California’s Central Valley—one of the nation’s most critical agricultural regions and home to over 1.3 million people—is prone to flooding. Mapping the extent of winter floods has been challenging for experts, however, because clouds can obscure the view of satellites.
Recent efforts to improve satellite flood mapping have been incorporated into a new study that offers insight into where winter flooding is occurring and inform how floodwaters can be used to replenish depleted aquifers.
The research, published in the Journal of Flood Risk Management, examined 20 years of satellite imagery to identify the extent and location of winter flooding in the region.
The midwinter months of December through February were found to have the highest likelihood of floods, particularly when atmospheric rivers brought heavy rains when soils were already saturated.
The study also identified areas where floodwaters fail to percolate through soils and offers suggestions for using the water to replenish rapidly depleting groundwater aquifers.
By examining insurance claim data and overlays of floodwaters and buildings, researchers also found that flood exposure was actually higher, by value, for buildings outside of officially designated flood boundaries. The study’s findings can be visualized in three interactive maps.
“We know that atmospheric rivers and winter precipitation are big drivers of flooding, and we can see that in stream flow gauge records,” said Christine Albano, ecohydrologist at DRI and lead author of the study.
“But we really had no data on how that water is dispersed across the landscape over the historical record, because cloudy winters obscure the view of Landsat imagery, which is only captured once or twice a month. By using daily MODIS imagery, we increase the odds of capturing a glimpse of the land surface.
“This fills an important gap in our understanding, because winter is the time when flood risks are greatest and when excess water is most available for groundwater recharge—so it is essential that we know where water is during this time of year.”
The researchers wanted to account for the extensive water management that occurs in the agricultural region, as well as the influence of atmospheric rivers. To do this, they combined the satellite imagery with precipitation and soil moisture data from upstream regions. This allowed them to identify where flooding occurs due to rainfall, rather than the intentional flooding sometimes used by water managers for purposes like flooding rice fields.
The Central Valley is known to be sinking at a rapid pace—with parts sinking over one foot per year—due to groundwater extraction. The maps offer a way to pinpoint where floodwaters exist and aren’t able to penetrate the ground surface.
Most of these areas are within 5km of soils with better permeability, the study found, and floodwaters could be redirected to these locations to recharge the aquifers below. Alternatively, the compacted soils in flooding areas could be tilled to better allow water to penetrate.
“We now have the methods and information we need to support ongoing water management efforts to redirect hazardous floodwater to key locations where depleted groundwater basins can be replenished so that rural communities and ecosystems have access to water in the dry season,” said Melissa Rohde, who co-authored the study.
“This is increasingly important as atmospheric river events intensify under a warming climate and local groundwater sustainability agencies work hard to achieve groundwater sustainability by 2040 under California’s Sustainable Groundwater Management Act.”
The tradeoff, however, is that the coarser resolution of MODIS imagery has limitations for urban areas, because the data can’t reliably distinguish between black asphalt and black water. The MODIS satellites also offer 20 years of data, rather than the 50+ years offered by LANDSAT, which means that some of the older, larger floods aren’t captured.
“We weren’t able to visualize some of the biggest floods, like in 1997,” Albano says. “But the smaller and more common floods are the ones impacting people living in the floodplains more frequently. Our maps offer a view of where the higher-frequency floods are occurring.”
The research methods can be replicated for other regions in the U.S. to identify flood risk and groundwater replenishment potential. In the future, Albano would like to utilize even newer satellites like the Sentinel constellation, which provide higher resolution, but which don’t yet offer more than a few years of data.
“By integrating Sentinel 1 and Sentinel 2 imagery with Landsat and MODIS data, we can create very dense map stacks of inundation information,” said Chris Soulard of the USGS, who co-authored the study.
“Sentinel 1’s radar technology allows for all-weather monitoring, while Sentinel 2’s optical imagery provides high-resolution insights into surface conditions. This combination of freely available image collections enables us to create a comprehensive and timely record of flooding events.”
The study’s interactive maps provide three ways to view the data:
- Monthly Landsat (1984–2023) and MODIS (2003–2023) surface water classifications based on the USGS Dynamic Surface Water Extent (DSWE) algorithm, which can be used to look at specific flooding events.
- Monthly frequencies of MODIS High Confidence Water (DSWEmod), which can be used to understand how often surface water occurs at different times of year.
- Probability of Precipitation-Driven Surface Water Occurrence, for different monthly precipitation amounts based on the results of this analysis.
More information: Assessing Causes and Consequences of Winter Surface Water Dynamics in California’s Central Valley Using Satellite Remote Sensing, Journal of Flood Risk Management (2025). DOI: 10.1111/jfr3.70080
Provided by Desert Research Institute