The Limits of the coming Soil Moisture Model in Prince William County

Quietly, Prince William County has engaged the USGS to develop a Soil-Water-Balance (SWB) model to estimate how development may affect groundwater. The model is expected to be finished by April 2028. The model cost only $320,000, but it’s use and value reflect that. The central problem is that this model is being advanced without the physical monitoring network needed to calibrate and verify its results. As a consequence, the forthcoming study may be useful as a broad land-use planning tool, but it should not be treated as a reliable measure of aquifer health or as a basis for confident policy decisions in heavily developing areas.

This limitation is particularly serious in Prince William County because the county spans multiple geological settings, including the Culpeper Basin, the Piedmont, and the Coastal Plain. In such a fragmented subsurface environment, surface-based estimates of infiltration cannot be assumed to reflect actual recharge of deep groundwater. Without direct field data, the model remains a theoretical approximation rather than a verified representation of underground conditions.

How the Model Is Being Evaluated Without New Physical Data

The USGS is not validating the model through new groundwater measurements. Instead, it is testing whether the model appears mathematically reasonable when compared with existing secondary data. That approach may offer a limited consistency check, but it is not the same as demonstrating that the model accurately captures how groundwater behaves in the field.

Streamflow baseflow separation. One indirect check uses decades of streamflow records from the Occoquan Watershed Monitoring Lab. Hydrologists can estimate baseflow, or the portion of streamflow sustained by groundwater during dry periods, and compare it to the model’s estimate of recharge. If the two are broadly similar, the model may be described as reasonable at a broad mathematical level. But this is still an indirect comparison: it does not show where recharge occurs, whether it reaches deep aquifers, or how groundwater behaves in specific development corridors. The weakness is even greater because OWML does not have stream monitoring data from the headwaters of Bull Run and the Occoquan, which are among the areas most vulnerable to sudden drying because of their geology. If the most sensitive headwater reaches are not directly monitored, then the model’s apparent consistency with downstream records cannot be treated as proof that groundwater-supported flows are being protected where risk is highest.

Historical data mining. A second form of support comes from preexisting records such as older Department of Environmental Quality data and well-completion logs from private wells. These sources may provide background context, but they are static, unevenly distributed, and not designed to function as a modern monitoring network. They cannot substitute for current, continuous field measurements.

Why Indirect Verification Is Inadequate in Prince William County

The weakness of the SWB approach is structural. At its core, the model is a surface water accounting exercise: rainfall minus evapotranspiration minus runoff is treated as recharge. That logic can be useful in simplified settings, but in a county with heterogeneous geology it risks confusing potential recharge at the surface with actual replenishment of the aquifer at depth.

1. The Bedrock Fracture Blind Spot

In western Prince William County, especially in the Culpeper Basin around Gainesville and Haymarket, groundwater movement depends heavily on fractures in shale and sandstone bedrock. Water that passes through the upper soil does not automatically become useful recharge to the deeper aquifer. It must also encounter pathways in the underlying rock.

The model cannot determine whether infiltrating water actually enters those fractures, stalls above relatively impermeable rock, or disperses into shallow perched zones that do little to sustain deeper groundwater supplies. This is a critical limitation because a high modeled recharge value may still coincide with weak real-world replenishment of the aquifer.

2. The Monitoring-Well Deficit

Any model is only as credible as the field data used to anchor it. In a county that spans two distinct geological provinces and supports thousands of residents on groundwater, credible calibration would require a distributed network of monitoring wells capable of tracking water levels across multiple locations and depths.

Prince William County currently has only two continuously monitored USGS groundwater wells—one in the west and one in the east. That leaves major parts of the county, including the central Piedmont and many of the most intense development corridors, without meaningful physical observation points. Even so, the limited monitoring that does exist already shows why direct data matter: USGS well 49V1 indicates a declining groundwater level over roughly the last 15 years. That kind of trend is precisely the sort of warning sign a soil-water-balance model cannot independently confirm or dismiss. With so little direct telemetry, the county cannot determine whether modeled recharge corresponds to actual water table behavior across the areas where the stakes are highest.

3. The Land-Cover Lag Problem

The SWB model also depends heavily on mapped land-cover conditions. In a rapidly changing county, this creates another source of uncertainty: model inputs can lag behind on-the-ground development. The natural lag  in observing the change in groundwater recharge from development is approximately the age of groundwater in the county. Which averages 15-20 years depending on the geology.

If the land-cover data are even one or two years out of date, the model may still treat a parcel as permeable open land when it has already been graded, compacted, or paved. In that circumstance, the model can overstate recharge precisely where impervious surfaces have already reduced infiltration. That makes the model especially risky as a basis for decision-making in fast-moving development areas.

What the Study Can and Cannot Support

What the Study IS Useful For	What the Study CANNOT Do
Mapping Loss of Infiltration: It mathematically proves exactly how many millions of gallons of potential rainwater infiltration are lost when a specific forest or farm is turned into a data center.	Measuring Aquifer Volume: It cannot tell you how much water is actually left in the underground bedrock, or how close local residential wells are to running dry.
A Macro-Planning Tool: It provides a broad, 10,000-foot view of watershed health to help guide general regional zoning regulations.	Local Site Safety: It cannot tell you if a new heavy industrial development will deplete or contaminate the specific fracture network feeding a neighboring subdivision's wells.

Conclusion: Why the Model Should Not Be Relied On Without Physical Data

The forthcoming study should be understood for what it is: a model of potential infiltration and land-use change, not a direct measure of aquifer condition. It may help illustrate how development reduces opportunities for recharge, but it cannot answer the most important policy question—whether groundwater storage is actually being sustained in the places where demand and development pressure are rising fastest.

For that reason, county leaders should not rely on the SWB model alone as evidence that groundwater resources are secure. In a fractured and uneven subsurface system, model outputs that are not grounded in current, distributed well data can create a false sense of confidence and obscure real depletion risks. That concern is heightened by two facts already visible in the available record: the most geologically vulnerable Bull Run and Occoquan headwaters are not directly covered by OWML stream monitoring, and USGS well 49V1 shows a long-term decline in groundwater level. Together, those gaps and trends underscore the same point: a model without adequate physical verification is not a dependable basis for development decisions. At minimum, any use of the model should be paired with a substantial physical monitoring program—including additional observation wells, headwater monitoring, and ongoing water-level measurements—before it is treated as a dependable guide for development decisions.

2026 Virginia Energy Plan

Every four years the Commonwealth is required to produce the Virginia Energy Plan. It is completed in the first year of every governor’s term. The 2026 Virginia Energy Plan (VEP) was released by Governor Abigail Spanberger’s administration on October 1, 2026. Each VEP  has a 10 year horizon and serves as the Commonwealth's energy roadmap.

While it is a policy framework rather than a binding regulatory document, it explicitly sets the state's legislative agenda to handle surging electricity demand—driven primarily by data centers—while remaining on a legally mandated path to a net-zero power sector by 2050. Unfortunately, what the VEP does is choose the assumptions that would give the exact result the Administration wanted; a paper-thin justification that a rapid clean energy transition is the most cost-effective path forward. In execution in the real world, this plan is unlikely to work and the gap between modeling assumptions and physical grid limits threatens to trigger severe reliability issues and dramatic, unmodeled rate hikes for Virginia consumers.

Virginia utilized advanced capacity expansion modeling to analyze the trade-offs of four distinct paths forward. The assumptions imbedded in the models have a tremendous impact on the results. For example, Virginia assumed a moderate annual growth rate which is lower than the PJM forecast. In truth, the actual growth rate will simply be how much of the forecast load growth from data centers gets connected.

In addition, a key factor in determining what scenario costs the least to Virginians is the social cost or carbon and the health impacts from burning fossil fuels. Climate modeling developed over the past 20 years—such as the EPA framework—relies on an what now appears to be an artificially low 2% long term discount rate to arrive at the central SCC value of $190 per metric ton. Right now the risk free 30 year rate is 5.4%. At a 2% discount rate, $1 million in climate damages 100 years from now is worth roughly $138,000 today. At a 5.4% discount rate, that same $1 million in future damages shrinks to a present value of just $5,200. So, by using a low discount rate you make carbon reductions more valuable.

The Core Challenge: Balancing Data Centers & The Grid

•  Rapid Demand Growth: Virginia is facing its fastest load growth in over 80 years. Data center electricity usage now accounts for roughly one-third of the state's total energy consumption.
• The "Cost-of-Service" Shift: The plan addresses rising retail utility bills by targeting infrastructure cost attribution. It explicitly mandates that large electricity users must pay for the specific transmission and infrastructure upgrades required to support their growth, protecting residential ratepayers from subsidizing industrial buildouts.
• The Virginia Clean Economy Act (VCEA): The VCEA mandates that the Commonwealth's major electric utilities transition to 100% renewable energy by 2045 or 2050 through a combination of solar, onshore and offshore wind, energy storage, and efficiency improvements.

The VEP focuses primarily on Virginia’s electricity system and the path to a carbon free electric power  under the Virginia Clean Economy Act (VCEA), with some additional consideration of transportation energy use and costs. Currently, Virginia’s largest energy use is electricity generation to power homes, businesses, and industry, with petroleum for transportation the second largest use. In 2024, Virginia’s electricity supply was primarily from natural gas, nuclear power, and electricity imports, with solar and offshore wind playing a growing role

Virginia has always imported a significant share of the electricity it uses annually. Since 2020, imports have grown to meet rising electricity demand primarily of data centers while retiring coal capacity. Virginia imports about 30-40% of its annual electricity use—the highest percentage in the nation—making regional transmission capacity and PJM Interconnection (PJM) market conditions important factors in the Commonwealth’s reliability and electricity costs.

High-voltage transmission infrastructure serves as the bulk delivery backbone of the Commonwealth’s electric grid, moving power from large-scale generation facilities to local distribution networks. Virginia’s transmission system is primarily comprised of 500 kV lines that move electricity across Virginia, anchored by interstate connections in Northern Virginia to PJM and in Western Virginia by two 765-kV lines operated by APCo that connect to AEP’s regional multi-state network.

In recent years, however, rapid load growth in Northern Virginia combined with permitting and construction delays on major regional transmission projects has significantly constrained interstate import capacity. PJM’s latest reliability modeling indicates that Virginia’s import demand will exceed its physical transmission capacity by 2028.

Which brings us to transmission. The modeled pathway costs explicitly exclude full transmission and distribution infrastructure costs- which are socialized to all customers. The 2026 VEP calculates only new generation capacity, energy storage, fuel, and the rest of the supply hardware. These baseline numbers omit critical cost components of distribution.  The VEP estimated costs exclude the transmission and localized distribution infrastructure investments. They also exclude federal tax credits or the guaranteed utility return on equity (ROE) that utilities add on top of infrastructure investments.

 The Four Modeled Energy Pathways

Each pathway evaluates a different supply mix to achieve net-zero power sector emissions.

Pathway	Core Strategy	Key Mechanics & Impacts
Current Policies	Baseline Compliance	Adheres strictly to the Virginia Clean Economy Act (VCEA) targets and state participation in the Regional Greenhouse Gas Initiative (RGGI).
Distributed Flexibility	Demand Management & Small-Scale Gen	Administration's Preferred Approach. Targets 10 GW of distributed energy (e.g., rooftop solar, batteries) by 2045 and mandates a 10% demand flexibility threshold for data centers. Cuts total system costs by roughly 10% ($37 billion) through 2050 by shaving peak demand.
Distributed Flexibility + In-State Priority	Localization	Reduces reliance on out-of-state power imports by half, limits utility-scale solar land footprint, and aggressively replaces it with in-state advanced nuclear and offshore wind.
Distributed Flexibility + No New Gas	Accelerated Elimination	Completely prohibits new gas generation infrastructure. Relies entirely on immediate clean firm resources and intensive battery storage.

 Key Strategies & Structural Reforms

Utility Procurement & Regulatory Reform: Proposes moving away from traditional cost-of-service regulations that reward monopolies (like Dominion Energy) purely for capital spending. Instead, it recommends performance-based utility incentives tied directly to cost-effective ratepayer outcomes.

Clean-First Deployments: Prioritizes dual-use land strategies, including rooftop solar, parking canopies, brownfield redevelopments, and agrivoltaics to minimize the rural land footprint.

Non-Combustion Bridges: To maintain reliability during severe peak grid strain, the plan permits up to 5 GW of shorter-lived, low-emission, non-combustion natural gas resources (such as linear generators or fuel cells) as a bridge, avoiding multi-billion dollar commitments to long-lived fossil fuel infrastructure.

 

Political and Legislative Friction

The plan faces immediate pushback as the administration prepares legislative proposals for the upcoming General Assembly session. Legislative critics argue that the plan relies too heavily on solar infrastructure, depends on "social costs" rather than hard utility bill metrics to calculate affordability, and fails to adequately insulate everyday ratepayers from grid upgrade expenses driven by tech companies.

To make any of the plan’s four pathways achievable, Virginia and PJM must build generation at an unprecedented, historic pace. However, the physical process of connecting new power to the grid is structurally broken. PJM’s "interconnection queue" (the waiting list for new power projects to get approved to plug into the grid) is severely backlogged. It currently takes an average of 5 years for a solar or battery project to move from proposal to operation. In addition, there is an equipment problem. There is a severe global shortage of high-voltage transformers, circuit breakers, and specialized labor. Lead times for critical grid transformers are currently 3 to 4 years, making the rapid scale-up required by the "Distributed Flexibility" pathway nearly impossible to execute.

The Solar Surge: Historically, Virginia has deployed roughly 0.6 GW of utility-scale and distributed solar per year. To meet the preferred path, that pace must jump immediately to 2.5 GW per year—a 4.1x acceleration.

Other assumptions that challenge reality:

• Battery storage requires an even steeper logistical steepening. Moving from a negligible historical baseline (~0.05 GW/year) to the target required pace of 0.50 GW per year demands a massive 10x acceleration factor.
• Meeting these specific buildout targets under current supply chain constraints remains the plan's highest risk hurdle.
• Even distributed resources like commercial solar require massive local zoning battles. In Virginia, local county boards are increasingly passing strict setback rules and moratoria on new solar and battery storage facilities due to rural land preservation concerns.
• The plan relies on up to 5 GW of advanced non-combustion gas (like hydrogen or fuel cells) to act as a bridge. However, these technologies are not yet manufactured or deployed at the commercial utility scale. This is counting on energy technology that is yet to be developed commercially. 
• SMRs are required in all the plans, but are still entirely in the planning and licensing phases in the United States. Commercial deployment is not expected until the mid-to-late 2030s at the absolute earliest. Relying on up to 17 unbuilt, unpermitted reactors to balance the grid is not realistic.

The biggest assumption is that Virginia can reliably import 30% of its annual electricity consumption from its neighbors while simultaneously choking off its own fossil-fuel generation.

Virginia does not import electricity in a vacuum; it imports it through PJM Interconnection, the regional grid network that spans 13 states. The 2026 Energy Plan treats PJM as an infinite, clean power supply but PJM’s own data shows the pool is drying up:

• While Virginia is electrifying and building data centers, neighboring states like Ohio, Illinois, and Pennsylvania are experiencing their own data center booms and factory electrifications.
• PJM has repeatedly warned that fossil-fuel plants are retiring faster than new clean generation can get permitted and built. PJM projects that by 2030, the region could face a reliable capacity shortfall of up to 10 GW during peak weather events.
• When neighboring states face their own supply crunches, they will restrict exports to keep their own lights on. Virginia cannot import power that does not exist.

The for all of the above reasons, the VEP’s model functions as a closed math problem rather than an executable engineering plan. By holding the import cap at 30% while drastically underpricing the power, ignoring the physical 5,992 MW transmission bottleneck, and utilizing what appears to be a very low discount rate of 2%, the model successfully outputs the exact result the administration sought: a paper-thin justification that a rapid clean energy transition is the most cost-effective path forward.

The 2026 Virginia Energy Plan is an exercise in political and environmental signaling, designed to show how a net-zero grid could theoretically coexist with a massive tech boom on paper. I am disappointed. 

PJM and Dominion Energy are treating it as exactly that: a theoretical paper. In the real world, they are executing a pragmatic, "all-of-the-above" strategy that relies on physical steel in the ground, conventional natural gas, large-scale transmission lines, and strict data center mandates to ensure that when a citizen or a tech company flips a switch, the power actually comes on.

The entire plan can be found at this link: 2026 Virginia Energy Plan -For Publishing.pdf

Winter Forecast from NOAA

From the National Oceanic and Atmospheric Administration,NOAA.  El Nino Looms Large over U.S. Winter Outlook

URL: https://www.noaa.gov/news-release/el-nino-looms-large-over-us-winter-outlook

NOAA’s U.S. Winter Outlook shows the current El Nino will likely influence the upcoming winter months (December 2026 to February 2027), with warmer-than-average temperatures expected across the northern U.S.; wetter-than-average conditions likely across the southern U.S., from California to Florida, and along the Eastern Seaboard; and drought improving for the majority of the country, including areas in the East, Southern Great Plains and Southwest. After, the devastating rain last weekend in Florida, I thought to take a look.

“As we know from decades of observations, a very strong El Nino will likely bring a southward and eastward shift to the jet stream for the contiguous U.S., which tilts the odds toward stormier weather across the South,” said Ken Graham, director of NOAA’s National Weather Service. “While higher chances of rain and snow may bring relief to drought-afflicted areas, too much precipitation on dry soil can fuel destructive flash floods and mudslides. Now is the time for communities and individuals to plan ahead for the upcoming winter.” 

Precipitation Outlook

• Wetter-than-average conditions are most likely across the southern tier, from the Southwest to the Southeast, and along the East Coast.
• Elevated odds for drier-than-average conditions for Hawaii, much of Alaska, the northern Mountain West, some areas in the Great Lakes and the western Pacific Islands.
• Areas in the Intermountain West, northern Plains, Ohio Valley and Puerto Rico / U.S. Virgin Islands have equal chances of below-average, near-average or above-average seasonal total precipitation amounts. 

They weren’t kidding about that forecast. Over last weekend more than two feet of rain fell along portions of Florida in just a day and a half, from Sunday afternoon to Monday evening.

As reported in the Washington Post an event like this so rare that it had a recurrence interval of more than 1-in-1,000 years in some areas. This could be a sign of what’s to come with the powerful El Niño this winter. Nothing like Florida is expected here (though it was not expected there, either), but we are forecast to have above average precipitation this winter and above average temperatures- so it’s rain.

 

Temperature Outlook

• Warmer-than-average temperatures are favored for northern California, the Pacific Northwest, Midwest, Mississippi Valley, much of the East, Alaska and for Puerto Rico / U.S. Virgin Islands.
• Near-normal temperatures are most likely for parts of the Southeast including Florida.
• Hawaii and much of the Southwest have equal chances of below-average, near-average or above-average seasonal mean temperatures.

Drought Outlook

• Drought conditions are forecast to improve or end in the Southwest, Southern Great Plains, Southeast and parts of the East Coast.
• Drought conditions are favored to develop or worsen across parts of the northern Rockies, along with northeastern Oregon and eastern Washington.

About NOAA's seasonal outlooks

NOAA’s seasonal outlooks provide the likelihood that temperatures and total precipitation amounts will be above-, near- or below-average, and how drought conditions are anticipated to change in the months ahead. The outlook does not project seasonal snowfall accumulations as snow forecasts are generally not predictable more than a week in advance. NOAA’s Weather Prediction Center updates its three-month outlook monthly.

 

Water Is One System: Understand the Cumulative Risk Before Development Proceeds

 

The central question is not whether development should ever occur. It is whether Prince William County should approve large-scale development outside the 2040 Comprehensive Plan before it understands the cumulative consequences for the Occoquan watershed—a drinking-water source for nearly one million people. When the potential harm is regional, long-lasting, and difficult or impossible to reverse, uncertainty is not a reason to proceed. It is a reason to pause.

Water does not recognize parcel lines, zoning districts, or regulatory silos. Groundwater, streams, wetlands, stormwater, treated wastewater, reservoir chemistry, and drinking-water treatment are parts of one connected hydrologic system. A decision that reduces groundwater recharge, increases polluted runoff, alters stream flow, or diverts treated return flow affects the same finite supply. Dividing those impacts among separate permits and agencies does not divide the physical risk.

That is why watershed-scale warnings must guide parcel-level decisions. Cumulative damage is created one rezoning, one permit, and one project at a time. No single approval may appear decisive, but together they can push the watershed beyond a threshold that cannot be undone. Evaluating each proposal in isolation obscures the central issue: the combined effect of all approved, pending, and reasonably foreseeable development.

The recently completed LIMA, land-use analysis, may actually understate that combined effect. Its development assumptions were based largely on past growth patterns during the 2013-1015 period of extremely low construction, and on the zoning and development framework reflected in the Comprehensive Plan. Yet, what is happening on the ground is major rezonings and proposals outside that framework. Longleaf, Alderwood, and Hawthorn at Kettle Run alone account for more than 1,200 acres, with additional rezonings advancing in the central Broad Run area and high density residential or data center rezonings under construction or approved in Gainesville, Haymarket and Mid County. If the model does not include the intensity, location, and timing of these exceptions, it cannot reliably describe the future being approved on a one off basis by the Board of County Superviosors.

The consequences are not limited to a single pollutant or stream segment. Clearing, blasting, mass grading, and soil compaction can permanently alter infiltration, groundwater recharge, and stream baseflow. Additional impervious cover increases runoff, sediment, nutrients, dissolved salts, and heat. Industrial cooling can consume or evaporate water that would otherwise remain in the system, while reduced or degraded return flows can weaken reservoir inflow during drought. These effects interact; they must be evaluated as a system.

Mitigation does not eliminate this concern. Stormwater ponds, filters, operating rules, and treatment upgrades can reduce some impacts, but they cannot recreate intact headwaters, uncompacted soil horizons, mature root systems, or the original subsurface pathways that store, filter, and slowly release water. Nor can conventional treatment easily remove every dissolved salt or persistent chemical introduced by more intensive development. Once the natural system is replaced, the public inherits the residual risk and the long-term cost.

This is also an issue of who benefits and who bears the consequences. Developers may capture the immediate financial gain from rezoning and construction, while water customers and taxpayers bear the cost of treatment upgrades, lost storage, drought vulnerability, and damage that may not appear until years later. Public officials should not transfer an irreversible regional risk to residents without first establishing, through credible cumulative analysis, that the watershed can absorb the proposed growth.

We Have Seen This Pattern Before

The rapid expansion of data centers offers a warning. Localities approved projects in a piecemeal fashion before fully accounting for their combined demand on generation, transmission lines, substations, land, and ratepayers. Only after many approvals were in motion did the scale of the strain on the power grid become clear. We should not repeat that sequence with water: approve first, discover the systemwide consequences later, and then ask the public to pay for emergency infrastructure or live with diminished reliability.

The wiser course is informed development within the Comprehensive Plan unless and until updated modeling accounts for the full buildout now being proposed, the interaction of water quantity and water quality, more intense storms and drought, industrial water use, wastewater return flows, and the combined effect of projects already approved or pending. The analysis must come before irreversible land disturbance, when public officials still have meaningful choices.

Water is one system. Before approving development that could permanently alter the Occoquan watershed, the County should understand the cumulative impact—not gamble a regional drinking-water supply so private developers can realize immediate gains while the public absorbs the long-term damage and costs.

 

Groundwater Recharge

Reager, J. T., Molotch, N. P., & Famiglietti, J. S. (2026). Net groundwater recharge in California occurs primarily during above-average snowpack years. Geophysical Research Letters, 53, e2026GL122396. https://doi.org/10.1029/2026GL122396

The below is excerpted from the research letter cited above.

Dr. John T. Reager, a water scientist at NASA Jet Propulsion Laboratory, and colleagues analyzed the GRACE and GRACE FO* satellite data and estimate how much groundwater recharge has occurred annually in California since 2004.  They found that the state’s aquifers have had a net gain in only six of 18 years, and years with lots of snow led to the biggest rises in underground water.

In the Central Valley, groundwater has been dropping for about 100 years as wells and pumps pull up water to irrigate crops and make up the difference when water allocations are cut. The historical maximum of 28 to 30 feet was measured southwest of Mendota in a famous USGS photo. Since that time other areas of the central valley have experienced over draft of groundwater and subsidence. The all resulted in the Sustainable Groundwater Management Act (SGMA), which has yet to prevent the overdrawing of groundwater.

“Nonetheless, California's Central Valley is one of the most heavily managed groundwater systems in the world, supplying much of the state's water use during dry periods and supporting a highly productive agricultural region. Long‐term groundwater withdrawals have led to sustained declines in aquifer storage over the past several decades (e.g., Liuet al., 2022), particularly during multiyear droughts when surface water availability is limited.”

Despite the importance of groundwater in buffering the region from drought, basin‐scale monitoring of groundwater recharge the satellite data from GRACE and GRACE‐FO observations reveal basin‐scale net groundwater recharge events in California are very rare. The net positive recharge years (when more water is added to groundwater stores than is pumped out) are strongly associated with above‐average snowpack years. Thus, future warming may reduce the likelihood of recharge‐capable years.

The data used in this study came from GRACE (Gravity Recovery and Climate Experiment) and its successor, GRACE-FO (GRACE Follow-On), are joint satellite missions between NASA and German research partners (DLR and GFZ) designed to track the movement of mass across the Earth by measuring changes in its gravitational field over time. The original GRACE mission operated from 2002 to 2017. GRACE-FO was launched in May 2018 (yeah SpaceX) after the first satellite had failed.

Unlike typical remote sensing satellites that observe the planet using reflected light or to look down at the surface. Instead, GRACE uses two identical satellites flying in tandem to act as a highly sensitive scale. Because water is incredibly heavy, its seasonal and multi-year movement constitutes the largest variable source of changing mass on Earth. GRACE and GRACE-FO provide critical monthly global maps of these regional variations, unlocking insight across years.

This years of data creates a time series of water data allows the researchers to derive the groundwater over time. By subtracting the measured surface water. The first derivative of a time data series represents the rate of change of the measured variable with respect to time. In physical terms, it allows the scientist to observe the recharge rate of groundwater.

Groundwater storage changes were estimated as the residual of the terrestrial water balance:

GWA(t) = TWSA(t) – SWEA(t) – SWA(t) – SMA(t) + E(t),

where GWA represents groundwater anomaly, derived from terrestrial water storage changes (TWSA) and sub-tracting out snow water equivalent changes (SWEA), surface water storage changes (SWA), and soil moisture storage changes (SMA). (We’ll leave off the discussion of the fudge factor E in this short discussion). The first derivative of a time data series represents the rate of change of the measured variable with respect to time. Thus, in physical terms, it measures the recharge of the water in the system.

By separating groundwater changes from other water stores such as snow, soil moisture, and reservoirs, they estimate when net groundwater recharge occurs and how large itis. The researcher found that basin‐scale net groundwater recharge is rare and has occurred in only 6 years since 2004. Importantly, years with groundwater recharge are more strongly linked to unusually large snowpack in the Sierra Nevada than to high rainfall alone. Even during the exceptionally wet winter of 2023, which produced record snowpack, only a small fraction of the total precipitation was retained as groundwater, and this gain did not offset long‐term groundwater losses. These results suggest that future groundwater replenishment in California depends on substantial snowpack, which is expected to decline as the climate warms. As more precipitation falls as rain rather than snow, opportunities for large‐scale groundwater recharge may become increasingly limited.

Groundwater recharge is a complex Earth system process that has been identified by the National Academy of Sciences as critically important  to our source waters and poorly monitored. Distinguishing between mechanisms of groundwater recharge is essential for assessing how groundwater recharge potential may evolve under a warming climate and that is expected to shift precipitation from snow toward rain, and land use change that may alter the mechanics of recharge and snow melt.

Though, land subsidence does not directly cause the disparity between rainfall and snowfall recharge, which is primarily governed by the physical delivery timeline of the water. However, decades of compaction significantly compound the issue by reducing the aquifer system's vertical permeability and capacity to absorb rapid surface water inputs. When decades of groundwater over-extraction cause land subsidence, the fine-grained clay layers (aquitards) within the aquifer system undergo irreversible compaction. This structural collapse dramatically alters how the subsurface interacts with different water delivery rate.

Net groundwater recharge reflects a balance between climatic inputs and anthropogenic withdrawals. During dry years with limited surface water availability, continued groundwater pumping results in net depletion, whereas during wet years, reduced pumping and enhanced infiltration allow for net recharge (e.g., Famiglietti et al., 2011). Recharge pathways in California include diffuse infiltration, managed recharge, and recharge associated with prolonged snowmelt and runoff from the Sierra Nevada. While basin‐scale GRACE observations cannot resolve individual pathways, the strong association between net recharge and high snowpack is consistent with enhanced focused and mountain‐front recharge during years with sustained snowmelt.

Several high‐precipitation years did not correspond to the largest recharge events, whereas water year 2023 coincided with record‐high Sierra Nevada snowpack, indicating that precipitation alone may be insufficient to produce basin‐scale net recharge. The magnitude of recharge in wet years further emphasizes the limits of basin‐scale recovery. Across the six net‐positive recharge years since 2004, annual net groundwater gains amount to approximately 20%–45% of April 1 Sierra Nevada snowpack.  The researchers found that only a portion of the snow pack was converted into net groundwater gain.

These results suggest that groundwater recharge is not only through total water input, but also through its control on the timing and persistence of water availability. Snowmelt provides a delayed and sustained input of water that extends into spring and early summer, when evaporative demand remains relatively low and infiltration conditions are more favorable. In this sense, snowpack acts as a gating condition for basin‐scale recharge, enabling sustained infiltration beyond the peak precipitation period rather than serving as an independent water source.

This is an important piece of work, the methodology can be used in other climates and geologies to determine groundwater recharge and its drivers.

Carbon Footprint of Northern Virginia

Loudoun County—followed by Prince William County—has undermined the climate progress made by the rest of Northern Virginia. MWCOG’s Mid-Course Review shows that most neighboring jurisdictions reduced greenhouse-gas emissions between 2005 and 2023, while emissions rose sharply in the two counties that approved the region’s largest concentrations of hyperscale data centers. The operating facilities already built, together with the enormous additional load in the development pipeline, are not merely local land-use decisions: they are reshaping electricity demand across Virginia and the wider PJM grid. Unless that growth is constrained or matched with additional carbon-free generation, storage, transmission, and enforceable clean-energy requirements, Virginia cannot remain on the decarbonization trajectory established by the Virginia Clean Economy Act (VCEA), and PJM’s transition away from fossil generation will be delayed.

Two Counties Reversed the Regional Climate Trend

From 2005 to 2023, Arlington and Fairfax counties cut gross emissions by 33% and 27%, respectively. Those gains were offset by Prince William County’s 19% increase and, above all, Loudoun County’s extraordinary 130% increase. Net emissions rose even faster in Loudoun and Prince William because development also removed forests and other carbon-sequestering land. The result is a stark regional inequity: communities that reduced energy use and emissions are seeing their progress eroded by data-center approvals concentrated in two Northern Virginia jurisdictions.

Loudoun County Created the Region’s Largest Carbon Load

Loudoun County is the principal source of the regional divergence. Its transformation into the world’s largest data-center market drove a 130% increase in gross emissions and a 145% increase in net emissions from 2005 to 2023, despite the region’s adopted climate targets. Loudoun’s land-use approvals converted what had been a manageable regional emissions-reduction challenge into a rapidly expanding electricity-demand problem that reaches far beyond the county’s borders.

·         Local approvals produced a permanent industrial load: The pandemic temporarily reduced transportation and office emissions, but data centers operate continuously. Loudoun’s community-wide emissions had already risen 54% by 2020 compared with 2005; by 2023, the accelerating data-center buildout had pushed the gross increase to 130%. This was not a short-lived post-COVID rebound. It was the predictable result of approving a new class of round-the-clock industrial electricity demand.

• Sector Concentrations: Because MWCOG applies a location-based electricity emissions factor, the electricity consumed inside the county maps directly to Loudoun’s inventory. Commercial grid electricity consumption within the Built Environment sector constitutes the vast majority of Loudoun's footprint, outstripping the combined emissions of the residential and transportation sectors.
• Carbon Sequestration Deficit: In 2020, Loudoun's forest canopy sequestered 430,000 metric tons of CO2 equivalent, offsetting roughly 7% of total emissions. However, ongoing land grading and industrial expansions have driven the county’s net emissions increase (+145%) higher than its gross increase (+130%), highlighting the loss of natural carbon sinks of the tree canopy.

Prince William County Copied Loudoun’s Example

Prince William County then followed Loudoun’s development model, approving major data-center corridors and adding another rapidly growing block of continuous industrial demand. This all took place right after the PWBOCS voted to approve climate goals for the County. By 2023, the county’s gross emissions were 19% above the 2005 baseline and net emissions were 22% higher. Instead of helping counterbalance Loudoun’s extraordinary increase, Prince William compounded it—further weakening the region’s ability to meet its collective climate commitments.

• The Multi-Year Trajectory: PWC's community-wide emissions grew from 4.16 million metric tons of CO2  equivalent in 2005 to 4.96 million metric tons of CO2 equivalent in 2018—a 19% increase prior to the pandemic. While per-capita emissions initially dropped 10% due to rapid population growth outstripping emissions pacing (+32% population vs +19% emissions), the post-lockdown industrial expansion stabilized the footprint at a persistent 19% gross and 22% net increase by 2023. Since that time data centers continued to come on line.
• Sector Breakdown Trends: Historically, PWC’s emissions were led by energy consumption (residential and commercial) at 54% and transportation at 38%. Post-lockdown data shows commercial grid consumption increasing its share of the pie, driven by the expansion of data center corridors along Route 234 and the Innovation Park tech zone.
• Canopy Degradation: PWC’s forest canopy provides net sequestration of more than 368,000 million tons of CO2 equivalent annually. Similar to Loudoun, massive land-use changes for new industrial campuses and right-of-way transmission lines have widened the gap between gross (+19%) and net (+22%) emission baselines.

The Consequence: Virginia and PJM Cannot Decarbonize on Schedule Under the Current Load-Growth Path

The consequences now extend well beyond county boundaries. The VCEA requires Dominion Energy to move toward 100% carbon-free electricity by 2045, but the data centers already operating—and the much larger pipeline under construction or approved in Loudoun and Prince William—are causing electricity demand to grow faster than clean generation and transmission can be added. Under that trajectory, new renewable resources must first chase new data-center load rather than displace existing fossil generation. The practical effect is to postpone coal and gas retirements, increase pressure for new gas-fired resources and imports, and make the statutory VCEA timetable increasingly difficult if impossible to achieve.

1.      Virginia’s problem becomes PJM’s problem: Dominion is part of the multi-state PJM system. When Northern Virginia’s continuous data-center load exceeds the pace at which Virginia can add carbon-free supply, PJM must commit additional generation somewhere in the regional fleet. Because marginal and reliability resources remain heavily dependent on natural gas and coal, local approvals in Loudoun and Prince William can prolong fossil generation and associated transmission investments across multiple states.

The approved pipeline locks in the conflict: This is no longer a question of whether individual facilities can purchase renewable-energy credits or claim annual clean-energy matching. The grid must supply these campuses every hour of every day. Unless data-center growth is tied to deliverable, additional, round-the-clock carbon-free power—and unless projects can be curtailed when that supply is unavailable—the approved pipeline will keep raising the amount of generation PJM must serve before it can retire fossil resources. On the present path, Loudoun and Prince William Board of County Supervisors have placed both Virginia’s VCEA schedule and PJM-wide decarbonization out of reach without major policy intervention.

EO 22 Changes the Conversation—Not Virginia’s Data Center Trajectory

Bottom line: Executive Order 22 marks an important change in how Virginia talks about data centers. For the first time, the Commonwealth formally acknowledges that rapid growth imposes statewide costs on the electric grid, water resources, communities, ratepayers, and the environment. But the order does not stop, cap, or materially slow development already in the pipeline. It changes the policy framework—not the trajectory.

 What EO 22 Changes Now

EO 22 shifts Virginia away from unqualified promotion of data centers and toward greater transparency, coordination, and accountability. Its principal immediate actions are to:

• end most state-agency nondisclosure agreements that conceal project incentives, resource demands, and community impacts;
• withdraw certain discretionary state assistance, expedited permitting, and site-readiness support from new projects of 25 megawatts or more;
• centralize state data center policy under the Office of the Chief Energy Officer; and
• direct agencies to study noise, backup generation, water use, emissions, grid reliability, cost allocation, workforce benefits, and community protections.

What EO 22 Does Not Change

EO 22 is not a moratorium and contains no statewide limit on data center capacity, square footage, or power demand. It does not revoke entitlements, cancel approved site plans, rescind vested rights, prevent local approvals, or direct Dominion Energy or PJM to deny service to entitled projects. Losing state assistance may add cost or delay, but it is unlikely to deter a hyperscale developer from a Northern Virginia site that already has zoning approval, fiber access, and a path to power.
Why the Approved Pipeline Matters

Loudoun County reports 233 data center buildings built or under construction,  and covering more than 56 million square feet, another 35.7 million square feet approved but not yet built—potentially about 115 additional buildings—and 20.7 million square feet still under review. Not every approved project will be constructed, but the pipeline is large enough to sustain years of construction, substations, transmission lines, generators, and additional water and power demand even if new approvals slowed sharply.

Prince William County has also accumulated substantial approved and pending capacity. Its September 2026 zoning amendment limits future expansion of the Data Center Opportunity Zone, but it preserves several pathways for properties and projects already in the development process. The amendment narrows future growth without eliminating the existing pipeline, and landowners have a deadline of December 21, 2026, to finalize their by-right status before it is eliminated. Other than the thousands of acres still in the overlay district, Prince William County has as of October 2026 :  64 completed data centers, 25 projects under construction,  53 approved awaiting groundbreaking, and 15 active land-use applications currently under staff and agency review. The completed data centers account for 13.4 million square feet and currently under construction are an additional 8 million square feet. 

A Policy Pivot, Not Yet a Growth-Control Policy

Virginia has actively recruited data centers since 2009 through tax incentives, economic-development assistance, and permissive local land-use policies. EO 22 breaks with the assumption that growth should be encouraged without first accounting for cumulative costs. That is a political and administrative shift.

Yet most of the order’s stronger provisions remain studies, planning tools, voluntary standards, agency recommendations, or proposals for future legislation. Transparency and coordination can improve decision-making, but they do not themselves constrain development. The order may affect projects not yet entitled; it is unlikely to produce a visible near-term decline in construction in Loudoun or Prince William.

What Would Bend the Curve

EO 22 will become a true growth-control policy only if it leads to enforceable measures: mandatory discretionary review of large facilities; cumulative regional impact analysis; binding limits tied to available power, water, air-quality, and watershed capacity; full assignment of infrastructure costs to the projects that cause them; authority to defer new utility connections where capacity is inadequate; tighter local zoning; limits on extensions or expansions under old entitlements; and reconsideration of tax incentives that continue to encourage growth.

Conclusion: EO 22 changes Virginia’s conversation by recognizing that data center growth carries statewide costs and requires statewide oversight. But the approved and entitled pipeline will continue to drive construction for years. Unless the order is followed by binding legislation, regulation, utility decisions, and stronger local land-use controls, Virginia will have changed its posture without changing its path.