Source-linked AI summary
Causal Attribution of Coastal Water Clarity Degradation to Nickel Processing Expansion at the Indonesia Morowali Industrial Park, Sulawesi
Sandy Hardian Susanto Herho, Alfita Puspa Handayani, Iwan Pramesti Anwar, Faruq Khadami, Karina Aprilia Sujatmiko, Doandy Yonathan Wibisono, Rusmawan Suwarman, Dasapta Erwin Irawan
TL;DR
Whether rapid industrialization at Indonesia’s Morowali nickel complex degraded adjacent marine water clarity remains unquantified. The study combines satellite ocean-color records, structural-break analysis, BSTS causal inference, and land-cover data, finding a post-2019 causal deterioration in nearshore clarity.
Problem
No comparable assessment has quantified whether Indonesia’s rapidly expanding nickel processing has degraded coastal water quality.
Method
The study combines multi-sensor K_d(490) records, consensus changepoint detection, BSTS counterfactual inference with Banda Sea controls, and satellite land-cover analysis.
Results
+0.676 × 10−2 m−1 (+14.38%, p = 0.012) was the estimated post-2019 causal increase in nearshore K_d(490) off Morowali.
Takeaways & Limitations
The 14.38% clarity loss shoaled the euphotic zone from 97.8 to 85.5 m in oligotrophic, highly biodiverse waters where turbidity may stress coral habitats.
Takeaways & Limitations
The 4 km satellite resolution integrates areas larger than individual plumes, attenuating the nearshore signal and making estimates conservative lower bounds.
Abstract
from arXiv · showhide
Indonesia's nickel ore export ban has driven rapid expansion of smelting and hydrometallurgical processing capacity at the Indonesia Morowali Industrial Park (IMIP), now the world's largest integrated nickel processing complex, on the coast of Central Sulawesi. Whether this industrialization has degraded the adjacent marine environment remains unquantified. We apply Bayesian structural time-series (BSTS) causal inference to a multi-decadal, multi-sensor satellite ocean color record of the diffuse attenuation coefficient at 490 nm, $K_d(490)$, to test for a causal link between IMIP expansion and nearshore turbidity change. A consensus structural breakpoint, a significant posterior causal effect estimated against a Banda Sea counterfactual, and a distribution-free placebo rank test collectively establish that coastal water clarity deteriorated after the transition from initial nickel pig iron production to hyper-expansion of high-pressure acid leaching facilities for battery-grade nickel. Satellite-derived land cover analysis independently corroborates this timing, showing substantial built-area growth and concurrent tree cover loss within the IMIP footprint. The resulting euphotic zone shoaling occurs in oligotrophic waters supporting high marine biodiversity, where even moderate optical degradation may impair coral photosynthesis and compress depth-dependent reef habitat. These findings quantify a marine environmental cost absent from Indonesia's mineral downstreaming policy discourse and demonstrate a transferable, satellite-based quasi-experimental framework for causal impact assessment at coastal industrial sites in data-limited tropical settings.
1 Introduction
Nickel demand for decarbonization is increasing environmental pressure, while Indonesia’s rapid laterite mining and processing expansion may threaten coastal water quality. This study tests whether industrialization at the Morowali coastline caused detectable nearshore water-clarity degradation using satellite observations and causal time-series methods.
- Motivation: Nickel demand is projected to rise substantially under decarbonization scenarios, although extracting battery-grade nickel causes severe and poorly quantified environmental harm.The paper frames nickel as a case study of the environmental costs associated with low-carbon technologies.
- Terrestrial impacts: Laterite nickel mining strips vegetation and topsoil across wide tropical forest tracts, while Indonesia’s nickel land footprint has grown dramatically.The passage attributes this land intensity to shallow ore bodies requiring open-pit extraction.
- Coastal impacts: Sediment runoff, tailings discharge, and riparian-buffer destruction can connect nickel mining and processing to coastal water-quality degradation.Comparable laterite mining in New Caledonia increased terrigenous sediment delivery and raised dissolved trace-metal concentrations in coastal waters.
- Ecological stakes: IMIP’s adjacent waters lie within the Coral Triangle, where turbidity can compress the euphotic zone and impair coral photosynthesis, calcification, and recruitment.The region also supports fisheries on which millions of coastal livelihoods depend.
- Study objective and approach: The study tests whether Morowali’s rapid industrialization caused statistically detectable nearshore water-clarity degradation using 1998–2024 Kd(490) records, changepoint detection, and BSTS causal modeling.The design compares an IMIP impact zone with a Banda Sea control zone and combines multiple changepoint algorithms.
2 Data
The study combines multi-decadal satellite ocean-color observations with terrestrial land-cover and ocean reanalysis data across coastal Morowali Regency, Central Sulawesi. The dataset centers on IMIP and compares its offshore impact zone with an open Banda Sea control zone.
- Study domain and sampling: The study domain spans 121.30–123.80°E and 1.80–3.80°S across inner Tolo Bay and the western Banda Sea margin, with IMIP near 122.16°E, 2.82°S.Smelting began in April 2015, followed by rapid capacity expansion after the January 2020 nickel ore export ban.
- Terrestrial land cover: 2017–2024 annual 10 m Sentinel-2-derived land-cover maps quantified terrestrial dynamics over the impact zone.The rasters covered approximately 478 km^2 per year, with water and cloud-flagged pixels masked before analysis.
- Ocean-color observations: 324 monthly Kd(490) fields from January 1998 to December 2024 were extracted from the CMEMS GlobColour merged ocean-color product.Area-weighted zonal means used 16 valid ocean pixels in the impact zone, 49 in the control zone, and 1467 of 2304 across the full domain.
- Oceanographic covariates: SST and SSS were extracted as monthly surface means from the 1/12° GLORYS12V1 global ocean eddy-resolving reanalysis over the same domains.GLORYS12V1 uses the NEMO ocean model, atmospheric boundary conditions, and ocean-observation assimilation through a reduced-order Kalman filter.
3 Methods
The study combines multi-year land-cover transition analysis with monthly, area-weighted ocean-color and hydrographic time series. It evaluates temporal trends, changepoints, causal counterfactuals, and distribution-free placebo confirmation using explicitly stated assumptions and computational procedures.
- Land-cover analysis: Annual 10 m Sentinel-2 LULC maps from 2017–2024 excluded water and clouds, retaining six terrestrial classes for seven consecutive transition matrices.The classes were trees, flooded vegetation, crops, built area, bare ground, and rangeland.
- Land-cover analysis: Transition matrices were aggregated across seven intervals, and three-level intensity analysis tested interval, category, and transition departures from uniform land conversion.Cohen’s h quantified practical significance, with |h| ≥0.8 classified as large.
- Oceanographic time series: Monthly area-weighted means of Kd(490), SST, and SSS were extracted for 324 observations, with isolated gaps linearly interpolated before averaging.Kd(490) was measured in ×10^-2 m^-1, SST in °C, and SSS in PSU.
- Oceanographic time series: The 324-month record was partitioned into pre-smelter, initial-operations, post-export-ban hyper-expansion, and full-record epochs for robust distributional summaries.The policy-aligned periods were before April 2015, April 2015–December 2019, January 2020 onward, and the full record.
- Trend analysis: Kd(490) trends were estimated with Theil–Sen slopes and assessed using Kendall’s τ with two-sided asymptotic-normal p-values separately by zone and epoch.The analysis used the median as the primary location measure and IQR and median absolute deviation for dispersion.
- Causal and changepoint analysis: Changepoints were assessed independently by zone using RBF-kernel segmentation with PELT, Binary Segmentation, and window-based detection, while causal attribution assumed the control captured shared basin-scale forcing.A nonzero difference-in-differences effect was therefore attributed to local perturbation under that identifying assumption.
4 Results
Results show rapid IMIP land-cover transformation alongside a post-May 2019 increase in nearshore Kd(490). BSTS estimates a positive causal effect relative to the counterfactual, supported by placebo testing and sensitivity analysis.
- Land-cover change: Built area increased from 12.26 km2 (2.56%) in 2017 to 46.18 km2 (9.65%) in 2024, while tree cover fell from 32.37% to 27.33%.This represented a +7.09 percentage-point built-area increase and a −5.04 percentage-point tree-cover loss.
- Land-cover change: All seven 2017–2024 year-pairs were classified as active, with St exceeding Uint = 1.10% yr−1 in every interval (χ2 = 32,221, df = 6, p < 0.001).Total changed area was 1,038,269 pixels (103.83 km2).
- Causal impact: The BSTS model estimated an average causal effect of ¯δ = +0.676 and a relative effect of +14.38% over the 68-month post-period.The post-period observed mean was 5.381 × 10−2 m−1 versus a counterfactual mean of 4.704 (95% CI: [4.179, 5.230]); the cumulative effect was ∆T = +46.00 and z = 2.522 (p = 0.012).
- Causal impact: None of 40 converged placebo tests exceeded the observed effect, yielding ˆprank = 0.000; all leave-one-out configurations retained a positive significant effect at α = 0.05.The all-covariates estimate was ¯δ = +0.676 (p = 0.012), while dropping control Kd(490) yielded +0.714 (p = 0.018).
5 Discussion
Independent evidence identifies May 2019 as a causal breakpoint in Morowali’s nearshore water-clarity degradation, coinciding with rapid industrial expansion and land-cover conversion. The resulting 14.38% increase in K_d(490) corresponds to 12.3 m of euphotic-zone shoaling in biodiverse, oligotrophic waters.
- Causal attribution: May 2019 marked a consensus breakpoint in nearshore K_d(490), with four independent evidence lines supporting causal local anthropogenic forcing.The evidence includes a multi-algorithm breakpoint, BSTS effect, placebo rank test, and stable leave-one-out sensitivity.
- Causal attribution: +0.676 × 10−2 m−1 (+14.38%, p = 0.012) was the BSTS posterior mean causal effect on nearshore K_d(490).The distribution-free placebo rank test yielded ˆprank = 0.000.
- Industrial timing: The May 2019 breakpoint aligned with the late-2018 transition toward HPAL facilities rather than April 2015 nickel pig iron commissioning.The initial smelter phase operated within a comparatively modest footprint.
- Land-cover corroboration: Built area expanded 3.8-fold from 12.26 km2 (2017) to 46.18 km2 (2024), while tree cover dropped by 5.04 percentage points.Over half of total land-cover change was exchange-type, consistent with organized industrial conversion.
- Ecological significance: The 14.38% increase in K_d(490) shoaled the euphotic zone from 97.8 to 85.5 m, a ∆Zeu = −12.3 m change in Coral Triangle oligotrophic waters.The passage notes that moderate turbidity increases may have disproportionate consequences for benthic communities adapted to high baseline irradiance.
- Identification and limitations: The BSTS design constructed a counterfactual from unaffected control-zone observations, while leave-one-out analysis found no single covariate drove the result.The 4 km GlobColour resolution integrates optical properties over areas larger than individual plumes, potentially attenuating the nearshore signal.
- Identification and limitations: τ = 0.179 (p < 0.001) indicates a weak but significant pre-intervention upward trend, possibly reflecting early construction, artisanal mining, or climatic drift.If anthropogenic, May 2019 represents a transition from gradual to acute degradation without invalidating the causal interpretation.
- Policy significance: The findings quantify a marine environmental externality from nickel downstreaming, with IMIP compressing comparable coastal transformation into barely a decade.The passage contrasts this compressed transformation with comparable western Pacific industries whose coastal impacts accumulated over decades.
6 Conclusion
Satellite-derived causal evidence shows that rapid industrialization along Indonesia’s Morowali coastline statistically degraded nearshore water clarity. The degradation emerged with HPAL-driven hyper-expansion, rather than initial smelter commissioning, and Kd(490) increased relative to a synthetic Banda Sea counterfactual.
- Conclusion: Satellite-derived causal evidence indicates statistically detectable degradation of nearshore water clarity after rapid industrialization along Indonesia’s Morowali coastline.The study identifies a causal effect using satellite observations and a synthetic counterfactual.
- Conclusion: Kd(490) increased substantially relative to a synthetic counterfactual constructed from contemporaneous Banda Sea controls.The counterfactual was based on contemporaneous controls from the Banda Sea.
- Conclusion: The consensus breakpoint coincided with HPAL-driven hyper-expansion and the imminent export ban, not initial smelter commissioning.The timing links the detectable change to the later expansion phase rather than the beginning of smelting operations.
Funding
The study received financial support from the ITB 3P Research Program and the University of California, Riverside.
- Funding came from the ITB 3P Research Program’s Talenta Unggul Scheme through ITB’s Directorate of Research and Innovation (Project ID: DRI.PN-6-64-2026) and UC Riverside’s Dean’s Distinguished Fellowship 2023.