Two Decades of Data Reveal a Hidden Split in South Florida’s Groundwater

Beneath the sawgrass and sloughs of South Florida lies one of the most closely watched water systems on Earth, and after twenty-six years of continuous monitoring, scientists have finally caught it changing its behavior in a way no one could previously quantify. A new analysis of groundwater elevations from 1997 to 2022, spanning the Water Conservation Areas, Everglades National Park, the coastal watersheds, and Florida Bay, has identified statistically significant regime shifts that divide the record into three distinct chapters: a stable period from 1997 to 2005, a decade of decline from 2006 to 2015, and a striking recovery from 2016 to 2022. What makes the finding remarkable is not simply that water levels rose and fell, but that the forces pulling them up and down turn out to be fundamentally different depending on where you stand in the landscape.

The research team, led by Ethiopia B. Zeleke of Florida International University together with colleagues from the South Florida Water Management District, applied an additive seasonal-trend decomposition filter to monthly groundwater records, separating each time series into trend, seasonal, and residual components. They then verified the timing of major transitions using the Chow test for temporal breaks, a statistical procedure that objectively confirms when the underlying behavior of a system has changed rather than merely fluctuated. Two breaks emerged with high confidence: one in 2006 and another in 2016. The first marked the onset of a sustained decline, the second a reversal that pushed groundwater elevations beyond anything seen in the earlier stable period.

The magnitude of these swings is substantial. In the Water Conservation Areas 3A and 3B, mean groundwater elevation during peak wet months fell by roughly 6 percent between the first and second periods, then rebounded by up to 16 percent in the third. Everglades National Park told an even sharper story. There, groundwater dropped by as much as 15 percent during the decline phase, particularly in the dry months of March through May, and then surged to levels exceeding the 1997 to 2005 baseline by up to 27 percent. The declining decade was punctuated by sharp depletion episodes in 2006 through 2008, 2009, 2011 and 2012, and 2015, each corresponding to documented regional droughts.

What drove the recovery is the study’s most consequential clue. The researchers compared cumulative groundwater deviations against two candidate influences: the natural water balance of precipitation minus evapotranspiration, and the volume of water deliberately released through the region’s managed canal and structure network. In the Water Conservation Areas, the correlation between groundwater elevation and managed flows reached 0.795, far exceeding the 0.618 correlation with natural drivers. In Everglades National Park the contrast was starker still, with managed water volume correlating at 0.86 against a mere 0.18 for the natural balance. The post-2016 recovery coincided with managed discharge rebounding from an average of 155 cubic feet per second during the lean years to 259 cubic feet per second afterward. In other words, human decisions about where to send water appear to have outweighed the whims of rainfall in determining what happened beneath the marsh.

The seasonal rhythm of the aquifer adds another layer of technical insight. Rainfall peaks in June, but groundwater in the conservation areas does not crest until October, a three-month lag that reflects the slow transmission of surface inputs through the highly porous surficial aquifer system. In the park, the lag shortens to two months. Seasonal correlations between surface water stage and groundwater elevation were nearly perfect, reaching 0.99 in both regions, confirming the tight hydraulic coupling that characterizes the Biscayne Aquifer, a karst limestone formation so transmissive that surface water and shallow groundwater in many places are effectively one continuous body. Extreme weather left its signature too: Hurricane Irma in September 2017 and Tropical Storm Arlene in 2005 both coincided with pronounced recharge pulses, and the three highest recharge years on record, 2022, 2021, and 2016, all fell within the recovery phase.

Mapping the system spatially revealed where the two diverging stories unfold. Empirical Bayesian kriging interpolation of long-term monthly averages showed groundwater elevations consistently highest in the north, reaching 10.5 feet above the NAVD 88 datum in East Collier, the L-28 Gap, and the conservation areas, then tapering toward and below sea level in the southern coastal watersheds of Taylor Slough, Cape Sable, and Lostmans Slough. Hydraulic gradients across the entire domain hover near 0.0001 feet per foot, a testament to the region’s almost nonexistent topographic relief, yet the flow directions they define remain remarkably stable between wet and dry seasons. Spatial autocorrelation analysis using Moran’s I produced values between 0.78 and 0.83, indicating that wells with similar elevations cluster tightly across the landscape rather than scattering randomly.

Clustering analysis split the monitoring network into two archetypes with profoundly different personalities. The upstream, high-elevation wells showed a seasonal range of 1.46 feet, peaking in October, and a maximum unsaturated depth of just 0.24 feet even in dry years. The coastal, low-elevation wells, by contrast, were saturated year-round with essentially no vadose zone at all, peaked a month earlier in September, and showed a compressed seasonal range of 0.99 feet. Despite similar long-term rates of elevation change, the coastal wells gained slightly more total elevation over the 26 years, and their behavior tracked cumulative sea level change with a correlation of 0.89 while showing essentially no relationship to the natural water balance. The conclusion is difficult to escape: in the freshwater interior, upstream recharge and managed flows dominate, while along the coast, the rising sea has become the governing hand, holding water levels above the ground surface even through dry periods.

The relationship between surface water and groundwater also proved to be spatially heterogeneous in unexpected ways. Across roughly 54 to 61 percent of the study area, depending on season, the two varied in a simple positive linear fashion, strongest in East Collier and the northern park. But coastal watersheds exhibited convex relationships, in which groundwater rises ever faster as surface stages climb, while the southeastern park and upper Taylor Slough showed concave behavior, where groundwater plateaus once surface stages pass a threshold. These nonlinear zones likely reflect local controls on permeability, topography, and tidal influence, and they warn that groundwater response to future management actions will not be proportional everywhere.

Florida Bay delivered perhaps the study’s most sobering numbers. A random forest regression model, trained on data from 2000 to 2016 and validated against 2017 through 2022 with an R-squared of 0.896, ranked six candidate drivers of bay water levels. Sea level emerged as overwhelmingly dominant, accounting for 88.3 percent of the model’s feature importance. Creek discharge contributed just 2.2 percent, rainfall 2.6 percent, and coastal groundwater elevation 2.4 percent. The bay’s water level has been rising at a statistically significant 0.0017 feet per month, and its seasonal peak has shifted from September to October after 2016, mirroring the sea level cycle of the adjacent Gulf of Mexico. The authors suggest sea level may act as an integrative proxy, capturing tidal inundation, saltwater intrusion, and shifts in the hydraulic gradient that can balance or even reverse groundwater discharge into the bay.

For the Comprehensive Everglades Restoration Plan, the largest wetland restoration effort in the world, the implications are double-edged. The good news is that controlled releases demonstrably raise groundwater levels in the freshwater interior, offering managers a powerful lever for buffering droughts, sustaining hydroperiods for wading bird prey, and slowing peat loss. The sobering news is that along the coast, no amount of local manipulation can outmuscle the ocean. Saltwater intrusion already extends 15 to 30 kilometers inland of Florida Bay, cutting off fresh groundwater discharge, and continued sea level rise threatens to accelerate the transition of sawgrass marsh into mangrove swamp. The study’s lasting contribution is a map of where human water management still holds sway and where the sea has taken over, a distinction that will shape restoration strategy for decades to come.

Subject of Research: Long-term regime shifts in South Florida groundwater elevation and the distinct influences of upstream recharge and coastal sea level

Article Title: Regime shifts in South Florida groundwater: distinguishing upstream recharge from coastal sea level influences (1997–2022)

Article References: Zeleke, E. B., Melesse, A. M., Price, R. M., Islam, T., Dessu, S., Strazisar, T., Krupa, S., Hoffart, A., & Sklar, F. (2026). Regime shifts in South Florida groundwater: distinguishing upstream recharge from coastal sea level influences (1997–2022). Environmental Monitoring and Assessment, 198(11), Article 1162. https://doi.org/10.1007/s10661-026-16000-x

Image Credits: AI Generated

DOI: 10.1007/s10661-026-16000-x

Keywords: groundwater, Everglades, Florida Bay, sea level rise, regime shifts, hydrology, Biscayne Aquifer, water management, saltwater intrusion, random forest, surface water, South Florida