Rejoinder Article by: J.Harsha
An article titled “Role of dams on the floods of August 2018 in Periyar River Basin” was published by Sudheer et al. (2019) in Current Science. A rebuttal was prepared and thanks to South Asia Network of Dams, Rivers and People (SANDRP), the same was published by SANDRP on 25th August 2020 (https://sandrp.in/2020/08/25/role-of-dams-in-kerala-floods-distortion-of-science/) for which Sudheer et al. (2019) has now furnished a response (https://sandrp.in/2020/09/19/response-of-sudheer-et-al-to-the-comments-by-mr-j-harsha-on-the-article-role-of-dams-on-the-floods-of-aug-2018-in-periyar-river-basin-kerala/).
In the rebuttal published by SANDRP, I had questioned the very basis of fitting HEC-HMS model for Periyar River Basin (PRB) by Sudheer et al. (2019), and also challenged the assumptions made by them, the methodology followed and the consequent voluminous inferences such as catchment response at Neeleshwaram (L2), virgin simulations, bank full discharges and particularly inferences that indicted nature for the flood calamity but exonerating the role of dams for the floods of Kerala in 2018.
A number of evidences were presented in the rebuttal against the premises of Sudheer et al. (2019).
In their response, Sudheer et al. (2019) have not refuted many of the evidences presented in my rebuttal against their premises. Few of the important evidences that they have not refuted along with the premises are shown in the Table-1 below:
Table-1: List of evidences presented in the rebuttal not refuted by Sudheer et al. (2019)
|Premises made by Sudheer et al. (2019)||Evidences presented against premise not refuted by Sudheer et al. (2019)|
|1||SCS-CN method and λ = 0.2 is valid in PRB.||SCS-CN method along with λ = 0.2 was recommended by United States Department of Agriculture (USDA) for watersheds of USA in 1954; Faizalhakim et al (2016) who questioned adoption of λ = 0.2 for watersheds in USA, Yuan et al. (2014), etc; different regional and climatic conditions represent different value of initial abstractions.|
|2||The land use/land cover in PRB derived from the data for the year 2010 is valid in 2018.||Spatial resolution of IRS AWiFS sensor is 56 m; Landsat 8 Band 1 to Band 7 is 30 m resolution. Landsat 8 analysis shown in rebuttal provides better estimate of vegetation cover than AWiFS.|
|3||The drainage pattern of the PRB is dendritic in nature.||Extensive quantitative analysis based on Meija and Niemann (2008); Howard (1967); slope map derived for PRB; Schumm (2000); basin morphometry, Minimum Bound Rectangle, calculation of junction angles, etc.|
|4||Central Water Commission (CWC) observed data at L2 is good, continuous without any data gaps and fit enough for validation of simulated data.||Analysis of CWC discharge data at Neeleshwaram (L2) for the period 1971-2017 shown in Table -1 and Table – 2 of the rebuttal. For the said period, 28% of data is computed not observed. Years 1991, 1992, 1993 are completely computed not obserseved. Years 2006, 2007, 2008 more computed data than observed data. No of years data is seasonal is 37. No of years data is year-wise is 6. Sudheer et al. (2019) neither has revealed the observed data for August 2018 nor have they provided stage-discharge curve at Neeleshwaram site.|
|5||Error in classification of Lateritic soils classified as HSG-D soil||Winterkorn and Chandrasekaran (1951) who state that laterite soils possess excellent drainability. Sudheer et al. (2019) statement in their Current Science article, “The percentage of gravel content in the laterite soil is relatively high” (Page-784).|
Some of the notable shift in the position found in the response of Sudheer et al. (2019) compared to their original position in Current Science article is presented in Table – 2 below:
Table-2: Notable shift in the position of Sudheer et al. (2019) found in response
|Premise and original position in Current Science||Notable shift in position of Sudheer et al. (2019) in their Response submitted to SANDRP|
|1||The drainage pattern of the PRB in general is dendritic in nature.||The basin supports diverse drainage patterns at various spatial scales. The major drainage patterns of the basin include dendritic, parallel, trellis and rectangular, where dendritic pattern was observed in gentle slope areas irrespective of the physiography. The parallel pattern is obvious in high slopes and trellis and rectangular patterns infer the structural controls.|
|2||Land use/land cover for the year 2010 is valid for August 2018 in PRB||They now admit about 0.5% change in vegetation cover in PRB and 16.51% change in built up area between the years 2010 and 2016 albeit without conducting any proper remote sensing study by themselves or without revealing any of the details of scholarly publication by National Remote Sensing Centre (NRSC)|
In their latest response submitted to SANDRP, Sudheer et al. (2019) continue to justify and mislead the readers with fallacious assumptions, errors in methodology and “cherry picked” information, hence few more evidences are presented here in this rejoinder in order to expose their distortion of science.
In order to justify the premise “SCS-CN method and λ = 0.2 is valid for PRB”, Sudheer et al. (2019) continue propounding “One method fits all “for all the river basins of the world (like “One size fits all”).
Pilgrim et al., (1982) states that catchment size influence runoff in a number of direct and indirect ways. Pilgrim et al., (1982) express doubts regarding the very concept of homogeneous hydrological regions and representative catchments. This refutes the contention of Sudheer et al. (2019) against export of SCS-CN method developed for watersheds in USA to Periyar river basin (PRB).
Mishra, Suresh Babu and Singh (2007) state that in the absence of clear guidelines, SCS-CN applies to small and mid-size catchments and its application to large catchments (say > 100 sq.mi. or 250 sq.km.) should be viewed with caution. But the area of Periyar river basin according to Sudheer et al. (2019) is 5398 sq km way way above the limitation of 250 sq km specified by Mishra, Suresh Babu and Singh (2007).
So, whether the catchment characteristics of PRB remain same as the catchment characteristics of Missouri river basin or Colarado river basin of USA or Danube river basin in Europe or even Indus river basin of India?
Figure-1 shows the river basins of different sizes in United States of America as well as river basins of India. The location of Periyar river basin is also indicated in Figure-1.
Contrary to arguments by Sudheer et al. (2019) for adopting λ = 0.2, there are plenty of literature available such as Karunanidhi et al. (2020) and Satheeshkumar et al. (2017) who have adopted λ = 0.3 for Lower Bhavani basin and Vaniyar sub-basin respectively located close to PRB in South India.
In a most recent study, Ling et al., (2020) have developed a watershed specific initial abstraction ratio λ for Wangjiaqiao watershed in China. According to Ling et al., (2020), the initial abstraction ratio λ varies from location to location, and therefore they developed watershed specific SCS-CN calibration method and λ was no longer fixed at 0.2. Instead they fixed λ = 0.043 specifically for Wangjiaqiao watershed. In contrast, by averaging λ = 0.1 and 0.3 for PRB, Sudheer et al. (2019) are averaging the conditions of catchment characteristics soil texture, soil surface moisture, conductivity, vegetative cover etc., that influence λ in PRB which is nothing short of huge approximation.
Sudheer et al. (2019) failed to mention any reference in their Current Science article for λ = 0.2, but in their response to SANDRP they now mention Mishra and Singh (2003) as a basis for adopting λ = 0.2. However, Mishra and Singh (2003) have never conducted any study for PRB and never recommended λ = 0.2 or average of λ = 0.1 and 0.3 for PRB. In fact, Mishra, Suresh Babu and Singh (2007) have modified the SCS-CN method and listed out a number of limitations of SCS-CN method.
Some of the outstanding limitations of SCS-CN method based on the findings of Mishra, Suresh Babu and Singh (2007) along with blunders committed by Sudheer et al. (2019) are shown in Table-3:
Table-3: Limitations of SCS-CN methodology & distortion of science by Sudheer et al. (2019)
|Limitations of SCS-CN method documented by Mishra, Suresh Babu and Singh (2007)||Distortion of science by Sudheer et al. (2019)|
|1||This method was found to be performing best in agricultural sites, fairly in range sites and poorly in forest sites.||Major part of the Periyar River Basin is covered by forest, which is spread across the upstream and central parts of the basin (Sudheer et al. 2019; Page – 784)|
|2||The method is not intended for continuous simulation; rather, only an event-by-event basis.||Sudheer et al. (2019) conducted continuous simulations using SCS-CN method from 1 August 2018 up to 31st August 2018, 31 days (Current Science, Page-787) instead of event by event basis (Two extreme rainfall events: 8-10 August, 2018 and 14-19 August, 2018). This is a blunder.|
|3||The success of the SCS method is mainly limited by the watershed size and to a lesser extent by the magnitude of runoff events. Actual distributions of the rainfall and infiltration rates would indicate whether or not the SCS method is appropriate for the watershed in question (Yu, 1998).||Sudheer et al. (2019) have ignored watershed size but give more prominence to runoff events. No distributions of infiltration rates have been adopted by Sudheer et al. (2019) for PRB. Therefore, they themselves are unaware whether SCS-CN method is appropriate for PRB or not.|
|4||Since the method was originally developed using regional data, some caution is recommended for its use in other geographic or climatic regions.||Sudheer et al. (2019) claim that SCS-CN method is suitable for PRB, a basin with diverse geographic and climatic region than the watersheds of USA based on which the method was developed (Refer fig-1).|
|5||The size of event (either rainfall or runoff) to which methodology is well suited is also not known.||Sudheer et al. (2019) have not substantiated how the current rainfall event in PRB is suited for their methodology of using SCS-CN method. Yet, they justify that the method is suitable for PRB.|
|6||Central Unit for Soil Conservation, Soil Conservation Division, Ministry of Agriculture, Government of India in 1972recommended a λ value of 0.3 for all regions of India, except for black cotton region for which it is 0.1 under AMC II (normal) and III (wet) conditions.||Sudheer et al. (2019) who have run HEC-HMS for AMC III condition (Page-787 of Current Science) have not followed the recommendations of Central Unit for Soil Conservation, (1972), Ministry of Agriculture for Indian catchments. Mishra and Singh (2003 and 2007) do not recommend averaging of λ = 0.3 and 0.1. No evidence provided to show that the catchment response of a river basin is average of the ratio of actual infiltration to potential maximum retention of the PRB averaged.|
|7||Heggen (1981) and Srinivas et al. (1997) illustrated relative runoff estimation by CN Nomograph for λ = 0.2 and λ = 0.3 or 0.1 (Indian catchments) respectively||But, Sudheer et al. (2019) has used λ = 0.2 instead of λ = 0.3 or 0.1(specified for Indian catchments)|
|8||In the absence of clear guidelines, it is assumed to apply to small and mid-size catchments. Its application to large catchments (say > 100 sq.mi. or 250 sq.km.) should be viewed with caution.||The catchment area of PRB is 5398 sq km which is far far greater than 250 sq km.|
|9||The modified MS model suggested by Mishra et al. (2004a) performs far better than existing SCS-CN model||Sudheer et al. (2019) adopted the model specified by Mishra and Singh (2003)|
|10||As the method was formulated on the basis of small catchment measurements, not on the point scale measurement, it is likely to be revisited in the future, due to the booming of point scale process.||SCS-CN method is not absolute, ultimate model and magic bullet that can represent catchment response without error across any catchment across the world forever. Sudheer et al.(2019) have failed to invent any representative model for catchment response for PRB on the lines of Mishra, Suresh Babu and Singh (2007) or Ling et al. (2020).|
Let a syllogism be constructed with an example of false premise-false conclusion:
“If Virat Kohli is the Prime Minister of India, then his age is less than 25 (Premise)
Virat Kohli is the Prime Minister of India (Premise),
Therefore, Virat Kohli’s age is less than 25 (Conclusion)”
In the above syllogism, the logic is true, but the premises are false and conclusion is also false. As we all know that Virat Kohli is neither Prime Minister nor his age is less 25. But this is how Sudheer et al. (2019) have proceeded from making false premises in their study related to floods in PRB and ended up making a number of inferences that exonerates role of dams but indicts nature alone for the cause of Kerala floods that occurred in 2018.
Let the above syllogism be modified replacing with one of the premise-conclusion adopted by Sudheer et al. (2019): PRB is dendritic drainage basin.
“If PRB is dendritic drainage basin, then HEC-HMS model fit to PRB (premise of Sudheer et al. 2019)
PRB is dendritic drainage basin (Premise)
|Therefore, HEC-HMS model fit to PRB (Conclusion)”
The most notable admission and climb down by Sudheer et al. (2019) in their latest response submitted to SANDRP is the shift in their position with respect to the drainage pattern of PRB. In their article published in Current Science, Sudheer et al. (2019) had “claimed” that PRB possess dendritic drainage basin but nowhere mentioned prevalence of other patterns parallel, trellis and rectangular pattern in PRB (Page-781). They adopted the premise that PRB is a dendritic drainage basin because they deliberately wanted to fit HEC-HMS to PRB. Obviously, Sudheer et al. (2019) states that HEC-HMS is a semi-distributed model that is capable of running continuous and event simulations in dendritic watershed systems (Page-783).
So, in the above syllogism, the premise that PRB is dendritic drainage basin has now changed. The quantitative analysis shown in my rebuttal proves that PRB is not dendritic. So, the conclusion that HEC-HMS will fit to PRB is false and misleading. Therefore, the output from such a model and the voluminous inferences are nothing short of scientific misrepresentation.
Similarly, another syllogism w.r.t SCS-CN method is constructed as below:
“If SCS-CN method with λ = 0.2 is valid in PRB, then the model output represents catchment response of PRB” (premise of Sudheer et al. 2019)
SCS-CN method with λ = 0.2 is valid in PRB (Premise)
Therefore, the model output represents catchment response of PRB (Conclusion)”
The evidences shown in the rebuttal, the unrefuted evidences shown in Table-1, evidences presented in this rejoinder and the exhaustive limitations stated by Mishra, Suresh Babu and Singh (2007) shown in Table-3 clearly demonstrate that SCS-CN method with λ = 0.2 cannot be simply extrapolated by averaging λ = 0.1 and λ = 0.3 as has done by Sudheer et al. (2019). It is nothing short of scientific fraud. This also raises credibility of their affiliated institutions.
Therefore, the conclusion that HEC-HMS model output represents the catchment response of PRB is false. Rather it represents the catchment response of a fictitious river basin.
Notably, the contradiction with respect to Hydrological Soil Group D/Class D soil classification in the downstream areas and presence of lateritic soil in downstream areas of PRB persists. In Current Science, page-784, they clearly mention that, “The percentage of gravel content in the laterite soil is relatively high” but they assign HSG –D/Class D (HSG-D is high clay content but low infiltration soil group) for the downstream areas where they also state lateritic soil is prevalent. This contradicts their statement made in their response submitted to SANDRP that A and B Groups are attributed to gravel and sand dominated textures. Seems Sudheer et al. (2019) have seriously messed up their study in PRB.
With respect to the premise, “Land use/land cover for the year 2010 is valid for August 2018 in PRB” there is a notable shift in the position of Sudheer et al. (2019) in their response. They now admit about 0.5% change in vegetation cover in PRB and 16.51% change in built up area between the years 2010 and 2016 albeit without conducting any proper remote sensing study by themselves or without revealing any of the details of scholarly publication by National Remote Sensing Centre (NRSC). According to Vandana et al. 2020, the number of quarrying sites in Periyar-Chalakudi basin has increased to 525 with many modification and disappearance of natural drainages. Without conducting any remote sensing study Sudheer et al. (2019), belonging to elite academic institutions, are prophesying land use/land cover change in PRB between 2010 and 2018. The outcome of such prophecies is distorted value of CN numbers and distorted runoff output through an unfit HEC-HMS model.
Sudheer et al. (2019) seems to be more impressed by the reputation of the government agencies such as CWC, NRSC, NBSS & LUP than science. For their remark made in their response, “..that the data provided by the CWC is unreliable, is a concern that is to be addressed to the CWC and the Government directly, as it is purely a policy issue”. Yes, I agree but that has huge implications for the authors’ study and conclusions. As a matter of fact, my analysis of the CWC discharge data from 1971-2017 at Neeleshwaram (L2) published in SANDRP dated 25th August 2020 shows extensive data gaps, missing data and enormous computed data that clearly disproves the theory of Sudheer et al. (2019) that “any researcher could rely on the data provided on the public domain by the CWC for any study related to water resources development and management in the country”. In fact, Sudheer et al. (2019) should have disclosed the poor data quality of the CWC at Neeleshwaram (L2) along with the hydrograph published in Current Science (Fig-2, Page-783), if they truly were honest to facts and science in their study.
Summing up, the rebuttal and this rejoinder prove beyond doubt that this study by Sudheer et al. (2019) is a bundle of scientific misrepresentations, widespread distortion of science, sub-standard study and therefore the inferences particularly their claim of higher contribution of runoff by tributaries at Neeleshwaram and exonerating the role of dams in the Kerala floods are not only inconclusive but fictitious.
Implications of scientific misrepresentations and scientific fraud:
Sub-standard research and scientific misrepresentations can cause serious implications to the flood management in India. One such example of scientific misrepresentation is given below:
In 2010, a British ex-physician Andrew Wakefield was held guilty for publishing fraudulent study in Journal Lancet that claimed link between Measles, Mumps and Rubella (MMR) vaccine and increase of autism and/or pervasive development disorder in children. The impact of this research became so widespread that the rate of intake of MMR vaccine dropped as parents were concerned with the health of their children (Sathyanarayana Rao and Andrade, 2011) and according to British Society of Immunology (accessed in 2020), the effect is felt even in 2018 during the outbreak Measles in UK.
Floods cause massive death and destruction of livelihoods. According to Narayanan and Thakur (2019) between 7th August – 20th August 2018 a total of 504 people died and 23 million people were directly affected with estimated US $ 2.83 billion damage to the economy during Kerala floods. Floods have multiple causes that include not only rainfall intensity and catchment response but also anthropogenic causes such as prudent implementation of rule curve, reservoir level & basin level mismanagement, lack of timely release of water and untimely release of water. Unplanned reservoir releases have caused floods in downstream areas in the past. For example: In 1988, Punjab experienced unprecedented floods due to releases from Bhakra Dam (Vijender, 2019). In another case of massive floods in Surat city in 2006, the role of Ukai dam mismanagement in the flooding of the city Surat has been first reported by SANDRP (2006) and extensively documented by Mavalankar and Srivastava (2008). The performance audit report no.4 of 2017 regarding flood management and response in Chennai and its suburban area by CAG, 2017 states inter alia that Chembarambakkam reservoir did not had scientific inflow forecast system, did not had emergency action plan and the outflow was more than inflow leading to unsustained release of water into Adyar river during the unfortunate floods of 2015.
The above cited instances show that poor water governance and reservoir mismanagement is a reality in India. CWC Director (Flood Forecasting) states faulty reservoir operation, breach of embankments for the cause of anthropogenic or man-made floods in discussion on DD NEWS on Aug 10, 2020. There is a need to improve the operation and reservoir management across India with all transparency and improved governance. India has more than 5700 large dams and hundreds of thousands of medium and small dams. Although reservoirs are operated largely by state governments, the level of transparency in operation and management of any of the reservoirs, regular updating the rule curve, the sedimentation rate, the loss of live storage in aging dams and the functional deterioration of all types of dams is dismal.
Therefore, the role of dams during any floods needs non-partisan diagnosis with more of region specific methodologies (similar to what Ling et al. 2020 has done with for Wangjiaqiao watershed in China), so as to identify the avoidable contribution of reservoirs in the cause of floods or accentuation of intensity of floods. This non-partisan diagnosis is essential in order to rectify the cause before another flood strikes and cause devastation. In contrast, it has become a pattern of sorts in India that whenever floods strike across downstream of series of reservoirs, the dam managers are quick to assign the responsibility on the nature. This is partly done to escape accountability and/or the eventual shame it brings not only to the specific dam managers but to their entire organization. So, the vested interests unite to thwart such accountability due to any probable accusation regarding the role of dams in any flooding across India however miniscule the role is. And this is where the scientific misrepresentations, manipulations and scientific frauds come in handy to help exonerate the role of dams in the cause or accentuation of floods.
In a nutshell, readers, legal, audit and investigation agencies should read the outcome of any such study claimed by any reputed institutions with a lot of skepticism and investigate the evidences. This is clearly demonstrated with Sudheer et al. (2019) where they are fitting HEC-HMS model based on false premises when in fact the conclusion is HEC-HMS model and SCS-CN method cannot be fit into PRB.
J Harsha (Director, Central Water Commission, Government of India, Chennai, India, Email: firstname.lastname@example.org; Mobile: +91 9902018682)
[Views expressed in this rejoinder are personal.]
- DD NEWS. (Aug 10, 2020). Time: 19.05-19.35 hrs. https://www.youtube.com/watch?v=Cx-_fPXy0bM
- Karunanidhi, D., Anand, B., Subramani, T and Srinivasamoorthy, K. (2020). Rainfall-surface runoff estimation for the Lower Bhavani basin in south India using SCS-CN model and geospatial techniques. Environmental Earth Sciences. Springer. 79. 335. https://doi.org/10.1007/s12665-020-09079-z
- Ling, L., Yusop, Z., Yap, W., Tan, W.L., Chow, M.F and Ling, J.L. (2020). A calibrated, watershed-specific SCS-CN method: Application to Wangjiaqiao watershed in the Three Gorges Area, China. Water, MDPI. 12 (60). doi: 10.3390/w12010060.
- Mavalankar, D and Srivastava, A.K. (2008). Lessons from Massive Floods of 2006 in Surat City: A framework for applications of MS/OR techniques to improve dam management to prevent flood. Research and Publications. W.P.No 2008-07-06. Indian Institute of Management. Ahmedabad.
- Mishra, S.K.; Suresh Babu, P.; Singh, V.P. (2007). SCS-CN method revisited. In Advances in Hydraulics and Hydrology; Water Resources Publications: Littleton, CO, USA.
- Narayanan, A and Thakur, G.M. (2019). Natural calamities and household finance: Evidence from Kerala floods. Draft available at https://www.isid.ac.in/~epu/acegd2019/papers/AbhinavNarayanan.pdf
- Pilgrim, D.H., Cordery, I and Baron, B.C. (1982). Effects of catchment size on runoff relationships. Journal of Hydrology. Elsevier, 58: 205-221.
- Satheeshkumar, S., Venkateswaran, S and Kannan, R. (2017). Rainfall-runoff estimation using SCS-CN and GIS approadh in the Pappiredipatti watershed of the Vaniyar sub-basin, South India. Modeling Earth Syatems and Sciences. Springer. 3:24. doi: 10.1007/s40808-017-0301-4
- Sathyanarayana Rao, T.S and Andrade, C. (2011). The MMR vaccine and autism:Sensation, refutation, retraction and fraud. Indian J Psychiatry. 53 (2): 95-96.
- SANDRP (South Asia Network on Dams, Rivers and People). (2006). Accessed
- Sudheer, K.P., Murty Bhallamudi, S., Narasimhan, B., Thomas, J., M. Bindhu, V., Vema, V., & Kurian, C., Role of dams on the floods of August 2018 in Periyar River Basin, Kerala. Current Science, 2019, 116(5), 780–794. Retrieved from https://www.currentscience.ac.in/cs/Volumes/116/05/0780.pdf
- Vandana, M., John, S.E., Maya, K.K and Padmalal. (2020). Environmental impact of quarrying of building stones and laterite blocks: a comparative study of two river basins in southern Western Ghats, India. Environment Earth Sciences. Springer. 79 (14). doi:10.1007/s12665-020-09104-1
- Vijender. (2019). Floods in Punjab, India, case study 2010. International Journal of Research and Analytical Reviews. 6 (1): 145-153.