Near Real-Time MODIS Global Flood Product Now Available from NASA’s LANCE
New NRT Global Flood product from LANCE will continue production of data critical for detecting floods around the globe.
By Joseph M. Smith, NASA EOSDIS Science Writer
Whether caused by heavy rains, storm surge, high tide, or a failing dam or levee, floods occur throughout the world and can have devastating consequences for people, ecosystems, and economies. Floods can happen in minutes or span several hours and last days, weeks, or longer. Further, flooded areas can be difficult to access and dangerous to work in, which makes it difficult to acquire information about a flood’s extent and impact. Therefore, the ability to accurately detect and assess the extent of floods via remote sensing is critical to mitigating, monitoring, and responding to these destructive events.
NASA's Goddard Space Flight Center in Greenbelt, Maryland, has worked to meet this need with the Near Real-Time (NRT) Global Flood Mapping product, which it created in 2011. This product was developed through a partnership between the Dartmouth Flood Observatory (which has since relocated to the University of Colorado, Boulder), and a team at Goddard. In 2017, NASA's Applied Sciences began transitioning the NRT global flood product to NASA’s Land, Atmosphere Near real-time Capability for EOS (LANCE), with additional support from NASA’s Earth Science and Data Information System Project (ESDIS) and LANCE MODIS, a division within LANCE that processes and distributes NRT data from the Moderate Resolution Imaging Spectroradiometer (MODIS) instruments aboard NASA’s Aqua and Terra satellites.
In March, the new and improved flood mapping product, the MODIS NRT Global Flood Product (also known as the MODIS Aqua+Terra Global Flood Product L3 NRT 250m, or MCDWD_L3_NRT), was released in LANCE. The legacy flood mapping product will be discontinued at some point in 2022.
Like its predecessor, the new product provides daily global flood maps based on imagery from the MODIS instruments aboard NASA’s Aqua and Terra satellites. Two benefits of the new LANCE product include more robust production and lower latency (by a few hours) meaning it is available to users in a more-timely manner. The product is also being made available in NASA’s Global Imagery Browse Services (GIBS) and NASA Worldview, which allows users to browse imagery and data more easily than possible with the legacy product.
“Having the product in GIBS and Worldview will allow users to more easily browse the products than they could in our legacy system, where there would be a snapshot of the tile, but they’d have to look tile-by-tile and download it to a computer and then into a GIS application, so I think for the end user that’s a huge advantage,” said Dan Slayback, research scientist in Goddard’s Biospheric Sciences Laboratory.
Like the legacy product, the LANCE product is derived from the NRT MODIS Surface Reflectance datasets and the process of generating it involves three key steps: applying the water detection algorithm to each MODIS observation (from both the Aqua and Terra satellites); compositing water detections over time, to reduce errors (principally from cloud shadow false positives) and more rigorously identify water; and differentiating flooding from locations of expected surface water. Given that the best product for a given flood event depends on four unpredictable factors — the specific location or area of interest, cloud cover, spatial extent of likely flood water, and the likely duration of flooding — the best product for a user depends on the particular event characteristics and the user’s specific needs. For casual browsing, the 3-day composite product is often the most useful.
“The product is based on optical data, which means that the sensor does not see through the clouds, so it is unable to detect water on the surface,” said Slayback. “What the water compositing algorithm does is composite or add-up water observations over 2 or 3 days so, when the clouds move, you might see down to the surface and capture the water. You have a chance within a day for the clouds to move and, over 2 or 3 days, you have even more chances.”
There are, however, some important differences between the LANCE and legacy flood products. Among the most significant for current users are a change in the file’s data values, to accommodate a “recurring flood” pixel value (currently not yet populated). Another is that the product tile grid now uses a standard MODIS naming convention (so the tile covering the mid-Atlantic region is named h10v05 instead of 080W040N in the legacy system; see below figure). The product is also distributed in a different file format, as one Hierarchal Data Format (HDF) file per tile, per day, that includes all the flood products, whereas the legacy product is delivered in individual GeoTIFF files per product per day. However, along with the core HDF product, LANCE is also distributing individual GeoTIFF files for each composite product. Tile and pixel sizes) are also slightly different. Each HDF file contains 12 raster layers. These include four flood layers (1-Day, 1-Day CS, 2-Day, and 3-Day), along with ancillary layers that allow users to construct alternative composites of their own. (Users should note that the 1-day product is not yet available in Worldview.)
Given these differences, the MODIS NRT Global Flood Product will be rolled out in stages. The current beta release replicates the legacy product and is intended to allow users to get accustomed to the new product without experiencing any major changes in the data product itself. Therefore, both the new and legacy products will be available simultaneously for a period of 3 months (April through June, 2021).
“We’re making them both available for a transition period so users who are used to the old product have some time to get the new product because it’s different,” said Slayback. “The product itself has different pixel values in it, so if users have any sort of automated display routines for it, they’ll have to adapt them.”
Workflow adjustments aside, there can be little doubt that the product’s users, be they emergency response personnel looking for flood extent for a current event, academic researchers interested in flood extents for modeling or other studies, or insurance agencies looking to update flood zones, will benefit from the more-timely data, more robust processing, and easier browsing accompanying the new global flooding product from LANCE.
In fact, Frederick Policelli, a member of the Goddard Hydrological Sciences Laboratory, said he recently received some feedback from the World Food Program, the largest humanitarian aid organization and winner of the 2020 Nobel Peace Prize, which uses the product to help plan its relief efforts.
“They said that they use our product quite a bit,” he said. “They want to know where it’s not flooding so that they can station supplies and their people to aid communities. They don’t want to have their supplies flooded out and their people compromised, complicating the situation even more.”
Following the three-month overlap period, the legacy product will be retired. Slayback added, however, that the legacy product will be kept and the files it generated will remain accessible as long as necessary.
“Having [the product] in LANCE will allow us to back-process the archive,” said Slayback. “So, we’ll have historical data, which is really helpful to compare current floods to past floods in the same locations. That’s a big advantage and we’re hoping to run that over the next year in different stages.”
There are other improvements planned for the product as well. In addition to improving the product’s quality, there will be an update to the reference water map the product uses to determine whether detected water is normal or expected (i.e., an existing lake or river), or flood water. The product currently uses the original version (2009) of the Terra MODIS Land Water Mask, but this has become increasingly out of date as normal surface water extents have changed over the past 1-2 decades.
“The reference water map has gotten out of date in many places so, where rivers have changed or lakes have changed, or new reservoirs have been constructed, those get reported as floods when it’s really normal, expected water,” Slayback said. “We don’t have a layer that tells us that. So, we’ll be generating a new reference water layer from the product’s own data, running at least 10 years back and then compiling it.”
In addition, Slayback and his team plan to add what he refers to as a “recurring flood flag,” which will be able to differentiate "normal" or expected annual flooding, from more unusual events.
In the meantime, users can download the beta release of the MODIS Global Flood product via the LANCE website and imagery are available via GIBS and Worldview. Users are advised to view the 2- and 3-day flood products in Worldview, as this will enable them to rapidly browse through recent dates and view the product along with MODIS corrected reflectance imagery, which will make clouds and any anomalies in the product easier to see.
LANCE is part of NASA’s Earth Observing System Data and Information System (EOSDIS) and distributes NRT data products from almost a dozen satellite-borne instruments, typically within three hours of data acquisition. EOSDIS provides end-to-end capabilities for managing NASA’s Earth science data. These data represent some of the most complex and diverse Earth science data sets on the planet from satellites, aircraft, field measurements, and numerous other EOSDIS programs. The primary services provided by EOSDIS are data archive, management, and distribution; information management; product generation; and user support services. These services are managed by NASA’s ESDIS project.
For More Information:
To download data from the MODIS NRT Global Flood Product, visit the EOSDIS LANCE website.
Discover and access NASA NRT data using LANCE
View the MODIS NRT Global Flood Product as a layer in Worldview. Please Note: Currently, the NRT Global Flood product is populated in Worldview only at the end of the day. This is being addressed and should be resolved by mid-December 2021.
Published October 25, 2021
Page Last Updated: Nov 3, 2021 at 12:24 PM EDT