Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Ajmal Roshan has a total of 32 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Objective Sub-Domain Selection Substantially Improves Sensitivity of Spatial Distribution Models of Heterogeneous Groundwater Arsenic Hazard in Bihar, India

    by Mohammad Mehrabi, Yang Han, Laura A. Richards, Ajmal Roshan, Maiko Sakamoto, David A. Polya

    Spatial heterogeneity of data is often a crucial obstacle to the optimal implementation of geospatial machine learning models, especially in larger study areas. In this research, a novel framework is proposed to systematically partition a heterogeneous area (Bihar, India) into more homogeneous sub-domains (SDs) in order to more accurately predict the spatial distribution of groundwater arsenic hazard. Fed by expert-selected key predictive variables (PVs), the algorithm performs local feature importance analysis to cluster the data points accordingly, followed by an automatic demarcation that results in generating SDs that are statistically significantly separated and more homogeneous than the full domain. Accuracy assessment of the used random forest model showed that while Sub-domained and Full-domain Models could both exhibit a high accuracy (e.g., respective AUCs (area under the curve) of 0.95 and 0.93), a Sub-domained Model achieved about 30% higher Sensitivity (0.85 vs. 0.56) while maintaining a high Specificity (0.93). Shapley Additive Explanations (SHAP) was also applied to interpret the predictive model and to explore modelled As-PV relationships. The consistency of the model comparison results was also demonstrated by internal and external cross-validations. The findings confirm that building geospatial models upon heavily heterogeneous data can lead to hidden loss of accuracy metrics, such as Sensitivity, the underestimation of which may be of more practical importance than Specificity. The proposed sub-domaining framework was demonstrated here to be an effective partial solution for this specific application. It is noted, however, that improvements in Sensitivity may be dependent upon several project-specific parameters, including the set of PVs selected and the degree of heterogeneity of the study area with respect to the target variable and processes controlling the distribution of its value.

  • Open Access

    Participatory Science Approaches for Water Quality Monitoring in Bihar, India Through International School‐College‐University Partnerships

    by Laura A. Richards, Samuel Addison, Ajmal Roshan, Rama Chakraborty, Biswajit Chakravorty, Arun Kumar, Ashok Ghosh, David A. Polya

    Provision of safely managed drinking water is a key pillar in achieving public health. Particularly in regions where water quality data may be limited, participatory science approaches offer the potential to bridge data gaps with allied benefits such as community engagement and knowledge exchange. Here, we explore the suitability of participatory science approaches for monitoring water quality in the Indian state of Bihar. A simplified methodology for water sample collection was adapted, with local input, for participatory science approaches. Volunteer participants, ranging from secondary school students to professionals, collected ∼600 water samples, following training. Samples were analyzed for arsenic (As), uranium (U), iron (Fe), manganese (Mn), fluoride (F − ), and nitrate (NO 3 − ). Comparisons were made between samples collected via this participatory science approach and samples obtained via conventional approaches in a geographically similar sampling frame. The comparisons revealed that the exceedance of contaminants against reference guideline values were broadly similar (<10%–15%) between approaches. However, more nuanced differences were observed when comparing cumulative distribution curves of concentration or geochemical associations. There was no statistically significant difference for As distributions arising from participatory science versus conventional sampling, but there were statistically significant differences for other parameters. Differences were plausibly because of the spatial heterogeneity of groundwater chemistry, sample handling/preservation and/or sites. Optimized selection of participating institutes' locations could improve the spatial representativeness of future participatory science initiatives. Results show that integrating participatory science approaches for water quality monitoring offers benefits of knowledge exchange and higher resolution data sets, particularly in data‐scarce regions.


  • Open Access

    Does Fire Increase Soil Selenium Mobility? Evidence from Thermal Alterations of Seleniferous Soils

    by Ajmal Roshan, Ajit Kumar, Prachi Joshi, Olivier Proux, Harald Neidhardt, Ashis Biswas

  • Open Access

    Microbial Contaminants in Groundwater Arsenic Mitigated and Piped Community Drinking Water Supplies in Bihar, India: Implications on Water Security

    by Ajmal Roshan, Gillian E. Clayton, David A. Polya, Arun Kumar, Ashok Ghosh, Anne-Marie Glenny, Majid Sedighi, Suzie M. Reichman, Meenakshi Arora, Laura Richards

  • Open Access

    From ground to home: Evidence, extent and potential controls on groundwater sourced household drinking water contamination in Bihar, India

    by Ajmal Roshan, David A. Polya, Meenakshi Arora, Arun Kumar, Ashok Ghosh, Anne-Marie Glenny, Majid Sedighi, Suzie M. Reichman, Laura A. Richards

    Groundwater, a major source of available freshwater 1, is central to meeting the drinking water needs globally 2, and in the eastern Indian state of Bihar 3. Groundwater in parts of Bihar however, has elevated levels of geogenic inorganic contaminants 4, with more recent studies also indicating microbial contamination 5 and presence of antimicrobial resistance (AMR) genes 6 in shallow and deeper depth groundwater. However, studies investigating contamination of groundwater-derived household drinking water at the point of consumption (PoC), as compared to the point of source (PoS), are limited in Bihar. Paired water samples (n = 39) were collected from groundwater wells at the PoS, and the corresponding household drinking water containers at the PoC; and tested for arsenic (As), Escherichia coli (E. coli) and extended spectrum beta-lactamase (ESBL) producing E. coli (ESBL-Ec) - as an exemplar of inorganic chemical contaminant, microbial contaminant, and AMR indicator organism, respectively. Results indicate samples exceeding the 10 µg/L WHO provisional guideline value for As to be same (at 5 %) between PoS and PoC. However, presence of E. coli increased from 46 % to 59 % (McNemar’s p = 0.27) while that of ESBL-Ec rose from 10 % to 46 % (McNemar’s p < 0.01) between PoS and PoC samples. 74 % of the households reported collection and storage of water prior to consumption, of which 72 % reported the storage vessels to be covered, and another 69 % reporting cleaning of the container’s multiple times a day. However, more than half of the sampled households owned livestock within premises, had kids under the age of five, and just 13 % reported any kind of treatment (including boiling) being done to the groundwater prior to drinking. No improvement in water quality was observed from PoS to PoC for E. coli. On the contrary, deterioration in water quality from PoS and PoC was indicated based on ESBL-Ec (Wilcoxon signed rank test; p < 0.05). As expected of inorganic contaminants, no significant shift (at the 0.05 level) in overall concentration was observed for As between PoS and PoC. The results call for sustainable management of groundwater sources, and improvements in delivery and treatment of contaminated groundwater prior to consumption so that potable drinking water is ensured at the end-user stage.References: [1] Weblink: https://portals.iucn.org/library/sites/library/files/documents/2016-039.pdf; [2] DOI: 10.1016/C2018-0-03156-4; [3] Weblink: https://nhm.gov.in/uhc-day/Session%202/NFHS-5%20State%20Factsheet%20Compendium_Phase-I%20%281%29.pdf; [4] DOI: 10.3390/ijerph17072500; [5] Roshan et. al., AGU 2025; [6] 10.1016/j.envpol.2024.124205. This work was supported by MR/Y016327/1 (to LAR et. al.,) and Cookson Scholarship (to AR).


  • Open Access

    Portable XRF (PXRF) Analysis of Agricultural Soils in Central Bangladesh - Approaches to Quantify Anthropogenic Trace Element Inputs

    by Farah T. Ahmed, Ajmal Roshan, Rory Cracroft, Laura A. Richards, David A. Polya

    The widespread distribution, non-degradable nature, and detrimental effects of many trace elements (TEs) render a significant ecological and public health issue on a global scale (Zhao et al., 2014). TEs in soil may have geogenic or anthropogenic origins – quantifying the latter may be of considerable utility in (a) identifying sources; and (b) rapid qualitative risk assessments in the absence of detailed mineralogical and petrographic data. Bangladesh, one of the world’s most densely and highly populated countries, though agriculturally intensive, has numerous sources of soil TEs pollution, particularly in the highly urbanised and industrialised Dhaka-Gazipur-Narayanganj corridor (Islam et al., 2024).In this study, PXRF was used to semi-quantitatively analyse a range of agricultural soils from central Bangladesh, with a view to quantifying and determining the spatial distribution of TEs of anthropogenic origin. Soil samples (n = 102) were collected from agricultural lands in seven districts of central Bangladesh across 8 different river plains (Meghna, Dhaleshwari, Old Brahmaputra, Buriganga, Ichamoti, Kaliganga, Shitalakshya, and Turag River). Soil samples were analysed by PXRF for trace element content, and the method was validated by using Certified Reference Materials (CRMs: NCS DC73309, NIST 2710A and NIST 1646A). The recovery rates for trace elements were within acceptable limits (< 100 ± 20%). Recovery rates for Si and Al were poor, and the estimated SiO2 and Al2O3 were assessed from measured Rb using the relationships identified for Meghna River sediment by Hossain (2019). The concentration range of trace elements (in mg/kg) in agricultural soil were found as 10 – 65 (Pb), 5 – 20 (As), 30 – 300 (Zn), 15 – 70 (Cu), 15 – 75 (Ni), 8 – 25 (Cd), 40 – 150 (Cr), 40 – 200 (V), 10 – 40 (Sb), 8 – 35 (Sn), 21000 – 60000 (Fe), 150 – 1000 (Mn), 2100 – 5700 (Ti), 1600 – 73000 (Ca), 8300 – 27000 (K), 2700 – 17000 (Mg), 125 – 520 (Zr), 30 – 200 (Sr), 70 – 180 (Rb), 10 – 25 (Nb) and 10 – 30 (Th). FAO guideline values for soils (FAO, 2004) were found to be exceeded, notably for Sn (58 % of samples), Cr (47 %), V (32%), Sb (27%), Cd (20%), Ni (15%) and Pb (2%). Preliminary qualitative determination of anthropogenic contributions to the soils was made following corrections for (a) grain size, and (b) river-valley-dependent provenance of geogenic sources and notably identified some soils for which Cr and/or Pb in particular were predominantly of anthropogenic origin. Further analysis will be carried out to enable more detailed TEs source apportionment. Keywords: Agriculture, Soil quality, Trace elements contamination, Portable XRF.Acknowledgement: FTA acknowledges a STFT, Ministry of Science and Technology (Bangladesh) scholarship. AR acknowledges a Cookson Scholarship. RC acknowledges a PGRTA Studentship from the University of Manchester (UoM). We thank all the technical support staff in the Manchester Analytical Geochemistry Unit at the UoM for the PXRF and laboratory facilities. References1. Zhao et al. (2014). https://doi.org/10.1016/j.scitotenv.2013.09.086 2. Islam et al. (2024). https://doi.org/10.1016/j.envres.2024.118551 3. Hossain (2019). https://doi.org/10.1002/gj.3595 4. FAO (2004). https://faolex.fao.org/docs/pdf/est97999E.pdf

  • Open Access

    Performance of field-installed groundwater arsenic remediation systems in Bihar, India: Role of source water chemistry, technology type and other factors

    by Ajmal Roshan, David A. Polya, Arun Kumar, Ashok Ghosh, Anne-Marie Glenny, Majid Sedighi, Suzie M. Reichman, Meenakshi Arora, Laura A. Richards


  • Open Access

    Co-occurrence of antibiotics, antimicrobial resistance genes and wastewater indicators in surface waters near Bangkok, Thailand: Characterization, Distribution & Controls

    by Laura Richards, George J. L. Wilson, Ajmal Roshan, Farah T. Ahmed, Mariel Perez-Zabaleta, Santiago Nicolás Otaiza-González, Sara Rodríguez-Mozaz, Zeynep Cetecioglu, David A. Polya

    Aquatic pollution from emerging contaminants, including antibiotics and antimicrobial resistance (AMR) genes, is an important environmental concern particularly pertinent in megacities such as Bangkok, Thailand, impacted by rapid urbanization and massive water demand. Using a suite of environmental and hydrogeochemical tracers including inorganics and organics, nutrients, metal(loids), select antibiotics and AMR genes [1, 2], we characterize the distribution and spatial patterns of a range of contaminants in a ~ 150 km transect of the Chao Phraya River Basin in Thailand capturing areas both upstream and downstream of Bangkok. A range of antibiotics and AMR genes were identified in parts of the transect and downstream trends are investigated. Co-occurrence between selected antibiotics and AMR genes was not statistically significant, although other significant hydrogeochemical relationships (e.g. between pH and selected AMR genes) were observed, suggesting complex controls and selection pressures. Comparisons are made with the types and concentrations of similar compounds detected in other major river and groundwater systems near other rapidly developing cities in South Asia (e.g. Patna, India) [3-5]. This work highlights the added interpretive value of a comprehensive range of analytes and provides insight on the potential co-occurrence of antibiotics, antimicrobial resistance genes and wastewater indicators that may be observed in surface waters in such settings.Acknowledgements: We acknowledge support from a UKRI ODA allocation (via UoM to DP et al), a UoM-KTH strategic partnership seedcorn award (to LAR & ZC), a UKRI Future Leaders Fellowship (MR/Y016327/1 to LAR et al), a DKOF (to LAR), Cookson Studentship (to AR), and the Resistomap team.References: [1] Larsson & Flach, Nature Reviews Microbiology, 2022, https://doi.org/10.1038/s41579-021-00649-x; [2] Hutinel et al., Science of the Total Environment, 2022, https://doi.org/10.1016/j.scitotenv.2021.151433; [3] Wilson et al., Environmental Pollution, 2024, https://doi.org/10.1016/j.envpol.2024.124205; [4] Richards et al., Environmental Pollution, 2023, https://doi.org/10.1016/j.envpol.2023.121626; [5] Richards et al., Environmental Pollution, 2021, https://doi.org/10.1016/j.envpol.2020.115765.

  • Open Access

    Comparison of the distribution of groundwater remediation units and contaminant (arsenic, iron, fluoride) distribution in Bihar, India for improved water security and management

    by Ajmal Roshan, David A. Polya, Arun Kumar, Ashok Ghosh, Anne-Marie Glenny, Majid Sedighi, Suzie M. Reichman, Meenakshi Arora, Laura A. Richards


  • Open Access

    Artificial Intelligence Modelling to Support the Groundwater Chemistry-Dependent Selection of Groundwater Arsenic Remediation Approaches in Bangladesh

    by Ruohan Wu, Laura A. Richards, Ajmal Roshan, David A. Polya

    Groundwater arsenic (As) still poses a massive public health threat, especially in South Asia, including Bangladesh. The arsenic removal efficiency of various technologies may be strongly dependent on groundwater composition. Previously, others have reported that the molar ratio [Fe]−1.8[P][As], in particular, can usefully predict the potential efficiency of groundwater As removal by widespread sorption/co-precipitation-based remediation systems. Here, we innovatively extended the application of artificial intelligence (AI) machine learning models to predict the geospatial distribution of [Fe]−1.8[P][As] in Bangladesh groundwaters utilizing our analogous AI predictions for groundwater As, Fe, and P. A comparison between the predicted geospatial distribution of groundwater As and [Fe]−1.8[P][As] distinguished high groundwater As areas where (a) sorption/co-precipitation remediation technologies would have the potential to be highly effective in removing As without Fe amendment, as well as from those areas where (b) amendment with Fe (e.g., zero-valent Fe) would be required to promote efficient As removal. The 1 km2 scale of the prediction maps provided a 100-fold improvement in the granularity of previous district-scale non-AI models. AI approaches have the potential to contribute to informing the appropriate selection and amendment of appropriate groundwater contamination remediation strategies where their effectiveness depends on local groundwater chemistry.