. This can help improve the understanding and modeling of hydrological processes and phenomena.
  • Climate science: The new model can contribute to the monitoring and attribution of climate change impacts on GWS. This can help assess the vulnerability and resilience of water systems to climate variability and change.
  • Sustainable water management: The new model can support the planning and management of water resources at different scales. This can help optimize the allocation and use of water for various purposes, such as irrigation, hydropower, drinking water supply, and ecosystem services.
  • Hazard prediction: The new model can enhance the forecasting and early warning of hydrological hazards, such as floods and droughts. This can help reduce the risks and damages caused by these events on human lives, livelihoods, and infrastructure.


The new model for measuring global water storage developed by Gou and Soja is a breakthrough in the field of satellite hydrology. It combines satellite observations with hydrological models using a novel deep learning approach to produce high-resolution and accurate estimates of GWS anomalies. The model promises significant benefits for various domains that depend on GWS information, such as hydrology, climate science, sustainable water management, and hazard prediction.

Recent Blog : Mitigating Climate Change: The Power of Nature-Based Solutions

Leave a Comment