Comment : Getting the Methodology Wrong for Analysing the Hydrological Changes in Watersheds

This article is a crit ique on the paper ‘Spatial characterization of long-term hydrological change in the Arkavathy watershed adjacent to Bangalore, India’, by Penny et al . (2018), published in the journal of Hydro logy and Earth System Sciences (volume 22, issue number 1). The article h ighlights choice of inappropriate methodology and faulty assumptions by 10 the authors for analysing the watershed scale hydrological changes, generating misleading results and inferences.


Introduction
Globally there is an increase in demand for water, mo re so in countries located in semi-arid to arid reg ions.In India, which is largely semi-arid, many large river basins such as Krishna, Cauvery and Pennar are on the verge of closure or are already closed, with no discharge into the natural sinks (oceans ) in low rainfall years.In order to manage water in such basins, it is important to first identify the factors causing growth in demand for water and also the long -term changes and in water supply fro m catchments and aquifers (runoff and groundwater).Th e long-term changes in runoff can mainly be because of rainfall change, land use changes (leading to increased evapo -transpirative demand causing reduction in runoff generation), and groundwater intensive use in areas where aquifer discharge contributes to runoff in the form of base flows.Changes in groundwater availability fro m natural recharge can be caused by changes in infilt ration of water fro m precipitation and natural discharge fro m aquifers.Apart fro m understanding the aggregate level changes, it is also important to analyse the spatial patterns, especially in the upstream and downstream parts of the basin as most of the time it is the people in the la tter that get affected by the hydrological changes occurring upstream.
One such attempt to exp lain spatial pattern of long-term hydrological changes has been made by Penny et al. (2018)

Misplaced concerns about data unavailability
A few recent studies that have focused on water management issues in India, have reported lack of time series data sets on hydrology and geo-hydrology and questioned the reliability of data availab le fro m official agencies (see for instance, Srinivasan et al., 2015;Penny et al., 2018).Often it is claimed that official agencies are unable to process and manage data properly.In some cases, raw data collected directly fro m the farmers on scientific phenomena that occurred as long back as 40-45 years, based on recall method, is considered to be more reliab le by researchers.While validated data might not be available for free on websites, they are certainly availab le with the central and state agencies in charge of monitoring surf ace water flows and groundwater in the basin.The point is one has to make an effort to reach the concerned agencies, collect and interpret the data in the best possible way.
Moreover, the situation with regard to hydrological data availability and their reliab ility has improved remarkably over the past 10-15 years in India.After the World Bank supported Hydrology Project (init iated in 1995), most of the historical data related to meteorology, hydrology, reservoir operations, water quality and groundwater are digitised in many States through establishment of water data centres (WDC).Presently, nine major states in India, in cluding Karnataka where the Arkavathy

Selected literature review
The authors (Penny et al., 2018) claim that no attempt has been made in India to analyse hydrological process at the watershed, sub-basin and basin scale and making it appear that their study is perhaps the first such attempt to do so.In the process, they have conveniently ignored a large body of peer-reviewed research work already undertaken not only at the watershed scale (for instance by Garg et al., 2012;James at al., 2015), but also at the basin level (for instance by Gosain et al., 2010;Ku mar, 2010).So me of these studies are highly empirical in nature and covered Narmada river basin (Ku mar and Singh, 2001;Ku mar et al., 2005;Ku mar 2010;James et al., 2015), Ganga river basin (Anand et al., 2017), Brah maputra and Kosi basins (Gosain et al., 2010).Simp ly disregarding the existing studies on river basin management, instead of critiquing the approach, methodology and findings of those studies if they find them to be lacking in any way, is poor scholarship on the part of the authors.

Wrong unit of analysis
Fro m the discussions presented in the previous sections, it is clear that there is no dearth of reliable official data on hydrometeorology and groundwater for Arkavathy watershed and since it is available for a long time period, various agencies are monitoring and managing data well.Strangely, to assess the hydrological changes, Penny et al. ( 2018) have considered changes in water spread area in the tanks where inflows are not gauged and not the large reservoirs in the watershed of where inflows are gauged, the reason being that reservoirs are actively managed for p roviding water for urban and agricultural uses.
However, the whole purpose of using this approach was to identify changes in tank water extent which can be attributed to changes in tank inflows.Authors have used volumetric water balance equation for estimat ing the tank inflows (equation 1 on page 600 in Penny et al., 2018).
If the aim of the authors was to analyse the hydrological changes due to changes in land use, the reservoirs which are gauged and for which reliab le data on inflows are available would have been a much better choice.A mass balance equation could have been used to estimate the actual inflows based on data on change in storage and outflows at different time intervals as data on rainfall at the reservoir s ite, releases fro m the dam, water losses and reservoir water level are availab le with the water resources department of the state.This data is also available in dig itised form with the Karnataka State WDC, wh ich is operated by Water Resources Department, Govern ment of Karnataka for surface water and Groundwater Directorate, Karnataka for groundwater (Table 1).Thus use of tanks, which are un-gauged, as a unit of analysis has actually adversely affected the confidence level of the model outputs and increased the uncertainty in the results as the tank water spread area may not be an accurate representation of the hydrological alterations happening in the watershed caused by land use changes and other factors, due to the reason that many complex factors (infiltration and evaporation) over and above the inflows affect the water spread area.

Faulty assumptions and inferences
Some assumptions and inferences in the paper by Penny et al. (2018) are a litt le hard to co mprehend.A few of them are discussed herewith.First, a water balance equation is used to estimate the inflow into the ungauged tanks.For this, it is assumed that the init ial storage in the tank is zero.However, the aerial photo provided in the paper shows that there is wat er in the tank before the onset of the first runoff event.
Second, the authors have assumed outflows fro m the tank to be negligible.This is based on observations on a few tanks and may be true for years with lo w rainfall.In years with normal and high rainfall, tanks will have overflow which usually enters the downstream tank in cascade, a common occurrence in southern India.In Arkavathy watershed, there is very high interannual variability in rainfall.The rainfall can be as high as 1400 mm against an average rainfall of about 800mm (Fig. 2).
in a watershed named Arkavathy in Cauvery river basin.The watershed lies in the state of Karnataka.The study claims to have used a methodology which is appropriate for ungauged catchments.Though the authors have noble ideas, the study methodology, assumptions, results and inferences all have serious shortcomings .Findings based on such shortcomings have limitations when it comes to making useful policy inferences which are discussed in the subsequent sections of the article.Hydrol.Earth Syst.Sci.Discuss., https://doi.org/10.5194/hess-2018-187Manuscript under review for journal Hydrol.Earth Syst.Sci. Discussion started: 15 May 2018 c Author(s) 2018.CC BY 4.0 License.
watershed studied by Penny et al. (2018) falls, have fully operational WDC.Penny et al. (2018) have raised serious concern that official data on surface water flows and groundwater in the watershed were not available and hence justified use of Remote Sensing (RS)/ GIS based methodology which they claim is suitable for ungauged catchments.However, as per the basin report by the M inistry of Water Resources, River Development and Ganga Rejuvenation (Government of India, 2014), middle part o f Cauvery river basin having an area of about 57,281 sq.km (where Arkavathy watershed falls) has 20 hydrological observations sites, 159 meteorological stations and 612 groundwater observation wells which are maintained by Central agencies alone (Fig. 1).In Arkavathy watershed, there are three stream gauging stations maintained by Central Water Co mmission (CWC) and about 50 groundwater observations wells maintained by Central Ground Water Board (CGW B).State agencies have more nu mber of observations sites.Therefore, it is not clear which data paucity the authors are referring to.

Figure 1 :
Figure 1: Cauvery River Bas in Map Showing Different Sub-basins and Spread of Hydrological Observation Sites maintained by CWC (Source: Government of India, 2014) 5