key: cord-1008356-pu2ecwtu
authors: Patel, Priyank Pravin; Mondal, Sayoni; Ghosh, Krishna Gopal
title: Some respite for India's dirtiest river? Examining the Yamuna's water quality at Delhi during the COVID-19 lockdown period
date: 2020-07-14
journal: Sci Total Environ
DOI: 10.1016/j.scitotenv.2020.140851
sha: 0ab492b83b8ffeadd819b563edb2180618e9d436
doc_id: 1008356
cord_uid: pu2ecwtu

Abstract The Yamuna's stretch within Delhi is considered as the dirtiest river reach in India and despite numerous restoration plans, pollution levels have risen unabated. However, the enforcement of a nationwide lockdown due to the ongoing COVID-19 pandemic can possibly provide a ray of hope. We analyze the lockdown's impact on the water quality status of this stretch using a combination of measured parameters and satellite image derived indices. Class C Water Quality Index estimates of nine stations indicate an improvement of 37% during the lockdown period. The Biological Oxygen Demand and Chemical Oxygen Demand values reduced by 42.83% and 39.25%, respectively, compared to the pre-lockdown phase, while Faecal Coliform declined by over 40%. Similar analysis of 20 major drains that meet the Yamuna revealed declining effluent loads and discernable improvements in drain contaminant status were ascertained via a hierarchical cluster analysis. Reach-wise suspended particulate matter content, turbidity and algal signatures were derived from multi-temporal Landsat-8 images of prior and ongoing lockdown periods for 117 channel segment zones. These parameters also declined notably within most stretches, although their extents were spatially varied. While the partial/non-operational status of most industries during the lockdown enabled significant reduction in effluent loads and a consequent betterment in the river water quality, its spatial variations and even deterioration in some locations resulted from the largely undiminished inflow of domestic sewage through multiple drains. This study provides an estimate of possible river recovery extents and degree of improvement if deleterious polluting activities and contaminants are regulated properly.

The COVID-19 global pandemic, caused by the Novel Coronavirus, is considered to be one of the most virulent diseases to have afflicted humankind. SARS-CoV-2 virus cases were first detected in December, 2019, in China's Hubei province (WHO 2020a) , being subsequently declared as a Public Health Singh et al. 2018 ), while models like QUAL2E/QUAL2Kw and STREAM-II have been used to simulate different pollution scenarios to determine the maximum permissible pollutant discharge (Paliwal et al. 2006; Sharma and Singh 2009; Walling et al. 2014; . Only one study has linked satellite image derived parameters with measured pollution loads in this stretch of the Yamuna (e.g. Said and

Hussain 2019), with this method presenting possibilities for continuous monitoring of the entire stretch (for further information on the Yamuna's water quality degradation, see Supplementary Section 1). The effectiveness of the Yamuna Action Plan (YAP I and II) was examined by Sharma and Kansal (2011b) , with their results suggesting that all attempts so far have been unsuccessful and that the river fails to maintain the minimum ecological flow necessary to sustain aquatic life (as also reported by Upadhyay et J o u r n a l P r e -p r o o f extent of water quality enhancement obtainable by periodic suspension of polluting industrial activities and identifies river reaches that benefit the most from this.

Water quality assessments gauge whether or not the available water is suitable for a specific use ( J o u r n a l P r e -p r o o f are located on the Yamuna itself while the last station at Jaitpur is located on the Agra Canal, that offtakes from the Yamuna at Okhla. As this canal carries a portion of the Yamuna's flow and is also linked to it through a smaller channel at Jaitpur, we have considered it to be representative of the Yamuna's water quality characteristics in its lower reaches within the Delhi NCT.

The time period primarily examined is from January to April of 2020. The data obtained for 6 January, 13 February and 13 March, 2020, was taken to represent the pre-lockdown state while that of 6 April), which can be taken to respectively denote conditions in the early and latter phases of the imposed first lockdown phase (25 March to 14 April, 2020) that was operative in the country. The water quality status of the 20 drains flowing into the Yamuna at Delhi was also obtained from the DPCC for the same time period as the river dataset. This was examined to discern the effluent quality meeting the river and gauge if the lockdown has had any effect on the area's drains besides its effects on the main channel.

Statistical attributes, trend analysis as well as box-whisker plots were computed and visualized using MS-Excel. Among the frequently used normalization techniques (i.e. ranking, distance to target, Zscore, min-max and proportionate normalization) (Nardo et al. 2009 ), the Decimal Point Normalization (also sometimes referred to as the Floating Point Number) technique was employed to rank the drains during the prior-lockdown and lockdown periods, based on the resulting additive aggregation (arithmetic mean) scores (Tate 2012; Tofallis 2014) . In this normalization method, the data attributes are transformed by simply moving the decimal points of the original data (García 2015) and the maximum number of J o u r n a l P r e -p r o o f water quality parameters mostly within themselves while the image derived indices were examined in the same manner, and so for the greater part of the analyses these two datasets were mutually exclusive and did not skew the results. Since there are no other similar datasets for any other dates within the examined time period, this analysis was perforce based on whatever information was available.

Image pre-processing was done on the ACOLITE (Atmospheric Correction for OLI 'lite '-Dogliotti et al. 2018 ) platform, which enables quick processing of multiple scenes (Yunus et al. 2020) and the direct derivation of some studied parameters (see ACOLITE User Manual-RBINS-REMSEM 2019 

The Yamuna's outline within the Delhi NCT was obtained from the OpenStreetMap dataset, overlain on Google Earth imagery and edited manually to obtain accurate banklines. The thalweg was prepared similarly and divided along its length into 117 segments of 500 m length each. These were overlain on J o u r n a l P r e -p r o o f markedly from that of the adjacent water pixels) from skewing the derived values, such features were digitized and masked during the database updation.

In the lockdown phase (i.e. 6th April and 14th April, 2020), six monitoring stations for Class B (about 67%) and 8 monitoring stations for Class C (about 89%), out of the nine stations overall, recorded significant improvements in their WQI (Table S3 and Figure 2a ), in comparison to the prior-lockdown phase (i.e. 6th January, 13th February and 13th March, 2020). On the whole, the improvement was 10%

for Class B and more prominently for Class C (37%). The difference of the WQI computed above in comparison to that recorded during the same months in the previous year (i.e. April 2019) was clearly discernable, particularly for Class B (Table S3) .

A more than 40% improvement in the WQI for both Class B and C was noticed for the monitoring stations at Nizamuddin Bridge (m6), Okhla (m7), Shahdara (m8) and Jaitpur (Agra Canal) (m9), all of which are located in South Delhi, downstream of the ITO Barrage ( Figure 2 ). The closure of almost all industries during the lockdown has likely resulted in minimal effluent discharge and thereby lessened the level of contamination in the Yamuna (as has been reported in leading dailies-Gandhiok 2020; PTI 2020). Nevertheless, even during the lockdown phase, the recorded water quality at Palla (m1), Kudesia Ghat (m4) and the ITO Barrage (m5) shows an increase in contamination. Possibly, as these locations are in north and central Delhi [areas that are either extensively cultivated (m1) or heavily built-up (m4 and m5)], even during the lockdown period, partly treated or untreated agricultural runoff and domestic wastewater continues to contaminate the river via the many drains that debouch into the Yamuna herein (zone demarcated by grey background in Figure 1b ).

While substantial precipitation can reduce the river's contaminant load by enabling greater dilution, the rainfall received at Delhi in the period following the lockdown's commencement was quite meagre [only 9.8 mm for the entirety of April, 2020-Indian Agricultural Research Institute (IARI), New J o u r n a l P r e -p r o o f could be obtained for April, 2020), almost no precipitation was received during this time period. Delhi had received a lot of unseasonal rainfall during early-to-mid-March (https://weather.com/en-IN/india/news/news/2020-04-03-delhi-wettest-march-ever-records-589-rainfall-weather) due to several Western Disturbances that passed over the region. Yet despite this substantial amount of rainfall (a total of 174.6 mm of rainfall was received as recorded by the IARI), the pollution load in the Yamuna was still high, whereas despite rainfall being minimal in April, 2020, we could still observe the reductions in the concentrations of the polluting parameters and a betterment in the WQI for the datasets of 6th April, 2020 and 14th April, 2020. Moreover, the examined stretch of the River Yamuna lies entirely within the National Capital Territory of Delhi, which is one of the most densely urbanised locales in India. As such, almost the entire stretch of the river here is actually affected almost wholly by anthropogenic factors and the influence of natural causes on water quality variation are quite minimal. This is even more so since the study was conducted during the dry season (and thereby minimal surface dilution of the contaminant load from meteoric water sources). Furthermore no tributaries, apart from sewer drains, come and meet the Yamuna River within this stretch. Thus the reductions observed in the polluting parameters for the 6th April, 2020 and 14th April, 2020 datasets and the corresponding betterment in the WQI may be taken to directly arise from the lockdown situation.

Despite significant improvements at most stations during the lockdown period, the Yamuna's water quality in the NCT remained far beyond permissible limits ( Figure S1 ). Only two stations namely Palla (m1) and Surghat (m2) recorded WQI values below 100 throughout the pre-lockdown and lockdown periods and only at Surghat could the water quality be accorded as being non-polluted, with J o u r n a l P r e -p r o o f

The BOD and COD values have been considerably attenuated during the lockdown in comparison to their respective pre-lockdown amounts ( Figure 2b ). Their mean concentrations (considering all monitoring stations for both parameters) have changed by much as -42.83% (net reduction of 11.82 mg/l) and -39.25% (net reduction of 32.06 mg/l), respectively (Table S5, column 7, 8, 11 and 12) . When compared against the previous year's levels, their reduction amounts were 19.78% and 10.45%, respectively. For monitoring stations downstream of the ITO Barrage (m6, m7, m8 and m9), the magnitude of reduction in both these parameters was more than 40% during the lockdown phase. Yet, the water quality criteria in terms of both the BOD and COD parameters, from Khajori Paltoon Pool (m3) downstream, does not meet the desirable standards for Class B or Class C.

While the FC parameter showed a substantial variation (>40% reduction) between the prior and during lockdown phases in four monitoring stations (Table S5 , column 14), its overall variation does not support any explicit trend (Figure 2b ), since at Palla (m1), Kudesia Ghat (m4) and ITO Barrage (m5), it got amplified by about +2223%, +202% and +70% respectively. Such substantial boosting up of the faecal contamination may result from untreated domestic waste discharging from the drains at these sites.

Notably, a total of 5 (d2, d3, d4, d5 and d6) and 7 (d7, d8, d9, d10, d11, d25 and d26) drains merge in the immediate upstream reaches of Kudesia Ghat (m4) and ITO Barrage, respectively. As the COVID-19

virus has been shown to be present in wastewater and excreta (Amirian 2020 

The pH value has increased slightly in the upstream section of Khajori Paltoon Pool (m3) but decreased faintly at its downstream section (from m4 onwards) in the lockdown period (Table S5, column 3 and 4). Its overall pattern demonstrates a slight rising trend ( Figure 2b ) and at all stations pH values have remained >7 (basic) in both the prior-to-lockdown and during-lockdown periods. The entry of organic matter from anthropogenic sources into river systems in urban areas is again a possible cause for such an alkaline nature of the stream, as has been documented in several studies (e.g. Peters 2009;

Mocellin and Magro 2011).

Significant differences were reported (Table S7 , columns 10, 11, 14 and 15) in the mean COD and BOD values between the pre-lockdown and during-lockdown dates for the 20 drains meeting the River Yamuna within its Delhi NCT stretch, for which the required data are available. Their respective overall percentage deviations were as much as -32% and -36%, with this attributable to industrial closure during the lockdown. However, such marked differences were not observed for the pH and Total Suspended Solids (TSS also referred to as SPM) concentrations (Table S7, and d24 (Indrapuri Drain) (Table S7 ). Despite the reduction in industrial effluents (of 35.9 MLD), a BOD load of 260 TPD was still expected to be discharged from domestic sewage (CPCB 2020). Apart from the domestic waste contribution that keeps levels up, another possible reason is that from January to April 

The HCA elicited the correspondence between drains based on four parameters (pH, TSS, COD and BOD), similar to the approach of Gautam et al. (2013) and during-lockdown stages. Seemingly, their effluents have been somewhat limited during the lockdown phase (possibly from retarded industrial and restricted daily commercial/residential activities within their catchment zones), diminishing pollution loads. Cluster 2 corresponds to 3 drains (d3, d4 and d12) in the pre-lockdown period and 7 drains (d1, d10, d11, d13, d14, d22 and d23) in the lockdown period.

Interestingly, d3, d4, d7, d10 and d12 are the least polluted drains in both these time slots (Table S9) .

Therefore Cluster 2 indicates the 'low-polluted' group. Cluster 3 contains 8 drains for the pre-lockdown (d2, d16, d17, d19, d21, d22, d23 and d24) and 4 drains (d2, d17, d21 & d24) for the lockdown slots and the lowest water quality level was recorded in them in both phases, i.e. these comprise the 'most'

polluting group (Table S9) . These drains likely receive effluents from mostly domestic sources and from industries that may have perforce operated during the lockdown period. Notably, due to the imposed restrictions, the number of drains in this cluster reduces by 50% from its pre-lockdown membership. The drains under Cluster 3 are of the highest priority for regulation and treatment, especially those that are members in both the pre-lockdown and during-lockdown phases (drains d2, d17, d21 and d24). 

The image-derived water quality parameters were mapped at the reach-level for eight different dates from 2017 to 2020 for the indices derived via band ratioing (NDTI, NDVI and FAI), while the SPM and turbidity levels were mapped for two days within the lockdown phase. Both temporal and spatial variations are apparent. These outputs allow estimates of the river's state beyond the first lockdown phase (which ended on 14th April, 2020-the date up to which the measured water sample data is available), and during the subsequent lockdown phases.

The NDTI, NDVI and FAI parameters were mapped on each of the dates for all 117 reach segments ( Figure 5 shows the respective outputs for 2020, while those of previous years are given in Figure S6b ) is similar to that for the SPM load.

The Dogliotti_2015 output (Figure 12c ) shows more elevated levels of turbidity in the channel segments as compared to the Nechad_2009_2016 method, particularly in the upper reaches, possibly due to its switching-algorithm extraction method (see Supplementary Section 3). However, the reductions seen are equally sharp for both these data products ( Figures S6b and S6c) , with a 60-80% decrease in the turbidity levels almost throughout the entire studied stretch within the lockdown period. While such marked reductions may be seemingly surprising (despite reports highlighting the 'sparkling' clarity of the Yamuna-Gandhiok 2020), at this moment we do not have an alternate measured dataset to validate this.

As pointed out before, our primary aim has been to document whether any changes occurred due to the lockdown instead of extremely accurate estimates of parameters and a similar systemic bias in image processing for both the pre-and during-lockdown periods would not affect this intention significantly.

Moreover, with the Dogliotti_2015 output having a much wider range than the Nechad_2009_2016 output, the percentage variation between its 29.03.2020 and 16.05.2020 results is more pronounced.

Instead of just comparing the 2019 and 2020 datasets, we felt it was more prudent to average the NDTI values for the years 2017-2019 ( Figure S2 ) and compare them against those derived for similar time stamps in 2020, to dampen any episodic effects in the older datasets. Since the image dates were not exactly the same, we designated four 'Days' for this comparative analysis (Figure 7) . Day 1 corresponds to early-mid March, Day 2 to late March, Day 3 to late April and Day 4 to mid-May (thereby denoting conditions prior to the lockdown, during its early phase, during its middle phase and towards its end phase). No marked changes were discerned in the Day 1 output, as expected, since environmental conditions could be expected to be quite similar in all the four years, while the onset of the lockdown had J o u r n a l P r e -p r o o f occurred too recently to significantly affect results in the Day 2 dataset. The Day 3 output showed the most reductions in reach-wise mean NDTI levels, after more than a month of the lockdown. While the Day 4 dataset also shows marked reductions (around Okhla), the lessening extents are relatively lower elsewhere as compared with Day 3. By mid-May, 2020, the Indian Government had lifted lockdown measures and some industries had resumed, depending on the COVID-19 infection rate and case numbers in different areas. With some semblance of normal life returning, we see its effect on the channel quality, with much lower reductions in the mean NDTI values for this phase in some of the channel reaches.

Pearson product moment correlations were derived between the measured and image-extracted parameters for the nine sampling stations and their reaches (Table 1) 

The lockdown has had significant impacts on the water quality of the Yamuna within its Delhi NCT stretch, with enhanced WQI ratings and a significant decline in the BOD and COD levels. However, the FC had increased at a few sites probably due to livestock excreta and domestic sewage (which remained unabated) inflows. Drain effluents also revealed similar reductions in pollutant loads while the most polluting drains in this reach were identified. A marked drop in the river's turbidity and SPM was also apparent. Some data limitations have constrained the performed analyses but the fact that the Yamuna's water quality has bettered, could be clearly established. However, despite such improvements, the WQI status could not meet the prescribed CPCB standards. The framework adopted in this study can be easily transferred to not only examine similar lockdown effects on the stream quality of large rivers in other areas, but also for continuous monitoring from integrated measured sample and image-extracted datasets, which can extend and improve spatio-temporal water quality assessments. The COVID-19 global pandemic is a once in a generation occurrence that is currently plaguing the world. Yet, one of its outfalls also presents a similarly once in a generation opportunity to re-comprehend and redesign existing frameworks and put in place robust mechanisms to cleanse one of India's most polluted rivers and the nation's other similarly afflicted watercourses.

J o u r n a l P r e -p r o o f Note: * Correlation is significant at the 0.05 level (2-tailed); ** Correlation is significant at the 0.01 level (2-tailed). Only those nine reaches were considered in the above correlation, within which were located the water sample measurement stations. There is a slight dissonance between the dates of the satellite-image derived parameters and the measured water sample dates. The former is from the 29.03.2020 dataset while the latter is from the 06.04.2020 dataset. T_Nch2016 and T_Dog2015 respectively denote the turbidity values obtained via the Nechad (2009 Nechad ( , re-updated in 2016 and Dogliotti (2015) Table S1 ; the WQI standard ranges and CPCB classification are given in Figure S1 ; the deviations and associated key statistics of the WQI's Box-and-Whisker plots are given in Tables S3 and S4; -05.03.2017 , 08.03.2018 , 11.03.2019 , 13.03.2020 Day 2-21.03.2017 Day 2-21.03. , 24.03.2018 Day 2-21.03. , 27.03.2019 Day 2-21.03. , 29.03.2020 

Temporal and spatial variations of water quality in the Jinshui River of the South Qinling Mts

Satellite Monitoring of Inland and Coastal Water Quality: Retrospection, Introspection, Future Directions

On the detectability of adjacency effects in ocean color remote sensing of midlatitude coastal environments by SeaWiFS

On the relation between NDVI, fractional vegetation cover, and leaf area index

Toxic foam pollutes India's sacred Yamuna River. abcNews

Surface Water-Quality Modeling

Studies on the Water Quality of a Water Body at Asansol Town

Eutrophication: Causes, Consequences, and Controls in Aquatic Ecosystems

Optimal parabolic section with freeboard

Monitoring of Alpine Rivers: the Italian experience

Control Remote Sensing of the Ocean, Sea Ice, and Large Water Regions

Calibration and validation of a generic multisensor algorithm for mapping of total suspended matter in turbid waters

Land Use and Land Cover Map of Delhi-Bhuvan Thematic Data Product. Land Use / Land Cover Analysis-Third Cycle. National Remote Sensing Centre, Department of Space, Government of India

Creating legal rights for rivers: lessons from Australia

Water quality modelling of the river Yamuna (India) using QUAL2E-UNCAS

Water Quality Status of Yamuna River and its Toxic Effects on Humans

Water Quality Parameters of River Yamuna in Delhi after 20 years of the Yamuna Action Plan

Water quality parameters of River Yamuna in Delhi after 20 years of the Yamuna Action Plan

A simple optical model to estimate suspended particulate matter in Yellow River Estuary

COVID-19: The environmental implications of shedding SARS-Cov-2 in human faeces

A Report on Toxicity Load of Yamuna River in Delhi. Toxics Link

Applications of Remote Sensing and GIS in water Quality Monitoring and Remediation: A State-of-the-Art Review

Water quality assessment of River Yamuna from origin to confluence to River Ganga with respect to Biological water quality and Primary Water Quality Criteria

Royal Belgian Institute of Natural Sciences

Remote Sensing Techniques to Assess Water Quality

Assessment of water quality: a case study of River Yamuna

Water quality assessment of Yamuna in Baghpat region

Development of an overall index of pollution for surface water based on a general classification scheme in Indian context

A river and the riverfront: Delhi's Yamuna as an in-between space

Pollution of River Yamuna: the Delhi story

Water quality analysis of River Yamuna using water quality index in the national capital territory

The status and effects of the Yamuna Action Plan (YAP)

Water quality modeling for urban reach of Yamuna river

DO-BOD modeling of River Yamuna for national capital territory, India using STREAM II, a 2D water quality model

Pollution Study of River Yamuna: The Delhi Story

Water quality monitoring of Yamuna River by using GIS based Water Quality Index in Delhi

Effect of restricted emissions during COVID-19 on air quality in India

Sewage Treatment and Marsh Regeneration. The Yamuna River Project

No decentralised STP in Delhi even after four years of green court order

Assessment of surface water quality using multivariate statistical techniques: a case study of the Fuji river basin

Lockdown impact: Ganga water in Haridwar becomes 'fit to drink' after decades

How River Yamuna cleaned itself in 60 Days of coronavirus lockdown. NDTV, 25th May

Assessment of water quality on the Yamuna river using principle component analysis: a case study

How lockdown has been a gift for Ganga. Times of India

Environmental flow for the Yamuna River in Delhi as an example of monsoon rivers in India

Automated algorithm for extraction of wetlands from IRS Resourcesat LISS III data. International Archives of the Photogrammetry

Nature of distribution of mercury in the sediments of the River Yamuna (tributary of the Ganges)

Multi-source high-resolution satellite products in Yangtze Estuary: cross-comparisons and impacts of signal-to-noise ratio and spatial resolution

Social vulnerability indices: A comparative assessment using uncertainty and sensitivity analysis

Indian under COVID-19 lockdown

Add or multiply? A tutorial on ranking and choosing with multiple criteria

Statistical approaches to evaluate the aquatic ecosystem qualities of a significant mining area: Emet stream basin (Turkey)

Automated water surface temperature retrieval from Landsat 8/TIRS. Remote Sensing of Environment 237

Advantages of high quality SWIR bands for ocean colour processing: Examples from Landsat-8. Remote Sensing of Environment

Acolite for Sentinel-2: Aquatic applications of MSI imagery

Atmospheric correction of metre-scale optical satellite data for inland and coastal water applications

Assessment of seasonal and polluting effects on the quality of river water by exploratory data analysis

Prediction of water quality from simple field parameters

Impact of Climate Change on River Water Quality: A Case Study on the Yamuna River

Water quality monitoring and evaluation using remote sensing techniques in China: a systematic review

humans-a-growing-number-of-voices-say-ye WHO: World Health Organization, 2020a. Novel coronavirus (2019-nCoV) Situation Report 1. World Health J o u r n a l P r e -p r o o f WHO: World Health Organization, 2020b. Coronavirus disease 2019 (COVID-19) Situation Report 51. World Health Organization

Dataset on assessment of River Yamuna, Delhi, India using indexing approach

Interceptors Sewers Strategy: The Post-Industrial Delhi

Retrieval of suspended sediment concentrations using Landsat-8 OLI satellite images in the Orinoco River (Venezuela)

COVID-19 and surface water quality: Improved lake water quality during the lockdown

Hindu festival in India marred by a river of toxic foam and a blanket of killer smog

A simple and effective method for monitoring floating green macroalgae blooms: a case study in the Yellow Sea

Solid Waste Management-Zero Waste City. The Yamuna River Project

Coupled approach for radiometric calibration and parameter retrieval to improve SPM estimations in turbid inland/coastal waters