Landside tritium leakage over through years from Fukushima Dai-ichi nuclear plant and relationship between countermeasures and contaminated water

The Fukushima Dai-ichi Nuclear Power Plant (FDNPP) accident in 2011 released large amounts of radionuclides (estimated ~520 PBq), primarily iodine-131, cesium-134/137, and noble gases. Tritium (3H, T1/2 = 12.3 y) was also released but, due to its low beta energy, received less attention immediately after the accident and environmental data remain limited in Japan. Tritium in boiling water reactors is produced mainly by ternary fission and neutron reactions on boron, lithium, and deuterium. Post-accident tritium inventory estimates vary: initial reports suggested ~1.81×10^13 Bq, whereas later TEPCO estimates indicate ~3.4×10^15 Bq across Units 1–3, with substantial amounts in storage tanks, reactor buildings, debris, and potentially released to the environment by 2016. Tritium can be released via ocean, atmosphere, and groundwater pathways. Early observations documented oceanic and atmospheric releases, including elevated tritium in precipitation and surface waters one month after the accident, well above natural background. However, the groundwater pathway is less understood. This study investigates continuous detection of tritium in groundwater on the land side of FDNPP (about 30 m from the site boundary) from 2013 to 2019, explores hydrogeological origins using 87Sr/86Sr as a tracer, and evaluates the relationship between site countermeasures and observed tritium levels.

Prior work quantified tritium releases to the North Pacific (0.1–0.5 PBq) and detected elevated tritium in precipitation (up to 1342 TU ≈ 158 Bq/L) and surface waters soon after the accident, confirming atmospheric deposition as a significant early pathway. Natural tritium levels in Japan are typically 1.1–7.8 TU (≈0.13–0.92 Bq/L). Studies of tritium in Japanese precipitation and ocean waters, as well as reactor production pathways, provide context for expected inventories and environmental behavior. Previous TEPCO reports and task force assessments estimated reactor inventories and distribution among tanks, buildings, and debris. Groundwater flow modeling suggests flow from the mountain side to the ocean. Strontium isotopes (87Sr/86Sr) have been used in hydrogeology to trace water-rock interactions and mixing, motivating their use here to assess groundwater origins and connectivity. However, detailed assessments of sustained land-side groundwater tritium contamination over multiple years have been lacking.

- Study area and sampling: Environmental groundwater (sump water) samples were collected in Okuma Town, Fukushima Prefecture, Japan, adjacent to the North Pacific, from late 2013 to 2019. Sump water was sampled directly from a pipe embedded in a cliff at the FDNPP site boundary (17 m above sea level), enabling direct groundwater collection without contact with soil or vegetation. Additional reference samples included flowing-well waters ~500 m from the plant and spring water on the shoreline. Published TEPCO monitoring data (≈17,000 points) for wells (groundwater bypass wells No.1–12; observation wells near H4 area E1–E14 and wellpoint F1; H6 area G1–G3) were compiled to contextualize local groundwater tritium trends. - Tritium analysis by LSC: Suspended matter was removed by filtration (Advantec 5C). Following MEXT protocols, ~50 mg KMnO4 and 50 mg Na2O2 were added to 50 mL of filtered water and distilled for 1 day. Aliquots (10 mL water + 10 mL UltimaGold LLT cocktail) were counted by liquid scintillation (Perkin Elmer Tri-Carb 3180 TR) with external standard quench correction (133Ba). Typical quench (tSIE) was 300–400, with ~23% detection efficiency. Counting time per vial was 24 h (1-h cycles). Detection limit was ~1.5 Bq/L. - Tritium analysis by 3He ingrowth: Selected samples were distilled and sealed in stainless steel bottles after complete degassing. After several months, ingrown 3He from tritium decay was measured using a Helix-SFT noble-gas mass spectrometer (AORI, University of Tokyo). Detection limit ~0.003 Bq/L; analytical uncertainty ~10%. Decay corrections were applied. Results agreed with LSC measurements. - Strontium isotope and concentration analysis: Water from two locations (cliff and shoreline) was processed for Sr concentration and isotopic ratios. Sr was separated using DGA and Sr extraction resins, then measured by TIMS (Phoenix X62, Isotopx) for 87Sr/86Sr and 90Sr/86Sr (static for 90Sr, multi-dynamic for stable isotopes). An Agilent 8800 ICP-MS measured stable Sr concentrations. Procedures included acid digestion (8 M HNO3), column chemistry, and loading onto degassed Re filaments with TaF5 activator. Results were normalized to NIST-987 (87Sr/86Sr = 0.710248) and expressed also in delta notation. 90Sr contamination was assessed; results were below detection limits (ND; DLs ~120–185 mBq/kg). - Countermeasures timeline considered: surface grouting (from Oct 2014), sub-drain pumping (from Sep 2015), land-side frozen soil wall (from Mar 2016), and sea-side impermeable wall (from Oct 2015). Potential correlations with tritium in sump water were evaluated qualitatively. - Constraints: TEPCO did not disclose groundwater bypass well flow rates, limiting calculation of absolute tritium fluxes in pumped well waters.

- Persistent land-side groundwater tritium: Sump groundwater sampled at the site boundary (cliff) from 2013–2019 consistently exceeded natural levels, with 3H ranging 15–31 Bq/L (average ~20 Bq/L), indicating a sustained FDNPP source. - TEPCO well data context: Groundwater bypass wells upstream of FDNPP showed heterogeneous tritium levels. Well No.10 (south side) rose from ~10 Bq/L (Jun 2014) to >3000 Bq/L (Apr 2016), then declined to ~1400 Bq/L by 2019; Well No.11 reached ~700 Bq/L (Jun 11, 2019); Well No.12 showed a monotonic decrease from a 2014 peak. Variability indicates complex groundwater flow. - Tank leak signatures inside site: Near the H4 tank leak area (Aug 2013), observation well E1 peaked at ~790 kBq/L (Oct 17, 2013) and trended downward thereafter; wells near the leak (E1, E10) remained higher than others, consistent with tank leak influence. In contrast, H6 area wells (G1–G3) did not show a clear correlation to the 2014 leak event. - Lack of atmospheric/surface inputs during study period: From 2013–2019, no tritium above natural levels was detected in air, precipitation, or nearby river waters within ~1 km, suggesting the sump groundwater tritium was not from contemporaneous atmospheric deposition or surface water infiltration. - Background reference waters: Flowing-well waters ~500 m from the plant had very low tritium (0.003–0.01 Bq/L by 3He ingrowth), below typical natural levels in Japan (0.13–0.92 Bq/L), indicating residence times on the order of decades (~≥40 years), and serving as local background. - Strontium isotopes indicate distinct groundwater pathways: 87Sr/86Sr at the cliff (0.705948) and shoreline (0.707525) differed significantly (delta values −6.054 vs −3.834 relative to NIST-987), implying different hydrogeological origins and a non-simple flow path over the impermeable layer. No Sr contamination was detected at either site (ND; DLs ~120–185 mBq/kg). - Countermeasures: No direct correlation was observed between the timing of water management measures (grouting, sub-drain pumping, frozen wall, sea-side wall) and the tritium radioactivity in the sump water. - Regulatory and inventory context: FDNPP’s tritium effluent limit is 1500 Bq/L (lower than the typical Japanese limit of 6.0×10^4 Bq/L), implying >1.2×10^12 L dilution would be needed for higher-activity waters. Post-accident tritium inventory estimated by TEPCO totaled ~3.4×10^15 Bq across Units 1–3, with substantial amounts in tanks and buildings and ~1.8×10^15 Bq released outside the reactor or in debris by Mar 2016.

The sustained detection of tritium at ~20 Bq/L in land-side groundwater at the FDNPP boundary demonstrates a persistent subsurface pathway for plant-derived tritium to move off site, addressing the central question of whether land-side groundwater leakage has occurred post-accident. The heterogeneity in TEPCO well data and the high tritium near known tank leak areas support at least partial contribution from 2013–2014 tank spills (H4, H6), while the widespread detection of tritium in multiple wells and the distinct 87Sr/86Sr signatures between cliff and shoreline springs indicate complex, stratigraphically controlled flow over an impermeable layer, consistent with a broader subsurface distribution of tritium originating from the initial accident and on-site contamination. The lack of elevated tritium in air, precipitation, and nearby rivers during 2013–2019 argues against ongoing atmospheric or surface-water inputs as the dominant source for the observed sump groundwater. Although the land-side frozen wall and sea-side sheet piles were intended to limit groundwater movements, the absence of a clear temporal correlation between these countermeasures and sump tritium suggests that established subsurface pathways may bypass or be only partially influenced by these interventions. The findings underscore the importance of strengthening land-side monitoring (in addition to ocean-side) and maintaining vigilance, as established subsurface routes could facilitate further leakage, particularly under perturbations such as earthquakes or heavy rainfall events.

This study provides the first multi-year documentation (2013–2019) of continuous, above-background tritium in land-side groundwater at the Fukushima Dai-ichi site boundary, averaging ~20 Bq/L. The evidence points to two plausible, non-mutually exclusive sources: infiltration from the 2013–2014 contaminated tank leaks and broader subsurface tritium disseminated across the site following the 2011 accident, with migration over impermeable layers toward the boundary. Strontium isotope ratios reveal distinct hydrogeologic flow paths between the cliff and shoreline springs, indicating complex subsurface hydrology. No direct linkage was found between the timing of on-site water management countermeasures and the tritium levels in boundary sump water. Given the persistence of land-side tritium and the potential for future mobilization, surveillance for both ocean-side and land-side leakage should be enhanced, including denser monitoring networks, better hydrogeological characterization, and disclosure of well flow rates to estimate fluxes. Future research should quantify tritium flux (not just concentration), integrate tracer tests and hydrogeological modeling to resolve flow paths and residence times, and assess how seismic or extreme weather events might alter subsurface transport.

- Lack of flow rate data for groundwater bypass and observation wells (not publicly disclosed by TEPCO) precluded calculation of absolute tritium fluxes and mass balances. - The study design relies on opportunistic boundary sump sampling and compiled TEPCO well concentrations; spatial coverage is limited and may miss heterogeneities. - Insufficient temporal/spatial resolution to definitively attribute the boundary tritium to specific sources (tank leaks vs. accident-derived subsurface contamination) or to confirm transport along specific segments of the impermeable layer. - No direct hydrologic connectivity was established between the cliff and shoreline springs despite strontium isotope contrasts; thus, inferred flow paths remain conceptual. - Detection limits for some radionuclides (e.g., Sr) constrain the ability to detect very low-level co-contaminants and to use multi-isotope fingerprints comprehensively.