Water Pollution and Treatment: Types, Causes, Major Incidents and the Chemistry of Clean Water

Water pollution means contaminants have entered a waterbody in amounts that degrade water quality, harm human health or disrupt ecosystems. This guide separates three questions that are often blurred together: what kind of pollution is it, where did it come from, and what can realistically be done about it. It ends with a decade of major incidents and the treatment chemistry that engineers actually rely on.

1. What counts as water pollution?

Pollution occurs when contaminants exceed a waterbody's capacity to assimilate them, or when concentrations reach levels that damage a designated use such as drinking, irrigation, recreation or aquatic habitat. A value only becomes meaningful when compared with the right benchmark, in the right units, for the right water use.

2. Types by pollutant

The most intuitive split is by what the pollutant is, because this directly determines both the harm and the treatment route.

Characteristic sub-forms are often treated separately: organic pollution (oxygen demand), heavy-metal pollution, pesticide pollution, oil pollution, thermal pollution and sensory pollution (colour, turbidity, odour).

3. Types by source and pathway

This split drives regulation and the choice of where to intervene.

• Point source — a single identifiable outlet such as a factory discharge pipe, a municipal wastewater plant outfall or a tanker spill. Easier to monitor and permit.
• Non-point (diffuse) source — contamination from broad or hard-to-trace areas: agricultural runoff, urban stormwater, fertilizer washed off lawns. Concentrations are usually lower but total loads are huge, and end-of-pipe interception alone cannot solve them.
• Internal (in-situ) source — pollutants accumulated in bottom sediment that are released back into the overlying water, producing "secondary pollution" long after the original discharge stopped.

By waterbody, pollution is also divided into surface water (rivers, lakes, reservoirs, seas) and groundwater. Groundwater is especially difficult: contamination is hidden and, once established, extremely slow and costly to reverse.

4. What causes water pollution?

Natural factors

Special geological conditions, rock weathering, volcanic activity, soil erosion, wind-blown dust, natural plant decay and atmospheric deposition can all raise contaminant levels. These are usually regional and cyclical, but they are rarely the main driver of the pollution problems people report.

Human factors — the main cause

Real waterbodies almost never carry one pollutant type from one source. A common chain: industrial point sources add metals and toxic organics; domestic sewage adds nutrients and pathogens; agricultural runoff adds a seasonal pulse of nutrients and pesticides. Together they raise the nutrient load, trigger algal blooms, and decomposition then strips oxygen and kills fish — while contaminated sediment releases pollutants again under low-oxygen conditions. This is why treatment usually combines source control, diffuse-load reduction and internal-load removal.

5. A decade of major incidents, 2015–2025

An incident is treated here as "major" if it caused deaths or large-scale supply interruption, crossed provincial, national or environmental-media boundaries, or had multi-year effects and triggered litigation or systemic reform.

Tailings dam failures: the deadliest category

On 5 November 2015 the Fundão dam in Minas Gerais collapsed, sending about 50 million m³ of iron-ore waste into the Doce River. Nineteen people died, hundreds were displaced, more than 600 km of river was affected and the slurry reached the Atlantic. The dam was operated by Samarco, jointly owned by Vale and BHP. In October 2024 the companies and the Brazilian government reached a settlement worth about 170 billion reais (~US$32.7 billion); in November 2025 the UK High Court ruled BHP liable.

The pollution did not end when the mud settled. Research published in October 2025 showed that iron oxides in the tailings dissolve under long-term anoxic conditions in estuary wetlands, re-releasing lead, cadmium, copper, nickel and chromium into soil and groundwater, where crops take them up. Local bananas exceeded FAO cadmium limits and cocoa exceeded copper and lead limits; for children under six, the health-risk index from eating local bananas alone was already above 1. Heavy-metal pollution can produce "delayed exposure" that surfaces a decade or more later.

On 25 January 2019 the Brumadinho dam, also operated by Vale, collapsed, releasing about 12 million m³ of mining waste and killing more than 270 people; some remains were never recovered. Years later, 85% of local households reported drinking-water impacts and 75% reported supply and quality problems; among adolescents, 52.3% had elevated blood manganese, 12.2% elevated blood lead and 28.9% elevated urinary arsenic.

On 28 March 2020 the No. 4 overflow shaft at the Luming mine tailings pond in Yichun, Heilongjiang, China, collapsed. About 2.32–2.45 million m³ of tailings containing 89.39–117.53 tonnes of molybdenum leaked, pushing molybdenum above limits along roughly 340 km of the Yijimi and Hulan rivers. The Tieli No. 1 waterworks stopped abstraction and about 68,000 people were affected. A joint investigation found the incident was caused by substandard engineering quality; the company was fined 13.261 million yuan, partly because it took no action for five hours after detecting the leak.

The common pattern is clear: the risk source is a static facility, warnings existed, and they were treated as tolerable. Tailings ponds, phosphogypsum stacks and wastewater lagoons are high-consequence, low-probability facilities. When they fail, the downstream river receives decades of accumulated pollution at once.

Drinking-water system failure: Flint

In April 2014 Flint, Michigan, switched its water source from Detroit's system to the corrosive Flint River to save money, without adding the required food-grade phosphate corrosion inhibitor. The river water attacked ageing lead pipes and released lead into tap water. Testing in 2015 found lead in some homes at 13,200 ppb, against a US EPA standard of 15 ppb. Between 6,000 and 12,000 children had elevated blood lead, with possible permanent neurological and cognitive damage; the same system fostered Legionella, with 90 confirmed pneumonia cases and 12 deaths in 2014–2015. In January 2016 the Obama administration declared a federal emergency.

What makes Flint distinctive is that the contamination was not an external spill: the supply system itself manufactured it, through a combination of ageing infrastructure, regulatory failure and a disadvantaged community bearing the cost. It directly drove the US Lead and Copper Rule Improvements published in October 2024, requiring nationwide lead-pipe replacement within ten years.

Nutrients and algal blooms: the Oder River

In August 2022 a mass fish kill struck the Oder, which forms part of the Poland–Germany border. A joint European Commission analysis confirmed about 360 tonnes of dead fish and ecological impact along 500 km, almost certainly caused by a large toxic algal bloom. The causal chain was high-salinity wastewater discharge combined with drought and low water levels: salt-tolerant toxic golden algae (Prymnesium parvum) bloomed and produced toxins that killed fish in bursts.

In August 2024 the Gliwice Canal, a tributary, again saw about 77 tonnes of dead fish from a golden-algae bloom; Poland injected hydrogen peroxide at the confluence to kill more than 90% of the algae, while stressing that this was an emergency measure rather than systematic treatment, and announced plans for wastewater desalination. The lesson is that climate change rewrites pollution thresholds: the same salt discharge may be diluted to harmlessness in a wet year and trigger a bloom in a dry, low-flow year. This is why treatment increasingly considers hydrological conditions, not just discharge concentrations.

Marine oil and radioactivity

MV Wakashio: on 25 July 2020 the Japanese bulk carrier ran aground off Mauritius, broke apart on 6 August and spilled more than 1,000 tonnes of the roughly 4,000 tonnes of diesel and fuel oil aboard; the hull split in two on 15 August. The spill site was next to the Blue Bay wetland and other sensitive ecosystems, and Mauritius declared an environmental emergency. The investigation found the vessel had been on autopilot and unmonitored for hours before grounding; the chief officer was jailed for 20 months for endangering navigation.

Piney Point: in March–April 2021 a phosphogypsum stack at an abandoned Florida phosphate plant leaked through a damaged liner and displacing wall, forcing about 814 million litres of nitrogen- and phosphorus-rich wastewater into Tampa Bay and the evacuation of more than 300 nearby homes. Tampa Bay had been a model of marine ecosystem recovery since the 1990s; the nutrient pulse was a direct stress test of that recovery and a harmful-algal-bloom risk.

Fukushima: on 24 August 2023 Japan began releasing treated water from the Fukushima Daiichi nuclear plant. About 1.34 million tonnes were stored in roughly 1,000 tanks; the 2023 plan was four releases totalling about 31,200 tonnes, with about 7,800 m³ in the first 17 days. Because rainwater and groundwater keep contacting the melted fuel debris, 90–140 tonnes of contaminated water are generated each day, and the release is planned to last at least 30 years. The dispute is not only about tritium: the water contains more than 60 radionuclides, including carbon-14 (half-life ~5,370 years) and iodine-129 (half-life ~15.7 million years), and many experts argue risk cannot be judged on tritium concentration alone. In February 2024 the plant's purification equipment leaked again, releasing about 5.5 tonnes of water containing caesium, strontium and other radioactive material.

Hokkaido arsenic: in July 2023 a geothermal survey site at Rankoshi, Hokkaido, continuously emitted steam; water samples showed arsenic at about 2,100× the Japanese drinking-water standard and 420× the agricultural standard. While transferring the arsenic-laden water through pipes, Mitsui Oil Exploration leaked about 2,000 litres and then another 10,000 litres; 17 people reported ill health from the steam.

Cross-border input: the Jialing River

On 5 May 2017 monitoring in Guangyuan, Sichuan, found abnormal water quality at the Jialing River's entry into the province, with thallium at the Xiwan waterworks intake 4.6× the national limit (0.0001 mg/L). Guangyuan's urban supply was cut for 36 hours. The source was traced to direct discharge of thallium-bearing wastewater from a Hanzhong zinc/copper mining company in Ningqiang County, Shaanxi — an upstream, cross-provincial, one-off pollution slug. Guangyuan activated 30,000–40,000 tonnes/day of emergency supply and delivered water by fire truck and sanitation vehicle to undersupplied areas.

The case shows that polluters and victims often sit in different administrative jurisdictions. One upstream direct discharge can force a downstream city to stop its taps, making cross-border monitoring, early warning and joint enforcement a critical weak point.

6. What the decade shows

• Facility failure, not routine discharge, is the deadliest. The two Brazilian dam failures, the Luming tailings pond, the Piney Point pond and a Latvian sewage-plant wall collapse all involved storage or treatment facilities failing under extreme conditions and releasing decades of accumulated pollution at once.
• The lag of pollution is badly underestimated. Heavy metals at Mariana reached the food chain 11 years later; Brumadinho residents still had elevated metals seven years on. An incident's "end" is usually only the end of emergency response — ecological and health effects are just beginning.
• Climate change is rewriting pollution thresholds. At the Oder, drought and low flow turned a routine high-salinity discharge into a 360-tonne fish kill; the 2025 Yunmeng Xida canal failure in China was officially attributed mainly to sustained drought leaving no upstream inflow, reduced self-purification and some domestic sewage mixing in.
• Disclosure and accountability usually advance only after disaster. Flint drove nationwide lead-pipe replacement, Mariana drove cross-border class actions and a multi-billion-dollar settlement, and the Oder pushed Poland and Germany to coordinate through the International Commission for the Protection of the Oder. Institutional progress is almost always disaster-driven.

A note on scope: incidents verifiable at global scale cluster in 2015–2023; 2024–2025 searches mostly returned regional or local water-quality events rather than new incidents of the same magnitude as the table above.

7. How water pollution is treated

Treatment design starts from two questions: what is the pollutant (which fixes the technology) and where does it enter (which fixes the governance and interception system). Real projects rarely use one technology; they chain source reduction, process interception, end-of-pipe treatment and internal/ecological restoration. China's programme for urban waterbodies near communities is summarised as source control and interception, internal-load treatment, ecological restoration and maintaining live flow.

By pollutant type

Physical pollutants. Suspended solids are removed by physical separation — bar screens and sieves for large debris, grit chambers for sand and grit, sedimentation tanks for settleable solids by gravity, and dissolved-air flotation for light solids and oil. Oil is handled with separators, demulsification and flotation, usually with three-phase separation and flocculation ahead of biological units. Thermal pollution is essentially energy management: cooling towers, recirculating cooling water, deep-water discharge and distributed outfalls. Radioactive pollution requires dedicated collection, decay storage, evaporation, ion exchange and solidification — it must not enter ordinary sewage systems.

Chemical pollutants diverge sharply by form:

• Heavy metals — the mainstream route is "decomplexation → precipitation → deep adsorption". Chemical precipitation (hydroxide or sulfide) works under conventional conditions, but is unstable in high-salinity, high-ammonia or complexed wastewater. Where metals are complexed with citrate, tartrate or ammonia, hydroxide precipitation fails and must be preceded by decomplexation — Fenton or sodium hypochlorite oxidation — then neutralisation and precipitation, then chelating-resin adsorption. A typical multi-barrier process is Fenton oxidation + neutralisation/precipitation + chelating-resin adsorption. High-value metals such as nickel and cobalt are better recovered as raw material than disposed of as hazardous waste.
• Acids, alkalis and salts — neutralisation with lime or caustic soda for acidic water and acid for alkaline water; high-salinity wastewater needs evaporation, membrane concentration and salt separation, often "pretreatment + evaporation + crystallisation" to approach zero liquid discharge.
• Refractory organics (phenols, BTEX, halogenated hydrocarbons, pesticides, pharmaceutical intermediates) — biological treatment alone is inefficient, so advanced oxidation first opens the rings and raises biodegradability: ozone, O₃/H₂O₂, UV/H₂O₂, Fenton. This is followed by hydrolysis-acidification + contact oxidation + MBR. High-strength liquid waste can be sharply reduced by low-temperature vacuum evaporation before biology.
• Nitrogen and phosphorus — biological nutrient removal is mainstream: A/O, A²/O and SBR achieve nitrification–denitrification through anoxic–aerobic cycling and biological phosphorus removal through anaerobic–aerobic cycling. Deep treatment can use denitrification filters (over 90% ammonia removal), MBR for more stable effluent, and reverse osmosis (over 95% salt rejection, with pretreatment) where reuse or tight standards require it.

Biological pollutants. The core is disinfection plus preventing recontamination: coagulation, sedimentation and filtration remove most pathogen carriers, then chlorine, chlorine dioxide, ozone or UV disinfect. Hospital wastewater must be separately collected and pretreated (with residual-chlorine control) before entering municipal sewers; drinking-water sources rely on isolation zones, buffer strips and strict total-load control.

Eutrophication is a composite problem where cutting nutrients alone often fails. The sequence is source control and interception (the precondition — otherwise continued input cancels the investment), emergency bloom control (physical barriers, mechanical harvesting, flushing), and ecological restoration (restore aquatic vegetation, optimise fish populations, raise transparency). Practice has used a three-dimensional system of floating ecological islands, bio-media to enhance self-purification, bottom aeration with submerged plants forming an "underwater forest", and filter-feeding fish.

By source type

Industrial point sources. Industrial parks must have centralised wastewater treatment with automatic monitoring, and firms must hold environmental impact approvals and discharge permits — gaps here are the root of much pollution. The technical route is "one plant, one plan": pretreatment (equalisation, oil separation, coagulation–flotation) → biological treatment (hydrolysis-acidification, contact oxidation, MBR) → deep treatment (RO, ion exchange) → compliant discharge or reuse. The key principle is segregation: mixing high-strength, metal-bearing, oily and rinse streams turns recoverable resources into hard-to-treat hazardous mixtures.

Municipal sewage. The standard three-stage flow: primary (screens, grit removal, primary sedimentation); secondary (activated sludge in aeration tanks, secondary sedimentation) — the main battleground for nutrient removal; and tertiary/deep treatment (coagulation, media filtration, membrane MF/UF/RO, ozone or UV-based oxidation) for residual organics, solids, colour and trace toxics. The decisive work is often outside the plant: separate storm and sanitary sewers, and sewer rehabilitation. Infiltration, misconnection and direct discharge are the real reasons many urban waterbodies return to black and odorous conditions.

Agricultural and rural diffuse sources. Diffuse pollution cannot be end-of-pipe intercepted; the approach is "reduce, intercept, absorb". Reduce: soil testing and formula fertilisation, organic fertiliser replacing 20–40% of chemical nitrogen, fertigation raising fertiliser use efficiency by over 40%, green pest control (light traps, sticky boards, pheromones, biocontrol). Intercept: ecological ditches, buffer strips and floating-island systems — Taicang monitoring showed efficient ecological ditches cut nitrogen and phosphorus export by about 15%. Absorb: straw use as fertiliser, feed, substrate or growing medium; livestock manure via crop–livestock integration and centralised treatment for dense smallholder areas. Rural domestic sewage is better served by small constructed wetlands, stabilisation ponds or land treatment, or by connection to town sewers, than by copying large urban plants.

Internal sources. Sediment dredging directly removes accumulated organics and pollutants and cuts secondary release; in black-odour treatment it is often combined with sediment disposal. Dredging must control depth and disturbance, or it can cause a burst of pollutant release. In-situ capping and passivation with zeolite, modified clay or calcium-based material suppress phosphorus release at lower cost than dredging but with limited duration. Aquaculture should move to non-fed ecological systems with treated tailwater and scientific use of pond sediment.

Groundwater. Once contaminated, groundwater is extremely hard to restore and needs a different logic. Pump-and-treat is most common and controls plume spread, but does not fully remove contaminants sorbed to the aquifer matrix and requires long operation. Permeable reactive barriers filled with zero-valent iron, zeolite or activated carbon need no continuous energy and cost less, but can clog and have limited life under high concentrations. In-situ chemical oxidation or reduction suits refractory organics; nanoscale zero-valent iron dechlorinates chlorinated solvents and reduces, adsorbs and co-precipitates metals. In-situ bioremediation is cheap but slow; monitored natural attenuation is cheapest but needs long verification and is unsuitable for high-risk sites. In every case the prerequisite is cutting off the source — nitrate from fertilizer infiltrating with irrigation, or leachate from a failed landfill liner — otherwise remediation never catches up.

Practical pitfalls

• Define the pollutant before choosing the process. "COD exceedance" may mean suspended solids, refractory organics or reducing inorganics; the routes differ completely. Bench tests on real water samples — evaporation, resin or activated-carbon adsorption — should validate feasibility before design.
• Do not expect one technology to solve combined pollution. Real waterbodies carry industrial point sources, domestic sewage and agricultural diffuse loads together; combined processes and multiple barriers are required.
• Source control always precedes restoration. If interception is incomplete, ecological restoration spending is cancelled by continuing input; this is especially true for groundwater.
• Distinguish "compliant discharge" from "ecological recovery". The first is delivered by treatment plants; the second needs hydrology, sediment and biological communities together. They are not substitutes.
• Operation and maintenance decide the real result. Leaking sewers, idle facilities and missing monitoring are the main reasons well-designed schemes underperform.

8. Key chemicals used in treatment

Two layers matter: chemicals dosed to make reactions happen (coagulants, oxidants, precipitants) and materials that remove pollutants by surface action (activated carbon, zeolite, ion-exchange resins). Engineering practice usually means the first when it says "dosing", but the two work together.

A quick lookup by target

9. Reading a water pollution story responsibly

• Identify the exact pollutant, its chemical form, the unit and the benchmark before drawing conclusions.
• Separate point, diffuse and internal sources: the same concentration can have very different control options.
• Check whether an incident is acute (spill, facility failure) or chronic (PFAS, nitrate, sediment release), because the response horizon differs by decades.
• Treat treatment as a chain, not a single process; source control usually matters more than the last polishing step.
• Keep the original report, method, station and date alongside any number you quote.