Pollution Datasets Research Articles

Prophet forecasting model: a machine learning approach to predict the concentration of air pollutants (PM2.5, PM10, O3, NO2, SO2, CO) in Seoul, South Korea.

Amidst recent industrialization in South Korea, Seoul has experienced high levels of air pollution, an issue that is magnified due to a lack of effective air pollution prediction techniques. In this study, the Prophet forecasting model (PFM) was used to predict both short-term and long-term air pollution in Seoul. The air pollutants forecasted in this study were PM2.5, PM10, O3, NO2, SO2, and CO, air pollutants responsible for numerous health conditions upon long-term exposure. Current chemical models to predict air pollution require complex source lists making them difficult to use. Machine learning models have also been implemented however their requirement of meteorological parameters render the models ineffective as additional models and infrastructure need to be in place to model meteorology. To address this, a model needs to be created that can accurately predict pollution based on time. A dataset containing three years worth of hourly air quality measurements in Seoul was sourced from the Seoul Open Data Plaza. To optimize the model, PFM has the following parameters: model type, changepoints, seasonality, holidays, and error. Cross validation was performed on the 2017–18 data; then, the model predicted 2019 values. To compare the predicted and actual values and determine the accuracy of the model, the statistical indicators: mean squared error (MSE), mean absolute error (MAE), root mean squared error (RMSE), and coverage were used. PFM predicted PM2.5 and PM10 with a MAE value of 12.6 µg/m3 and 19.6 µg/m3, respectively. PFM also predicted SO2 and CO with a MAE value of 0.00124 ppm and 0.207 ppm, respectively. PFM’s prediction of PM2.5 and PM10 had a MAE approximately 2 times and 4 times less, respectively, than comparable models. PFM’s prediction of SO2and CO had a MAE approximately five times and 50 times less, respectively, than comparable models. In most cases, PFM’s ability to accurately forecast the concentration of air pollutants in Seoul up to one year in advance outperformed similar models proposed in literature. This study addresses the limitations of the prior two PFM studies by expanding the modelled air pollutants from three pollutants to six pollutants while increasing the prediction time from 3 days to 1 year. This is also the first research to use PFM in Seoul, Korea. To achieve more accurate results, a larger air pollution dataset needs to be implemented with PFM. In the future, PFM should be used to predict and model air pollution in other regions, especially those without advanced infrastructure to model meteorology alongside air pollution. In Seoul, Seoul’s government can use PFM to accurately predict air pollution concentrations and plan accordingly.

Exploring the relationship between air pollution and meteorological conditions in China under environmental governance

Extensive studies have been carried out on the impact of human activities on air pollution, but systematic investigation on the relationship between air pollutant and meteorological conditions is still insufficient, especially in the context of China’s site scale and recent comprehensive environmental pollution control. Here, we used a spatial interpolation technology to establish a set of data sets of pollutants and meteorological elements that are spatially matched at 896 stations in China to reveal the air pollutant-meteorological interactions between 2014 and 2019. We found that air pollution and meteorological elements have obvious seasonal and regional characteristics. Over the last few years, the concentration of most air pollutants in China has dropped significantly except for O3. The increase in O3 concentration was closely related to the decrease of particulate matter and NO2 concentration. The concentration of most air pollutants was affected by meteorological conditions, but the level of impact depended on the type of pollutants and varied across regions. The concentration of air pollutants at most stations was significantly negatively correlated with wind speed, precipitation and relative humidity, but positively correlated with atmospheric pressure. As the latitude increases, the impact of temperature on the concentration of air pollutants becomes more obvious. To effectively control air pollution, it is further urgent to reveal the relationship between air pollution and meteorological conditions based on long-term daily or real-time data.

Using the MLR and Neuro-Fuzzy Methods to Forecast Air Pollution Datasets

The forecasting of time series data is essential by following statistical and intelligent techniques. Particular matter (PM10) is a time series dataset used to scale the air pollution as a dependent variable while there are many types of pollutants used as independent variables. MLR model has been used as a traditional linear approach to forecasting PM10 data. Combining NF as a nonlinear intelligent method with MLR in a hybrid MLR-NF method has been proposed for improving PM10 forecasts and handling the nonlinearity of datasets. The forecasting results reflected that the hybrid method outperformed the traditional method. Although a multiple linear regression (MLR) model has been used for air quality forecasting depending on several meteorological variables in many recent studies, the MLR model is unable to identify the nonlinear pattern of these types of data. Malaysian datasets of PM10 and several climate pollutants will be studied in this paper. The objective of this study is to forecast PM10 and obtain the best results and minimum forecasting error. In this paper, the dependent variable will be forecasted by using traditional and intelligent methods. MLR has been used as a traditional method to forecast PM10. Neuro-fuzzy (NF) in the adapted copy, which calls the adaptive neuro-fuzzy inference system (ANFIS) is combined with MLR and used as an intelligent method to forecast PM10. The results reflect that MLR-NF outperformed MLR for forecasting PM10 data. In conclusion, MLR-NF can be used to forecast PM10 for more accurate results compared to traditional methods.

Canonical Correlation Analysis Based Hyper Basis Feedforward Neural Network Classification for Urban Sustainability

People give more importance concerning the overall quality of the modernized ecosystem. The pollution of air is one of the significant problems to be resolved as it restricted the ecological transformation of the modernized ecosystem. Therefore, it is fundamental to evaluate the implication of these ecological issues to enhance the urban ecosystem. This vital purpose of this research is to propose a canonical correlation analysis based hyper basis feedforward neural network classification (CCA-HBFNNC) model for evaluating sustainable urban environmental quality. The CCA-HBFNNC model initially acquires a large size of U.S. air pollution dataset as input. Then, a canonical correlative analysis based feature selection algorithm is applied in the CCA-HBFNNC model to select the key pollutant features, which bear fundamental implications to the modernize air pollution to maintain the level of urban sustainability. After the feature selection process, the CCA-HBFNNC model applies the HYPER BASIS FEEDFORWARD NEURAL NETWORK CLASSIFICATION (HBFNNC) algorithm in order to classify input air data based on chosen pollutants features. During the classification process, the HBFNNC algorithm used three critical layers namely hidden, output and input layers for efficiently categorizing each input data as higher or lower pollution level with higher accuracy. If the level of air pollution on the urban environment is higher, finally CCA-HBFNNC model significantly reduces the pollution level. In this way, the CCA-HBFNNC model attains improved urban sustainability levels when compared to sophisticated operation. An experimental evaluation of the CCA-HBFNNC model is determined in terms of CCA-HBFNNC model, time complexity and false-positive rate in consideration of the diversified number of air data retrieved from the big data sets. An investigational result shows that the proposed CCA-HBFNNC model can increases the sustainability level and minimizes the time complexity of urban development when contrasted with contemporary works.

A Data-Driven Framework for Deploying Sensors in Environment Sensing Application

Sensor deployment routines for environment sensing applications make various assumptions about the underlying spatiotemporal field. These assumptions render the deployment ineffective in a practical scenario. This article proposes a two-step process: initially, the sensors are deployed based on geographical covariates. Then, after a fixed period, the data collected from sensors are used to find optimal locations for sensors. The spatiotemporal representation of sensor values has been modeled as the sum of a systematic trend component and a residual process. The trend component is modeled as the sum of deterministic functions, and the residual component is modeled using support vector regression. The locations with maximum support vector count in the residual model are identified as optimal for the deployment of sensors. The method can be used for both static and dynamic deployments. The proposed strategy has been applied to a specific case study of air pollution dataset.

Deposition processes over complex topographies: Experimental data meets atmospheric modeling

The present paper describes the assessment of the atmospheric deposition processes in a basin valley through a multidisciplinary approach based on the data collected within an extensive physico-chemical characterization of the soils, combined with the local meteorology. Surface soil cores were collected on a NNW-SSE transect across the Terni basin (Central Italy), between the Monti Martani and the Monti Sabini chains (956 m a.s.l.), featuring the heavily polluted urban and industrial enclave of Terni on its bottom. Airborne radiotracers, namely 210Pb and 137Cs, have been used to highlight atmospheric deposition.We observed an increased deposition flux of 210Pb and 137Cs at sites located at the highest altitudes, and the associated concentration profiles in soil allowed to evaluate the role of atmospheric deposition. We also obtained a comprehensive dataset of stable anthropogenic pollutants of atmospheric origin that showed heterogeneity along the transect. The behavior has been explained by the local characteristic of the soil, by seeder-feeder processes promoted by the atmospheric circulation, and was reconciled with the concentration profile of radiotracers by factor analysis. Finally, the substantial impact of the local industrial activities on soil profiles and the role of the planetary boundary layer has been discussed and supported by simulations employing a Lagrangian dispersion model.

Comparative Analysis of Machine Learning Regression Algorithms on Air Pollution Dataset

Air pollution has both acute and chronic effects on human health, affecting a number of different systems and organs. Examining and protecting air quality has become one of the most essential activities for the government in many industrial and urban areas today. Air pollutants, such as carbon monoxide (CO), sulfur dioxide (SO(2)), nitrogen oxides (NOx), volatile organic compounds (VOCs), ozone (O(3)), heavy metals, and respirable particulate matter (PM2.5 and PM10), differ in their chemical composition, reaction properties, emission, time of disintegration and ability to diffuse in long or short distances. The main objective of this paper to build a model for predicting Air Quality Index(AQI) of the specific cities using various types of machine learning algorithms namely Multiple Linear Regression, K Nearest Neighbours(KNN), Support Vector Machine(SVM) and Decision Tree. And also evaluate and compare the performance of every algorithm based on their accuracy score and errors. Air Pollution dataset is publicly available on different government sites. The implementation phase dataset is divided as 80% for the training of different models and the rest of the dataset is used for testing the model.

Assessment and prediction of surface ozone in Northwest Indo-Gangetic Plains using ensemble approach

The earth’s surface ozone levels are becoming very significant due to their negative impact on human health, vegetation and climate. In this study, the methodology based on ensemble approach embodied linear and nonlinear behaviors was developed. It was applied for prediction of ozone concentration using dataset (2013–2016) of gaseous pollutants (O3, CO, NOx, MHC, TNMHCs) and meteorological variables as input variables. The daily O3 max/O3 min ratio of 10.9 marks the peculiar ozone pollution in the area. The fourteen prediction algorithms and their possible combinations of ensemble models were employed in this paper. Compared with individual models, the ensemble model approach showed an index of agreement of 0.91, the accuracy of 95.5% and mean absolute error of − 0.001 ppb between the predicted and observed diurnal cycle and daily averaged data of the year 2016 for benchmark analysis.

Improved recurrent generative adversarial networks with regularization techniques and a controllable framework

Generative Adversarial Network (GAN), a deep learning framework to generate synthetic but realistic samples, has produced astonishing results for image synthesis. However, because GAN is routinely used for image datasets, regularization methods for GAN have been developed for convolutional layers. In this study, to expand these methods for time-series data, which are one of the most common data types in various real datasets, modified regularization methods are proposed for Long Short-Term Memory (LSTM)-based GANs. Specifically, the spectral normalization, hinge loss, orthogonal regularization, and the truncation trick are modified and assessed for LSTM-based GANs. Furthermore, a conditional GAN architecture called Controllable GAN (ControlGAN) is applied to LSTM-based GANs to produce the desired samples. The evaluations are conducted with sine wave data, air pollution datasets, and a medical time-series dataset obtained from intensive care units. As a result, ControlGAN with the spectral normalization on gates and cell states consistently outperforms the others, including the conventional model, called Recurrent Conditional GAN (RCGAN).

Comparisons between Logistic Regression and Support Vector Machine for Air Pollution Datasets Forecasting

Particular matter (PM10) studying and forecasting is necessary to control and reduce the damage of environment and human health. There are many pollutants as sources of air pollution (Co, So2, O 3, Nox, No, Wind Speed, and Ambient Temperature) may effect on PM10 variable. PM10 and the pollutant variables have been taken from the meteorological station in Kuala Lumpur, Malaysia. All of these variables classified as nonlinear data. Logistic regression (LR) model can be used for modeling and forecasting these multivariable datasets. LR is one of linear statistical methods, therefore it may reflect inaccurate results when used with nonlinear datasets. To improve the results of forecasting, support vector machine (SVM) method has been suggested in this study. The results in this study reflect outperforming for SVM method comparing to LR. In conclusion, SVM forecasting can be used for more accuracy with nonlinear multivariate datasets when PM10 is as dependent variable. Keywords: Logistic Regression (LR), Support Vector Machine (SVM), Particular Matter (PM10), Forecasting, Air Pollution

Using Logistic Regression with Time-Stratified Method for Air Pollution Datasets Forecasting

Particular matter (PM10) studying and forecasting is necessary to control and reduce the damage of environment and human health. There are many pollutants as sources of air pollution may effect on PM10 variable. Studied datasets have been taken from the Kuala Lumpur meteorological station, Malaysia. Logistic regression (LR) is built by using generalized linear model as a special case of linear statistical methods, therefore it may reflect inaccurate results when used with nonlinear datasets. Time stratified (TS) method in different styles is proposed for satisfying more homogeneity of datasets. It includes ordering similar seasons in different years together to formulate new variable smoother than their original. The results of LR model in this study reflect outperforming for time stratified datasets comparing to full dataset. In conclusion, LR forecasting can be depended after datasets time stratifying to satisfy more accuracy with nonlinear multivariate datasets in which PM10 is to dependent variable.

Labelling Data for Correlation Pollution Dataset by Using Machine Learning

The significant problem in the environment is the pollution of air which affects all the areas in our world. The solution of this problem is excluding by using Internet of Things (IoT) technology which designed system that monitors the parameters that affect the pollution in the environment. The handling and operation of large quantities of information from a wide range of sources is challenges in IoT technology. The purpose of this study is to decrease the dimensions in IoT systems so we suggest two models for IoT system and also aims to build rules for Air quality classification by defining pollution levels. In this study, we suggested machine learning techniques to predict an air pollution levels based on a dataset consisting of daily weather, also we simplified our classification models to solve this problem by using binary classification which is to classify the air pollution level into “High pollution” and “Simple pollution” class labels. The decision tree J48 Algorithm gives the best outcomes in the first model than the second models.

Chemical characteristics and source apportionment of PM2.5 using PMF modelling coupled with 1-hr resolution online air pollutant dataset for Linfen, China

The chemical species in PM2.5 and air pollutant concentration data with 1-hr resolution were monitored synchronously between 15 November 2018 and 20 January 2019 in Linfen, China, which were analysed for multiple temporal patterns, and PM2.5 source apportionment using positive matrix factorisation (PMF) modelling coupled with online chemical species data was conducted to obtain the apportionment results of distinct temporal patterns. The mean concentration of PM2.5 was 124 μg/m3 during the heating period, and NO3− and organic carbon (OC) were the dominant species. The concentrations and percentages of NO3−, SO42−, and OC increased notably during the growth periods of haze events, thereby indicating secondary particle formation. Six factors were identified by the PMF model during the heating period, including vehicular emissions (VE: 26.5%), secondary nitrate (SN: 16.5%), coal combustion and industrial emissions (CC&IE: 25.7%), secondary sulfate and secondary organic carbon (SS&SOC: 24.4%), biomass burning (BB: 1.0%), and crustal dust (CD: 5.9%). The primary sources of PM2.5 on clean days were CD (33.3%), VE (23.1%), and SS&SOC (20.6%), while they were CC&IE (32.9%) and SS&SOC (28.3%) during the haze events. The contributions of secondary sources and CC&IE increased rapidly during the growth periods of haze events, while that of CD increased during the dissipation period. Diurnal variations in the contribution of secondary sources were mainly related to the accumulation and transformation of corresponding gaseous precursors. In comparison, contributions of CC&IE and VE varied as a function of the domestic heating load and peak levels occurred during the morning and evening rush hours. High contributions of major sources (CC&IE and SS&SOC) during haze events originated mainly from the north and south, while high contribution of a major source (CD) on clean days was from the northwest.

Peak Outpatient and Emergency Department Visit Forecasting for Patients With Chronic Respiratory Diseases Using Machine Learning Methods: Retrospective Cohort Study.

BackgroundThe overcrowding of hospital outpatient and emergency departments (OEDs) due to chronic respiratory diseases in certain weather or under certain environmental pollution conditions results in the degradation in quality of medical care, and even limits its availability.ObjectiveTo help OED managers to schedule medical resource allocation during times of excessive health care demands after short-term fluctuations in air pollution and weather, we employed machine learning (ML) methods to predict the peak OED arrivals of patients with chronic respiratory diseases.MethodsIn this paper, we first identified 13,218 visits from patients with chronic respiratory diseases to OEDs in hospitals from January 1, 2016, to December 31, 2017. Then, we divided the data into three datasets: weather-based visits, air quality-based visits, and weather air quality-based visits. Finally, we developed ML methods to predict the peak event (peak demand days) of patients with chronic respiratory diseases (eg, asthma, respiratory infection, and chronic obstructive pulmonary disease) visiting OEDs on the three weather data and environmental pollution datasets in Guangzhou, China.ResultsThe adaptive boosting-based neural networks, tree bag, and random forest achieved the biggest receiver operating characteristic area under the curve, 0.698, 0.714, and 0.809, on the air quality dataset, the weather dataset, and weather air quality dataset, respectively. Overall, random forests reached the best classification prediction performance.ConclusionsThe proposed ML methods may act as a useful tool to adapt medical services in advance by predicting the peak of OED arrivals. Further, the developed ML methods are generic enough to cope with similar medical scenarios, provided that the data is available.

The characterization of fine particulate matter downwind of Houston: Using integrated factor analysis to identify anthropogenic and natural sources

The interpretation of large air pollution datasets involves a great deal of complexity. To gain a better understanding of the complicated relationships and patterns within datasets, we perform factor analysis. Between December 2015 and December 2017, fine particulate matter (PM2.5) samples were collected at a suburban site northeast of the Houston metropolitan area, TX. A total of 233 filter samples were analyzed for chemical composition. The average of all PM2.5 samples consisted of 38.1% inorganic ions, 28.9% elements, 29.1% organic carbon, and 3.7% elemental carbon and other organic materials. Principal component analysis and positive matrix factorization were utilized to identify eight factors: regional aerosols, biomass burning, gasoline combustion, industry, crustal material, incineration, marine dust, and fireworks. The first three contributed more than 70% of the total PM2.5 mass. The receptor models also captured the impact of fireworks and classified it as a source of PM2.5 over Houston. To identify the origins of air masses transporting PM2.5 to the site, we applied the NOAA hybrid single-particle Lagrangian integrated trajectory model and performed a cluster analysis of back trajectories and determined six cluster source regions: the Gulf of Mexico, the Southeast, two midwestern clusters, the Pacific Northwest, and the Southwest. The results of our analysis show that during the summer months, marine and crustal sources were often associated with an onshore flow from the Gulf of Mexico and that four clusters covering 38% of the West Liberty area were strongly influenced by trajectories originating from biomass burning. The results of this study represented a variety of sources that affect the PM2.5 over the Houston metropolitan area. The quantified contributions of these sources could provide policymakers with useful information for developing more efficient control systems and making more effective decisions to cope with the harmful effects of ambient air pollution.

Performance Analysis of KMeans and KMediods Algorithms in Air Pollution Prediction

Air pollution is a major part of human health problems in many cities. Air pollution can cause many negative effects on the environment. The most basic solution for air pollution is humane should have responsible habits and also have to use more efficient devices to predict and control the atmosphere pollution. The nearly everyone important objective of this effort is to analyze and predict the atmosphere smog using data mining techniques. And help to take inevitable carriage steps or decision for protect the future generation from the rapid increase of air pollution. To turn raw data into useful information, Data mining technique is used by many companies. Data mining extract the hidden useful information from the air pollution dataset. It helps and supporting human making decision human, being as responsible, and control/ avoids the air pollution as claimed by the severe level of pollution in the airspace city based. To analyze and predict simple and efficient clustering techniques such as K-Means and K-Medoids has been used.

Constructing a PM2.5 concentration prediction model by combining auto-encoder with Bi-LSTM neural networks

Air pollution problems have a severe effect on the natural environment and public health. The application of machine learning to air pollutant data can result in a better understanding of environmental quality. Of these methods, the deep learning method has proven to be a very efficient and accurate method to forecast complex air quality data. This paper proposes a deep learning model based on an auto-encoder and bidirectional long short-term memory (Bi-LSTM) to forecast PM2.5 concentrations to reveal the correlation between PM2.5 and multiple climate variables. The model comprises several aspects, including data preprocessing, auto-encoder layer, and Bi-LSTM layer. The performance of the proposed model was verified based on a real-world air pollution dataset, and the results indicated this model can improve the prediction accuracy in an experimental scenario.

Re-shaping the “original SIN”: a need to re-think sediment management and policy by introducing the “buffer zone” concept

The effects of long-standing industrialization processes and poor environmental management practices have often left a harmful legacy for marine-coastal sites worldwide, causing a wide range of unforeseen impacts on the ecosystem and on human health. A critical revision of available data from three highly contaminated Italian sites (Augusta Bay, Sicily; Cagliari Gulf, Sardinia; Pozzuoli Bay, Campania) revealed the crucial role of biogeochemical/physical dynamics and potential widespread delivery of contaminants as key components for a wider comprehensive sediment management. Datasets of organic pollutants (ΣPAHs, ΣPCBs), heavy metals (As, Cd, Co, Cr, Cu, Hg, Ni, Pb, V, and Zn), radionuclides, and grain size composition of marine sediments of all three sites were critically reviewed and seafloor bathymetric data were analyzed. A critical review of available physical and chemical information of sediments collected from the three marine sites provides a better understanding of the physical and biogeochemical dynamics of large-scale dispersion of contaminants in those areas, as well as new insights in terms of improved sediments management policies with actions of larger scale monitoring of the “buffer zone” confining with the national relevance and highly contaminated site to trace inputs and accumulation effects of pollutants deriving from the point-source. This study demonstrates how the presence of polluted sediments on the coastal areas can have, also after the closure of anthropogenic activities, a potential impact at a large scale. Then to better evaluate the effective impact, we suggest the adoption of a buffer zone contiguous to the national relevance sites (SIN) area. While SINs are optimized to define the state of the art of the near-source pollution, buffer zones may help to outline the rate of delivery of pollutants to the deep sea. Buffer zones must be designed by taking into account the local hydrodynamics, sedimentology, and geomorphology, and should be monitored, even if at a lower resolution, like SINs.

Measurement of mercury, other trace elements and major ions in wet deposition at Jomsom: The semi-arid mountain valley of the Central Himalaya

Abstract South Asian pollutants can be transported and deposited via wet/dry deposition to the remote areas of the Himalayas and could pose a serious threat to the mountain ecosystems. Therefore, in order to understand the concentrations, fluxes, seasonal variation and origin of the mercury (Hg), major ions and trace elements, precipitation samples were collected during 2012–2013 from a data gap region, Jomsom, the high elevation semi-arid mountain valley in the central Himalayas. The volume-weighted mean (VWM) concentrations of ions followed the order of Ca2+ > Mg2+ > Na+ > NH4+ > SO42− > Cl− > NO3− > K+. The concentration of Cd was lowest (0.07 μg L−1) whereas that of Fe was the highest (1073.59 μg L−1) in the precipitation samples. Wet deposition level of all the measured inorganic species was comparable to urban Lhasa but higher than those in remote alpine sites of the Tibetan Plateau (TP). This study shows that Hg and other inorganic constituents were higher in the non-monsoon season compared to monsoon due to enhanced washout of aerosols. Enrichment factor (EF), sea salt fraction, crustal and anthropogenic fractions, principal component analysis (PCA) and correlation coefficient analysis suggested that crustal dust and anthropogenic activities as the major sources of measured chemical species whereas the influence of sea-salt was minimal. In addition, local anthropogenic emissions were low suggesting that the majority of the pollutants could have been transported from the South Asian region to the high elevation mountains. Meanwhile, low precipitation and dry environment could have enhanced the concentrations of inorganic species in the arid region than other sites over the central Himalayas. This work adds new dataset of inorganic pollutants in wet precipitation and provides baseline information for an arid region environmental protection. However, there is a need for further long-term monitoring to understand the precipitation chemistry of the arid regions.

Spatio-temporal variability of water pollution by chlordecone at the watershed scale: what insights for the management of polluted territories?

Chlordecone, applied on soils until 1993 to control banana weevil, has polluted water resources in the French West Indies for more than 40years. At the watershed scale, chlordecone applications were not homogenous, generating a spatial heterogeneity of the pollution. The roles of climate, hydrology, soil, agronomy, and geology on watershed functioning generate a temporal heterogeneity of the pollution. This study questions the interactions between practices and the environment that induce such variability. We analyzed hydrological and water pollution datasets from a 2-year monitoring program on the Galion watershed in Martinique (French West Indies). We conjointly analyzed (i) weekly chlordecone (CLD) concentration monitored on 3 river sampling sites, (ii) aquifer piezometric dynamics and pollutions, and (iii) agricultural practices on polluted soils. Our results showed that chlordecone pollution in surface waters are characterized by annual trends and infra-annual variations. Aquifers showed CLD concentration 10 times higher than surface water, with CLD concentration peaks during recharge events. We showed strong interactions between rainfall events and practices on CLD pollution requiring a systemic management approach, in particular during post-cyclonic periods. Small sub-watershed with high CLD pollution appeared to be a substantial contributor to CLD mass transfers to the marine environment via rivers and should therefore receive priority management. We suggest increasing stable organic matter return to soil as well as external input of organic matter to reduce CLD transfers to water. We identified hydrological conditions-notably drying periods-and tillage as the most influential factors on CLD leaching. In particular, tillage acts on 3 processes that increases CLD leaching: organic matter degradation, modification of water paths in soil, and allophane clay degradation.