Climate Risks in Mining: Tailings Dams, Pits and ESG Compliance

Climate risk in mining is the chance that heavy rain, wind, drought or slope instability will compromise the safety of a tailings dam, the stability of a pit slope, or the cost of an operation. It is a management variable, not an EIA detail that gets filed away after the permit is issued. And it can be measured with local, hour by hour data.

• Brazil lost R$ 184 billion to climate events between 2022 and 2024, and 91% of those losses were uninsured (CNseg/EY).

• ANA identified 213 priority dams for safety management in the RSB 2026, across 19 states and the Federal District. Mining leads the list, with 55 structures (26% of the total).

• ANM Resolution 220/2025 published new rules for mining dams. They take effect on August 2, 2027, when they revoke ANM Resolution 95/2022.

• CVM Resolution 244/2026 revoked the mandatory nature of the CVM 193 report and adopted the *pratique ou explique* model. CVM 218 remains in force: whoever publishes the report is still required to follow CBPS 02.

• At a copper mining operation in Chile, the i4sea model captured 20 of 21 actual weather-related shutdown events, 95% sensitivity (technical data, with no published financial estimate).

I have been in a dam safety meeting where the question was not "is it going to rain a lot?". It was "how much is it going to rain, where, and will this change surface drainage and the structure's freeboard in the next few hours?". No one there wanted a capital-city bulletin. They wanted a number for that specific structure.

Operational climate risk is the chance that a weather or hydrological condition (heavy rain, flooding, wind, slope instability) will compromise the safety, continuity or cost of an asset, within a defined timeframe. In mining, this becomes concrete in three places: the tailings dam, the open-pit mine and the report that investors demand.

This article covers all three, in the order in which the dam engineer, the pit geotechnical engineer and the ESG compliance team need to make decisions.

What climate risks in mining are and why they became a compliance issue

Climate risk in mining is the combination of physical exposure (rain, wind, drought, slope instability) and the vulnerability of the asset exposed to that condition. It becomes a compliance issue because insurance companies, banks and investors started demanding proof of risk management before releasing coverage, credit or capital.

The number behind this shift is large and national in scope. Brazil recorded R$ 184 billion in climate losses between 2022 and 2024, across 67 significant events, and 91% of that amount was uninsured, according to a study by CNseg with EY presented at COP30 in 2025. This is a picture of the country, not of mining specifically: the percentage combines different sectors, regions and company sizes. For mining, the figure scales the exposure, it does not define how much of a specific event stays on a company's balance sheet. Coverage exists, and contracts have deductibles, limits and exclusions. What cannot be insured is the halted operation, the lost production and the time needed to resume. That stays with the company.

In my experience with mining company boards, climate risk only becomes a compliance issue when someone asks "is this auditable?". It is not enough to say the company monitors the climate. It is necessary to show the data, the source and the resulting decision, all documented. This requirement comes from insurers, lenders and, increasingly, from investors who read the sustainability report before allocating capital.

Climate compliance in mining, therefore, does not originate from a sector-specific law. It arises from the intersection of dam safety (regulated by ANM), water resource use (regulated by ANA) and capital market requirements (now under a *pratique ou explique* regime, as we will see below). All three fronts require the same basic input: reliable, local climate data with an auditable history.

Climate risks in tailings dams: what changes with ANM Resolution 220

Extreme rainfall affects tailings dam safety through three pathways. The first is pore pressure: infiltrating water reduces the effective stress of the embankment and, with it, the mobilized strength of the material.

The second pathway is overtopping, when flow exceeds the spillway's capacity or the freeboard is insufficient. The third is internal erosion, which responds mainly to seepage and the hydraulic gradient within the structure. Rain aggravates these conditions by raising the water table and pore pressure, but the relationship is not immediate: it depends on the zoning, filter and drainage system of each dam. Of the three pathways, overtopping is the most directly hydrological, and where rainfall forecasting delivers the clearest value.

ANM Resolution 220, published on October 16, 2025, consolidated mining dam safety rules into a single text. The regulation takes effect on August 2, 2027, the date from which it revokes ANM Resolution 95/2022. Until then, Resolution 95/2022 governs.

For Articles 76 and 77, enforcement begins earlier, on April 22, 2027. These are the provisions related to hypothetical dam failure studies and flood zone areas.

The transition has its own dates, and they matter for planning. Hypothetical failure studies and the delimitation of new flood zones and affected areas must be completed by July 30, 2027. Adaptation of Self-Rescue Zones has a deadline of December 26, 2029, under ANM Resolution 245/2026.

Regarding decommissioning, precision matters. The obligation to decommission dams raised using the upstream method comes from Law 14.066/2020. Resolution 220 addresses the completion of such works among the conditions it regulates and enforces. The obligation itself did not originate with it.

The scope ended up narrower than it appears. In the final text, ANM removed mining areas, processing areas and tailings deposit areas from the concept of structures associated with the dam. The regulation covers the embankment and its auxiliary structures. It does not cover the pit or the plant.

ANA, in the 2026 Dam Safety Report, identifies 213 priority structures for safety management, spread across 19 states and the Federal District. Mining leads the list, with 55 dams, 26% of the total. Under this criterion, "priority" means a conservation problem or a PNSB safety requirement not met by the operator.

The regulation specifies what must be done with the structure. It does not say when rainfall in that basin will put pressure on the embankment. A dam can meet every criterion and still receive a volume above what the original design accounted for, because the historical series used for sizing is old and the reference station is far from the dam.

Combining regulation and local data is what turns compliance into management. The regulation sets the criterion. Climate data calibrated to the basin determines how many hours or days of advance warning there is to act, and that difference is operational, not bureaucratic.

Climate risks in mining pits: slopes, drainage and failure

Heavy rain affects the stability of a mining pit by saturating the slope. Saturation reduces effective stress and, with it, the mobilized strength. In an unsaturated slope, the loss of suction has a similar effect, since suction functions as apparent cohesion. The material's angle of friction itself changes little.

The same volume of rain that a city absorbs without issue can saturate a slope already weakened by excavation. The difference lies in the hourly intensity and the sequence of events, not the monthly average.

INMET maintains historical rainfall series by station, but many pits are kilometers away from the nearest station, sometimes in a different microbasin. And ANM Resolution 220 covers mining dams, not pit slopes. No regulation solves the scale limitation of the data.

In operational practice, the risk of slope failure in an open-pit mine grows under three recurring conditions. Accumulated rainfall above the drainage design. Rain concentrated in a few hours, with monthly volume within the average. And a sequence of events without enough time for desaturation between them. None of these three show up in a regional average bulletin.

I have already covered this topic in more depth in the article on slope stability and anticipating landslide risk in mines, where I detail the drainage model combined with precipitation forecasting. The central point bears repeating here: anticipating the event costs less than reacting to it.

ESG and climate reporting in mining after CVM 244

CVM Resolution 244, of May 29, 2026, revoked Article 2 of CVM Resolution 193 and item III of its Article 5. Article 2 was the provision that made it mandatory for publicly traded companies to prepare and disclose a sustainability-related financial information report, under the IFRS S1/S2 standard.

CVM 218 remains in force, and this tends to be misunderstood. It is this resolution that makes CBPS Technical Pronouncement No. 02 mandatory, which internalizes IFRS S2. In practice: filing the report is no longer mandatory; whoever publishes it is still required to follow CBPS 02.

The new regime is *pratique ou explique*. CVM Resolution 244/2026 is already in force, but the requirement to publicly justify the choice not to file the report takes effect starting January 1, 2027. From then on, a company that decides not to file must communicate that choice to the market, in its own disclosure. It does not stay silent. It is required to publicly explain why it does not report, and that exposure carries a reputational cost.

There is also a timing issue. The original obligation would have applied to fiscal years starting on or after January 1, 2026. Resolution 244 arrived before the first cycle closed, so some companies that were preparing to report never ended up doing so under the mandatory regime.

And there is a legal challenge underway. A lawsuit before the 30th Federal Court of Rio de Janeiro seeks to reinstate the mandatory requirement. The request for urgent injunctive relief was denied, and the regulation remains applicable while the merits of the case are still pending judgment.

i4sea | Climate intelligence for mining slopes. See how climate intelligence and on-site monitoring anticipate landslides on mining slopes, reducing shutdowns with 48 to 72 hours of advance warning.

I covered the timeline of this change, including the reasons for the revocation and what became optional versus mandatory, in the article on what changes for those who were preparing to report climate risk. It is worth reading for anyone who already had a structured process in place and now needs to decide whether to keep it.

My view, after talking with boards that have already invested in reporting: maintaining the climate measurement process is cheaper than rebuilding it when the next requirement arrives. The data that feeds the report is the same data that feeds the operational decision for the dam and the pit. Discarding that structure means losing the most expensive asset to rebuild: the historical record.

What public forecasting does not capture at the pit and dam level

What is called a "25 km forecast" is, in practice, a product distributed on a grid of about 25 km, generated from global models with a native resolution of 9 to 13 km. ECMWF's IFS runs at close to 9 km. NOAA's GFS runs at close to 13.5 km. The 0.25 degree grid was the output delivered to users. On October 20, 2026, ECMWF began also publishing IFS products on a 0.1 degree grid. These are distinct concepts: the delivery grid became finer, while the model's native resolution remains close to 9 km. A smaller grid improves data access; it does not create detail that the model does not resolve.

Both models are the foundation of global numerical forecasting. They do a good job of anticipating synoptic patterns, cold fronts and cyclones, and ECMWF is often cited as the most accurate at the medium range. The problem starts when this scale is used to decide whether a specific slope will become saturated in the next 12 hours.

The strongest argument in favor of 1 to 3 km resolution is convection. A 13 km model does not resolve the convective cloud: it parameterizes it, estimating convective rainfall using simplified formulas. Below about 3 km, convection starts to be represented explicitly, albeit still with limitations. High resolution improves rainfall placement. It does not close the stability equation on its own: slope saturation depends on rainfall intensity and duration, antecedent moisture and the site's geotechnical characteristics. A few hours of convective rain is one of the scenarios that stress the structure, not the only one.

This does not mean a 1 to 3 km model can tell the difference between the bottom of the pit and the top of the slope. A pit can range from a few hundred meters to about 2 km in extent, which means it fits within one or two grid cells.

Neither alternative eliminates spatial variability within the pit or the basin. A rain gauge represents a single point, and statistical downscaling calibrated by station reduces error without eliminating uncertainty. What the combination does is increase the representativeness of local data, and that caveat applies to any provider. The real gain comes from the combination: regional model plus on-site station.

This is how i4sea works: proprietary climate reanalysis combined with a regional numerical model at 1 to 3 km resolution, calibrated with a station on the asset, for each monitored structure. This has been the operating format for more than ten years.

In a region of rugged terrain, common in Brazilian mining, a 25 km cell blends valley and mountain into the same average. The relevant scale is the basin that drains into the dam and the specific slope of the pit, together with the asset's own station.

How to turn climate risk into an auditable number

i4sea turns climate exposure into an auditable indicator by cross-referencing each asset's historical rainfall, flow and soil saturation data with the actual event that occurred, not with the regulation's theoretical threshold. The result is i4sea's risk index per asset, hour by hour, with the data source documented. The index is not the probability of failure: it is the probability of the weather condition that stresses the structure, cross-referenced with each asset's threshold and measured against the history of actual events. Failure is a conclusion drawn from geotechnical analysis, and that remains the engineering team's role.

At a copper operation in the Chilean Atacama, the model captured 20 of 21 actual weather-related shutdown events in the historical record analyzed, 95% sensitivity. This is a technical result, measured against the mine's own records, not against a generic risk assumption.

Sensitivity alone does not tell the whole story. A model that raises an alert every day reaches 100%. What separates a useful alert from noise is the false alarm rate and the critical success index. Before celebrating a capture rate, the question is how much noise it costs per correct event.

This type of indicator is missing from most content on climate risk and mining. Most describe the regulation (ANM, ANA, CVM) or the phenomenon (rain, drought, El Niño), and few show how to turn it into an auditable number for engineering or investors. I detailed this path in how to calculate physical climate risk as an auditable number for reporting.

You do not control the climate, but you can manage the risks. In mining, that means having data for your dam's basin and your pit's slope, not the state capital's average, and having that data documented for when the compliance question comes.

Frequently asked questions

References