Driving Processes of the Niland Moving Mud Spring: A Conceptual Model of a Unique Geohazard in California’s Eastern Salton Sea Region

Abstract

The Niland Moving Mud Spring, located near the southeastern margin of the Salton Sea, represents a rare and evolving geotechnical hazard. Unlike the typically stationary mud pots of the Salton Trough, this spring is a CO₂-driven mud spring that has migrated southwestward since 2016, at times exceeding 3 m per month, posing threats to critical infrastructure including rail lines, highways, and pipelines. Emergency mitigation efforts initiated in 2018, including decompression wells, containment berms, and route realignments, have since slowed and recently almost halted its movement and growth. This study integrates hydrochemical, temperature, stable isotope, and tritium data to propose a refined conceptual model of the Moving Mud Spring’s origin and migration. Temperature data from the Moving Mud Spring (26.5 °C to 28.3 °C) and elevated but non-geothermal total dissolved solids (~18,000 mg/L) suggest a shallow, thermally buffered groundwater source influenced by interaction with saline lacustrine sediments. Stable water isotope data follow an evaporative trajectory consistent with imported Colorado River water, while tritium concentrations (~5 TU) confirm a modern recharge source. These findings rule out deep geothermal or residual floodwater origins from the great “1906 flood”, and instead implicate more recent irrigation seepage or canal leakage as the primary water source. A key external forcing may be the 4.1 m drop in Salton Sea water level between 2003 and 2025, which has modified regional groundwater hydraulic head gradients. This recession likely enhanced lateral groundwater flow from the Moving Mud Spring area, potentially facilitating the migration of upwelling geothermal gases and contributing to spring movement. No faults or structural features reportedly align with the spring’s trajectory, and most major fault systems trend perpendicular to its movement. The hydrologically driven model proposed in this paper, linked to Salton Sea water level decline and correlated with the direction, rate, and timing of the spring’s migration, offers a new empirical explanation for the observed movement of the Niland Moving Mud Spring.

1. Problem Statement

Geothermal features such as mud pots, mud volcanoes, and mud springs are common along the margins of California’s Salton Sea. These features, typically stationary, are driven primarily by carbon dioxide (CO₂), with minor contributions from gases like hydrogen sulfide, ammonia, and methane which seep upward from subsurface sources below the water table [1,2,3]. The source of the CO₂ remains uncertain. Ref. [2] suggests that it may originate either from plutonic gases released by shallow magma bodies known to occur in the region, or from the hydrothermal alteration of carbonate rocks into greenschist facies metamorphic rocks. Around 2011, an unusual phenomenon emerged near Niland, California; a large CO₂-driven mud spring began migrating southwest (Figure 1 and Figure 2). Now known as the Niland Moving Mud Spring or “W9a,” and referred to as the Moving Mud Spring or W9a henceforth, this migrating feature has posed significant threats to critical infrastructure, including the Union Pacific Railroad mainline tracks, petroleum pipelines, fiber-optic lines, and California State Route 111 (Figure 1 and Figure 2).

The region lies within the Salton Trough, a tectonically active pull-apart basin shaped by the interaction of the San Andreas, Imperial, and Cerro Prieto faults. This geologically complex zone is characterized by high geothermal gradients, thick sedimentary basins composed of Colorado River alluvium and lacustrine sediments, and extensive subsurface gas migration [1]. Mud pots and mud volcanoes form when CO₂ and other gases entrain groundwater through fine-grained, saturated sediments, producing fluid-filled pots or, when more viscous, cone-shaped mud volcanoes. Unlike artesian springs, which are driven by hydraulic pressure and confined aquifers, these gas-pressurized systems respond to the flux and pressure of rising gases [1].

The predecessor spring feature, W9, which does not appear in 2002 aerial imagery, likely formed or reactivated around 2005 (Figure 2). It later evolved into a second migrating spring, W9a, as W9 began expanding after 2012 and separated in 2016, with W9a diverging and moving southwest [4]. Initially advancing just a meter or two per year, W9a’s rate of movement accelerated to nearly 3 m per month for a short period of time. By 2021, W9a had migrated more than 100 m from W9’s 2016 location [5]. It reportedly remains the only known example of a fast-moving, gas-driven mud spring of this scale [4,5,6,7], although small amounts of movement of mud pots or mud springs in other areas has occasionally been noted, over decades [8]. Ref. [4] suggests that the movement is facilitated by a pressurized subsurface conduit of CO₂ and liquefied sediments, forming a deformable pathway of least resistance.

Due to additional containment measures, particularly spring dewatering and local degassing, W9a’s movement slowed significantly after initial rapid movement. By 2023, W9a appeared to stabilize, and as of 2025, it remains relatively stationary, although the mud spring’s surface dimensions have continued to expand. The original W9 spring no longer exhibits visible surface activity, while W9a remains the primary threatening feature (Figure 2). Although currently stabilized, its long-term behavior and potential risk to surrounding infrastructure remain uncertain.

2. Background: Purpose, Study Area, and Physical Characteristics

2.1. Purpose of Study

This paper proposes a new conceptual model for the movement of the W9a Moving Mud Spring, based on recent hydrochemical and environmental isotope data that identify the spring’s shallow water source. The primary research question is: What are the physical and geochemical controls on the migration of the W9a spring, and to what extent is shallow groundwater versus deeper sources responsible for its movement? To address this question, we test two hypotheses: (1) the spring water is sourced from shallow, recently recharged groundwater rather than deep geothermal aquifers, and (2) changes in hydraulic head, driven by Salton Sea level decline, control the lateral migration of CO₂ and entrained sediments to the surface.

These findings augment the framework developed by [2,3] by adding a hydrologic model, offering a more complete understanding of the geochemical and physical processes driving its continued migration. Earlier models of the Moving Mud Spring’s southwestward migration focused on a tilted subsurface conduit or “soil pipe,” through which CO₂-charged water rises. Gas buoyancy drives upward flow, while internal erosion or “sapping” causes sediment collapse within the conduit, which can block or redirect flow and shift the surface vent without changing the deep source [2,3]. Gravitational settling removes sediment from higher portions and deposits it downslope, gradually displacing the vent. Movement would stop once the spring aligns vertically with its deep source, with sediment behavior governed by particle size, cohesion, and water velocity. Ref. [2] likened this to streambed aggradation, linking sediment dynamics to flow conditions.

A “hydra model” was also proposed, where pressurized water and CO₂ find multiple near-surface pathways as overburden sediments liquify [3]. Local collapses block flow, redirecting discharge to new branches and creating alternating surface vents, allowing lateral “jumping” of mud springs and mud pots. Depictions of these models and related figures can be found in [3] in open source literature. According to [2], movement could also occur along small, unmapped faults or desiccation cracks in ancient Lake Cahuilla and earlier lake sediments, although no major faults are known to exist at the site parallel to the direction of spring movement [2,4]. Moisture lineaments south of State Route 111 may indicate fault-related permeability zones [2], but could also reflect changes in sediment texture and leakage from nearby ponds and canals (Figure 1 and Figure 2).

The model presented here complements these hypotheses while emphasizing shallow groundwater changes as the primary driver. Faults descending to great depth would affect only gas movement, as isotope and geochemical data show the spring water is shallow. Spring migration correlates with a 4.1 m drop in Salton Sea levels since 2002, suggesting changes in hydrostatic pressure and lateral groundwater flow may influence CO₂ pathways and surface expression. This highlights the critical role of evolving shallow hydrologic conditions in driving geotechnical hazards in geothermal terrains. By explicitly framing the research question and hypotheses, this study provides a structured test of prior conceptual models and identifies measurable criteria for future investigations.

2.2. Study Area

The study area includes the Wister Unit and agricultural lands surrounding the unit. The Wister Unit of the Imperial Wildlife Area, managed by the California Department of Fish and Wildlife (CDFW), is a network of seasonally flooded wetlands near Niland, California (Figure 1). Constructed levees divide the area into ponds filled with Colorado River water via the Coachella Branch Canal, supplemented by agricultural drain flow. These wetlands provide critical habitat for migratory waterfowl and support recreational activities such as duck hunting and birdwatching [9]. Water is typically added in fall and winter to coincide with migration and hunting seasons. Some ponds are filled earlier to accommodate early migrants. The Salton Sea’s retreating shoreline has exposed large areas of playa, creating both dust hazards and opportunities for wetland restoration [10,11]

East and southeast of the Moving Mud Spring, extensive agricultural lands form part of one of the most productive irrigated regions in the United States. Supplied primarily with imported Colorado River water delivered via the Coachella Branch of the All-American Canal, farms in this area cultivate a wide variety of crops across the alluvial and lake plains. Major commodities include alfalfa, cotton, wheat, lettuce, sugar beets, and several types of melons, while newer plantings of olives and other specialty crops are becoming increasingly common. The combination of fine-textured soils, engineered irrigation systems, and extensive drainage networks enables intensive crop production in this arid environment.

2.3. Physiography, Drainage, and Climate

Physiographically, the project area lies near the northeastern margin of the Imperial Valley within the Salton Trough, a segment of the Basin and Range Province that extends across inland California, much of the southwestern United States, and northern Mexico. The Salton Trough includes the Imperial Valley to the south and the Coachella Valley to the north, forming a low-lying structural depression bounded by the Coyote and Jacumba Mountains to the west, the Chocolate and Orocopia Mountains to the northeast, the Sand Hills and Cargo Muchacho Mountains to the southeast, and the U.S.–Mexico border to the south [12]. The elevated margin of the valley in the eastern part of the study area is known as the East Mesa.

Major canals in the area include the Coachella Canal, a branch of the All-American Canal, and the East Highline Canal, along with subsidiary lateral canals that support agriculture and the Wister Unit (Figure 3). Originally unlined, the Coachella Canal lost water through seepage into alluvial sediments [13]. In 1979, the first 79 km of the canal were lined with concrete, with the downstream end of lined canal extending due east of Niland [13]. By 2007, the remaining 59 km, including the segment near the study area, were lined, substantially reducing seepage into the underlying alluvial aquifer [14,15]. The entire lined stretch of the Coachella Branch Canal illustrated in Figure 3 was completed in 2007.

The region has a hot desert climate, with mean monthly air temperatures ranging from 12.7 °C in December to 33.1 °C in August [12]. Average annual precipitation at Niland is approximately 73 mm, mostly occurring between December and March and increases with elevation to 100 to 150 mm year⁻¹ in the adjacent Chocolate Mountains [12,16]. Potential evapotranspiration is extremely high, ranging from 2000 to 2500 mm year⁻¹ in the Wister Unit to 2500 to 2800 mm year⁻¹ in the surrounding mountains [12,17]. Natural groundwater recharge generally occurs only during rare, intense storm events that generate flows in ephemeral channels and drainages [15,18].

2.4. Geology

The Salton Trough is a tectonically active pull-apart basin undergoing crustal thinning and subsidence due to extensional tectonics. The study area within the Salton Trough includes both the San Andreas and Imperial fault zones, which are structurally linked and together influence regional tectonics and groundwater flow. Continued subsidence in this topographically low area has led to the accumulation of thick sediments, producing an estimated vertical relief of over 4250 m through faulting, folding, and warping [12,18]. The geology and geomorphology of the region have also been shaped by repeated high stands of prehistoric Lake Cahuilla, which left lacustrine sediments and well-defined paleo-shorelines (Figure 3). The Salton Trough, a structural and topographic depression, serves as the landward extension of the Gulf of California [19,20,21]. Formed in late Cenozoic time by seafloor spreading along the northern section of the East Pacific Rift [22,23,24], the trough is separated from the Sea of Cortez by a low-relief divide composed of Colorado River Delta sediments [19,25]. It comprises a crystalline basement complex filled with Cenozoic sedimentary rocks up to 6100 m thick above crystalline bedrock [23]. Precambrian basement rocks intruded by Mesozoic plutons and Tertiary volcanic rocks form the foundation, particularly visible in the southern Chocolate Mountains [26,27,28]. Overlying this complex is a sequence of Tertiary and Quaternary sedimentary deposits emplaced in shallow-marine, lacustrine, or saline-lake environments [29,30,31].

The Imperial Formation, a Late Miocene to mid-Pliocene unit, is the oldest lacustrine deposit identified in the basin, with marine sediments and fanglomerates up to 1200 m thick [31,32]. Above it, the mid-Pliocene to lower-Pleistocene Palm Springs Formation reaches 2100 m (6890 ft) in thickness and contains lacustrine and fanglomerate alluvial facies [31,33]. Successive deposits include those of ancient Lake Borrego and Lake Brawley, followed by more recent gravels, sands, and silts overlain by lacustrine sediments from historic Lake Cahuilla [33,34,35]. Ref. [31] also documented several large lakes filling the basin in modern times, present in 1849, 1861, and 1891, with the Salton Sea partially filling the basin since 1906, referred to in this paper as “the great flood” [36]. Additional surficial and near-surface deposits, such as Quaternary alluvial fans and ephemeral sand channels, extend from the Chocolate Mountains onto the floor of the Salton Trough. These deposits interfinger with Quaternary lake sediments, forming well-defined shorelines, including the one mapped just below the Coachella Canal [13] (Figure 3).

Within the study area, the Wister Unit and surrounding areas span about 127 km² and occupy the northeastern margin of the Salton Trough. The geological formations oucropping in the area are a subset of the overall geological formation in the Salton Sea area (Figure 3). The study area is underlain by a mix of young and older sediments and volcanic rocks (Figure 4, Table 1). Recent alluvium (Qa) of gravel, sand, silt, and clay dominates river channels and foothills, with local dunes and windblown sands. Lake Cahuilla deposits (Ql) and the Brawley Formation (Qc) reflect past lake and deltaic environments, containing clays, silts, sands, gravels, and evaporites. Recent volcanic rocks (Qb) include rhyolite, pumice, and obsidian domes interbedded with lake deposits, along with small Quaternary rhyolite plugs (v). Older formations such as the Borrego Formation (QTb) preserve lacustrine claystones and minor sandstones with fossil mollusks and ostracods. Beneath these lie much older granitic, metamorphic, and carbonate rocks (bc) ranging from Mesozoic to Precambrian in age. To the northeast, the Chocolate Mountains expose Tertiary and older igneous and metamorphic rocks whose piedmont slope consists of poorly sorted alluvial and fluvial deposits with sparse vegetation [12,18].

The Salton Trough is one of the most geothermally active regions in North America [1]. High heat flow from relatively shallow magma bodies, part of the broader tectonic and volcanic system at the Pacific–North American plate boundary, produces an elevated geothermal gradient that alters shallow sediments [37] and supports numerous surface geothermal expressions such as mud pots, mud volcanoes, fumaroles, and CO₂ vents [38,39,40]. These features are concentrated near the southern and eastern edge of the Salton Sea, including parts of the Wister Unit where fault systems and active geothermal fields converge [1].

Figure 3. Geologic map of the study area on the northeastern margin of the Salton Trough. The map displays Quaternary alluvium (Qa), Lake Cahuilla deposits (Ql), Borrego Formation (QTb), recent volcanic rocks (Qb), and older granitic, metamorphic, and carbonate basement rocks (bc). These units reflect the interplay of repeated lacustrine, volcanic, and alluvial processes that shaped the modern landscape (modfied from [40]).

Figure 3. Geologic map of the study area on the northeastern margin of the Salton Trough. The map displays Quaternary alluvium (Qa), Lake Cahuilla deposits (Ql), Borrego Formation (QTb), recent volcanic rocks (Qb), and older granitic, metamorphic, and carbonate basement rocks (bc). These units reflect the interplay of repeated lacustrine, volcanic, and alluvial processes that shaped the modern landscape (modfied from [40]).

Figure 4. Generalized stratigraphic column of the mapped geologic units in the study area. The column shows the vertical succession from the oldest basement rocks (bc) through Borrego (QTb) and Brawley (Qc) Formations to Lake Cahuilla deposits (Ql), recent volcanic rocks (Qb, V), and surficial alluvium (Qa). This sequence highlights the transition from older lacustrine and deltaic strata to the younger volcanic and alluvial deposits exposed today (modified for study area rocks only, from [40]).

Figure 4. Generalized stratigraphic column of the mapped geologic units in the study area. The column shows the vertical succession from the oldest basement rocks (bc) through Borrego (QTb) and Brawley (Qc) Formations to Lake Cahuilla deposits (Ql), recent volcanic rocks (Qb, V), and surficial alluvium (Qa). This sequence highlights the transition from older lacustrine and deltaic strata to the younger volcanic and alluvial deposits exposed today (modified for study area rocks only, from [40]).

Table 1. Geologic descriptions of units depicted in Figure 3 and Figure 4 (modified from [40]).

Map Unit	Geologic and Lithologic Descriptions
Qa	Recent Alluvium: Gravel, sand, silt, and clay. Occurs along river channels and on foothills above the ancient shorelines of Lake Cahuilla. Windblown sands and local dunes are grouped in this unit
Qb	Recent Volcanic Rocks: Obsidian, rhyolite, and pumice composing volcanic domes on the southeast shore of Salton Sea. Locally interbedded obsidian flows and lake deposits
Ql	Lake Cahuilla Deposits: Clay, silt, sand, beach gravel, and evaporite deposits of the former extensive lake, includes older lake beds above ancient shoreline of Lake Cahuilla, and locally undifferentiated alluvium
v	Quaternary rhyolite plugs along Salt Creek and other areas of Salton Sea
Qc	Brawley Formation: Grouped with Ql on Figure 3, Geologic Map. Red-gray claystone, siltstone, sandstone, and pebbly gravel deposits
QTb	Borrego Formation: Light gray claystone and minor amounts of buff sandstone of lacustrine origin; contains a lacustrine fauna of minute mollusks, ostracods, and rare foraminifera
bc	Granitic rocks. Pre-Cretaceous schist. Gneiss, limestone, schist, and granitic rocks ranging in age from Mesozoic to Precambrian

Table 1. Geologic descriptions of units depicted in Figure 3 and Figure 4 (modified from [40]).

Map Unit	Geologic and Lithologic Descriptions
Qa	Recent Alluvium: Gravel, sand, silt, and clay. Occurs along river channels and on foothills above the ancient shorelines of Lake Cahuilla. Windblown sands and local dunes are grouped in this unit
Qb	Recent Volcanic Rocks: Obsidian, rhyolite, and pumice composing volcanic domes on the southeast shore of Salton Sea. Locally interbedded obsidian flows and lake deposits
Ql	Lake Cahuilla Deposits: Clay, silt, sand, beach gravel, and evaporite deposits of the former extensive lake, includes older lake beds above ancient shoreline of Lake Cahuilla, and locally undifferentiated alluvium
v	Quaternary rhyolite plugs along Salt Creek and other areas of Salton Sea
Qc	Brawley Formation: Grouped with Ql on Figure 3, Geologic Map. Red-gray claystone, siltstone, sandstone, and pebbly gravel deposits
QTb	Borrego Formation: Light gray claystone and minor amounts of buff sandstone of lacustrine origin; contains a lacustrine fauna of minute mollusks, ostracods, and rare foraminifera
bc	Granitic rocks. Pre-Cretaceous schist. Gneiss, limestone, schist, and granitic rocks ranging in age from Mesozoic to Precambrian

2.5. Hydrogeology

The study area lies at the boundary between the East Salton Sea Groundwater Basin and the Imperial Valley Groundwater Basin, with State Route 111 roughly marking the divide between the two groundwater basins (Figure 1 and Figure 3). The East Salton Sea Basin covers the northeastern Imperial Valley, including East Mesa and alluvial deposits of the Chocolate Mountains, while the Imperial Valley Groundwater Basin encompasses agricultural lands and the Wister Unit wetlands and wildlife complex [12,41].

Both basins are dominated by unconsolidated Quaternary and semi-consolidated Tertiary–Quaternary alluvium. Valley-fill deposits consist of coarse sands and gravels of high permeability interbedded with silts and clays, which become more prevalent near the Salton Sea. Boundaries of the study area are defined by Chocolate Mountains to the east and north and Salton Sea to the West (Figure 1 and Figure 3). Imperial Valley lies on the southern side of the study area. Aquifers in East Mesa and Imperial Valley are typically unconfined, derived from Colorado River or ancient Salton Sea lake deposits [15,42]. The East Salton Sea Basin is classified as “C” by [43], indicating insufficient data to compute a groundwater budget. The hydrogeology of the Imperial Valley portion of the study area is less understood, with water-bearing strata increasingly saline and limited to almost no wells, with the exception of a few water level monitoring wells. Fine-grained sediments increase toward the Salton Sea, reflecting lacustrine deposition, while sandy layers in the lake shore sediments originate from ephemeral arroyo runoff [43,44,45].

A geothermal test well drilled about 1 km east of Niland reached 1036 m below ground surface (bgs) but had non-economic temperature gradients. Temperature logs in the test well suggest a shallow aquifer at 12 to 15 m bgs, with deeper zones uncharacterized [12]. Historically, East Mesa recharge was dominated by seepage from the unlined Coachella Canal, estimated at about 1.5 million m³/yr along unlined sections between 1979 and 2007 [13,46]. With canal lining, direct recharge from the Coachella Canal is negligible, although the East Highland Canal and smaller laterals remain unlined and contribute to shallow groundwater recharge (Figure 3). Prior to canal construction and agricultural development, recharge was derived primarily from runoff from the Chocolate Mountains and direct precipitation runoff into arroyos [15,23,47,48].

Discharge occurs mainly through agricultural drains, seepage to springs along the Salton Sea, interstitial aquifer flows beneath Salton Sea, and limited pumping. Basinwide extraction from the East Salton Sea Groundwater Basin was estimated at just 7400 m³/yr in 1952 [44] and remains low today [12]. Extraction in the Imperial Valley portion of the study area is negligible due to the absence of domestic or irrigation water wells and increasing groundwater salinity near Salton Sea. Groundwater underflow moves across State Route 111 toward Salton Sea, though its magnitude is unknown.

Canal seepage historically created a groundwater mound exceeding 21 m on East Mesa [18]. Since lining, water levels have declined. USGS, monitoring well 11S/15E-23M, 16 km southeast of Niland, dropped from 6.3 m bgs in 1979 to about 15 m bgs by 2020 [12]. Current data suggest groundwater elevations remain highest in the mountains and East Mesa, declining toward Imperial Valley and the Salton Sea, broadly following topography but locally influenced by faults and transmissivity variations in the saturated strata [12]. A 1993 composite hydraulic head map [13] aligns water-table contours with both the regression of the Salton Sea shoreline and the trajectory of the Niland Moving Mud Spring (Figure 5; compare to Figure 2).

Groundwater quality is generally poor, dominated by sodium-chloride or sodium-sulfate types, with TDS ranging from 356 to 51,600 mg/L with most wells far above the 500 mg/L secondary standard [41]. The Niland landfill, located about 5.6 km north of Niland, California on Cuff Road, has groundwater occurring at approximately 13 m below ground surface, generally flowing southwest. The site is monitored with one upgradient well and two downgradient wells to assess potential impacts on groundwater quality. Semiannual monitoring between 1994 and 2008 indicated Total Dissolved Solids in the upgradient well ranged from 23,900 to 35,400 mg/L, averaging 27,432 mg/L [41] of water near unlined canals is typically fresher [12]. These data suggest the presence of buried evaporites several kilometers east of the mapped boundary between lake deposits and recent alluvium (Figure 3).

Mud pots, mud volcanoes, geothermal springs, and mud springs are hydrogeologically significant, revealing interactions between deep subsurface processes and surface conditions. These features often emit gases such as CO₂, and their formation and migration can be influenced by tectonics, heat flow, and human activities including groundwater extraction, irrigation return flows, and surface-water diversions [1,15]. The Niland Moving Mud Spring is a notable example, affecting road and rail infrastructure. Historically, nearby geothermal wells were exploited for CO₂ extraction for dry ice production and other industrial uses, including geothermal energy and lithium-brine extraction, highlighting both the economic potential and hydrologic complexity of the region [1,49]. Anthropogenic impacts on groundwater and land use continue to influence the behavior and distribution of these mud features, underscoring their significance for hydrogeological studies and land management.

Near the Moving Mud Spring, subsurface studies indicate Pleistocene lakebed sediments with groundwater upwelling from ~60 m depth [4]. Two test wells drilled by Union Pacific near the spring encountered clay-rich sequences with interbedded sands and silts to depths of 244 m and 123 m but did not intersect significant conventional aquifers [2].

3. Methods of Investigation

3.1. Field Methods

Water samples were collected from a variety of sources, including Colorado River Project canal water, agricultural drains, groundwater discharge pipes, a geothermal spring, and the Moving Mud Spring. The primary objective was to compare temperature, dissolved solids, and, most importantly, stable water isotope values to help determine the origin of the Moving Mud Spring, whether it is fed by local groundwater, imported Colorado River water, or a geothermal source. A tritium sample was also collected for possible modern dating of groundwater issuing from the spring. Carbon-14 dating was not performed due to the extremely high CO₂ content in the mud spring which would impact age dating results.

Most surface drain samples were collected during the winter of 2025, when discharge is primarily from groundwater baseflow, with minimal contributions from canal lateral releases (e.g., for vector control) or agricultural tailwater return flows [50]. Latitude and longitude were recorded at all sampling locations. Groundwater discharge pipes flow continuously and did not require purging. Depending on site conditions, drain samples were collected either as single grab samples or via width- and depth-integrated methods. Geothermal and mud spring samples were collected using an extendable pole grab sampler.

Specific conductance and temperature were measured in the field using portable meters. Surface water discharge was measured using Marsh-McBirney Model 2000 flowmeters with top-setting wading rods, following USGS protocols with an accuracy of approximately ±5% [51,52]. In steep or otherwise unsuitable channels, discharge was estimated using stream width, depth, and two to three velocity measurements, with accuracy within ±20 to 25%. Discharge from groundwater pipes was measured using timed bucket or bin collections, accurate to within ±15%.

All samples for hydrochemical and isotopic analysis were collected in new high-density polyethylene (HDPE) bottles with tight-sealing caps to prevent headspace. Bottles were triple-rinsed with deionized water prior to field use and then rinsed with the sample water before collection. Each bottle was labeled with sample ID, date, target analytes, and preservation method. Stable isotope samples were collected in 60 mL vials, while tritium samples were collected in 500 mL polyethylene bottles, both filled to eliminate headspace.

3.2. Laboratory Methods

Isotopic measurements were conducted at the Environmental Isotope Laboratory at the University of Arizona in Tucson. The δ¹⁸O and δ²H values were determined using an automated gas-source isotope ratio mass spectrometer (Finnigan Delta S). For δ²H analysis, water samples were reacted with chromium metal at 750 °C in a Finnigan H/Device attached to the mass spectrometer. δ¹⁸O values were measured by equilibrating water samples with CO₂ at 15 °C in an automated equilibration device also coupled to the mass spectrometer. All measurements were standardized using international reference materials VSMOW and Standard Light Antarctic Precipitation [53]. The analytical precision was 0.9‰ or better for δ²H and 0.08‰ or better for δ¹⁸O [54]. Tritium concentrations were analyzed by liquid scintillation counting on Quantulus 1220 spectrophotometers (PerkinElmer: Turku, Finland) using 0.18 L water samples that had undergone electrolytic enrichment, with a detection limit of 0.7 tritium units (TU).

Total dissolved solids (TDS) for all samples were estimated using temperature-compensated specific conductance and the equation TDS = 0.67 × specific conductance. This coefficient was derived by [55] from a linear regression of general minerals analysis data for groundwater samples collected at Dos Palmas Preserve, located 33 km northwest of the Moving Mud Spring. Based on that regression, TDS estimates from specific conductance measurements are considered accurate within approximately 6.5% of the true value.

The salinity and isotopic parameters in this study were used to characterize the Moving Mud Spring water type and compare it to potential water sources in the region. Stable isotope data were used for distinguishing imported Colorado River water from locally recharged groundwater or geothermal sources, based on established isotopic signatures [56]. Total dissolved solids and temperature measurements were used to associate spring and groundwater chemistry with salinity sources and to infer the geologic origin of drain water, groundwater, and spring water. Tritium was analyzed to assess the age of water at the Moving Mud Spring.

4. Results

4.1. Temperature

Drain water temperatures ranged from 14.8 °C to 36.1 °C, reflecting clear seasonal variation. The highest temperatures (29.2 °C to 36.1 °C) were recorded during the summer months, while the cooler values (14.8 °C to 15.7 °C) were observed in winter (

Table 2. (a) Water samples collected in this study showing agricultural drain data. Map index in Figure 1 can be used to find locations of the data samples by map index number. (b) Water samples collected in this study showing agricultural groundwater discharge pipe data. Map index in Figure 1 can be used to find locations of the data samples by map index number. (c) Water samples collected in this study showing project canal water, moving mud spring and Slab City Hot Spring data. Map index in Figure 1 can be used to find locations of the data samples by map index number.

(a)
Map Index	Sample Location	Sample Type	Date Sampled	Flow (liters/minute)	Temp. ˚C	TDS (mg/L)	δ18O per mille	δ2H per mille
1	Setmire at 111 Road Drain	Drain Water	23 June 2023	38	31.8	8041	−5.7	−69.5
2	English Rd Above 111 Road Drain	Drain Water	25 March 2025	2536	29.2	1614	−11.7	−95.7
3	Beal and English Drain	Drain Water	19 January 2025	11	15.7	3296	−11.6	−96.0
4	Noffsinger and English Drain	Drain Water	25 March 2025	934	31.2	1088	−12.2	−98.0
5	Gillespie Drain Above Moving Mud Spring	Drain Water	6 July 2023	83	36.1	2634	−7.4	−76.9
6	Pound and 111 Road Drain	Drain Water	19 January 2025	606	14.8	2582	−11.9	−96.7
7	Schrimpf Above 111 Highway at Tracks Drain	Drain Water	25 March 2025	681	29.8	2681	−11.4	−94.0
(b)
Map Index	Sample Location	Sample Type	Date Sampled	Flow (liters/minute)	Temp. ˚C	TDS (mg/L)	δ18O per mille	δ2H per mille
8	Pipe at Drain at English Road Above 111 Road	Ground Water	25 March 2025	114	23.5	1952	−12.2	−97.9
9	Beal T Drain Pipe	Ground Water	21 July 2023	34	28.8	2802	−10.0	−90.3
10	Hazard Pipe	Ground Water	6 July 2023	76	27.6	24,937	−8.7	−83.8
11	Pound and Blair—Big Pipe	Ground Water	21 July 2023	57	28.1	4791	−11.8	−96.6
12	Pound and Blair—Small Pipe	Ground Water	21 July 2023	95	29.0	3685	−11.4	−94.6
13	Pound and Burke Pipe	Ground Water	21 July 2023	114	28.9	2301	−11.1	−93.8
14	Noffsinger and Pound—Pipe	Ground Water	21 July 2023	227	28.2	6305	−10.1	−87.6
(c)
Map Index	Sample Location	Sample Type	Date Sampled	Flow (liters/minute)	Temp. ˚C	TDS (mg/L)	δ18O per mille	δ2H per mille
15	Setmire at 111 Rd Project Water	Project Canal Water	23 June 2023	1632	31.1	794	−11.0	−93.0
16	Alcott and Tracks Project Water	Project Canal Water	19 January 2025	Very high flowing supply canal at least 11,500 L/min	14.7	878	−12.6	−99.4
17	Niland Moving Mud Spring	Gaseous Mud Spring	23 June 2023	80 to 107 L/min. 1	28.3	18,070	−5.4	−66.3
17	Niland Moving Mud Spring	Gaseous Mud Spring	1 October 2023	80 to 107 L/min. 1	27.5	17,889	−5.1	−65.3
17	Niland Moving Mud Spring	Gaseous Mud Spring	20 January 2025	80 to 107 L/min. 1	26.5	18,693	−5.4	−63.6
18	Slab City Hot Spring	Thermal Gaseous Hot Spring	23 June 2023	Drains into lateral canal, no estimate available	41.1	2186	−7.7	−64.5

¹ near constant flow rate reported in [2,4] and downgraded by [3] to ~80 L/min by year 2020.

). This pattern suggests that ambient seasonal temperatures strongly influence drain water thermal conditions. Shallow groundwater temperatures collected from drain pipes ranged from 23.5 °C to 29.0 °C. The highest reading, 29.0 °C, was recorded during the peak of a very hot summer, indicating a strong influence from seasonal heat and shallow burial depth. Slight warming may also have occurred during sample collection, as water was obtained from drainage pipes at the sides of channels using a bucket, and carried up to the measurement station, potentially allowing for minor temperature increases. The lowest value (23.5 °C) reflects cooler seasonal conditions (Table 2). Overall, the data suggest that shallow groundwater in this system is thermally responsive to ambient surface temperatures. Canal water temperatures showed strong seasonal variation, with a summer measurement of 31.1 °C recorded on 23 June 2023, and a much cooler winter temperature of 14.7 °C recorded on 19 January 2025. This wide range reflects direct exposure of canal water to ambient air temperatures, consistent with unshaded, surface-level flow conditions typical of open canal systems in arid regions.

Temperatures measured in the Moving Mud Spring indicate relatively stable thermal conditions compared to surface water sources even as these were grab samples and not in situ samples collected in the spring. Recorded values include 28.3 °C on 23 June 2023 (summer), 27.5 °C on 1 October 2023 (fall), and 26.5 °C on 20 January 2025 (winter). The modest seasonal variation, only about 2 °C, suggests a strong influence from subsurface water surfaces but not of geothermal origin, rather a groundwater source of moderate depth probably not exceeding one hundred meters, buffering the pond from ambient temperature swings typical of surface water bodies (Table 2). Due to collapse hazards near the Moving Mud Spring, the temperatures were not collected insitu in the spring, but rather in a bucket filled with spring water, and the brief delay in temperature measurements in transit away from the hazardous spring might have created small temperature change based on air temperature at the time of sampling.

The Slab City Hot Spring, a thermal gaseous spring, sampled on 23 June 2023, registered a temperature of 42.7 °C. This more elevated temperature, compared to nearby surface and shallow groundwater sources, reflects a geothermal influence (Table 2). The spring’s persistent heat and gas emissions mark it as a distinct hydrothermal feature in the region, and found in this study to be fed by deeper subsurface flow fields of local water origin.

4.2. Total Dissolved Solids

Two canal project water samples had TDS concentrations ranging from 794 to 878 mg/L, with an arithmetic average of 836 mg/L, consistent with previous measurements of Coachella Canal water in this area, delivered from the Colorado River at Imperial Dam. In contrast, drain water TDS values ranged from 1088 to 8041 mg/L, with 75% of values falling between 1614 and 2634 mg/L, and an overall arithmetic average of 2991 mg/L (Table 2, Figure 6). This reflects variable salinization of drain water due to groundwater enrichment and the leaching of salts from irrigated agricultural fields. Some drains may also contain tailwater return flows or project water, which may partially dilute or enrich individual samples. Samples collected from irrigation drainage pipes containing 100% groundwater exhibited TDS concentrations ranging from 1952 to 24,937 mg/L. Seventy-five percent of these values ranged from 2301 to 6305 mg/L, with an arithmetic average of 6396 mg/L. The progressive enrichment of TDS from canal water to groundwater to drain water reflects cumulative salt loading from irrigation practices, subsurface leaching, and the upward or lateral seepage of saline groundwater from subsurface collection pipes and groundwater beneath the water table into surface drain collection systems (Table 2, Figure 4).

The Slab City Hot Spring yielded a single TDS value of 2186 mg/L, while the Niland Moving Mud Spring produced three measurements ranging from 17,889 to 18,693 mg/L, with an arithmetic average of 18,217 mg/L. These elevated values in the Niland Moving Mud Spring are consistent with saline groundwater sourced from evaporite-influenced strata and are distinctly higher than surface or shallow subsurface waters associated with irrigation (Table 2, Figure 4).

4.3. Stable Water Isotopes and Tritium

Stable water isotope data from all drain and shallow groundwater samples align along an evaporation trajectory originating from imported Colorado River project water. The Global Meteoric Water Line [57,58] is plotted for comparison. In contrast, the Slab City Hot Spring exhibits isotopic values consistent with locally recharged geothermal water, sharing characteristics with geothermal artesian well waters located about 17 km to the northwest [59] (Figure 7).

Unexpectedly, the Moving Mud Spring plotted within the same isotopic range as evaporated Colorado River water. This finding prompted a second round of sampling a few months later, which confirmed the initial results. A third sampling campaign, conducted to collect tritium data, further corroborated the pattern, the tritium level measured 5 Tritium Units (TU), indicating that the water is modern and likely recharged within the past 2 to 6 decades. The oldest plausible recharge scenario is that the groundwater contains residual bomb tritium from the late 1950s or 1960s, which has since decayed to the observed level of 5 TU. For context, water from the Coachella Canal, which supplies the region, was found to carry background tritium levels of 6.0 TU near Dos Palmas Oasis in 2007 [60].

Taken together, the stable isotope and tritium results indicates that the water emerging from the Moving Mud Spring originates from imported Colorado River water (Figure 7). This finding has important implications for understanding the dynamics of the Niland Moving Mud Spring and its connection to regional surface hydrology and irrigation infrastructure.

Also shown in Figure 7 is a circled group of Hot Artesian Wells, sampled in the geothermal spa area about 17 km northwest of the Moving Mud Spring [59]. These wells represent locally sourced, older groundwater with no detectable tritium and radiocarbon values of about 3.5 percent modern carbon (PMC). The Slab City Hot Spring, plotted alongside, shows isotopic values consistent with locally recharged geothermal water althrough slightly shifted parallel to the Global Meteoric Water Line from Hot Artesian Wells, and is clearly distinct from all of the imported waters following the evaporation trend (Figure 7). The isotopic shift in Canal Project Water, shown by date sampled, reflects extremely low storage in Lake Mead and other Lower Colorado River Basin reservoirs during summer 2023. Storage conditions improved somewhat after that period, following wetter years and increased snowmelt runoff from Colorado.

4.4. Time Series Image Mapping of Moving Mud Spring and Salton Sea Shoreline

The predecessor spring feature, W9, which does not appear in 2002 aerial imagery, likely formed or reactivated around 2005 (Figure 1 and Figure 2). It later evolved into a second migrating spring, W9a, as W9 began expanding after 2012 and separated in 2016, with W9a diverging and moving southwest (Figure 8). Initially advancing just 1 to 2 m per year over a few weeks, W9a’s rate of movement accelerated to nearly 3 m per month for a brief period [4]. By 2021, W9a had migrated more than 100 m from W9’s 2016 location [2] (Figure 8).

Geotechnical surveys confirmed a consistent southwestward trajectory toward vital infrastructure. By 2018, W9a neared the Union Pacific Railroad Mainline, prompting emergency mitigation efforts. These included drilling decompression wells, installing a 43 m long steel sheet pile barrier, and pumping up to 151,000 L of fluid per day to reduce pore pressure [2,4,5]. Riprap was added to resist erosion, and temporary “shoofly” rail detours were constructed adjacent to the original rail mainline. Despite these interventions, W9a breached the containment area, forming sinkholes and advancing beneath the original track alignment as well as the western shoofly railline [2,5]. Once W9a had passed the eastern shoofly, the railline was then stabilized as the permanent alignment near the old rail mainline, with W9a now moving to the west toward State Route 111 (Figure 8).

By late 2018, the mud spring had expanded into a 2230-square-meter basin. Subsurface erosion scoured sediments more than 6.7 m deep, continuing to expose road, rail, and other infrastructure to risk [61,62,63]. At the same time, the California State Transportation Agency (Caltrans), launched a multi-million-dollar, multi-phase response to protect State Route 111, including the installation of drainage systems, CO₂ diversion structures, and a realigned roadbed west of the affected zone [5,62,63,64]. In parallel, Kinder Morgan rerouted its petroleum pipeline at a cost of $3 million [65], while Verizon and AT&T relocated fiber-optic cables [2,5,66].

Due to additional containment measures, particularly spring dewatering and local degassing, W9a’s movement slowed significantly (Figure 8). By 2023, W9a appeared to stabilize, and as of 2025, it remains relatively stationary, although the mud spring’s surface dimensions have continued to expand (Figure 8). The original W9 spring no longer exhibits visible surface activity, while W9a remains the primary threatening feature. Although currently stabilized, its long-term behavior and potential risk to surrounding infrastructure remain uncertain. Figure 8 illustrates the key developments discussed between 2002 and 2025. The references cited above provide very detailed documentation of the engineering interventions, stabilization efforts, and impacts on transportation and utility infrastructure. Figure 8 also illustrates the progressive retreat of the Salton Sea shoreline alongside the migration of the Moving Mud Spring. A particularly noteworthy correlation emerges between the timing and sub-parallel movement of the shoreline and the mud spring from the initial appearance of W9 (coinciding with the enactment of the Quantification Settlement Agreement; [67]) to the later separation of W9 and W9a during the period of most rapid sea-level decline. The correlation model complements existing conceptual models for mud spring movement, which [2] acknowledge are not strongly anchored in empirical data. The movement of the mud spring and the retreat of the shoreline align with the hydraulic head gradient observed in the study area (compare Figure 5 and Figure 8).

5. Discussion

As a newly observed kinetic phenomenon that has no apparent historical or spatial analog, the movement of the mud spring presents distinct challenges for developing conceptual models. Consequently, any explanatory framework, including that proposed in this study, should be regarded as a provisional analysis. The data indicate that the Salton Sea shoreline recession (Figure 2 and Figure 8), the orientation of the Moving Mud Spring (Figure 2 and Figure 8), and the groundwater hydraulic gradient (Figure 5) all trend in the same general direction. Notably, the timing of the shoreline recession and the onset of mud spring movement appears to coincide.

The combined temperature, salinity, stable water isotope, and tritium data provide strong evidence that the water emerging from the Moving Mud Spring originates from Colorado River irrigation supplies. Initial isotope results show an evaporation trajectory consistent with Colorado River water, leading to two possible sources: residual water from the 1906 Salton Sea flood inflows or more recent irrigation-derived inputs. The detection of tritium at 5 TU clearly indicates a modern origin, as water from the 1906 event would no longer contain detectable tritium.

Isotopic signatures from drain waters and shallow groundwaters are generally similar to the project water baseline, with most samples showing only slight evaporation (Table 2, Figure 7). A subset of samples, including the Moving Mud Spring, the Setmire site at 111 Rd Drain, Gillespie Drain above the Moving Mud Spring, and Hazard Pipe, are more isotopically evaporated than the norm (Table 2). However, these samples do not display a consistent linear relationship between physical evaporation and salinity enrichment. Because all waters originate as Colorado River imports with typical TDS of 750–850 mg/L (Table 2), this suggests that salinity changes are driven primarily by dissolution of precipitated salts in soils and evaporite lake deposits rather than by evaporation alone.

This interpretation is illustrated using the Moving Mud Spring as an example. Prior studies in arid zones have quantified the relationship between δ²H and salinity using an evaporation–salinity enrichment coefficient [68]. Ref. [69] found δ²H increases by 0.65‰ for each 1% water loss by evaporation, while [70] reported 0.78‰. Using an average enrichment rate of 0.70‰ δ²H per 1% evaporative water loss, the δ²H enrichment between Moving Mud Spring samples (−67.3‰, −65.3‰, −63.6‰) and project canal waters (−93.0‰, −99.4‰) averages 30.9‰. This implies approximately 44% evaporative water loss between recharge and discharge at the Moving Mud Spring. Assuming an initial salinity of 830 mg/L TDS, physical evaporation alone would yield a concentration of ~1485 mg/L TDS at the spring, over an order of magnitude lower than the observed ~18,000 mg/L TDS (Table 2). These calculations demonstrate that dissolution of evaporite minerals along the groundwater flow path is the dominant process enriching the groundwater in salts before discharge at the Moving Mud Spring. For empirical comparison, Sample 5 from the Gillespie Drain, approximately two-thirds as evaporated as the Moving Mud Spring water (Table 2), contains 2634 mg/L TDS, with an initial feed salinity of about 750 to 800 mg/L TDS for both the Gillespie Drain and the Moving Mud Spring. These data indicate that physical evaporation alone cannot explain the observed salinity increase in the Moving Mud Spring; instead, interaction with evaporite minerals appears to be a more significant source of groundwater salinity.

Geologic and geophysical data are consistent with this interpretation. High-resolution seismic S-wave reflection data from [4] identified a geophysical anomaly about 60 m below ground surface, interpreted as a groundwater upwelling zone. Ref. [4] also reported Pleistocene lakebed sediments beneath the site, consistent with regional lacustrine units. Drilling data cited by [2] indicated plastic, clayey units throughout much of the stratigraphy to 244 m depth. These lithologies, often indicative of evaporitic lake deposits rich in gypsum and occasionally halite, align with the spring’s measured salinity (Table 2; Figure 6). The fine-grained, clay-rich units become more abundant toward the Salton Sea, reflecting the area’s lacustrine sedimentation. Interbedded sandy layers in the two drill logs probably derive from arroyo runoff transporting coarser sediments from nearby mountain slopes. Ref. [2] describe these strata as “dry,” but more accurately they are low-permeability layers.

The Moving Mud Spring temperature (~28 °C) is consistent with shallow subsurface groundwater movement, likely following a flow path from the land surface to depths of no more than 50 to 100 m. This supports the interpretation that the groundwater issuing from the spring is not connected to the region’s deeper, high-temperature geothermal systems at depths of 300 to 2000 m or more [40]. Isotope data confirm that the Moving Mud Spring is a shallow, saline groundwater discharge system sourced from modern imported Colorado River irrigation water. As this groundwater migrates through the subsurface, it interacts with evaporite-rich lakebed deposits, reaching ~18,000 mg/L TDS. While the groundwater is shallow and of modern origin, the gas driving the mud spring is sourced from much deeper formations. This decoupling between upwelling gas and groundwater source is significant, as the dynamics of shallow groundwater flow and gas upwelling differ from those of deep geothermal fluids, which typically behave as fixed surface features (Figure 9).

The isotope evidence also clarifies why the Moving Mud Spring remains at ambient temperatures: it is derived from near-surface Colorado River irrigation water and does not circulate to depths where significant geothermal heating occurs. This contrasts with nearby features such as the Slab City Hot Spring, where elevated temperatures and locally derived recharge, probably from mountainous areas to the east, indicate much deeper groundwater circulation (Table 2). Ref. [1] noted that, similar to the Wister mud pots located about 1.5 km west of the Moving Mud Spring, water and mud temperatures are not noticeably elevated. It is possible that the Wister mud pots, like the Moving Mud Spring, issue percolating Colorado River irrigation water that has not circulated deeply enough for geothermal heating.

A conceptual understanding of groundwater movement in the study area includes three components: (1) older deep groundwater recharged more than 5000 years ago in the Chocolate Mountains or other eastern ranges; (2) shallow groundwater derived from imported Colorado River water, confirmed by isotopic and geochemical data from shallow wells and agricultural drains in the Wister Unit; and (3) CO₂-charged fluids that contribute to geothermal features at the Slab City Hot Spring. Declining Salton Sea water levels are lowering shallow aquifer levels in some areas, especially near the Salton Sea, altering flow directions and recharge–discharge dynamics across connected-groundwater systems (Figure 8 and Figure 9).

No known faults or structural features align with the mud spring’s movement; major structural trends, including significant fault systems, run perpendicular to its path [2]. The concurrent movement of the shoreline and mud spring may represent the strongest potential cause–effect relationship observed to date. Stabilization efforts since 2018 probably have slowed and nearly arrested the movement of W9a, artificially affecting recent correlations. Ref. [1], citing personal communication from A. Hernandez in 2007, described how managed wetland cycles influence mud pot activity: “Some of the ponds are filled with fresh water in late summer to accommodate migrating birds… In 2006, new gas vents developed in water delivery channels.” Given that Salton Sea decline had already begun between 2002 and 2006, the appearance of new mud pots may be linked to changing sea level and water table hydraulic gradients. The 4.1 m drop in Salton Sea level between 2003 and 2025 (Figure 2 and Figure 8) altered hydraulic head laterally and vertically across shallow and deeper aquifers. The Sea acts as a regional, shifting hydrogeologic boundary and ultimate discharge zone for groundwater. These head changes particularly would be expected to affect shallow aquifers characterized by lateral flow between semi-pervious fine-textured layers (Figure 9). Changing hydraulic heads between the Moving Mud Spring and Salton Sea could influence lateral gas migration with flowing groundwater, as well as the formation of new mud pots, especially close to areas where geothermal plants withdraw and reinject groundwater under high pressure.

Although not depicted in Figure 9, a saltwater wedge similar to those in coastal aquifers, simplified by the Ghyben–Herzberg principle [71], probably exists beneath and adjacent to the Salton Sea. Because Salton Sea, salinity now exceeds modern ocean water by ~55%, density-driven downward intrusion of hypersaline Sea water into underlying aquifers is plausible where the underlying groundwater is less dense. In many areas where the shoreline stood in 2003, playa surfaces now overlie dry sediment with the water table more than 2 m below ground surface [59].

6. Limitations of the Proposed Model and Reconciliation with Prior Analysis of Aquifer Conditions near the Moving Mud Spring

Groundwater dynamics in the eastern Salton Sea region, particularly within the Wister Unit, remain poorly understood compared with the Coachella Valley at the northwestern end of the Sea. The volume and significance of groundwater and geothermal discharge into adjacent wetlands have not been quantified, though emerging evidence suggests these inputs may be important in forming and sustaining localized wetland systems [15]. Hydrologic interactions between groundwater and surface water are largely undocumented, and no recent piezometric maps (post-2000) exist [15]. The continued decline in Salton Sea water levels is likely altering regional groundwater gradients [15].

Evidence of groundwater discharge near the Salton Sea includes gypsum “tepee-like” structures on both the eastern and western shores [15] and historic accounts of numerous springs [72], some of which may re-emerge as the shoreline retreats [15]. Yet groundwater data remain sparse, especially along the eastern lake margin. Limited USGS data indicate a shallow water table near the Sea, with spatial variability influenced by complex faulting [13,15]. The potentiometric surface map for the study area (Figure 5) shows solid contour lines approaching Salton Sea, even though well control was not included with the contours [13]. Managed flooding of newly exposed playa surfaces, such as through Wister Unit Wetlands Enhancement Programs, may locally slow water-table decline, but the 4.1 m drop in Salton Sea level and exposure of terrain west of the Moving Mud Spring are reshaping both shallow and deeper groundwater hydraulic gradients.

Current hydraulic head data are critically needed to evaluate temporal changes in aquifer systems as the Sea recedes. At the Moving Mud Spring, the presence of modern water sourced from imported Colorado River flows is well supported by tritium and stable isotope data (Figure 7). This modern signature provides a strong basis for interpreting spring dynamics. The observed correlation between spring movement and declining sea levels remains a provisional hypothesis that requires further testing through spatial and temporal groundwater monitoring in the study area.

Ref. [2] observed that “flow at the two springs, W9 and W9A, varied, but water production remained constant, which suggests an unchanging deep source whose gas and water find their way to the surface by any route possible.” They reported that total discharge remained constant regardless of which vent was active, supporting their “hydra” model of spring migration. Ref. [3] subsequently reported a 25% reduction in flow between 2019 and 2020, possibly attributable to installation of spring degassing apparatus. However, the results of this study indicate a deep groundwater source is implausible. Isotopic data show that the spring discharges modern water (<60 years old) imported via Colorado River irrigation canals, evidenced by tritium (~5 TU) and Colorado River-type stable isotope signatures. Such water cannot originate from a deep, isolated aquifer. Ref. [2] also speculated that the local water table near the spring is patchy and discontinuous, citing two deep test wells drilled about 30 m from W9A reported as “dry.” Two boreholes drilled 45 m from the spring reportedly encountered saturated mud only in the upper 1.5 m underlain by interbedded dry clayey sediments and sand to 30 m depth.

These findings raise questions about hydrogeologic conditions around the Moving Mud Spring and point to the need for more comprehensive subsurface characterization. While localized dry or semi-perched zones may exist, especially in fine-grained lakebed sediments, regional evidence suggests that a patchy or disconnected water table is not typical across the 2 km radius surrounding the spring. Parts of the Wister Unit and all adjacent farmlands are underlain by engineered drainage systems, perforated pipes, groundwater discharge pipes (tile drains), and open agricultural drains designed to maintain unsaturated crop rooting zones (Figure 10). These systems rely on lateral groundwater movement, which requires moderate hydraulic continuity and sufficient vertical infiltration rates. While perched conditions may occur locally, the ongoing function of the drainage network implies a generally connected, mobile shallow aquifer to ensure good drainage and crop health (Figure 10).

Many drains are located in fields near the Moving Mud Spring (Figure 10). Measured flows (Table 2) far exceed spring discharge rates, for example, W Drain yielded 2536 L/min in March 2025 from >3.6 m depth (Figure 1 and Figure 10, Table 2). Niland Drain 1, along Gillespie Road about 170 m north of the spring, has frequently held drainage water in most years from 2002 to 2023 (Figure 8) and was flowing in summer 2023 when sampled 2.2 km above the spring (Figure 1, Table 2). Drain wetting is assumed to be caused mostly by groundwater seepage flows (Figure 8). These observations are difficult to reconcile with widespread 30 m thick dry strata directly beneath the land surface within a 2 km radius of the spring.

Ref. [4] proposed that the source of CO₂ at the Moving Mud Spring is thousands of feet below the surface. They suggested that as gas migrates upward it interacts with deep groundwater to form a carbonated fluid mixture that helps drive fluid and gas to the surface, and that near the land surface CO₂ may mix with shallow groundwater to generate clay-rich mud pots and mud volcanoes. This paper demonstrates that the water involved in the Niland mud volcano is derived exclusively from imported Colorado River water, which lacks the capacity to circulate to depths below ~50 to 100 m. Even those depths seem unrealistic, given that the Coachella Canal that was unlined until 2007 is only about 8 km east of the spring and 80 m higher in elevation.

If a patchy, irregular shallow water table exists and the upper 30 m is dry, the movement of imported Colorado River water must be influenced by some active hydrogeologic pathway carrying modern recharge to greater depth than 30 m (Figure 11). In this intermediate zone, probably a maximum of 50 to 100 m, vertical and horizontal aquifer connectivity would have to allow relatively young, imported water to migrate toward the spring. Given such dynamic movement, the 4.1 m drop at the Salton Sea boundary would presumably increase groundwater velocity and flux beyond 2005 flow rates. The models proposed in this paper to explain spring movement could operate just below a shallow water table at <30 m depth (Figure 9) or at intermediate depths of 50 to 100 m (Figure 11).

Regarding the potential water table connection between the Moving Mud Spring and the Salton Sea, ref. [73] developed a conceptual framework for preferential groundwater pathways in alluvial deposits using Monte Carlo simulations and Gibbs’ distributions. This framework provides a valuable lens for interpreting groundwater flow in aquifers like those in the study area. The aquifer in the study area is composed of heterogeneous Colorado River alluvium and Chocolate Mountain fanglomerates to the east, interbedded with fine-grained lacustrine sediments mixed with medium-grained shoreline and ephemeral channel facies, intersected by a dense network of canals and drains oriented perpendicular to the Salton Sea shoreline. This configuration resembles the high-conductivity “efficient” pathways predicted by [73] under low Gibbs’ exponents, in which low-entropy conduits transmit water through permeable pathways with minimal energy dissipation.

A complex hydrogeologic picture of the study area emerges when system hydraulics, drainage infrastructure, and depositional history are considered together. The Salton Sea shoreline has advanced and retreated repeatedly, producing coarser-textured, shoreline-parallel facies that are now buried and can intercept groundwater flow fields oriented perpendicular to surface drainage. Predevelopment ephemeral tributaries from the Chocolate Mountains extended perpendicular to the Salton Sea, depositing coarse sands and gravels that graded into medium-to-fine-grained sediments closer to ancestral lake margins. These buried channels are interbedded with fine-textured lacustrine deposits and crosscut by more permeable shoreline facies, creating an irregular orthogonal network of higher-permeability units encased in tight lake sediments.

Although the precise geometry of this permeable “hatchwork” is unknown, these deposits likely form the main conduits for shallow subsurface flow, consistent with [72] energy-minimization concept of groundwater movement through intersecting permeable pathways and discharge boundaries at Salton Sea. The resulting flow patterns are complex yet energy-efficient. The intersection of ephemeral channel sands perpendicular to the shoreline with shoreline-parallel channel sands, embedded within a fine-grained lacustrine matrix, provides a conceptual model explaining the possible hydraulic connection between the Moving Mud Spring area and the dynamic, receding Salton Sea. The abundance of hydraulic head data used to construct the potentiometric surface map (Figure 5) is insufficient to capture all of these intricate hydrogeologic details. Nevertheless, these complexities should be considered when interpreting hydrochemical and groundwater isotope data, as well as overall groundwater system dynamics.

Drainage infrastructure performance and isotope data (Table 2) imply a relatively mobile, connected shallow groundwater regime at least regionally across much of the study area where data were collected (Figure 1); however, the heterogeneity of the underlying sediments and possible preferential pathways remain poorly characterized. The link between seasonal irrigation, wetland management practices, and changes in mud spring behavior also merits closer examination. These open questions underscore the need for additional data, aquifer testing, and possible modeling before long-term forecasts or understanding of spring behavior can be made with confidence.

Previous studies have proposed several different mechanisms for the Niland Moving Mud Spring’s activity, ranging from deep geothermal or magmatic inputs along hidden faults, to purely structural controls such as tilted conduits, to hydra-like rising models that follow sediment stress-deformation pathways [2,3]. The data presented here reconsiders some of these models by suggesting a shallow, modern groundwater origin for the spring’s water, yet they do not rule out the possibility of structurally or lithologically guided pathways. A key limitation of previous models is that previous authors attempt to account for the behavior of a phenomenon widely regarded as very rare, if not unique. Although their models propose mechanisms to explain the observed spring movement at the mud spring site, they do not address why similar mechanisms are apparently absent or inactive elsewhere under comparable conditions, raising the question of whether this is the only setting in which such processes operate. The model proposed in this paper is a cause-and-effect framework in which the spring’s movement is driven by rapid changes in hydraulic head conditions associated with the decline of the Salton Sea. Although the new conceptual model emphasizes hydrologic boundary conditions associated with Salton Sea decline, it should be viewed as a provisional hypothesis.

In sum, none of the models proposed for spring movement by [2,3] or in this paper have been proven. The dynamics of the Moving Mud Spring remain incompletely understood, and further investigation is required. Aquifer characterization of deep and shallow systems, through water-level measurements, aquifer testing, and possibly tracer experiments, would be highly valuable. Installing shallow monitoring wells within a 2 km radius of the spring could help determine whether a continuous water-table aquifer is present, its hydraulic gradients, and other flow dynamics. Furthermore, whether the Niland Moving Mud Spring will remain relatively stationary, or reactivate despite mitigation efforts, remains to be seen.

Figure 10. The Wister Unit and surrounding farmland are underlain by engineered drainage infrastructure, including perforated pipes, groundwater discharge pipes (tile drains), and open drainage channels, designed to maintain unsaturated root zones for crops. These systems rely on lateral groundwater movement and suppression of the water table beneath plant rooting zones, which requires sufficient lateral hydraulic conductivity and soil infiltration rates in the shallow subsurface. Although perched water may occur locally in fine-grained alluvium, the consistent performance of the drainage network suggests a generally connected and mobile shallow aquifer system. The upper image shows dense placement of agricultural drains near the Moving Mud Spring, while the lower image captures high seepage flow in W Drain, and significant groundwater discharge from two tile drains. Upper image modified from [74]; lower photo by the author of this paper.

Figure 10. The Wister Unit and surrounding farmland are underlain by engineered drainage infrastructure, including perforated pipes, groundwater discharge pipes (tile drains), and open drainage channels, designed to maintain unsaturated root zones for crops. These systems rely on lateral groundwater movement and suppression of the water table beneath plant rooting zones, which requires sufficient lateral hydraulic conductivity and soil infiltration rates in the shallow subsurface. Although perched water may occur locally in fine-grained alluvium, the consistent performance of the drainage network suggests a generally connected and mobile shallow aquifer system. The upper image shows dense placement of agricultural drains near the Moving Mud Spring, while the lower image captures high seepage flow in W Drain, and significant groundwater discharge from two tile drains. Upper image modified from [74]; lower photo by the author of this paper.

Figure 11. This highly idealized conceptual figure considers a rather hypothetical scenario in which the upper 30 m of sediment near the mud volcano is dry and relates to the entire regional strata being dry up to 30 m, except for isolated perched groundwater lenses, requiring a deeper, active hydrogeologic pathway to transport modern Colorado River water to depths necessary to reach the moving mud volcano. Refs. [2,3] reported dry strata in test holes drilled to 30 m near the Moving Mud Spring. Such a pathway would need both vertical and horizontal aquifer connectivity to explain the presence of young groundwater at these depths and appearing quickly (<than a few decades) at the Moving Mud Spring. If this idealized mechanism exists in any form, the 4.1 m drop in Salton Sea level could enhance groundwater velocities at depth, potentially accelerating groundwater flow velocity and flux well beyond rates observed in 2005. This model, incorporating a conceptual “pipe” described in [2,3] could only capture groundwater originating from irrigation with imported Colorado River water, based on water isotopes collected in this study that prove the Moving Mud Spring issues imported Colorado River water.

Figure 11. This highly idealized conceptual figure considers a rather hypothetical scenario in which the upper 30 m of sediment near the mud volcano is dry and relates to the entire regional strata being dry up to 30 m, except for isolated perched groundwater lenses, requiring a deeper, active hydrogeologic pathway to transport modern Colorado River water to depths necessary to reach the moving mud volcano. Refs. [2,3] reported dry strata in test holes drilled to 30 m near the Moving Mud Spring. Such a pathway would need both vertical and horizontal aquifer connectivity to explain the presence of young groundwater at these depths and appearing quickly (<than a few decades) at the Moving Mud Spring. If this idealized mechanism exists in any form, the 4.1 m drop in Salton Sea level could enhance groundwater velocities at depth, potentially accelerating groundwater flow velocity and flux well beyond rates observed in 2005. This model, incorporating a conceptual “pipe” described in [2,3] could only capture groundwater originating from irrigation with imported Colorado River water, based on water isotopes collected in this study that prove the Moving Mud Spring issues imported Colorado River water.

7. Conclusions

This study proposes a new conceptual model for the migration of the Moving Mud Spring near Niland, California, based on hydrochemical, isotopic, and thermal data and water level decline in Salton Sea. Earlier models focused primarily on trying to identify structural or sediment deformation controls such as tilted conduits or unmapped faults, while this approach highlights the critical influence of groundwater fluctuations modulated by decline in Salton Sea water levels. The integration of these data sets focused on the geochemical and physical processes driving the spring’s southwestward movement, an increasingly urgent issue given the spring’s proximity to vital road and rail infrastructure.

Previous conceptualizations of the Niland Mud Spring migration generally assumed a deep groundwater source feeding the spring and did not incorporate the role of changing hydrologic boundary conditions linked to sea water declines. The new data presented in this study show that the water issuing from the Moving Mud Spring is modern and shallow in origin. Stable water isotope and tritium analyses place the spring’s water within the isotopic signature of imported Colorado River water, recharged within the past several decades, and distinguish it clearly from deeper geothermal or local groundwater sources that are much older. Tritium levels around 5 TU confirm this recent recharge. This finding contradicts any earlier assumptions or prior models that implied a deep aquifer source (e.g., >500 m) as the water source and indicates that the spring’s dynamics are closely linked to surface hydrology and imported irrigation water.

Thermal measurements reveal that shallow groundwater and drainage waters exhibit strong seasonal temperature variations consistent with ambient air temperatures, ranging from about 14.8 °C in winter to over 30 °C in summer. In contrast, the Moving Mud Spring itself maintains a relatively stable temperature between 26.5 °C and 28.3 °C across seasons, indicative of buffering by moderate-depth subsurface groundwater, likely less than 100 m, rather than geothermal heating. Nearby known geothermal springs such as the Slab City Hot Spring register significantly higher temperatures (~42.7 °C) and local origin based on stable water isotopes, confirming a distinct deeper hydrothermal source separate from the Moving Mud Spring’s water.

Hydrochemical analyses show a progressive increase in total dissolved solids from the canal water, through drain waters, through groundwater reflecting the cumulative effects of agricultural irrigation, leaching, and salinization. The Moving Mud Spring exhibits particularly elevated TDS values (around 18,000 mg/L), consistent with saline groundwater sourced from evaporite-influenced strata, suggesting complex subsurface mixing and salt accumulation processes. Despite this salinity, the water’s isotopic and chemical signature confirms its shallow, modern origin rather than deep geothermal sources.

Time-series imagery reveals that the Moving Mud Spring began migrating southwest around 2012 to 2016, accelerating from a slow annual rate to nearly three meters per month at peak movement, eventually spanning over 100 m by 2021. This movement closely parallels the retreat of the Salton Sea shoreline over the same period, which has dropped approximately 4.1 m since 2002 (Figure 8). The strong temporal and spatial correlation between shoreline recession and spring migration suggests a cause-effect relationship where declining lake levels alter hydraulic gradients, increasing lateral groundwater velocities and shifting the pathways of CO₂-charged water rising to the surface. Although mitigation efforts since 2018, such as decompression wells, sheet pile barriers, and extensive infrastructure stabilization, may have largely arrested spring migration for now, the underlying hydrologic forces remain active, and the spring’s long-term behavior remains uncertain.

The decline of the Salton Sea acts as a regional hydrogeologic boundary condition, gradually changing the lateral and vertical hydraulic heads in shallow and deeper aquifers as the lake level falls (Figure 9 and Figure 11). These changes presumably affect groundwater flow velocities and pressure gradients in sedimentary layers with variable permeability. Such evolving groundwater dynamics can redirect gas and water migration pathways and trigger surface expression changes, that might include the formation of new mud pots or mud volcanoes, as documented in historical observations around the Wister Unit. Seasonal wetting and drying cycles in agricultural and managed wetland areas may further influence these processes by altering local saturation and soil stability, contributing to the episodic appearance and movement of these interesting, but usually non-threatening features.

The presence of a generally connected shallow aquifer system near the Wister Unit is supported by agricultural drainage efficiency and measured discharge rates far exceeding those of the mud spring (Supplemental). These observations call into question notions of a patchy or disconnected water table, suggesting instead a mobile shallow groundwater regime. Uncertainty increases toward the Salton Sea due to the limited availability of groundwater control data in that area. To the extent that it exists, hydrologic connectivity, combined with declining sea levels, could drive increased groundwater velocities, and associated spring migration proposed in this study. Whether the Moving Mud Spring will reactivate or remain stationary due to aggressive mitigation efforts is uncertain.

This research may suggest that the Moving Mud Spring is controlled not solely by structural features or deep geothermal gas inputs but significantly by the shallow hydrologic system responding to the Salton Sea’s retreat. The integration of isotopic, chemical, thermal, and geospatial data provides robust evidence for a shallow, modern groundwater source whose flow dynamics are modulated by shifting hydraulic gradients linked to surface water decline. These findings add a novel conceptual model to help understand spring migration and highlight the need for comprehensive groundwater monitoring, aquifer characterization, and integrated management of the water resources of the study area.

Recommendations

Further work is needed in several areas. Future investigations should prioritize establishing a coordinated network of shallow and deep monitoring wells around the spring and adjacent wetlands to continuously measure water levels, hydraulic gradients, and aquifer connectivity. Coupled with tracer tests and aquifer-performance testing, this network would improve understanding of groundwater flow paths and velocities. Long-term thermal and chemical monitoring across seasons and years would help reveal interannual variability and the influence of climate or anthropogenic changes on the system. Such multidisciplinary efforts are critical to unraveling the complex interplay of geologic, hydrologic, and human factors shaping geothermal phenomena in this tectonically active, environmentally sensitive region, and will inform risk-mitigation and infrastructure-protection strategies while adding detail to understanding the contribution of groundwater to the Wister Unit wetlands

Expansion of groundwater level monitoring, together with high-resolution geophysical surveys, will further constrain subsurface geologic framework and transient changes in hydraulic gradients. Such data can provide independent evidence for migration pathways and allow the identification of zones where fluid or gas inputs vary seasonally or episodically.

Additional emphasis should be placed on integrating gas geochemistry with flux monitoring to strengthen interpretations of fluid and gas migration pathways. Supplementary measurements such as δ¹³C-CO₂, ³He/⁴He ratios, CO₂ fluxes, and flux time series could provide a more complete picture of the temporal and spatial variability of gas release. These datasets would directly test the hypothesis of shallow water source–deep gas source coupling and clarify the degree to which gas transport controls fluid movement within the system.

Finally, continuing the assessment of salinization in the study area will require the application of additional conservative and isotopic tracers. Analyses of Cl/Br ratios, Sr, S, N, and C isotopes, and trace element geochemistry can help discriminate between evaporative concentration and mineral dissolution processes, as well as illuminate mixing between distinct aquifers or between meteoric, connate, and magmatic fluids. Incorporating these approaches will reduce existing uncertainties and provide a stronger geochemical framework for interpreting groundwater–gas interactions.