Towards Modeling and Predicting Water Table Levels in California

Water is precious. Current trends suggest it may become increasingly scarce. Water table levels and their predictions can be used to help manage this most important resource.  Using data sourced from NASA’s GRACE satellites, direct measurements, and LANDSAT snowpack images, we have developed a modeling protocol to predict water table levels from the late Spring to early Fall coinciding with the dry season and heavy agricultural water consumption.

Our protocol combines two independent approaches: 1) predicting future water table levels based on previous values and historical trends, and 2) a composite image-processing method which predicts future water table levels based on snowpack data from the LANDSAT in previous months, as well as other data gathered from GRACE satellites. This protocol yielded accuracy over 96% - a marked improvement over a 90% baseline from previous work. The method successfully maintains error rates that do not trend upward over time.