Skip to main content
U.S. flag

An official website of the United States government

Evapotranspiration (ET) presents a difficult problem.

Evaporation from the Earth’s surface and transpiration from the leaves of plants accounts for about 60 percent of the water budget – a budget that has implications for agriculture and food security the world over.

map showing locations of flux towers in evapotranspiration study
This figure from “Operational Global Actual Evapotranspiration: Development, Evaluation, and Dissemination” a recently-released publication authored by EROS scientists Gabriel Senay, Stephanie Kagone and Naga M. Velpuri, shows the location of the flux towers used to check the accuracy of the EROS-based evapotranspiration products.

But ET is not as easy to track as the water we can see. Rainfall is measured with rain gauges, river flow with streamgages. Snowfall can be measured with a tool as simple as a ruler. But water vapor is an invisible gas, best measured by expensive flux towers.

That simple reality – that the largest part of the water budget is both invisible and hard to measure – helps explain why remote sensing scientists have poured so much effort into building and improving satellite-based ET maps in recent years, using tools like the Operational Simplified Surface Energy Balance (SSEBop) model created by EROS scientist Gabriel Senay. 

It also explains why Senay’s global ET products have gained traction since their distribution began in 2014 through the Famine Early Warning Systems Network (FEWS NET) data portal. The portal offers maps of both ET anomalies – deviations from the norm — and actual ET in millimeters. These are updated weekly and available for download at no cost. For Continental Africa alone, around 150 users a month visit the portal to check for ET anomalies.

Anomalies and actual ET figures are both valuable. Anomalies can send a warning about impending drought; actual ET can guide decisions on how much water to release for irrigation and where. Until now, however, users haven’t had a peer-reviewed study to tell them how accurate the ET estimates are and how best to use them in either context.

“People are using this product, but there was really no publication behind it,” Senay said. “They were relying on a methods publication for the SSEBop model applied in the U.S. or other location-specific cases. But this is a global product, and people don’t know how it was generated, what kind of accuracy to expect, and how to remove a bias if they need to. Those kinds of qualifying statements weren’t there.”

The new publication aims to fill that gap. Senay is the lead author of “Operational Global Actual Evapotranspiration: Development, Evaluation, and Dissemination,” which was released this spring as an open access publication.

In it, Senay and his EROS co-authors compared their satellite-derived ET with ET readings from 12 flux towers around the globe. They found that the ET product captures 87 percent of seasonal variability across the globe, making it appropriate for drought monitoring as-is. The actual ET showed regional biases, they found, but that a one-time correction could account for those and improve its value as a water budgeting tool.

“Bias can be dangerous when you don’t know it,” Senay said. “But once you know it, you know it does not change from year to year. It’s a systematic error, like if your measuring stick is off by one inch. Once you know that, you just subtract that inch.”

In countries without access to quality ET data, the publication offers more confidence to decision-makers who might use it to make water management decisions. For Senay and his team, pulling back the curtain on the global data’s methods and biases in an open access paper amounts to a call for the kind of constructive criticism and input that can improve the product.

“It’s like hiring many, many more scientists for free,” Senay said. “It’s painful when they tell you it doesn’t work, but in the end it’s a very good thing.”

That Senay’s model performed well tracks with the expectations of hydrologist Amy McNally of NASA and the U.S. Agency for International Development (USAID). SSEBop scored well in a previously-published study that compared the results of nine satellite-derived ET models.

McNally has used SSEBop-modeled data in comparisons with land surface models, as well as for other unpublished comparisons of modeled ET. Data quality, regular updates and open approach taken by the EROS team and its partners makes the SSEBop approach a particularly useful contributor to the global conversation on ET as a driver of decision.

“In addition to its good performance, as shown in a peer review literature, this product is updated routinely, publicly available, and has global coverage,” McNally said. “That is really important for the hydrologic research community trying to better understand the water cycle, the science applications community exploring real world problems, as well as organizations that routinely monitor conditions to support decision makers.”

Publishing the results of an accuracy assessment helps get the word out to those groups, McNally said, which boosts the likelihood that the data will be folded into other models and improve a variety of applications.

“This information on vegetation water use could be used for the monitoring and evaluation of programs interested in water conservation and water use efficiency, as well as crop production estimation and forecasting,” McNally said.

The spirit of collaboration – perhaps best characterized by the OpenET project on which Senay is a collaborator – will become especially important in the near future. The current ET maps on the FEWS NET portal are built from data collected by the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor on NASA’s Aqua satellite. MODIS is nearing the end of its life, which means the ET estimates from SSEBop and other models that rely on it will need to be produced using data from its replacement: the Visible Infrared Imaging Radiometer Suite (VIIRS).

Senay hopes the new publication generates feedback that will help smooth the transition.

“Perhaps there are some areas where we need to be really careful and re-parameterize the model,” Senay said. “The feedback will really help us do that when we move to our next version.”

Related Content