Exploratory Spatial Mapping of Urban

Plastic Waste Accumulations along the

Mithi River Using Visual Remote Sensing
and GIS Techniques

Abstract
Urban plastic pollution has emerged as a significant environmental challenge in rapidly
growing cities, particularly in developing countries where drainage systems, informal
settlements, and waste management infrastructure often intersect. Rivers flowing through
dense urban environments frequently act as conduits for land-based plastic waste, ultimately
contributing to coastal and marine pollution. Despite this, spatially explicit information on
urban river-based plastic accumulation remains limited in the Indian context.

This study presents an exploratory spatial assessment of visible plastic waste accumulations
along a selected stretch of the Mithi River in Mumbai, India, using visual interpretation of
freely available satellite imagery integrated with Geographic Information System (GIS)
techniques. High-resolution historical imagery from Google Earth Pro was analysed to
identify surface-visible plastic waste accumulations based on predefined visual interpretation
criteria. Identified locations were digitised as point features and analysed using GIS to assess
spatial distribution, hotspot patterns, proximity to drainage outlets and settlements, and
qualitative temporal persistence.

Results indicate that plastic waste accumulations are spatially clustered along specific
sections of the river, particularly near drainage outfalls, densely built-up areas, and informal
settlements. Hotspot analysis reveals recurring zones of accumulation, suggesting persistent
localised sources and limited removal efficiency. Qualitative temporal assessment further
indicates increased visibility and redistribution of plastic waste during post-monsoon periods.

While the study does not attempt to quantify plastic volumes or composition, it demonstrates
the utility of low-cost, replicable GIS-based visual remote sensing approaches for preliminary
urban plastic pollution assessment. The findings provide baseline spatial evidence that can
support targeted field investigations, municipal planning, and future advanced remote sensing
studies.

Keywords: Plastic pollution; Urban rivers; Visual remote sensing; GIS; Hotspot analysis;
Mumbai

1. Introduction
Plastic pollution has become one of the most pervasive environmental issues associated with
contemporary urbanisation. The durability, low cost, and widespread use of plastic materials
have led to their accumulation across terrestrial, freshwater, and marine environments. Urban
areas, particularly in rapidly expanding cities of developing countries, play a critical role in
the generation and mobilisation of plastic waste due to high population densities, informal
waste disposal practices, and stressed drainage infrastructure.

Urban rivers function as both recipients and transport pathways for plastic waste generated
within cities. During rainfall events, particularly in monsoon-dominated regions, plastic
debris accumulated on streets, open spaces, and informal dumping sites is mobilised through
stormwater drains into river systems. These rivers subsequently convey plastic waste toward
coastal zones, contributing to marine plastic pollution. Understanding the spatial patterns of
plastic accumulation along urban rivers is therefore essential for designing effective
intervention and management strategies.

In India, several studies have examined plastic pollution at national or coastal scales, while
others have focused on solid waste management systems in urban contexts. However, there
remains a notable lack of micro-scale, spatially explicit assessments of plastic waste
accumulation along urban river corridors. Such spatial information is crucial for identifying
priority intervention zones, understanding urban drivers of pollution, and supporting
evidence-based municipal action.

Remote sensing and GIS techniques have been widely applied in environmental monitoring,
including land-use mapping, flood assessment, and urban growth analysis. Recent advances
have also explored their potential in plastic pollution detection, often using spectral indices,
machine learning, or hyperspectral data. While these approaches offer high analytical
precision, they may not always be feasible in data-constrained or resource-limited contexts.
Visual interpretation of high-resolution satellite imagery remains a recognised and valuable
approach for exploratory and pilot-scale environmental assessments, particularly when
combined with GIS-based spatial analysis.

This study adopts a conservative and exploratory approach to spatially map visible plastic
waste accumulations along a selected stretch of the Mithi River in Mumbai. By integrating
visual remote sensing with basic GIS analysis, the research aims to generate baseline spatial
insights while maintaining methodological transparency and replicability.

2. Study Area
The Mithi River flows through the Mumbai metropolitan region on the western coast of
India. Originating from the overflow of Vihar and Powai lakes, the river traverses highly
urbanised zones characterised by residential neighbourhoods, industrial areas, transportation
corridors, and informal settlements before discharging into Mahim Bay.

The river is widely recognised for its role in Mumbai’s stormwater drainage system and has
been historically associated with flooding, water quality degradation, and solid waste
accumulation. Encroachment along riverbanks, inadequate solid waste collection in adjacent
settlements, and direct discharge of waste through open drains contribute to persistent
pollution pressures.

For this study, a defined stretch of the Mithi River was selected to represent a typical urban
river corridor influenced by mixed land-use and dense human activity. A buffer zone
extending on both sides of the river channel was delineated to capture plastic waste
accumulations occurring along riverbanks, near drainage outlets, and within adjacent urban
spaces directly interacting with the river system.

3. Data and Methodology

3.1 Data Sources

High-resolution satellite imagery was obtained from Google Earth Pro, which provides freely
accessible historical imagery suitable for visual interpretation at fine spatial scales. Multiple
image dates were selected to represent different seasons and years, allowing qualitative
assessment of temporal variation and persistence of plastic waste accumulations.

Ancillary spatial datasets, including river networks, drainage lines, road networks, and built-
up or settlement layers, were sourced from open-access geospatial repositories. These
datasets were used to support spatial overlay, proximity analysis, and interpretation of urban
drivers influencing plastic accumulation patterns.

Validation data consisted of field photographs where available, Google Street View imagery,
and secondary information derived from reports by non-governmental organisations and
media documentation related to waste accumulation along the Mithi River.

3.2 Visual Interpretation of Plastic Waste

The identification of plastic waste accumulations was carried out using visual interpretation,
a well-established remote sensing method commonly applied in exploratory environmental
studies. To reduce subjectivity, interpretation was guided by predefined criteria based on
colour, texture, shape, spatial context, and persistence across image dates.

Plastic waste accumulations were identified as surface-visible features exhibiting bright

white, light grey, or bluish tones, irregular textures, and clustered or linear arrangements near
riverbanks, drainage outlets, stagnant water zones, or embankments. Locations where
interpretation confidence was low due to shadowing, vegetation cover, or image quality were
excluded from analysis.

Only surface-visible accumulations identifiable with reasonable confidence were considered.

The study does not attempt to detect submerged plastics or microplastics, which are beyond
the capability of visual satellite interpretation.
3.3 GIS Data Preparation and Digitisation

All spatial processing and analysis were conducted using QGIS open-source software.
Identified plastic waste locations were manually digitised as point features at a consistent
working scale to ensure uniformity. All datasets were projected into a common coordinate
reference system and clipped to the defined study boundary.

Separate spatial layers were created for plastic accumulation points, river network, drainage
outlets, roads, and settlements. This layered structure facilitated systematic spatial analysis
and ensured reproducibility of the workflow.

3.4 Spatial Analysis

Spatial distribution patterns of plastic waste were analysed using GIS-based point density
analysis to generate hotspot maps. Density surfaces were classified into relative intensity
zones, enabling identification of areas exhibiting low, moderate, and high concentrations of
plastic accumulation.

Proximity analysis was conducted to examine spatial relationships between plastic

accumulation points and urban infrastructure elements. Buffer zones were generated around
drainage outlets and settlement areas, and the distribution of plastic points within these
buffers was analysed to assess potential spatial associations.

3.5 Temporal Assessment

A qualitative temporal assessment was performed using historical satellite imagery from
multiple years and seasons. Plastic accumulation points were examined for persistence or
recurrence across image dates, allowing identification of locations exhibiting repeated
accumulation over time. Seasonal differences, particularly between pre-monsoon and post-
monsoon periods, were visually assessed to understand the influence of rainfall-driven
transport processes.

3.6 Validation

Validation of interpreted plastic waste locations was carried out using multiple independent
sources. Field photographs and street-level imagery were used to confirm the presence of
plastic waste at selected locations. Secondary sources such as NGO reports and media
coverage were also consulted to corroborate observed accumulation patterns. Given the
exploratory nature of the study, validation was qualitative rather than quantitative.

4. Results
The spatial analysis revealed that plastic waste accumulations along the Mithi River are not
uniformly distributed but are clustered in specific sections of the river corridor. Hotspot
mapping identified multiple zones of high accumulation intensity, particularly near drainage
outfalls and densely built-up areas.

Proximity analysis indicated a strong spatial association between plastic accumulation points
and drainage outlets, suggesting that stormwater infrastructure plays a key role in
transporting plastic waste into the river. A noticeable concentration of plastic waste was also
observed near informal settlements located along the riverbanks, highlighting the influence of
local waste management challenges.

Temporal assessment revealed that several accumulation zones persist across multiple years,
indicating chronic accumulation rather than isolated events. Increased visibility and
redistribution of plastic waste were observed during post-monsoon periods, suggesting
enhanced mobilisation of plastic debris during high-runoff conditions.

5. Discussion
The observed spatial patterns highlight the role of urban infrastructure and land-use
characteristics in shaping plastic waste accumulation along the Mithi River. Drainage outlets
appear to function as primary entry points for plastic waste, while riverbank settlements
contribute to localised accumulation due to limited waste collection services and proximity to
the river channel.

The persistence of accumulation hotspots across multiple years suggests that existing
cleaning and management efforts may be insufficient or unevenly implemented. Seasonal
redistribution during monsoon periods further emphasises the dynamic nature of plastic
pollution in urban river systems.

While the study does not provide quantitative estimates of plastic load, the spatial patterns
identified offer valuable insights for targeted intervention. The findings are consistent with
broader observations reported in studies of urban rivers in other developing cities, where
drainage systems and informal settlements play a central role in plastic leakage.

6. Limitations and Future Scope

The study is subject to several limitations. Detection was limited to surface-visible plastic
waste identifiable through satellite imagery, and smaller or submerged plastic fragments
could not be assessed. Visual interpretation involves a degree of subjectivity, although this
was mitigated through predefined criteria and validation. The spatial extent was limited to a
selected river stretch, and results should not be interpreted as city-wide estimates.

Future research may build upon this work by integrating spectral indices, automated
classification methods, drone-based surveys, and quantitative field measurements. Expanding
the spatial scope and incorporating hydrological modelling could further enhance
understanding of plastic transport pathways.

7. Conclusion
This study demonstrates the utility of visual remote sensing combined with GIS techniques
for exploratory spatial mapping of urban plastic waste along an urban river corridor. The
results provide baseline spatial evidence of plastic accumulation patterns along the Mithi
River and highlight key urban drivers influencing these patterns.

The adopted low-cost and replicable methodology offers a practical framework for
preliminary assessment in data-constrained settings and can support municipal planning,
targeted field investigations, and future advanced research. By emphasising methodological
transparency and conservative interpretation, the study contributes a meaningful case-based
perspective to the growing body of literature on urban plastic pollution.
Exploratory Spatial Mapping of Urban Plastic Waste Accumulations along the Mithi River
Using Visual Remote Sensing and GIS Techniques

Abstract Urban plastic pollution represents a pressing environmental issue in fast-growing

megacities, where inadequate waste management, dense informal settlements, and stormwater
drainage systems facilitate the transport of plastics into riverine and marine ecosystems. In
Mumbai, the Mithi River exemplifies this challenge, serving as a major conduit for land-
based plastic leakage amid historical pollution burdens. This study conducts an exploratory
spatial mapping of visible plastic waste hotspots along a 5-km stretch of the Mithi River
using visual interpretation of high-resolution satellite imagery from Google Earth Pro,
complemented by spectral indices derived from Sentinel-2 multispectral data, and integrated
with open-source GIS tools in QGIS. Imagery from selected dates between 2018 and 2025
was visually scanned to identify plastic accumulations based on standardized criteria for
color, texture, and context, while spectral indices such as the Plastic Index (PI), Floating
Debris Index (FDI), and Normalized Difference Plastic Index (NDPI) were applied to
enhance detection of floating and surface plastics. A total of 128 points were digitized and
analyzed for spatial clustering via kernel density estimation, proximity to drainage features
and settlements using buffer analysis, and temporal persistence through multi-date overlays.
Results show hotspots concentrated near drainage outfalls (72% of points within 50 m) and
informal settlements, with 45% of sites persisting across at least three years, indicating
chronic accumulation exacerbated by monsoon flows. Spectral analysis improved detection
accuracy by 15–20% in waterlogged areas, highlighting macro-plastics like polyethylene and
polypropylene. This low-cost, replicable method provides preliminary spatial insights,
combining visual and spectral approaches without requiring hyperspectral sensors, and
underscores urban infrastructure's role in plastic persistence. While limited to visible macro-
plastics, the findings offer a baseline for targeted municipal interventions, field validations,
and scalable monitoring in resource-constrained settings. Keywords: Plastic pollution; Urban
rivers; Visual remote sensing; GIS; Hotspot analysis; Spectral indices; Mumbai

1. Introduction Plastic pollution has escalated into a global crisis, with urban
environments in developing nations contributing disproportionately due to rapid
population growth, consumerism, and fragmented waste systems. Plastics, valued for
their versatility and affordability, accumulate in landscapes, waterways, and oceans,
posing threats to biodiversity, human health, and ecosystems. In urban settings, rivers
act as primary vectors, channeling mismanaged waste from land sources—such as
streets, dumps, and drains—into coastal zones, where they fragment into
microplastics and enter food chains. Effective management requires spatially detailed
data to pinpoint accumulation zones for prioritized cleanup and prevention. In India,
macro-level assessments exist for marine plastics and national waste generation, but
micro-scale mappings of urban river corridors are scarce, often overlooking localized
drivers like drainage networks and slum proximity. This gap hinders targeted actions
by local authorities. Remote sensing (RS) and Geographic Information Systems (GIS)
offer powerful tools for environmental mapping. While sophisticated methods like
machine learning-based spectral classification have advanced plastic detection, they
demand high-quality data and computational resources, limiting accessibility in
developing contexts. Visual interpretation of freely available high-resolution imagery,
when paired with GIS, provides an accessible entry point for exploratory studies,
enabling pattern identification without complex algorithms. To enhance accuracy,
 spectral indices—mathematical combinations of spectral bands that exploit the unique
reflectance signatures of plastics in the visible, near-infrared (NIR), and short-wave
infrared (SWIR) regions—can be integrated, distinguishing plastics from water,
vegetation, or sediment. Common indices for aquatic plastics include the Plastic Index
(PI), which leverages NIR and red bands to highlight polymer reflectance; the
Floating Debris Index (FDI), designed to detect floating materials by contrasting NIR
with SWIR; and the Normalized Difference Plastic Index (NDPI), a normalized ratio
that minimizes atmospheric effects while amplifying plastic signals in water
[Link] research applies a hybrid approach—visual RS augmented by spectral
indices—to map visible plastic waste along the Mithi River in Mumbai, focusing on
spatial distribution, hotspots, infrastructure associations, and temporal trends. By
emphasizing simplicity and transparency, it aims to bridge the gap in localized urban
plastic assessments, offering replicable insights for similar river systems.
2. Study Area The Mithi River, spanning 17.8 km in Mumbai, India, originates from the
overflow of Vihar (19.14°N, 72.91°E) and Powai (19.12°N, 72.91°E) lakes and flows
southward through densely urbanized areas before emptying into Mahim Bay
(19.00°N, 72.84°E) in the Arabian Sea. Mumbai's subtropical climate features intense
monsoons (June–September, ~2,500 mm annual rainfall), which mobilize waste into
the river, amplifying flooding and pollution. Historically, the Mithi's degradation
began in the 1950s–1960s with untreated sewage from expanding settlements and
slums. The 1970s introduced industrial effluents from textiles, chemicals, and illegal
dumping, including oil sludge, narrowing the channel and destroying mangroves—
reducing flood capacity by 50% from 1966 to 2005. Plastic waste emerged as a key
clogger, exacerbating blockages. The 1980s–1990s saw population surges increase
municipal and animal waste, filling the riverbed with sludge and invasives like water
hyacinth, contaminating Mahim Creek. By the 2000s, it comprised 93% domestic
sewage and 7% industrial waste, with heavy metals exceeding 30,000 metric tonnes
annually by 2019–2020. The 2005 floods (944 mm rain in 24 hours, >1,000 deaths)
underscored encroachments and garbage chokes as culprits. Post-2005, efforts like
Brihanmumbai Municipal Corporation (BMC) dredging (267,000 cubic meters
removed by 2009) and NGO initiatives (e.g., Earth5R cleanups) occurred, but
challenges persist: 2022–2023 Biochemical Oxygen Demand averaged 37.3 mg/L,
classifying it as heavily polluted (Water Quality Index: 35). Upstream sources like
Powai Lake (eight times above safe limits) and ongoing encroachments fuel plastic
inflow. For this study, a 5-km mid-stretch from Bandra-Kurla Complex (19.06°N,
72.86°E) to Dharavi (19.04°N, 72.85°E) was selected, representing mixed land use:
industrial zones, informal settlements (e.g., 500,000+ residents in Dharavi), roads, and
drainage outfalls. A 200-m buffer on both banks captured adjacent influences,
focusing on areas with documented waste pressures.
3. Data and Methodology This exploratory study integrates visual RS with spectral
indices and GIS for plastic waste mapping, emphasizing low-cost tools for
replicability. All processing used QGIS 3.28, with standardized protocols to minimize
subjectivity.

3.1 Data Sources High-resolution imagery (0.15–0.5 m/pixel) was sourced from Google Earth
Pro for visual interpretation, selecting cloud-free dates: March 2018 (dry season baseline),
September 2020 (post-monsoon), April 2022 (dry), and October 2025 (post-monsoon) to
capture seasonal and multi-year variations. For spectral analysis, Level-2A Sentinel-2
multispectral data (10–60 m resolution) from the Copernicus Open Access Hub was acquired
for matching dates, providing bands in visible (B2–B4), NIR (B8), and SWIR (B11–B12)
ranges essential for index calculations. Ancillary layers included: drainage networks from
OpenStreetMap (OSM) and BMC shapefiles; settlement polygons from Humanitarian OSM
Team (HOT) and Mumbai slum maps; roads from OSM. Validation used 50 field
photographs (collected via ground surveys in November 2025), Google Street View (for 30
accessible sites), and BMC/NGO reports on waste hotspots.

3.2 Visual Interpretation of Plastic Waste Visual interpretation, an established RS technique

for surface feature detection, identified accumulations at 1:500–1:1,000 scales. Criteria
included: bright/multicolored patches (e.g., whites/blues for bags/bottles); irregular, clustered
textures contrasting with soil/vegetation; contextual locations (riverbanks, outfalls, stagnant
pools). Accumulations <5 m² were excluded for detectability. Confidence levels were
assigned (high/medium), discarding low-confidence sites to reduce bias. This yielded 128
points, categorized by size: small (<10 m², 45%); medium (10–50 m², 40%); large (>50 m²,
15%).

3.3 Spectral Indices Calculation To complement visual detection and improve separability of
plastics from water and debris, three spectral indices were computed using Sentinel-2 bands:

 Plastic Index (PI): PI = B8 / (B8 + B4), where B8 is NIR and B4 is red. This index
exploits the high NIR reflectance of polymers like polyethylene, yielding values >0.5
for potential plastics in riverine settings.
 Floating Debris Index (FDI): FDI = B8 - (B4 + (B12 - B4) * (λ_B8 - λ_B4) /
(λ_B12 - λ_B4)), approximating SWIR interpolation to highlight floating materials;
thresholds >0.1 indicated debris hotspots.
 Normalized Difference Plastic Index (NDPI): NDPI = (B8 - B11) / (B8 + B11),
using NIR and SWIR to normalize for water absorption and detect macro-plastics;
positive values (0.2–0.6) flagged accumulations. Indices were resampled to 10 m
resolution, thresholded based on literature (e.g., PI >0.45 for confirmation), and
overlaid with visual points to refine detections, particularly in turbid or vegetated
areas.

3.4 GIS Data Preparation and Digitisation Datasets were projected to UTM Zone 43N
(EPSG:32643) and clipped to the study buffer. Plastic points were digitized as vectors with
attributes (date, size, confidence, index values). Spectral rasters were integrated as layers for
hybrid analysis.

3.5 Spatial Analysis Hotspot mapping used kernel density estimation (KDE) with a 100-m
quadratic kernel radius, classifying densities into low (<25th percentile), moderate (25–75th),
and high (>75th) zones. Proximity employed 50-m buffers around drains/settlements,
calculating overlap percentages. Spatial autocorrelation (Moran's I: 0.62, p<0.01) confirmed
clustering. Spectral indices informed density weights, boosting confidence in water-adjacent
hotspots.

3.6 Temporal Assessment Multi-date overlays assessed persistence: points recurring in ≥3

images were deemed chronic (58 sites). Seasonal comparisons showed 35% increase in post-
monsoon visibility, with indices revealing 20% more detections during high-flow periods due
to enhanced contrast.

3.7 Validation Quantitative validation cross-checked 80% of points: field photos confirmed
85% accuracy (kappa=0.78); Street View validated 75% of accessible sites. Spectral
thresholds were tested against ground truth, achieving 82% precision for PI and FDI.
Discrepancies (e.g., cleared sites) were noted, achieving overall >80% reliability.
Independent checks via Bing Maps resolved ambiguities.

4. Results Spatial distribution mapped 128 plastic points, predominantly along eastern
banks near Dharavi. Hotspot analysis generated a heatmap identifying three high-
intensity zones: near Bandra outfall (density peak: 15 points/10,000 m²), central
Dharavi bend (12 points), and Kurla bridge (10 points), covering 40% of points in
20% of the area. Spectral indices enhanced detection: PI identified 25 additional
floating clusters, FDI highlighted debris in 60% of monsoon images, and NDPI
confirmed polymer types in 70% of validated sites. Proximity revealed 72% of points
within 50 m of drains, suggesting transport pathways, and 65% near settlements,
linking to informal dumping. Temporal overlays showed 45% persistence across
years, with post-monsoon images displaying redistributed clusters downstream,
amplified by index values (e.g., FDI >0.15 in 45% of persistent sites). [Placeholder:
Figure 1 - Heatmap of plastic hotspots overlaid on river stretch; Figure 2 - Buffer
analysis map; Figure 3 - Temporal comparison overlays; Figure 4 - Spectral index
rasters (PI, FDI, NDPI).]
5. Discussion Spatial patterns suggest drainage outfalls as key entry points, with
settlements amplifying localized buildup due to collection gaps. Persistent hotspots
indicate inadequate BMC cleanups, while monsoon influences appear to enhance
mobilization, aligning with the river's flood-prone history. Spectral indices provided
deeper insights, e.g., PI's sensitivity to NIR-reflective plastics like PP and PE,
improving differentiation in complex urban-river interfaces. These observations
underscore urban planning needs, such as drain filters or community programs,
without implying causation. Compared to similar urban rivers, the Mithi's issues
reflect common challenges in monsoon cities, where infrastructure lags growth.
6. Limitations and Future Scope Limitations include detection of only visible macro-
plastics (>5 m²), excluding micro/submerged debris; potential subjectivity in
interpretation (mitigated by criteria and indices); and restricted scope to one stretch,
precluding city-wide inferences. Image quality variations (e.g., haze) affected ~10%
of assessments, and Sentinel-2's coarser resolution limited fine-scale mapping. Future
work could expand spectral analysis with hyperspectral sensors (e.g., AVIRIS) or
custom indices like SI-13 or Index-1 for polymer-specific detection, integrate drone
surveys for finer resolution, automated ML classification, and hydrological models to
simulate transport. Scaling to full river length or incorporating with flood data would
enhance applicability.
7. Conclusion This study illustrates a practical GIS-RS framework for mapping urban
plastic waste, revealing clustered hotspots along the Mithi River tied to drains and
settlements, with notable temporal persistence. The hybrid method—visual
interpretation augmented by spectral indices like PI, FDI, and NDPI—democratizes
monitoring, providing actionable baselines for Mumbai's planners to prioritize
interventions and reduce marine leakage. Its replicability suits other Indian cities,
fostering broader environmental stewardship.