You are on page 1of 54

Accepted Manuscript

Research papers

A modeling approach to establish environmental flow threshold in ungauged


semidiurnal tidal river

A. Akter, A.H. Tanim

PII: S0022-1694(18)30070-2
DOI: https://doi.org/10.1016/j.jhydrol.2018.01.061
Reference: HYDROL 22543

To appear in: Journal of Hydrology

Received Date: 18 November 2017


Revised Date: 3 January 2018
Accepted Date: 29 January 2018

Please cite this article as: Akter, A., Tanim, A.H., A modeling approach to establish environmental flow threshold
in ungauged semidiurnal tidal river, Journal of Hydrology (2018), doi: https://doi.org/10.1016/j.jhydrol.2018.01.061

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1 A modeling approach to establish environmental flow threshold in
2 ungauged semidiurnal tidal river
3

4 A. Akter1,2*, A. H.Tanim1,
1
5 Center for River, Harbor & Landslide Research (CRHLSR), Chittagong University of
6 Engineering & Technology (CUET) Chittagong 4349, Bangladesh.
2
7 Department of Civil Engineering, Chittagong University of Engineering & Technology (CUET)
8 Chittagong 4349, Bangladesh.
9 * Corresponding author. E-mail: aysha_akter@yahoo.com; http://aakter.weebly.com
10

11 Abstract
12 Due to shortage of flow monitoring data in ungauged semidiurnal river, ‘environmental flow’
13 (EF) determination based on its key component ‘minimum low flow’ is always difficult. For EF
14 assessment this study selected a reach immediately after the Halda-Karnafuli confluence, a
15 unique breeding ground for Indian Carp fishes of Bangladesh. As part of an ungauged tidal river,
16 EF threshold establishment faces challenges in changing ecological paradigms with periodic
17 change of tides and hydrologic alterations. This study describes a novel approach through
18 modeling framework comprising hydrological, hydrodynamic and habitat simulation model. The
19 EF establishment was conceptualized according to the hydrologic process of an ungauged semi-
20 diurnal tidal regime in four steps. Initially, a hydrologic model coupled with a hydrodynamic
21 model to simulate flow considering land use changes effect on streamflow, seepage loss of
22 channel, friction dominated tidal decay as well as lack of long term flow characteristics.
23 Secondly, to define hydraulic habitat feature, a statistical analysis on derived flow data was
24 performed to identify ‘habitat suitability’. Thirdly, to observe the ecological habitat behavior
25 based on the identified hydrologic alteration, hydraulic habitat features were investigated.
26 Finally, based on the combined habitat suitability index flow alteration and ecological response
27 relationship was established. Then, the obtained EF provides a set of low flow indices of desired
28 regime and thus the obtained discharge against maximum Weighted Usable Area (WUA) was
29 defined as EF threshold for the selected reach. A suitable EF regime condition was obtained
30 within flow range 25~30.1 m3/s i.e., around 10 - 12% of the mean annual runoff of 245 m3/s and
31 these findings are within researchers’ recommendation of minimum flow requirement.
32 Additionally it was observed that tidal characteristics are dominant process in semi-diurnal
Page 1 of 53
33 regime. However, during the study period (2010-2015) the validated model with those reported
34 observations can provide guidance for the decision support system (DSS) to maintain EF range
35 in an ungauged tidal river.
36
37 Keyword: tidal process, flow components, habitat simulation, ungauged rivers.
38
39 1. Introduction
40 ‘Low flow’ is a seasonal phenomenon during prolonged dry weather and could be resulted from
41 the human regulation on natural stream. The low flow of a stream could result adverse effect on
42 the entire ecosystem and ‘Environmental flow’ (EF) can be an approach to mitigate this issue.
43 EFs are defined as quantity, quality and timing of water flow to sustain estuarine and freshwater
44 ecosystem and maintain stakeholders’ livelihood (Brisbane Declaration, 2007). In two decades
45 EF management becomes a vital issue for the river management as flow restoration serves biotic
46 function to drive every step in the life cycle of the ecological habitat (Arthington et al., 2006,
47 Fanaian et al., 2015, Waddle, 2012, Poff and Zimmerman, 2010). To sustain freshwater
48 ecosystem at regional scale Ecological Limits of Hydrologic Alterations (ELOHA), an
49 ecologically coherent framework was developed for the effective EF management (Poff et al.,
50 2010, Richter, 2010, Richter et al., 2012). ELOHA comprises of two processes i.e. scientific and
51 social, to reinforce hydrologic alteration with ecosystem management, and confirmed social
52 participation for EF implementation. This flexible ELOHA framework can fit with a regional
53 scale application following four steps i.e. (a) basin wide hydrologic and ecologic database
54 development, b) identification of distinctive flow regime with hydrologic indices having
55 ecological significance, c) hydrologic alteration determination comparing current and idealized
56 habitat conditions, d) link up ecological response with obtained hydrologic alteration. Overall,
57 the ELOHA framework has dependence on extensive hydrologic and ecologic database although
58 those may often unavailable for ungauged stream (Poff et al., 2003). Regional regression analysis
59 can be applied for the ungauged stream while reference flow gauges are available with
60 continuous historical records (Zhang et al., 2016). However, applying this method is rather
61 difficult task in a trans-boundary tidal river basin with limited reference flow gauge information.
62 This hindrance might be overcome using a hydrologic model to investigate long term flow
63 characteristics with identification of ecologically relevant hydrologic indices (Richter et al.,

Page 2 of 53
64 2012). EF assessment particularly in an ungauged tidal regime faces a number of difficulties, for
65 instance, topographical datasets play a significant role in hydrodynamic results as well as
66 prediction accuracy of river discharge (Saksena and Merwade, 2015). On the other hand,
67 hydrodynamic process of tidal river is a complex phenomenon relies upon the upstream
68 freshwater inflow and downstream tide. The estuary ecology, flood routing and the relevant
69 parameters were governed by the periodic changes of water level or tide pattern viz. semidiurnal,
70 diurnal or mixed of both (Sun et al., 2015).
71
72
73 To conduct EF assessment, hydrologic model should consider land use change and
74 hydrodynamic model should pose responsive behavior in groundwater interaction with surface
75 water and therefore should consider seepage loss. Generally, EF assessments in a river carried
76 out using four methods i.e. representing in hydrological methodologies, habitat simulation
77 method, hydraulic rating and holistic methodologies. All of these methods are dependent on long
78 term flow monitoring whereas in a tidal river basin ecological processes are regulated by five
79 basic components i.e., magnitude, frequency, duration, timing and changing rate of hydrologic
80 conditions during both high and low tides (Poff et al., 1997, Aguilar and Polo, 2016). Numerical
81 models can be applied for a particular regime to avail those historical databases and Hydrologic
82 Engineering Center’s River Analysis System (HEC-RAS) is one of the frequently used models in
83 this regard. Recently, Hickey et al. (2015) introduced Ecosystem Functions Model (HEC-EFM)
84 to translate flow statistical results of HEC-RAS in an ecology function relationship to assess
85 aquatic habitat suitability of river. On the other hand, Mouton et al. (2007) presented an
86 integrated modeling approach combining a 1D hydraulic model based on the HEC-RAS software
87 and the fish habitat module of CASiMiR, a fuzzy logic-based eco-hydraulic modeling system, to
88 simulate and assess ecological effects of physical habitat changes in rivers. Due to capability of
89 backwater simulation Tsihrintzis et al. ( 2007) has adopted Storm Water Management Model
90 (SWMM) model for management decisions on coastal lagoon through evaluating alternative
91 scenarios to increase tidal flushing and water residence time. To evaluate spawning suitability of
92 Chinese sturgeon in Yangtze River basin in China, a generalized additive method with a 2D
93 hydrodynamic model showed well response (Yi et al., 2016). Recently, based on the fish
94 preference curves a combination of hydrodynamic and heat transfer model was coupled to obtain

Page 3 of 53
95 habitat suitability (Yao et al., 2015). An artificial neural network was also applied to forecast the
96 habitat suitability and EF assessment in coastal, harbor and river environment (Watts et al., 2011,
97 Olden et al., 2004). Unfortunately those data driven methods involved with flow regime
98 characteristics, are often fail to apply on ungauged streams with limited historical dataset. EF
99 estimation is also possible from synthetic streamflow time series in deterministic simulation
100 approach. In this approach, rainfall-runoff models are usually involved to convert catchment
101 rainfall to a continuous streamflow (Smakhtin, 2001). To develop EF standards of an ungauged
102 stream a hydrologic model is an ultimate requirement for providing reasonable estimation of
103 flow to serve as a source of required long term database.
104 Establishing hydrologic alteration linkage with ecological response can be done following the
105 biodiversity behavior and EF sensitivity. The behavior of ecological habitat like carp fish in tidal
106 regime has dependence on tidal cycle. Spawning of those species depends on high tide, low
107 discharge and moonlight conditions (Tsai et al., 1981). The water temperature and salinity
108 modify their migration pattern and they often prefer to a less tidal, brackish regime in late spring
109 for spawning (Lucas et al., 2001). Such ecological conditions can be prioritized in habitat
110 simulation model. Globally, among 207 methodologies of instream flow assessment in 44
111 countries, habitat simulation method is the second most frequently EF assessment method
112 (Tharme, 2003). This is the widespread adopted method applied frequently either in a knowledge
113 based or a data driven approach with quantitative results of combined suitability index ( ) and
114 introduced with application software like Physical Habitat Simulation System (PHABSIM) (Jung
115 and Choi, 2015, Aguilar and Polo, 2016, Fornaroli et al., 2016, Yi et al., 2014, Woo, 2010).
116 PHABSIM is capable to correlate hydraulic model with biological model that identifies suitable
117 habitat criteria based on the relationship of discharge and Weighted Usable Area (WUA) i.e.
118 surface area based on composite probability of  to indicate relative habitat suitability (Ayllón
119 et al., 2012). Alternatively, 2D hydraulic and habitat model like River 2D has been suggested in
120 channel bottom friction dominated scenario during low tide or low flow as an improvement of
121 EF prediction (Qin et al., 2016, Rivaes et al., 2015, Steffler and Ali, 2012, Lee et al., 2010).
122 Thus, considering sensitivity to microhabitat variables 2D models can perform better for spatial
123 distribution phenomena for WUA and  determination.
124 This study of EF assessment was primarily focused on its key component minimum flow
125 requirement of semi-diurnal Karnafuli River based on its ecological habitat for carp fish.

Page 4 of 53
126 Moreover, optimum meteorological factors along with the hydraulic conditions and water quality
127 parameters play crucial role in EF maintenance. Riverine ecology commonly impaired by
128 hydrologic alterations, but for tidal river the EF assessment and flow standards need further
129 intensive studies. This research aimed to develop a flexible modeling framework to establish
130 environmental flow threshold in ungauged semi-diurnal tidal regime through flow
131 characterization i.e high/low flow of tidal regime from the representative flow duration, and thus
132 finally obtain the optimum hydraulic conditions for suitable EF. Therefore, a comprehensive
133 modeling framework was required to establish following the fundamental EF assessment
134 concepts of ELOHA. So, initially a hydrologic model i.e. SWMM combined with hydrodynamic
135 model HEC-RAS to overcome several difficulties of existing EF assessment in ungauged streams
136 i.e. scarcity of long term flow characteristics, land cover/land use changes effect on streamflow,
137 seepage loss of channel, friction dominated tidal decay. Then, to define hydraulic habitat feature,
138 a statistical analysis on derived flow data was performed focusing key ecological habitat carp
139 fish. Furthermore, to characterize the ecological habitat behavior under several flow conditions
140 using habitat simulation model (PHABSIM and River2D) on WUA basis, those ecologically
141 relevant flow was examined and suitable hydraulic conditions were proposed. Finally, based on
142 combined habitat suitability index flow alteration and ecologic response relationship was
143 established.
144
145 2. Materials and Methods
146 2.1 Study area
147 Karnafuli River exists in the south-eastern part of Bangladesh, is the largest and most important
148 river in Chittagong region. Originating from the Lushai hills in Mizoram, India, the Karnafuli
149 flows 270 km southwest through the Chittagong Hill Tracts and Chittagong and finally disposes
150 to the Bay of Bengal. As per Bangladesh Meteorological Department (BMD) the annual rainfall
151 amount is 3000 mm of which 80% experienced during monsoon period (June-September).
152 According to the Chittagong Port Authority (CPA), on an average the width of the river is about
153 667 m, typical tidal range in a semi-diurnal period (12hr 28min) at the river estuary was observed
154 with tide level of 1.5m during low tide that reaches 5.5m during high tide. The water level
155 gradually passes through non tidal regime at Kaptai, located a distance of 62 km from the river
156 mouth, and show daily variation of 1.2m ~1.8m (Fig. 1). Based on semi-diurnal tidal nature, the

Page 5 of 53
157 hydrodynamic study considered lower 41 km of Karnafuli River and 12 km of lower parts of
158 Halda River, this covers a catchment area of about 310 km2. The hydrologic alteration of this
159 area was tested on the restoration site, selected at 32 km downstream of the Kaptai Dam upto
160 Karnafuli-Halda confluence based on the ecological indictors i.e. Major Indian Carps (Fig. 1).
161 So, the outlet discharge of Kaptai Dam has an influence on the aquatic community of this regime
162 and thus there is an influence on the lower part of the Halda River. This area usually treated as
163 nursery ground of major carps during spawning season (April-June) as those fishes migrates
164 from upper reach to this zone (Tsai et al., 1981; Akter and Ali, 2012). This is the only pure
165 Indian Carp breeding field in Bangladesh and possibly in South Asia. However, due to habitat
166 alteration the major carp production declined from 2470 kg in 1945 to 20 kg in the year 2004 and
167 then slightly increased to 47 kg in 2015 (MoEF, 2015). This declination is possibly due to
168 excess exploitation for supplying water to adjacent Chittagong city, rainfall pattern change
169 suitable climatic conditions for spawning, increasing salinity amount, changed geomorphology
170 by sand extraction and others.
171

Figure 1: (a) The redline showing watershed boundary in Chittagong region, (b) the watershed
and stream network (c) the spawning ground
172
173 2.2 Data preparation
174 To avail the historical flow time series for the spawning ground (Fig. 1), a combination of
175 rainfall-runoff and hydraulic model was adopted during the study period 2010-15. Hydrological
176 and meteorological parameters were derived to set up SWMM model (Table 1 and 2). The
177 rainfall-runoff model SWMM was engaged to acquire tributaries runoff contribution and then
178 applied as upstream boundary condition in HEC-RAS. Finally, tidal simulation in HEC-RAS was
179 performed with downstream boundary conditions of stage hydrograph.
180
181 Table 1: Details on data preparation for model setup
182
183
184 2.2.1 Hydrological parameters

Page 6 of 53
185 Hydrologic parameters were derived to incorporate in USEPA's SWMM. An extension of
186 Geographic Information System (GIS) called HEC-GeoHMS executed to delineate watershed
187 boundary and stream network through terrain preprocessing. Using ASTER GDEM as origin of
188 topography stream network was obtained through this process (Table 1). The area threshold
189 method was used to delineate watershed and stream network as the input DEM in a gridded
190 nature. The steps followed in HEC-GeoHMS terrain preprocessing was as: (a) DEM
191 reconditioning using Karnafuli River streamline, (b) sink filling, (c) flow accumulation grid
192 generation based on computed flow direction, (d) development of stream link grid based on flow
193 direction and stream grid, and (e) obtaining catchment grid along with associated stream link to
194 obtain watershed boundaries. Then, the model parameters i.e. slope (S) and imperviousness (I)
195 were derived from spatial data (Table 1) as well as from the available literatures (Table 2). The
196 width parameter refers to the width of overland flow path for sheet flow runoff and thus the
197 overland flow width FW (m) was determined as (Ahiablame and Shakya, 2016):
 =


198 (1)

199 Where A (m2) is the catchment area and L (m) is the length of the overland flow path, calculated
200 from obtained HEC-GeoHMS watershed analysis.
201
202 In this study, land use and land cover (LULC) map (Fig. 2a and b) were prepared
203 considering four land use criteria viz. vegetation, barren land (including cultivated lands without
204 crops), water body and built-up area through the supervised classification by ERDAS IMAGINE
205 software (Lillesand et al., 2008). In supervised classification, spectral signatures were collected
206 from specified locations (training sites) in the image by digitizing various polygons overlaying
207 different land use types. Spectral signatures were classified following the knowledge-based
208 expert classification systems depending on the reference Google earth maps to improve the
209 accuracy of the classification process. For a large study area this approach provides a reasonable
210 accuracy. Two satellite images (Table 2) were used to determine built-up area i.e., the
211 imperviousness percentage and curve number (Eq. 2). Kappa coefficient is an accuracy
212 assessment measure of predicted landuse pixel. Thus, the consistency between the classified
213 LULC map and reference maps (Google Earth images /Field data) represented by Kappa
214 coefficient, that was determined by classification error and found within the range between -1 to
215 +1 (Foody 2010). LULC in 2010 and 2015 were assessed by available high resolution Google

Page 7 of 53
216 Earth images on May 5, 2010 and April 4, 2015 respectively. Here, 30 random points for each
217 type of layer were sampled and compared with the available LULC. Thus, the overall accuracy
218 and Kappa Coefficient was respectively 81% and 0.79 for the LULC map 2010. Applying similar
219 approach for the LULC map 2015 overall accuracy and Kappa Coefficient was obtained 83.67%
220 and 0.82 respectively. However, a variation of barren land pixel was observed between two
221 LULC maps due to time period of image acquisition. From the field survey, it was reported that
222 the crop period of irrigation field responsible for these variation in vegetation and barren field
223 pixel. The mean slope % of each watershed has determined using zonal analysis (mean) in Arc-
224 GIS spatial analysis.
225
Figure 2: Hydrologic model (SWMM) input a) Landuse map of 2015, b) Landuse map of 2010, c)
Watershed wise mean slope analysis. ‘W’ denotes with corresponding watershed number
226
227
228 Table 2: SWMM model parameters
229
230 The Soil Conservation Service (SCS) method was selected for analysis as (i) this is the
231 commonly used method in watershed to quantify landuse effect on runoff and provides
232 reasonable results, (ii) its calculation is made easier by the fact that only a few variables need to
233 be estimated (hydrologic soil group, land use and slope) and (iii) despite its simplicity, it yields
234 reasonable results compare to the complex models (Lastra et al., 2008). Modified SCS equations
235 (Eq.2) to suit Indian subcontinents conditions were used in this study as (Suriya and Mudgal,
236 2012):
∑ CN × A 
CN =
∑ A
(2)

237 CNaw=the area weighted curve number of each sub-watershed


238 CNi = the curve number of each land use soil group in sub-watershed (Table 3)
239 Ai = area of each land use
240
241 Table 3: SWMM sub-watershed properties
242

Page 8 of 53
243 For the study period (2010-2015) daily meteorological data i.e., precipitation,
244 temperature, humidity, sunshine and wind speed were obtained from the Potenga Meteorological
245 Station (Fig. 4). CROPWAT 8.0 software was engaged in this study to calculate
246 Evapotranspiration (ETo) using Penman-Monteith Method (Eq.3).
0.408∆  −  + " %&'() *' +, − +- 
#$$

 =
∆ + " 1 + 0.34*' 
(3)

247 Where,
248 ETo = reference evapotranspiration [mm/ day], Rn = net radiation at the crop surface [MJ m-2
249 day-1], G = soil heat flux density [MJ m-2 day-1], T = mean daily air temperature at 2 m height
250 [°C], u2 = wind speed at 2 m height [m/s], es = saturation vapour pressure [kPa], ea = actual
251 vapour pressure [kPa], es–ea = saturation vapour pressure deficit [kPa], ∆ = slope vapour pressure
252 curve [kPa/°C], γ = psychrometric constant [kPa/ °C].
253
254 Thus, a daily time series of evaporation (mm/day) was obtained as well as the effective rainfall.
255
256 2.2.2 Hydraulic properties:
257 Channel topography of stream network was exchanged from DEM to prepare hydraulic
258 input using HEC-GeoRAS preprocessing. For hydrodynamic modeling using HEC-RAS, total 32
259 cross sections based on continuous field survey by CPA were engaged (Table 1). Due to the low
260 resolution (i.e. 30 m) of DEM, all cross sections were reconstructed based on surveyed cross
261 sections by the BWDB and CPA. Cross sections were assumed representative as uniform within
262 1000 m in tributaries. In pre-processing, the acquired DEM was used as grid layer and processed
263 to create channel geometry as this was required in SWMM. Cross sections were digitized
264 perpendicular to the path of flow in the channels. The bank lines were manually checked with
265 Google Earth as base map and they are manually relocated.
266
267 For channel seepage loss causes significant amount reduction of discharge. Aquifer to
268 channel flux was ignored as the groundwater table was assumed below the river bed. Moreover
269 the Karnafuli River is a perennial stream and so groundwater flux in streamflow can be ignored.
270 Seepage loss rate in each river cross section was computed using Eq.4 and then further converted
271 to seepage loss rate as per Eq.5 (Garg, 2007):

Page 9 of 53
B + D'/)
∆Q =
(4)
200
272 Where ∆Q = channel losses, m3/s per km length of channel
273 B = Bed width of the channel, m
274 D = Depth of water in the channel, m
275
∆Q × 3600 × 1000
q =
(5)
P×L
276 q is the seepage loss rate (mm/hr); P is wetted perimeter of the Channel (m) ; L is the conduit
277 length (m)
278
279 As the tide pattern of the Karnafuli River is semi-diurnal, 6 hr interval water level time series
280 were used as downstream boundary condition in the HEC-RAS during 2010-2015 (Table 1). To
281 account the river flow from upstream areas of watershed boundary at node CS-1 (Fig. 4), a 6 hr
282 discharge time series were collected (Table 1).
283
284 3. Model Set up
285 For the selected study area SWMM model was preferred to derive rainfall hydrograph and to
286 conduct hydrological routing for tributaries as tidal effect is not significant. Hydrological routing
287 model is advantageous where morphology varies at small scale and can also be used at larger
288 scale with limited topography information (Fekete et al., 2001). Thus, to observe the ecological
289 responses with hydrologic alteration a hydrologic model SWMM was used to derive upstream
290 contribution over river flow. For the Karnafuli River, 1D hydrodynamic model HEC-RAS was
291 used due to significant tidal effect (Fig.4). Again in this condition a detailed study on a
292 hydrological model would provide misleading information as these models assume uniform
293 velocity or a constant velocity throughout the large conduit, ignoring topographic variation
294 (Saleh et al., 2013). The dynamic mechanism of unsteady flow is also ease in simulation in a
295 hydrodynamic model. HEC-RAS unsteady flow simulation was carried to avail the sensitive
296 flow components. Then, the bidirectional hydrodynamic process of tidal river was analyzed
297 separately as unidirectional river while habitat simulation method was also engaged i.e.
298 PHABSIM and River 2D. Overall model set up process was shown in Fig. 3 and the measured
299 data was detailed in Table 1.

Page 10 of 53
300
301 3.1 Hydrologic model:
302 Most of the parts of Karnafuli River are ungauged and the available gauge data already involved
303 in the hydrologic model, so it is advantageous to integrate a rainfall runoff model with a
304 hydrodynamic model. The preliminary purpose of this integration is to define the upstream
305 boundary condition of hydrodynamic model HEC-RAS.
306
Figure 3: A schematic overview of model set up process
307
308 USEPA SWMM is a link-node model, used to route in the tributaries and thus it can
309 drive inflow hydrograph at desired gauge location of River (Tsihrintzis et al., 2007). Application
310 of SWMM found in real time optimal behavior as well as flood hazard risk mapping (Ilker et al.,
311 2008; Mohammad et al., 2016). The model has a wide variety of application including
312 hydrologic and hydraulic capabilities i.e., flow routing through closed and open conduit
313 networks of unlimited size, backwater effect simulation, surface ponding, flood modeling and
314 flood reduction with Low Impact Development (LID) (Kelly and James, 2009, Rossman, 2010).
315 Kenneth et al. ( 2008) described an automatic calibration method of SWMM model for large
316 watershed incorporating GIS with SWMM. This optimization method was applied to determine
317 the hydrologic parameters and described in section 2.2.1. The topography of the study area was
318 created by combining stream lines defined by HEC-GeoHMS, the watershed boundary was
319 created using HEC-GeoHMS and river center lines, cross section cut lines, bank lines those were
320 predefined by HEC-Geo RAS (Fig. 4). The inputs to the model are: (a) the geometry with the
321 numbering nodes and links as HEC-GeoRAS, (b) For each conduit lengths, maximum depths,
322 invert elevation, (c) Manning’s roughness coefficient, nc for each conduit, and (d) the
323 bathymetric information acquired from field survey (e) the inflow hydrograph at node CS-1(Fig.
324 4) to specify control (Fig. 3) of relative runoff contribution of from upstream areas to river flow
325 as Karnafuli has its origin outside watershed boundary.
326
327 Hydrological parameters (imperviousness, depression storage, width and Manning’s co-
328 efficient for overland flow, % slope) were incorporated in the model interface. Dynamic wave

Page 11 of 53
329 routing method with a time steps of 30s was used to simulate the model for 6 year daily rainfall
330 time series (i.e., 2010-2015).
331
Figure 4: Coupled hydrologic and hydrodynamic model showing applied boundary conditions
on HEC-RAS model.
332
Figure 5: Hydrograph obtained from SWMM
333
334
335
336 3.2 Hydrodynamic model
337 A tidal river flow can rarely appear in steady condition, thus HEC-RAS unsteady flow analysis
338 was carried out for tide simulation of Karnafuli River. An inaccurate choice of time steps and
339 cross section spacing is the possible reason for model instability (Baldassarre et al., 2009).
340 Samuels (1989) suggested a guideline for optimal cross sections, i.e.:
∆: ≈ <= (6)
341 Where, B is bankfull surface width of the main channel (m); and k is a constant ranges from 10
342 to 20. To obtain optimal number of cross sections for unsteady flow simulation based on an
343 estimation of backwater flow (Samuels, 1989) as :
@ABC D
∆x < 0.2 D ≈ 0.2 E When  ' → 0
F (7)
E

344 A uniform cross section spacing of an interval 1000 m satisfies Eq. 7 for HEC-RAS river
345 geometry. Brunner ( 2006) described a method to select time steps (∆H) for medium to large
346 rivers as:
J
∆H ≤
(8)
20
347 Where, Tr is the time of rise of the flood wave (hr). Several time steps in incremental order (0.5,
348 1, 1.5, 2, 2.5, 3, 4 hr) were tested. Model stability was obtained at a time step of 0.5 hr and
349 chosen as model time steps.
350
351 As earlier node-link set to compute the hydrograph as previous steps, for the intended modeling
352 only Karnafuli River was considered in HEC-RAS platform. The hydraulic modeling in HEC-

Page 12 of 53
353 RAS needs upstream hydrograph as boundary condition and SWMM was coupled with the
354 model for producing runoff hydrograph (Fig. 5) to the connected tributaries. To ensure
355 homogenous hydraulic routing and channel geometry in both models i.e., SWMM and HEC-
356 RAS environments, the cross section spacing was tested at similar regular intervals (1000 m) for
357 both tributaries and main stream. Contraction and expansion coefficients were considered as 0.1
358 and 0.3 respectively following Brunner ( 2006). The simulation of hydrodynamic consists of 2
359 parts:
360
361 a) firstly, the HEC-RAS model run using steady state analysis to prepare initial
362 boundary condition of HEC-RAS unsteady flow; and
363 b) then the HEC-RAS model run for unsteady stage, the upstream boundary condition
364 was defined by flow hydrograph at cross section CS-1 and lateral inflow hydrographs
365 (Fig. 5) at the tributaries, those are connected to main stream (Fig 4). The
366 downstream boundary condition was defined as stage hydrograph at outfall (Fig. 4) to
367 simulate tidal flow during 2010-2015.
368
369 3.3 Habitat Simulation model
370 A number of factors i.e., water temperatures, water depth, substrate size, flow velocity,
371 suspended sediment concentration need to be considered as ecological habitat factor. One
372 dimensional PHABSIM and two dimensional River 2D (http://www.river2d.ualberta.ca) were
373 selected for habitat simulation. Both models are based on a hydraulic model that uses preference
374 curves considering eco-factors such as water depth, water temperature, flow velocity and
375 substrate size. River 2D model comprise of continuity equation and two components of the
376 momentum vector (Eqs. 9 ,10 and 11) (Steffler and Blackburn, 2002):
∂H ∂qN ∂qO
+ + =0
(9)
∂t ∂t ∂t
∂qN d ∂ g ∂ ' 1 ∂ ∂
+ UqN  + @VqO D + H = gH S$N − SVN  + X HτNN Z + @HτNO D
(10)
∂t dx ∂y 2 ∂x ρ ∂x ∂y
∂qO d ∂ g ∂ ' 1 ∂ ∂
+ @UqO D + VqN  + H = gH@S$O − SVO D + X @HτON DZ + @HτOO D
(11)
∂t dx ∂y 2 ∂y ρ ∂x ∂y
377 Where, H is average depth (m ); U and V are the average velocity (m/s) along x and y axes
378 respectively ; S0x and S0y are the frictional slope values; Sfx and Sfy are the frictional resistance

Page 13 of 53
379 values; qx and qy are the discharge components in the respective direction and τNN , τNO , τON , τOO
380 represent horizontal shear stress in different directions.
381 For this study, the bathymetry of spawning ground (Fig. 1) at a spatial scale varying 10-
382 30 m collected from 4 survey points (Table 1). To set up River 2D model the restoration reaches
383 were divided into 568 elements and 370 nodes to form a triangular irregular network (TIN). Such
384 finite element based model predicts the location of water edges from interpolation of three points
385 of adjacent TIN elements. A new node of interpolated bathymetry inserted in the areas of
386 significant topographic variations viz. steep banks, pools etc. This approach can orient smallest
387 elements on narrow necked down cross-sections and the largest TIN elements on the large pool
388 following Holmquist and Waddle (2013). A number of discharges obtained from the Flow
389 Duration Curve (FDC) were defined at inflow sections and depth-discharge relationship was
390 derived in outflow locations (Fig. 8). Such approach reduces the varying discharge effect on the
391 outflow section. During modeling, contribution of groundwater and tributaries losses were
392 ignored, as those were already accounted in the hydrologic modeling i.e. SWMM.
393
394 Similar cross section orientation was adopted both in HEC-RAS and PHABSIM considering
395 streams as rectangular cells with the constant depth and flow velocity. According to Wu and
396 Wang (2006), this assumption maintains parallel streamlines and neglect transverse eddy. As
397 flow characteristics i.e., flow depth, flow velocity and stage–discharge relationship were
398 obtained from HEC-RAS, these can be used for hydraulic modeling in PHABSIM. Mean column
399 velocity method was used to simulate cell velocity of the selected reach. Advantage of using this
400 model is the incremental analysis that can predict and quantify the impact of hydrologic
401 alterations on targeted species. Nikghalb et al. (2016) had developed an approach for HEC-RAS
402 1D channel average velocity to distribute for each cell in the PHABSIM. The cell velocity
403 (Vi)mod obtained from the average velocity (Eq. 12) the associated parameters are the Manning’s
404 roughness coefficient (n), hydraulic radius of each cell (HD or R) and weighted discharge (Q[ ) of
405 each cell :
406
∆Q
V \]^ = V +
(12)
nA

Page 14 of 53
C
bbbb
HD d
(13)
V = V̀ × a c
HD

Where HD = R = g ≈ h ≈
f f f
i

∆Q=Q[ ~Q klm n


Q = o V × A
(14)
[



407 Here, V is average velocity in each subsection having cross-sectional area A and hydraulic
408 radius HD . Q klm n is actual discharge, T is top width, b is width of bottom and P is effective
409 perimeter.
410
411 3.3.1 Estimation of Habitat Suitability Index (HSI)
412 Three species of the Indian major carps, Catla (Catla catla), Mrigala (Cirrhinus mrigala) and
413 Rohi (Labeo rohita), spawn in this region and reported as key ecological indicator. Thus,
414 environmental flow of the selected regime was investigated based on the habitat suitability of
415 those species. Fish and other aquatic organisms usually susceptible to a certain ranges of stream
416 velocity, water depth and bed substrate. Eco-hydraulic model i.e., PHABSIM have a key
417 assumption that consider habitat suitability of a regime based on the velocity, depth and substrate
418 size and the representative index are known as Habitat Suitability Index (HSI). Tsai et al. (1981)
419 investigated hydrological conditions and water quality conditions during spawning of major
420 carps at lower parts of the Halda River and from this study the Habitat Suitability Curve (HSC)
421 for the targeted species was developed (Fig. 6). HSI were obtained from the HSC curve (Fig. 6)
422 for the targeted species on a scale of 0 to 1(Im and Kang, 2011). HSI for the three carp species
423 was assumed similar from previous study recommendation (Akter and Ali.,2012; Lucas, et al.,
424 2001). The substrate size was determined from field survey and these can be engaged in channel
425 index determination. Field survey was conducted during spawning season at low tide period on a
426 sunny day. To assign the channel index the substrate size was visually determined at spatial scale
427 varying 10-30m. Those nodes locations were identified from common points of bathymetry.
428
Figure 6: Habitat Suitability Curves of major carp fishes based on (a) depth, (b) flow velocity,
and (c) channel index
429

Page 15 of 53
430
431 3.3.2 Estimation of ecological flow rate
432 Minimum environmental flow was obtained from the PHABSIM using HSCs that relate habitat
433 indicators, i.e., the WUA with the hydraulic variables (water depth and stream velocity). Then,
434 the optimum ecological flow rate was estimated through flow-WUA relationship following
435 Aguilar and Polo (2016). The WUA can be obtained by multiplying the area of each cell by its
436 suitability index value, as:


WUA = o A × C
(15)



437 Where q is an area of cell i and  is the combined HSI of cell i.


438 The combined HSI,  , is a suitability criterion with a combination of HSI of depth, velocity and
439 substrate size, i.e.,
 = γ v  × γ d  × γ c  (16)
440 Where, " u  γ d  and " v  are HSI function of cell i regarding velocity, depth of flow and
441 substrate size respectively.
442
443 3.4 Statistical analysis of the model outcome
444 SWMM generated hydrograph at tributary CST-1 was validated using available dataset from
445 BWDB stream gauge station (Fig. 5). Observed water level from the 2 tidal gauge stations were
446 collected and compared to evaluate model performance. Those stations were Khal no 18 (CS-32)
447 and Khal no 10 (CS-27) and they located respectively 5 km and 11 km from the mouth of estuary
448 (Fig. 4). Model performance was evaluated using a matrix of error statistics, the criteria to
449 evaluate the hydrologic and hydraulic model performances were: Percent bias (PBAIS), Root
450 Mean Square Error (RMSE), RMSE–observations standard deviation ratio (RSR) and Nash-
451 Sutcliffe efficiency (NSE) as:
∑z
 x − xy 
PBAIS = 100 ×
(17)
∑z x,

|∑z
 x − xy 
' (18)
RSR =
|∑z b '
 x − x 

Page 16 of 53
1
(19)
RMSE =  X\ − X] '
N

∑
 X  − X  
] \ '
NSE = 1 −
(20)
∑ bbb] '
 X  − X 
]

Here, Mean observed value, bXbb] =  ∑


 X 
 ]
452

453 N is the total data points of field observations, X\ is the modeled value and X] is the observed
454 value; the negative mean error indicates the model under-predicted the water level. This
455 statistical analysis is used to characterize the quality and evaluation of model performance.
456
457
458 Figure 7a: Comparison of simulated and Observed water levels at Khal no 18 obtained from
459 HEC-RAS
460
461 Figure 7b: Comparison of simulated and Observed water levels at Khal no 10 obtained from
462 HEC-RAS
463
464 Figure 7c: Comparison of Simulated and Observed discharge at CS-17 obtained from HEC-RAS
465
466 Figure 7d: Comparison of simulated and observed discharge at CST-1 obtained from SWMM
467
468
469 Table 4: Summary of HEC-RAS performances at available stations
470
471 NSE value ranges from -∞ and 1 and a perfect match between observed and simulated value is 1.
472 Hence, the model predicted flow show reasonable agreement as all stations show NSE above 0.9.
473 For validation, NSE value of two tide gauges stations i.e. CS-32 and CS-27 shows a decrement
474 from 0.99 to 0.93 indicates that water level prediction towards inland show slight deviation.
475 PBAIS values show the average tendency of the simulated value and found within ±10 (Table 4).
476 At flow gauge CST-1 positive value represents bias towards overestimation and negative value at
477 CS-17 indicates underestimation. This can be concluded that the SWMM prediction has slightly

Page 17 of 53
478 overestimation on flow and HEC-RAS underestimated discharge at selected gauge within
479 allowable limit. Sediment transport change cross sectional value by erosion or deposition,
480 showed influence on the simulated water levels or low flows prediction. Following Saleh et al.
481 (2013), this can be concluded that the topography obtained from the field survey should improve
482 at regular time interval considering geomorphological change. Thus, the model could predict
483 water level and discharge reasonably
484
485
486
487
488
489 4. Results
490 4.1 Estimation of Flow Duration Curve (FDC) and selected discharge as hydraulic habitat
491 feature
492
493 Karnafuli River has prominent semi-diurnal tide effect and from HEC-RAS simulation it was
494 observed that discharge reduce rapidly towards inland for instance during simulation period
495 maximum flow was found 4000 m3/s at outfall and it was reduced to 250 m3/s at CS-10, located
496 27km away from mouth (Fig. 4). Due to semi-diurnal fluctuations, during flood tide saline water
497 flow from sea to the inland. Thus flow in this direction was assumed as positive and opposite
498 appeared as negative flow (Fig. 7c). Establishing a stage discharge relationship in a tidal river is
499 difficult, as it usually took form of looped curve and for fewer water level fluctuation and
500 unidirectional analysis it can be adopted for a flood or ebb tide. The spawning ground was
501 located at a distance of 27 km from the sea mouth, stage –discharge relationship can be
502 established as water level ranges from 2.5-4 m. Flood tide was considered due to a unidirectional
503 flow is a prerequisite condition for PHABSIM hydraulic modeling. Flow Duration Curve (FDC)
504 during flood tide was obtained from a period of simulated flow (positive discharge) against a
505 cumulative percentage of flows. In an ungauged tidal river observed time series are less available
506 and flow variability is high, thus, simulated flow time series are desirable to derive a FDC. To
507 cover entire range of flow of a representative regime, FDC can provide a holistic concept from a
508 long term flow variability as well as unknown flow occurrence probability. However, HEC Data

Page 18 of 53
509 Storage System (HEC-DSSv2) model was used to perform statistical analysis from a flow time
510 series obtained from HEC-RAS during flood tide period. Using the math function of DSS tool
511 the discharges during 2010-2015 for the corresponding regime ranked according to the percent
512 flow exceeds or equaled and finally FDC curve were derived (Fig. 8). From FDC curve during
513 high tide the flow ranges were determined and a series of flow alternates were produced for
514 determination of suitable EF. Several discharges were selected and relative information about
515 their velocity and depth obtained from stage- discharge and velocity-discharge relationship
516 establishing reasonable correlation co-efficient obtained from flood tide (Table 5).
517
518
519
520 Figure 8: Flow Duration curve at different cross section of spawning ground
521
522 Table 5: Selected discharge for habitat simulation in spawning ground.
523
524 4.2 Estimation of Ecological Flow Rate
525 The obtained flow range was simulated in PHABSIM and River 2D to determine suitable EF
526 against the optimal WUA. Suitable discharge was obtained from PHABSIM that can be
527 determined for spawning season of major carps was 30.1 m3/s (Fig. 9 and Fig 10c). On the other
528 hand, River 2D predicts the optimum EF as 25 m3 /s. With the increase of discharge several parts
529 of reach attain high velocity and water depth and the spawning ground gradually lost the
530 appropriate situation of spawning except near bank as there are less water depth and velocity
531 (Fig. 10 d). Extreme flow event like flooding discharge interrupt spawning suitability of major
532 Indian carps. Initially, the WUA increases with discharge and while the maximum discharge of
533 25m3/s obtained this implies the suitable instream flow to maintain healthy ecology (Fig. 9).
534 From the selected discharge (Table 5), WUA found decreases except two flow case i.e. 75m3/s
535 and 200m3/s. The Ci at different discharge reveals the formation of different sizes riffle make this
536 spawning ground more suitable during low discharge (Fig. 10a, 10b and 10c).
537
538 Figure 9: Variation of Weighted Usable Area with discharge for Indian Major Carps
539

Page 19 of 53
Figure10: Combined Habitat Suitability Index obtained from River 2D in spawning ground
with incremental discharge
540
541 It has been already detected the flow components like magnitude, duration, frequency,
542 timing of low flows are closely related with different aquatic process of eco-hydrology. For
543 stream flow alterations hydrologic model SWMM was applied considering anthropogenic
544 changes like land use that are frequently ignored in conventional hydrologic modeling. There is
545 some discrepancy while low flow simulation as model overestimates low flow stage. This is due
546 to most of the hydraulic model is based on flood wave routing thus low flow event simulation
547 performances are not completely reliable. Multi-objective calibration approach of hydrologic
548 model can be applied for stream flow generations to capture entire flow regime characteristics.
549 Although all cross sections were acquired from BWDB and CPA surveyed dataset, stage and
550 discharge prediction from HEC-RAS was affected at different stations possibly due to sediment
551 deposition of those cross-sections. As a result, cross sections got shortened and predicted
552 discharges were less than actual discharge.
553
554 One of the aims of the study was to define long term flow time series and flow ranges of
555 ungauged tidal Karnafuli River where flow monitoring was not available. To determine the flow
556 ranges this model can also be applied for unidirectional river after calibration. The flow ranges of
557 flood tide (15.1 m3/s to 245 m3/s) were obtained more than the ebb tide period (23.1 m3/s to
558 187.5 m3/s). Due to inability to simulate bidirectional flow in PHABSIM, EF analyzed
559 considering flow ranges of flood tide as unidirectional river. WUA decreased sharply after the
560 discharge 30.1 m3/s and with the increase of discharge corresponding velocity and depth
561 increases which made Ci less. This resulted unsuitable flow for the major carps spawning. Less
562 turbulent place might be preferable to the major carps and areas of deep pool during spawning
563 season which located within 1000 m from upstream (Fig. 10a). Water depth, velocity and
564 substrate size were considered as dominant eco-factor for determining suitable ecological flow
565 rate. Other eco-factors i.e. water temperatures, suspended sediment concentration etc. was not
566 considered in model development.
567

Page 20 of 53
568 Variations in habitat predictions between PHABSIM and River2D habitat simulation model may
569 be occurred due to longitudinal variation of depth, velocity or velocity distribution while model
570 simulation. Again, inadequate substrate mapping may influence the obtained results from River
571 2D. Preference curve method of habitat simulation using PHABSIM and River 2D was preferred
572 in this study, this needs further development while engaging in EF assessment. Although this
573 method sensitive to preference curve and usually showed high HSI but variation in same trend
574 relatively other method such as generalized additive model, data-driven fuzzy logic model (Yi et
575 al., 2016). Thus, prediction of EF thresholds might not be affected by this method. Previous
576 studies reveal that the human activity with different hydraulic structures like dam, sluice gate,
577 bridges etc. influences the ecosystem (Yao et al., 2015, Sun et al., 2015, Boavida et al., 2015).
578 Thus, the recommended EF (25~30.1 m3/s at depth of 3.5-4.2 m) from long term stage-discharge
579 relationship has to maintain without any human interruption. Moreover the optimal
580 meteorological, hydrodynamic and several water quality parameters were suggested to maintain
581 EF (Table 6)
582
583 Table 6: Optimal conditions for targeted carp fish spawning to achieve EF
584
585 4.3 Kaptai Dam operational influence on habitat conditions
586 Kaptai Dam is an earth fill dam, covering a reservoir area of 777 sq. km and control annual
587 inflow runoff of approximately 15646 Million cubic meters with its 16 gated spillways (Noman
588 1997). From estimated FDC curve two high flows (200 and 225 m3/s) was tested to represent
589 peak released discharge impact on habitat condition (Fig. 10g and 10h). WUA was quantified
590 within 48000-61000 m2 for high upstream flow i.e. 200 m3/s, after then, a sharp decrease of
591 WUA was observed against further increment of discharge (Fig 10). The dam released discharge
592 distributed partially through Ichamoti River, a tributary of Karnafuli River and several
593 stakeholders i.e. Water treatment plants, Karnafuli irrigation projects and numerous industries. It
594 was difficult to determine the direct impact of dam released discharge on restoration site, as the
595 study area was 32 km downstream of Kaptai dam, unknown amount of water withdrawal of those
596 stakeholders from released discharge and inadequate data. Moreover due to absence of semi-
597 diurnal effect and dominant non-tidal regime this portion of Karnafuli River was beyond the
598 scope of this study, although indirect dam impact was considered to achieve flow hydrograph at

Page 21 of 53
599 CS-1 (Fig. 4, Table 1) as upstream discharge contribution. Existing reservoir operation rule are
600 divided into two parts; first part (August-October) utilizes the reservoir for the storage period to
601 refill the reservoir at potential level (33.4m above Mean Sea Level) for power generation and the
602 rest for the drawdown period November through July (Fig 11). During the spawning period of
603 carp fish (April-June) peak released discharge was found 400-450 m3/s and it can disperse the
604 restoration site providing fish access to floodplain. Present dam operation mode was set with an
605 objective to generate hydroelectric power and protect the downstream from riverine flooding.
606 Hence the integration of operation rule curve needs improvement to resemble suitable ecological
607 habitat condition. Hydrologic alteration induced by dam can be indicated by low flow or high
608 flow varied from their occurrence frequency, duration and pulse (Richter and Thamos, 2007).
609 Rapid fluctuations of water level have negative consequences on the brood egg, washing away
610 from spawning ground. Furthermore, such high discharge can be recommended after prolonged
611 dry weather to assist this restoration site reducing salinity amount and other water quality
612 parameter at suitable level. Dam released discharge should control during high tide to avoid the
613 high flow in restoration site. If the hydropower dam can be operated in daily basis at the same
614 rate of Kaptai reservoir inflow, the EF could be unchanged due to sudden high released
615 discharge.
616
617 Figure 11: Peak monthly released discharge from Kaptai reservoir
618
619 5. Conclusions
620 This EF assessment was carried with a modeling framework comprised of hydrological,
621 hydraulic and habitat models those are commonly used for water management. The model
622 framework can be applied in both unidirectional and tidal river after calibration and validation.
623 Moreover habitat simulation methods required considering more abiotic factors i.e. water quality
624 and meteorological conditions to predict EF ranges. To achieve EF optimum habitat conditions
625 were suggested combining present study and expert recommendations (Table 6). Fundamental
626 concepts of ELOHA Framework were followed, but conceptualized following hydrologic
627 process of an ungauged semi-diurnal tidal regime. Model performance was reasonable (below
628 10% error) and discrepancies probably arose from proper tidal simulation. Despite these
629 achievements few attempts were suggested for future EF assessment, as:

Page 22 of 53
630 • For sparsely gauged regions establishment of ELOHA framework is a concerning issue
631 due to prediction accuracy. Thus, regionalization of stream flow characteristics was done
632 on physical characteristics of watershed. Additionally prediction uncertainties can be
633 reduced by multi-objective optimization for hydrologic process (Zhang et al., 2008). As,
634 flow characteristics in semi-diurnal regime usually dominated by tidal characteristics, a
635 hydrodynamic model needs to integrate while simulating overall processes.
636 • In this study the flow regime characteristics of tidal Karnafuli River were derived with a
637 modeling framework, principally dominated by tidal nature. The method described in this
638 study will facilitate a model-based study to obtain historical flow that can be employed as
639 an assessment tool of ungauged river basin management. In fact the flow regime of tidal
640 river can be characterized by a spring tide and a neap tide occurred twice in a lunar month
641 and considering tidal variation in a year may be sufficient for simulation period.
642 Uncertainties should consider from extreme events i.e., storm surge and sea level rise as
643 the tide range has prominent influence over those source.
644 • As some parts of the watershed were already urbanized, the SWMM model was employed
645 to simulate the relative contribution of upstream runoff. For low flow simulation, the
646 modeling framework also consider the seepage loss from channel, land use change and
647 bottom friction domination in tidal decay. Applying a hydrologic model showed
648 reasonable representation in hydrologic alteration in this study.
649 • The statistical analysis on flow components was carried out to define the range of flow
650 regime. Finally, PHABSIM and River 2D were used to determine EF based on the
651 preference curve method. The minimum instream flow requirements for major carps
652 spawning was found 25~30.1 m3/s at depth of 3.5-4.2 m. The simulated EF during the
653 flood tide is 25~30.1 m3/s i.e., around 10 - 12% of mean annual runoff viz. 245 m3/s. This
654 outcome reasonably match with the minimum flow requirement as 10% flow of the mean
655 annual runoff defined by Acreman et al. (2000) for the managed flood and river flow .
656 • From water quality perspective the fish spawning of carp fish found sensitive to salinity
657 concentration. During dry weather the salinity intrusion may be reduced from the released
658 discharge of dam but should not exceed flow 100 m3/s. For habitat simulation River2D
659 considers spatial flow variability and prediction of habitat availability performed better

Page 23 of 53
660 compare to the PHABSIM. The prediction accuracy of habitat simulation requires
661 improvement using frequent fish sampling and survey on predicted hotspot.
662 These areas of improvement can briefly synthesized by two categories i.e. tidal hydrodynamic
663 and required dataset includes bathymetry, streamflow and carp fish habitat related ecological
664 information. Water managers and scientific community can work together to overcome those
665 limitations.
666
667 Acknowledgements
668 The research was supported by funds provided by Center for River, Harbor and Landslide
669 Research (CRHLSR), Chittagong University of Engineering and Technology (CUET),
670 Bangladesh. We express our gratitude to BWDB, CPA, BMD, IWM for providing bathymetric,
671 meteorological and validation data during the study.
672
673 References:
674 ACREMAN, M., FARQUHARSON, F., MCCARTNEY, M., SULLIVAN, C., CAMPBELL, K.,
675 HODGSON, N., MORTON, J., SMITH, D., BIRLEY, M., KNOTT, D., LAZENBY, J.,
676 WINGFIELD, R. & BARBIER, E. 2000. Managed flood releases from reservoirs: issues
677 and guidance. report to Department for International Development & World Commission
678 on Dams, Centre for Ecology and Hydrology, Wallingford, UK.
679 AGUILAR, C. & POLO, M. J. 2016. Assessing minimum environmental flows in nonpermanent
680 rivers: The choice of thresholds. Environmental Modelling & Software, 79, 120-134.
681 AHIABLAME, L. & SHAKYA, R. 2016. Modeling flood reduction effects of low impact
682 development at a watershed scale. Journal of environmental management, 171, 81-91.
683 AKTER, A. & ALI, M. H. 2012. Environmental flow requirements assessment in the Halda
684 River, Bangladesh. Hydrological Sciences Journal, 57, 326-343.
685 Aysha A. & Tanim A. H. 2017. Response of Salinity Distribution in River network with Tidal
686 Circulation in Well-Mixed Estuary. (under review)
687 ARTHINGTON, A. H., BUNN, S. E., POFF, N. L. & NAIMAN, R. J. 2006. The challenge of
688 providing environmental flow rules to sustain river ecosystems. Ecological Applications,
689 16, 1311–1318.
690 ARTHINGTON, A. H. & ZALUCKI, J. M. (eds.) 1998. Comparative Evaluation of
691 Environmental Flow Assessment Techniques: Review of
692 Methods. (Authors – Arthington, A.H., Brizga, S.O., Pusey, B.J., McCosker, R.O., Bunn, S.E.,
693 Loneragan, N.,
694 Growns, I.O. & Yeates, M.) .
695 ASCE & WEF 1992. Design and construction of urban stormwater management system. ASCE
696 Manual and reports of engineering practice, No.77.
697 AYLLÓN, D., ALMODÓVAR, A., NICOLA, G. G. & ELVIRA, B. 2012. THE INFLUENCE
698 OF VARIABLE HABITAT SUITABILITY CRITERIA ON PHABSIM HABITAT
699 INDEX RESULTS. River Research and Applications, 28, 1179-1188.
700 BALDASSARRE, G. D., CASTELLARIN, A., BRATH, A. & BATES, P. D. 2009. Optimal
701 Cross-Sectional Spacing in Preissmann Scheme 1D Hydrodynamic Models.

Page 24 of 53
702 BOAVIDA, I., SANTOS, J. M., FERREIRA, T. & PINHEIRO, A. 2015. Barbel habitat
703 alterations due to hydropeaking. Journal of Hydro-environment Research, 9, 237-247.
704 BRUNNER, G. W. 2006. HEC-RAS, River Analysis System User’s Manual, Version 4.0 Beta.
705 U.S. Army Corps of Engineers.
706 Brisbane Declaration. 2007. The Brisbane Declaration: environmental flows are essential for
707 freshwater ecosystem health and human well‐being. 10th International River
708 Symposium, 3–6 September 2007, Brisbane.
709 FANAIAN, S., GRAAS, S., JIANG, Y. & VAN DER ZAAG, P. 2015. An ecological economic
710 assessment of flow regimes in a hydropower dominated river basin: The case of the lower
711 Zambezi River, Mozambique. Science of The Total Environment, 505, 464-473.
712 FEKETE, B. M., VÖRÖSMARTY, C. J. & LAMMERS, R. B. 2001. Scaling gridded river
713 networks for macroscale hydrology: Development, analysis, and control of error. Water
714 Resources Research, 37, 1955-1967.
715 FOODY, G.M., 2010. Assessing the accuracy of land cover change with imperfect ground
716 reference data. Remote Sensing of Environment, 114(10), 2271–2285.
717 FORNAROLI, R., CANOBBIO, S., VIGANÒ, G., MEZZANOTTE, V., BOCCHIOLA, D.,
718 CABRINI, R., CONFORTOLA, G. & BOCCHIOLA, D. 2016. Effects of Future Climate
719 Change on a River Habitat in an Italian Alpine Catchment.
720 GARG, S. K. 2007. Irrigation Engineering and Hydraulic Structures, 21st Edition,88.
721 GILLEY, J. E., KOTTWITZ, E. R. & WIEMAN, G. A. 1991. Roughness Coefficients for
722 Selected Residue Materials. BiologicalSystems Engineering: Papers and Publications.
723 Paper 76.
724 HICKEY, J. T., HUFF, R. & DUNN, C. N. 2015. Using habitat to quantify ecological effects of
725 restoration and water management alternatives. Environmental Modelling & Software,
726 70, 16-31.
727 HOLMQUIST, J. G. & WADDLE, T. J. 2013. Predicted macroinvertebrate response to water
728 diversion from a montane stream using two-dimensional hydrodynamic models and zero
729 flow approximation. Ecological Indicators, 28, 115-124.
730 ILKER, T. T., KIJIN, N., JIABAO, G. & A., U. M. 2008. Real Time Optimal Monitoring
731 Network Design in River Networks. World Environmental and Water Resources
732 Congress 2008.
733 IM, D. & KANG, H. 2011. Two-dimensional physical habitat modeling of effects of habitat
734 structures on urban stream restoration. Water Science and Engineering, 4, 386-395.
735 IWM 2016 Chittagong Water Supply Improvement and Sanitation Project (CWSISP) Water
736 Resources Study: Groundwater Hydrology & Rainwater Harvesting (Packages: S/7)
737 JUNG, S. H. & CHOI, S.-U. 2015. Prediction of composite suitability index for physical habitat
738 simulations using the ANFIS method. Applied Soft Computing, 34, 502-512.
739 KAVVAS, M. L., CHEN, Z. Q. & ANDERSON, M. L. 2004. Modeling Low Flows on the
740 Cosumnes River.
741 KELLY, H. & JAMES, C. G. 2009. Conservative Design Rainfall Distribution.
742 KENNETH, M. W., JANET, B. & MICHAEL, K. S. 2008. Automatic Calibration of the U.S.
743 EPA SWMM Model for a Large Urban Catchment.
744 KREBS, G., KOKKONEN, T., VALTANEN, M., SETÄLÄ, H. & KOIVUSALO, H. 2014.
745 Spatial resolution considerations for urban hydrological modelling. Journal of
746 Hydrology, 512, 482-497.
747 LASTRA, J., FERNÁNDEZ, E., DÍEZ-HERRERO, A. & MARQUÍNEZ, J. 2008. Flood hazard
748 delineation combining geomorphological and hydrological methods: an example in the
749 Northern Iberian Peninsula. Natural Hazards, 45, 277-293.
750 LEE, J. H., KIL, J. T. & JEONG, S. 2010. Evaluation of physical fish habitat quality
751 enhancement designs in urban streams using a 2D hydrodynamic model. Ecological
752 Engineering, 36, 1251-1259.
753 Lucas, M. C., Baras, E., Thom, T. J., Duncan, A. and Slavík, O., 2001. The Stimulus and
754 Capacity for Migration, in: Migration of Freshwater Fishes. Blackwell Science Ltd,
755 Oxford, UK., pp. 14–65. doi:10.1002/9780470999653.ch2
Page 25 of 53
756 MoEF . 2015. Fifth National Report to the Convention on Biodiversity, Ministry of Environment
757 and Forests, Government of the People's Republic of Bangladesh. Available at:
758 https://www.cbd.int/doc/world/bd/bd-nr-05-en.pdf
759 MOHAMMAD, F., MOHAMMAD, K., ZAHRA, Z. & FOROUGH, A. 2016. Coastal Flood
760 Damage Estimator: An Alternative to FEMA’s HAZUS Platform.
761 MOUTON, A. M., SCHNEIDER, M., DEPESTELE, J., GOETHALS, P. L. M. & DE PAUW, N.
762 2007. Fish habitat modelling as a tool for river management. Ecological Engineering, 29,
763 305-315.
764 NIKGHALB, S., SHOKOOHI, A., SINGH, V. P. & YU, R. 2016. Ecological Regime versus
765 Minimum Environmental Flow:Comparison of Results for a River in a Semi
766 Mediterranean Region. Water Resources Management, 30, 4969-4984.
767 NOMAN, N. S. (1997); Optimal operation of the Karnafuli Reservoir, Bangladesh. MSc.
768 Thesis, Asian Institute of Technology, Bankok, Thailand, pp. 23-31.
769 OLDEN, J. D., JOY, M. K. & DEATH, R. G. 2004. An accurate comparison of methods for
770 quantifying variable importance in artificial neural networks using simulated data.
771 Ecological Modelling, 178, 389-397.
772 POFF, N. L., ALLAN, J. D., BAIN, M. B., KARR, J. R., PRESTEGAARD, K. L., RICHTER,
773 B., SPARKS, R. & STROMBERG, J. 1997. The natural flow regime: a new paradigm for
774 riverine conservation and restoration. BioScience 47, 769-784.
775 POFF, N. L. & ZIMMERMAN, J. K. H. 2010. Ecological responses to altered flow regimes: a
776 literature review to inform the science and management of environmental flows.
777 Freshwater Biology, 55, 194-205.
778 POFF, N.L., RICHTER, B.D., ARTHINGTON, A.H., BUNN, S.E., NAIMAN, R.J., KENDY,
779 E., ACREMAN, M., APSE, C., BLEDSOE, B.P., FREEMAN, M.C., HENRIKSEN, J.,
780 JACOBSON, R.B., KENNEN, J.G., MERRITT, D.M., O’KEEFFE, J.A.Y.H., OLDEN,
781 J.D., ROGERS, K., THARME, R.E., WARNER, A., 2010. The ecological limits of
782 hydrologic alteration (ELOHA): a new framework for developing regional environmental
783 flow standards. Freshw. Biol. 55, 147–170.
784 PUSHPALATHA, R., PERRIN, C., LE MOINE, N., MATHEVET, T. & ANDRÉASSIAN, V.
785 2011. A downward structural sensitivity analysis of hydrological models to improve low-
786 flow simulation. Journal of Hydrology, 411, 66-76.
787 QIN, L., LI, K., LI, Y., LIANG, R., CHEN, M. & HODGES, B. R. 2016. A habitat similarity
788 model based on vague sets to assess Schizothorax prenanti spawning habitat. Ecological
789 Engineering, 96, 86-93.
790 RIVAES, R., RODRÍGUEZ-GONZÁLEZ, P. M., ALBUQUERQUE, A., PINHEIRO, A. N.,
791 EGGER, G. & FERREIRA, M. T. 2015. Reducing river regulation effects on riparian
792 vegetation using flushing flow regimes. Ecological Engineering, 81, 428-438.
793 RICHTER, B. D., AND THOMAS, G. A. 2007. Restoring environmental flows by modifying
794 dam operations. Ecology and Society 12(1): 12.
795 RICHTER, B.D., 2010. Re-thinking environmental flows: from allocations and reserves to
796 sustainability boundaries. River Res. Appl. 26, 1052–1063.
797 RICHTER, B.D., DAVIS, M.M., APSE, C., KONRAD, C., 2012. A PRESUMPTIVE
798 STANDARD FOR ENVIRONMENTAL FLOW PROTECTION. River Res. Appl. 28,
799 1312–1321. doi:10.1002/rra.1511
800 ROSSMAN, L. A. 2010. Storm Water Management Model User’s Manual Version 5.0. Water
801 Supply and Water Resources Division National Risk Management Research Laboratory.
802 Cincinnati, OH. Chp3. Pg 34. .
803 SAKSENA, S. & MERWADE, V. 2015. Incorporating the effect of DEM resolution and
804 accuracy for improved flood inundation mapping. Journal of Hydrology, 530, 180-194.
805 SALEH, F., DUCHARNE, A., FLIPO, N., OUDIN, L. & LEDOUX, E. 2013. Impact of river
806 bed morphology on discharge and water levels simulated by a 1D Saint–Venant hydraulic
807 model at regional scale. Journal of Hydrology, 476, 169-177.
808 SAMUELS, P. G. 1989. BACKWATER LENGTHS IN RIVERS. Proceedings of the Institution
809 of Civil Engineers, 87, 571-582.
Page 26 of 53
810 SHRESTHA, R. R., PETERS, D. L. & SCHNORBUS, M. A. 2014. Evaluating the ability of a
811 hydrologic model to replicate hydro-ecologically relevant indicators. Hydrological
812 Processes, 28, 4294-4310.
813 SMAKHTIN, V. U. 2001. Low flow hydrology: a review. Journal of Hydrology, 240, 147-186.
814 SMAKHTIN, V. U. & ERIYAGAMA, N. 2008. Developing a software package for global
815 desktop assessment of environmental flows. Environmental Modelling & Software, 23,
816 1396-1406.
817 STEFFLER, P. & ALI, M. M. 2012. Effect of Friction on Spurious Oscillations in Open Channel
818 Modeling with Variable Bathymetry or Roughness.
819 STEFflER, P. & BLACKBURN, J. 2002. River2D : Two-dimensional Depth Averaged Model of
820 River Hydrodynamics and Fish Habitat. Introduction to Depth Averaged Modeling and
821 User’s Manual.UniversityofAlberta,Edmonton.
822 SUN, T., ZHANG, H., YANG, Z. & YANG, W. 2015. Environmental flow assessments for
823 transformed estuaries. Journal of Hydrology, 520, 75-84.
824 SURIYA, S. & MUDGAL, B. V. 2012. Impact of urbanization on flooding: The Thirusoolam
825 sub watershed – A case study. Journal of Hydrology, 412–413, 210-219.
826 TENNANT, D. L. 1976. Instream flow regimens for fish, wildlife, recreation and related
827 environmental resources. Fisheries 1:6–10.
828 THARME, R. E. 2003. A global perspective on environmental flow assessment: emerging trends
829 in the development and application of environmental flow methodologies for rivers. River
830 Research and Applications, 19, 397-441.
831 TSAI, C., ISLAM, M. N. & RAHMAN, K. U. M. S. 1981. Spawning of
832 majorcarpsinthelowerHaldaRiver, Bangladesh. Estuaries, 4, 127–138.
833 TSIHRINTZIS, V. A., SYLAIOS, G. K., SIDIROPOULOU, M. & KOUTRAKIS, E. T. 2007.
834 Hydrodynamic modeling and management alternatives in a Mediterranean, fishery
835 exploited, coastal lagoon. Aquacultural Engineering, 36, 310-324.
836 USGS. 2015. EarthExplorer [Online]. Available: http://earthexplorer.usgs.gov [Accessed]
837 WADDLE, T. J. E. 2012. PHABSIM for Windows User's Manual and Exercises: U.S.
838 Geological Survey Open-File Report 2001-340.
839 WATTS, M. J., LI, Y., RUSSELL, B. D., MELLIN, C., CONNELL, S. D. & FORDHAM, D. A.
840 2011. A novel method for mapping reefs and subtidal rocky habitats using artificial
841 neural networks. Ecological Modelling, 222, 2606-2614.
842 WOO, H. 2010. Trends in ecological river engineering in Korea. Journal of Hydro-environment
843 Research, 4, 269-278.
844 WU, W. & WANG, S. S. Y. 2006. Application of a Depth-Averaged 2-D Model in River
845 Restoration. World Environmental and Water Resource Congress 2006.
846 YAO, W., RUTSCHMANN, P. & SUDEEP 2015. Three high flow experiment releases from
847 Glen Canyon Dam on rainbow trout and flannelmouth sucker habitat in Colorado River.
848 Ecological Engineering, 75, 278-290.
849 YI, Y., CHENG, X., WIEPRECHT, S. & TANG, C. 2014. Comparison of habitat suitability
850 models using different habitat suitability evaluation methods. Ecological Engineering,
851 71, 335-345.
852 YI, Y., SUN, J. & ZHANG, S. 2016. A habitat suitability model for Chinese sturgeon
853 determined using the generalized additive method. Journal of Hydrology, 534, 11-18.
854 ZHANG, Z., WAGENER, T., REED, P., BHUSHAN, R., 2008. Reducing uncertainty in
855 predictions in ungauged basins by combining hydrologic indices regionalization and
856 multiobjective optimization. Water Resour. Res. 44.
857 ZHANG, Z., BALAY, J.W., BERTOLDI, K.M., MACOY, P.O., 2016. Assessment of Water
858 Capacity and Availability from Unregulated Stream Flows Based on Ecological Limits of
859 Hydrologic Alteration (ELOHA) Environmental Flow Standards. River Res. Appl. 32,
860 1469–1480. doi:10.1002/rra.2979
861
862

Page 27 of 53
Figure 1: (a) The redline showing watershed boundary in Chittagong region, (b) the watershed
and stream network (c) the spawning ground
863
Figure 2: Hydrologic model (SWMM) input a) Landuse map of 2015, b) Landuse map of 2010, c)
Watershed wise mean slope analysis. ‘W’ denotes with corresponding watershed number
864
Figure 3: A schematic overview of model set up process
865
866
Figure 4: Coupled hydrologic and hydrodynamic model showing applied boundary conditions
on HEC-RAS model.
867

Figure 5: Hydrograph obtained from SWMM


868
Fig 6: Habitat Suitability Curves of major carp fishes based on (a) depth, (b) flow velocity, and
(c) channel index

869 Figure 7a: Comparison of simulated and Observed water levels at Khal no 18 obtained from
870 HEC-RAS
871 Figure 7b Comparison of simulated and Observed water levels at Khal no 10 obtained from
872 HEC-RAS
873
874 Figure 7c Comparison of Simulated and Observed discharge at CS-17 obtained from HEC-RAS
875 Fig. 7d: Comparison of simulated and observed discharge at CST-1 obtained from SWMM
876
877
878 Figure 8: Flow Duration curve at different cross section of spawning ground
879
880 Figure 9: Variation of Weighted Usable Area with discharge for Indian Major Carps
881 Figure10: Combined Habitat Suitability Index obtained from River 2D in spawning ground with
882 incremental discharge

Page 28 of 53
883 Figure 11: Peak monthly released discharge from Kaptai reservoir
884

Page 29 of 53
885
886 Table 1: Details on data preparation for model setup
Data Description Engaged Location Acquisition Source
type model period
Digital ASTER Global Digital SWMM Catchment 17/11/2011 USGS
Elevation Elevation Model (ASTER area (2015)
Model GDEM) 30m resolution
(DEM)
Satellite 30 m resolution Landsat 7 SWMM Catchment December USGS
images and 8 remote sensing image area 2010 and (2015)
February
2015
Cross 32 cross sections were HEC-RAS Karnafuli 2012 CPA
sections employed for River and
hydrodynamic model set up its
tributaries
79 cross sections were SWMM Karnafuli Compiled CPA and
employed for River Bangladesh
from several
hydrodynamic model set up survey Water
periods Developme
(2000-2015)nt Board
(BWDB)
Spatial scale varying 10- PHABSIM Restoration Compiled CPA and
30 m collected and site from 2000- BWDB
River2D 2015
Water Stage hydrograph HEC-RAS Outfall 2010-2015 CPA
level (1 hr interval) and
SWMM
Water level (6 hr interval) HEC-RAS CS-27 1/7/2015 to CPA
validation (Khal no 31/12/2015
10)
Water level (6 hr interval) HEC-RAS CS-32 01/12/2015 CPA
validation (Khal no to
18) 21/12/2015
Discharg 6 hr interval data was HEC-RAS CS-1 2010-2015 IWM
e collected from temporary and (2016)
installed flow gauge and SWMM
calibrated MIKE 11 model
Temporary installed flow HEC-RAS CS-17 25/2/2014 to IWM
gauge 26/7/2014 (2016)
Monthly peak discharge SWMM CST-1 2/1/2010 to BWDB
information 26/6/2015
887
888
889
890

Page 30 of 53
891 Table2: SWMM model parameters
Parameter Reference Results Summery
Depression storage, Dp ASCE and WEF (1992) Dp= 0.39 mm for impervious
Krebs et al. (2014) area and Dp=4.18mm for
pervious area
Manning’s roughness coefficient Gilley et al. (1991) no=0.011~0.013
no for watershed runoff
estimation
Manning’s roughness BWDB nc=0.035 for Karnafuli River
coefficient, nc for Karnafuli and based on land use map
River and its tributaries 0.015~0.025 selected for flood
plain
892
893 Table 3: SWMM sub-watershed properties
b
Sub- Watershed % Imperviousness % Slopec Curve Numberd
watershed Area (km2)a
2015 2010 2015 2010
Name
W1 12.34 6.67 2.11 3.61 72 69
W2 18.98 3.48 0.94 4.23 73 70
W3 7.50 4.71 2.0 4.21 71 69
W4 37.32 3.66 1.74 4.44 61 54
W5 5.37 7.09 5.04 5.96 71 55
W6 16.0 4.88 1.31 3.97 63 57
W7 27.2 27.17 15.14 5.61 84 53
W8 7.17 17.3 16.75 4.07 68 51
W9 21.08 4.69 3.46 3.72 62 62
W10 20.9 53.66 44.19 8.34 83 50
W11 12.59 53.46 36.97 5.26 83 83
W12 8.19 14.62 1.58 4.23 61 60
W13 6.64 51.13 44.79 5.46 83 49
W14 29.06 64.6 58.64 5.3 67 49
W15 17.90 30.2 23.81 3.99 69 50
W16 10.0 11.77 5.61 3.76 61 53
W17 5.82 34.35 28.22 3.93 84 48
W18 14.46 17.88 4.84 3.11 73 70
W19 31.34 12.16 10.43 4.4 84 73
a
894 Calculated using HEC-GeoHMS tool
b
895 Built up area % of representative year.
c
896 Mean slope obtained from spatial analysis of GIS (Fig. 2c)
d
897 As per Eq. 2
898
899

Page 31 of 53
900
901 Table 4: Summary of HEC-RAS performances at available stations
Streamgauge Period Parameter PBAISb RSRc RMSEd NSEe
Stationa
Khal no 18 01/12/2015 to Water levelf -0.007 0.007 0.01 0.99
(CS-32) 21/12/2015
Khal no 10 1/7/2015 to Water levelg -7.25 0.15 0.5 0.93
(CS-27) 31/12/2015
CS-17 25/2/2014 to Dischargeh -12.29 0.32 211.97 0.90
26/7/2014
CST-1 2/1/2010 to Dischargei 9.42 0.33 3.13 0.97
26/6/2015
a
902 Figure 5
b
903 Calculated as Eq.17
c
904 Calculated as Eq.18
d
905 Calculated as Eq.19
e
906 Calculated as Eq.20
f
907 Figure 7a
g
908 Figure 7b
h
909 Figure 7c
i
910 Figure 7d
911
912
913
914
915
916

Page 32 of 53
917
918 Table 5: Selected discharge for habitat simulation in spawning ground.
Discharge (m3/s) Description
15.76 Minimum flow of flood tide in Karnafuli River during winter season
25 10% flow of annual runoff reported as per Tennant method (Tennant,
1976)
30.1,35.6 Minimum discharge of flood tide during monsoon season
60.66 Q95 reported according to Q95 method obtained from FDC (Arthington and
Zalucki, 1998). Q95 denotes the flow rate equaled or exceeded for 95% of
the time.
75 30% of mean annual runoff as per Tennant method(Tennant, 1976)
100, 152.82, 174.7 Discharges were selected in incremental order based on the change of stage
200, 225 High flow, selected based on flow rate equaled or exceeded for 2% of the
time.
919
920

Page 33 of 53
921
922 Table 6: Optimal conditions for targeted carp fish spawning to achieve EF
Parameter Optimal condition Reference
Spawning period From month April-June more particularly Tsai et al., (1981)
during the dates of full moon and new moon
Tide When tidal cycle changes from low tide to high Lucas, et al.,
tide flow velocity got reduces and this condition (2001)
is optimum for spawning.
Discharge Carps spawning obtained maximum WUA at Present study
discharge 25~30.1 m3/s. River2D predicts this
situation can be maintained upto 100 m3/s (Fig.
9)
Water level Suitable water depth 3.5-4.2 m should maintain
at restoration site to attain maximum habitat
suitability
Channel velocity Carp fish find favorable spawning conditions at
velocity 0.5-0.6 m/s.
Bottom substrate size Silt and sand are suitable bottom substrate for
channel spawning
Precipitation Fish spawning aided by rainfall or Tsai et al., (1981)
thunderstorms. Dependence on rainfall may
delay fish spawning.
Water Temperature Suitable air temperature was found 22-31°C. Tsai et al.,
and corresponding water temperature varies (1981); Akter
between 24-31°C and Ali, (2012)
Dissolved oxygen (DO) About 90% saturation DO level near 7.1 mg/l
pH 6.2 to 8.1 Tsai et al., (1981)
Turbidity 25-30 mg/l
Salinity During rainy season freshwater from runoff Akter and Tanim
reduces salinity concentrations. Spawning (2017)
occurs salinity concentrations ranging from 10-
40 mg/l
923
924

Page 34 of 53
925 • ‘Environmental Flow’ (EF) determination for an ungauged tidal rivers face challenges
926 • a noble approach described comprising of hydrological, hydrodynamic and habitat
927 simulation
928 • hydrodynamic model outcomes showed EF resulted a set of low flow indices
929 • The bidirectional hydrodynamic process was modelled using habitat simulation method
930 • The obtained discharge against maximum Weighted Usable Area (WUA) was defined as
931 EF
932
933
934

Page 35 of 53
935
936

Page 36 of 53
937
938

Page 37 of 53
939
940

Page 38 of 53
941
942

Page 39 of 53
943
944

Page 40 of 53
945
946

Page 41 of 53
947
948

Page 42 of 53
949
950

Page 43 of 53
951
952

Page 44 of 53
953
954

Page 45 of 53
955
956

Page 46 of 53
957
958

Page 47 of 53
959
960

Page 48 of 53
961
962

Page 49 of 53
963
964

Page 50 of 53
965

Page 51 of 53
966

Page 52 of 53
967

Page 53 of 53

You might also like