Water . 2017, 9, x doi: FOR PEER REVIEW www.mdpi.comjournal water [609819]

Water . 2017, 9, x; doi: FOR PEER REVIEW www.mdpi.com/journal/ water
Article 1
Mures river basin lag time analysis. The SCS -CN 2
versus R omanian national standard comparative 3
approach developed for small watersheds . 4
Abstract: Global water cycles and energy fluxes understanding is leading to better predictions of 5
land atmosphere interaction and local hydro -climates evolution. The water transfer time 6
determination from rainfall to runoff needs accurate measurements of river basin s hydrological 7
parameters. We analyzed the lag time value results of two different methodologies used for 54 8
Romanian small catchment areas . The focus of this study is lag time evaluation for an effective 9
implementation of the best methodology approach in our geographical space . Our research in 10
Mures river basin was developed using remote sensing technology maps , GIS and environmental 11
datasets in combination with field work on every drainage basin in order to assess the specific 12
morphological features and v alidate the land cover typology . We found that Soil Conservation 13
Service – Curve Number (SCS -CN) method is widely used according to USA landscape features 14
classification, not necessarily applicable to Mures basin characteristics. Our results show how the 15
official Romanian rational methodology can be improved and the limits of SCS -CN method. 16
Keywords: runoff ; SCS-CN method; rainfall ; lag time; curve number; rational method 17
18
1. Introduction 19
The local Mures r iver basin hydro -climates assessment and continuing observation is 20
integrating our research in a broader integrated regional system that allows us to understand water 21
and energy exchanges . Previous studies implies that small river basin models represent re asonably 22
good regional water cycles predictor [1]. 23
Transferring research information from one region to another and synthetizing the results at a 24
global scale is what the Global Energy and Water Exchanges (GEWEX) Hydroclimatology Panel 25
aims [2]. 26
Our paper is merging the GEWEX framework in terms of understanding, measuring and 27
predicting local and regional water cycles trough improved rainfall -runoff simulations for 28
hydrologic applications. 29
As SCS -CN method is increasingly seen as the assessment solution fo r land -atmosphere 30
interaction in many other countries, it is worth examining its impact in Romania, where the Rational 31
method is widely used for small watersheds evaluation [3]. 32
Various investigations [1-4] have explored the subject, analyzing the time interval between the 33
moment of culminant rainfall and maximum stream flow discharge. Reistetter and Russell (2011) 34
argued that o nly the rainfall quantity entering the catchment area is important. In our study w e used 35
as inputs in hydrological computational models of the 54 small watersheds the 60 minutes 36
maximum precipitation with the probability of returning once in 100 years. The land coverage and 37
soil types are remarkably affecting the stream flow [ 5]. We use d lag time in our research as a 38
benchmark in order to compare the SCS -CN and Romanian national standard methods. 39
According to the Romanian National Institute of Hydrology and Water Management (NIHWM) 40
methodology , the small river basins lag time is calculat ed using the Rational method [3]. Lag time 41
estimated by this iterative method was compared with t he lag time equation from SCS -CN, 42
calculated on watershed Curve Number basis . The curve number indicates the water retention 43
capacity of different land cover types [ 5]. 44

Water 2017 , 9, x FOR PEER REVIEW 2 of 9
Fan et al. (2013) explored the rainfall -runoff inter -connection and constructed a simulation 45
model based on the SCS-CN method [4]. Choi et al. (2013) developed a l and surface model (LSM) for 46
the small river basins stream flow prediction and monitorin g [6]. Hunink et al. (2017) used the 47
Normalized Difference Vegetation Index (NDVI ) coefficient parameterizations to assess the 48
hydrologic response sensitivity of a basin -scale hydrological model and time series data to calibrate 49
it [7]. Grimaldi et al. (2013) used the initial abstraction and volume s from SCS -CN methodology to 50
combine the infiltration equation and soil hydro -conductivity [8]. 51
Runoff estimation for small un -gauged river basins is difficult, requiring validation field trips. 52
Bozzano et al. (2017) combined field surveys with remote sensing techniques to integrate the Satellite 53
Advanced Differential Interferometry Synthetic Aperture Radar (A -DIn SAR) data in the small river 54
basin investigated area [ 9]. Remote sensing technology is increasing the on -site assessment accuracy. 55
In our study SCS CN -Romanian national standard lag time comparison methodology aimed to 56
identify the best watershed development prediction and provide solutions for river basins 57
integrated management. Both metho ds needed land cover , vegetation and soil evaluation using the 58
remote sensing available data . 59
Eilander et al. (2014) found that small watersheds can be outlined through optical imagery like 60
MODIS or Landsat for base map creation and then monitored with synthetic aperture radar (SAR) 61
imagery such as Envisat or Radarsat data using a Bayesian time series classifier [10]. 62
Representing the runoff parameter within a drainage basin, the curve number (CN) combines 63
all these geographical factors to generate a synthetic index . Huang et al. (2006) improved the SCS -CN 64
methodology by the slope inclusion into the CN method, analyzing rainfall -runoff data from an 65
experimental watershed, founding unacceptable differences between perceived and prognosticate d 66
runoff depth correlations to slope values [11]. 67
68
2. Materials and Methods 69
The digital elevation model (DEM) , land cover map from Corine Land Cover (CLC) project and 70
soil map (HWSD v. 1.1) were used in GIS to create the CN grids [12]. Alternatively, INIS Viewer and 71
Google Earth environmental datasets were analyzed for the small catchments morphological 72
parameters assessment. Remote sensing imagery was used to deduce vegetation cover, soil and 73
terrain slope [4]. 74
Jung et al. (2015) analy zed DEM vertical error and grid -box size for hydrological models, 75
finding a reliant connection to the stream flow and watershed morphological characteristics [13]. 76
The hydrological soil classes were adapted based on Harmonized World Soil Database (HWSD) 77
Version 1.1 layer (March 2009) . In GIS map algebra a curve number was estimated on each raster cell 78
by a ssigning specific average values of soil permeability, vegetation density and hill slope . The 79
results are integrating these parameters all over the rasters extents in order to generate the Mures 80
river basin CN map (Figure 1) and all 54 watersheds CN were also determined [5, 17 ]. 81
The river basins curve number was used in the calculations of the maximum potential water 82
retention (S) with the following f ormula: 83
S=0.0001xCN -10 (1) 84
Along the average gradient of hill slope in percentages (I) and the stream length in feet (L) we 85
obtained the SCS lag time in hours, using the lag time equation from the SCS -CN method [17]: 86
87
𝑇𝐿 𝑆𝐶𝑆=𝐿0.8(𝑆+1)0.7
1900 √𝐼𝑣 (2) 88
89

Water 2017 , 9, x FOR PEER REVIEW 3 of 9
90
Figure 1 Mures river basin CN map 91
The Romanian national standard methodology approach for the lag time calculation 92
differentiates an overland f low time and a stream flow time as main components [1]. In the first case , 93
the slope flow characteristics such as average length, hill slope gradient, roughness are generating 94
the lag time formula : 95
𝑇𝑜𝑣= √𝐿𝑣
𝑚𝑣√𝑖𝑣ℎ2 4 (3) 96
• Lv average length of slopes, meters; 97
• iv average gradient of slopes, m/km; 98
• mv a roughness coefficient of slopes; 99
• h average depth of water on the slopes, in mm/min. 100
101
For the second Romanian national standard method lag time component, the stream 102
characteristics like average stream gradient, roughness and length from spring origin t o control 103
section were considered to determine the stream flow lag time formula: 104
105
𝑇𝑠𝑡= 𝐿𝑎
𝑚𝑎√𝑖𝑎𝑄3/43 (4) 106
107
• La stream length from the spring to the control section, in meters; 108
• ia average gradient of stream, in m/km; 109
• ma a roughness coefficient of river bed; 110
• Q is maximum stream flow discharge, in m3/s. 111
112
Finally, the Romanian national standard lag time was estimated as 0.6 times of the water 113
concentration time, defined as runoff duration of a raindrop from the watershed hydraulically most 114
distant point to the control section [1]: 115

Water 2017 , 9, x FOR PEER REVIEW 4 of 9
TLR = 0.6∙(1.2 T st1.1 + T ov) (5) 116
117
Furthermore, this method lag time was compared with the SCS-CN methodology resulting lag 118
time , determined on 54 small watersheds of 0.22 km2 to 10.06 km2 surface areas from Mures river 119
basin. 120
The figures obtained calculating the runoff lag time using two different approaches were 121
analyzed taking into account on field estimated parameters like r oughness, land cover, soil and the 122
hydrographic netw ork characteris tics. Other input parameters were extracted from maps – direct 123
calculated or estimated from Romanian regional generalizations [14]. 124
125
3. Results 126
For a better understanding of the differences between the results of both methods we need to 127
compare the same type of parameters . Yu et al. (2000) pointed out the lag time importance as a runoff 128
function that needs to be taken into account when modeling runoff rate at small river basin scales 129
[15]. 130
3.1. S CS lag time to rational lag time 131
Best fit equation of lag time in SCS theory (Figure 2) as a power function of lag time in Romanian 132
national standard method is: 133
TL SCS = 0.8207 T L R0.9398 (6) 134
We found a significant c orrelation , but not as good as presumed (the correlation coefficient 135
value r = 0.7 626), while TL SCS variation intervals at givens T L R are considerably wide [1]. The causes of 136
this dispersion are consider ed to be the effect of inclusion of extra parameters (roughness, runoff 137
coefficient, discharge) in the Romanian national standard lag time equation. 138
139
Figure 2 SCS-CN lag time to R omanian national standard lag time 140
3.2. Lag time to area equations 141 y = 0.8207×0.9398
R² = 0.5816
0.010.020.030.040.050.060.070.080.0
0.0 25.0 50.0 75.0 100.0 125.0T lag SCS [min]
T lag R [min]TL SCS = f(TL R)

Water 2017 , 9, x FOR PEER REVIEW 5 of 9
We were also interested to find simple lag time estimation formula as function of watershed 142
area (Figure 3) to verify if similar variation type equation could be infered in the analyzed methods 143
[16]. 144
The results of our approach are visible different and the differences are increasing at higher 145
areas , indicating distinct rainfall – runoff processes of the hydrol ogical models : 146
147
TLR = 29.926 + 0. 062 F (R = 0.7445) (7) 148
TLSCS = 3.1049 F0.4208 (R = 0.7405) (8) 149
150
where lag times are in minutes and watershed areas (F) are in hectares. Differences between lag time 151
values are from 0.8 minutes (36.1 and 36.9 minutes in w20) to 67.5 minutes (113.2 and 45.7 minutes). 152
The lag time dispersed values represent the hydro -morphological parameters variability effects [16]. 153
As relativ e values, the best fit is 1.96% (50 to 51 minutes, w54) and the worst is 70.8%, (34.8 to 154
10.1 minutes, w39) in the smallest watershed of our analysis [1]. 155
156
Figure 3 Lag time versus Area (R omanian national standard and SCS -CN method) [16] 157
3.3. Reversed Romanian national standard CN to SCS CN 158
Critical information about the drain in SCS method is synthesized by Curve Number (CN) 159
parameter [17]. Considerably dependent on catchment area`s hydro -morphological -soil type 160
characteristics, the CN higher values il lustrates sharply increasing runoff volumes on direct 161
watershed surfaces [ 5]. Xu et al. (2016) presented a CN spatial distribution estimation methodology 162
using remote sensing and GIS in order to extract vegetation, soil and impermeable surfaces [1 8]. 163
Because only the SCS method involve s the use of CN, we performed a reverse calculation from 164
the Romanian national standard method computed lag time and from the hydro -morphological 165
parameters to find an equivalent CN [1]. Clustering below the expected equa lity CN R = CN SCS (the 166
red line in Figure 4) s how an undere stimation of CN R. 167
168

Water 2017 , 9, x FOR PEER REVIEW 6 of 9
169
Figure 4 Reverse Romanian national standard CN versus SCS CN 170
In a similar “what if” approach we performed a reverse calculation from the SCS method 171
computed lag time to find equivalen t slope and river bed roughness (Figure 5) . 172
173
Figure 5 Estimated river bed and slope roughness to rational CN 174
Charts of equivalent slope and river bed roughness (m aCN and m vCN) as functions of Curve 175
Number show a total different dispersion pattern [1]. Even if the correlation coefficients are poor, 176
except for the dependence of CN of slope roughness (r = 0,7998), i ts reflects different relationship s to 177
SCS CN than to CN Romanian national standard (CN R). It is an inverse variation of roughness with 178
CN R and a direct exponential rise with CN SCS (Figure 6). 179
180
Figure 6 Estimated river bed and slope roughness to rational CN 181

Water 2017 , 9, x FOR PEER REVIEW 7 of 9
Based on watershed average slope, forest cover percent and soil type from Mita and Muscanu 182
(1986) tables, the runoff coefficient was estimated [19]. At SCS CN values of 80 a large range of 183
runoff coefficient dispersion has been observed between 0.425 and 0.699 [1]. SCS CN does not show 184
any obvious dependence on the runoff coefficient. 185
An inverse variation of Romanian methodology lag time with runoff coefficient, as would be 186
expected, it is not at all clear because of large dispersion of points (r = 0,1643 at best fit power 187
trendline). At 52 minutes lag time duration, the range of runoff coefficient dispersion is between 0.52 188
to 0.7 (Figure 7) . 189
190
Figure 7 SCS Curve Number and lag time versus runoff coefficient 191
Kim and Kaluarachchi (2008) found that regression equations are generating river basins 192
regionalized parameters, considerably efficient for ungauged watersheds runoff estimation because 193
of the important correlation with the basin geographical characteristics, soils and land cover [20]. 194
195
4. Discussion 196
Although this study concentrates mainly on lag time determination using different research 197
approaches , the results are indicating that improvements c ould be made to the Romanian 198
methodology for small river basins analysis . 199
Our research has demonstrated that relevant hydro morphological factors estimated from SCS 200
methodology results are inconsistent , slig htly matching the Romanian Waters Administration 201
regional generalizations documents and specialized maps . Kowalik and Walega (2015) calculations 202
also showed that empirical SCS -CN values [17] are different in comparison with actually CN 203
observed numbers [ 21]. More research is required for direct on field measurements – in office 204
calculated values comparison , studied areas extension and modified parameters sensitivity 205
assessment . River basins runoff high -resolution simulations models and maps are needed for 206
improved predictions of the surface water energy fluxes [ 6]. 207
Carried out in Mures river basin on a quite significant number of small watersheds, our study 208
was designed to clarify these contradictions. The implications of our findings emphasized SCS -CN 209
meth od efficiency if correlation analysis between satellite imagery and available geographic data are 210
made. Fan et al. (2013) observed that the l inear s pectral mixture analysis (LSMA) image processing 211
technique generates river basins surface impermeability gra de for the composite Curve Number 212
calculations [4]. 213
It is generally agreed that deforestation affects the water cycle. Energy fluxes are rapidly 214
changing direction and intensity, water transfer time decreases , influencing the rainfall -runoff 215
processes. This can be a disastrous scenario situation example for ungagged small river basins with 216
local communities , where people are suddenly exposed to flood risk. Most of the 54 analyzed 217
watersheds from Mures river basin are confined to rural village s. Young and Carleton (2006) 218
demonstrated that CN methodology is expressing runoff as a function of rainfall and water 219
hydrological fluxes in catchment systems [22]. 220

Water 2017 , 9, x FOR PEER REVIEW 8 of 9
Posner et al. (2014) developed a new methodology for NASA Global MODIS Flood Mapping 221
produ cts integration to hydrological soil -water models for small watersheds flash flood protection, 222
stating the importance of the water balance and rainfall -runoff connections inside every small river 223
basin [23]. 224
Our research proved that land cover typology is an essential predictor for small catchment 225
areas hydrological evolution if it is permanently observed, validated on field and connected to the 226
calibrated watershed parameters and morphological features introduced in the geographical 227
information system database. Jeon et al. (2014) discovered that considerable errors can result from 228
the use of uncalibrated SCS -CN parameters for surface runoff estimation in ungauged watersheds 229
[24]. 230
231
232
5. Conclusions 233
Our research findings are indicating that the r ational me thod is considerably more adjustable to 234
Romanian Mures river basin geographical conditions than the SCS -CN method. For instance, t he 235
Romanian national standard lag time has a stronger dependence on the small catchment area 236
parameters while in SCS method the conection is slightly weaker. 237
Currently, the imagery validations by on field estimations of Google Earth or INIS Viewer data, 238
appear to substantially contribute to the Romanian NIWMH methodology advantage over the 239
SCS-CN method. On the other hand, t he human factor assessments on remotely located watersheds 240
parameters such as slope s and river bed morphological characteristics can be subjective and a 241
possible error source. 242
Runoff coefficient, roughness effects and o ther rainfall -runoff parameters from Romanian 243
national standard method are somehow averaged in SCS method and are “condensed” in the Curve 244
Number value. 245
Therefore, applying SCS CN method is simpler and faster, without the risk of inducing 246
subjective errors, but is affected by this intrinsic m ethod imprecision. 247
The findings were evaluated and consistent adjustments of the lag time determination method 248
were proposed for further research . New analysis is foreseen , hydrological parameters sensitivity 249
and measured values interpretation will be carried out . 250
251
252
Acknowledgments: We received funds for covering the costs to publish in open access from D . C. University, 253
Targu Mures, Romania . 254
Author Contributions: “S.A. conceived and designed the experiments; V.A.I. performed the experiments; V.M. 255
and S.A. analyzed the data; V.M. and S.A. wrote the paper.” 256
Conflicts of Interest: “The authors declare no conflict of interest." 257
References 258
1. Sarpe, C. A.; Voda, I. Small Watershed Hydrological Models – Lag Time Comparation. 3rd PannEx 259
workshop on the climate system of the Pannonian basin. Cluj-Napoca, Romania, 20 -21 March 2017; 260
Horvath, C., Croitoru, A. E., Guettler, I., Man, T. C, Bartok, B. Eds.; Romania, 2017. 261
2. GEWEX. Available online: https://www.gewex.org (accessed on 4 April 2017). 262
3. Ordinul MMDD nr. 976/2008 . Available online: https://lege5.ro (accessed on 2 November 2016). 263
4. Reistetter , J.A.; Russell, M. High -resolution land cover datasets, composite curve numbers, and storm 264
water retention in the Tampa Bay, FL region. Appl. Geogr. 2011 , 31, 740–747. 265
5. Fan, F.; Deng, Y.; Hu, X.; Weng, Q. Estimating Composite Curve Number Using an Improve d SCS -CN 266
Method with Remotely Sensed Variables in Guangzhou, China. Remote Sens. 2013 , 5, 1425 -1438. 267
6. Choi, H.I. Application of a Land Surface Model Using Remote Sensing Data for High Resolution 268
Simulations of Terrestrial Processes. Remote Sens. 2013 , 5, 6 838-6856. 269

Water 2017 , 9, x FOR PEER REVIEW 9 of 9

7. Hunink, J.E.; Eekhout, J.P.C.; Vente, J.D.; Contreras, S.; Droogers, P.; Baille, A. Hydrological Modelling 270
using Satellite -Based Crop Coefficients: A Comparison of Methods at the Basin Scale. Remote Sens. 2017 , 271
9, 174. 272
8. Grimaldi, S.; Petroselli, A.; Romano, N. Green -Ampt Curve Number mixed procedure as an empirical tool 273
for rainfall -runoff modelling in small and ungauged basins. Hydrol. Process. 2013 , 27, 1253 –1264. 274
9. Bozzano, F.; Mazzanti, P.; Perissin, D.; Rocca, A.; De Pari, P.; Discenza, M.E. Basin Scale Assessment of 275
Landslides Geomorphological Setting by Advanced InSAR Analysis. Remote Sens. 2017 , 9, 267. 276
10. Eilander, D.; Annor, F.O.; Iannini, L.; van de Giesen, N. Remotely Sensed Monitoring of Small Reservoir 277
Dynamics: A Bayesian Approach. Rem ote Sens. 2014 , 6, 1191 -1210. 278
11. Huang, M.; Gallichand, J.; Wang, Z.; Goulet, M. A modification to the Soil Conservation Service curve 279
number method for steep slopes in the Loess Plateau of China. Hydrol. Process. 2006 , 20, 579–589. 280
12. FAO/IIASA/ISRIC/ISS -CAS/J RC. Harmonized World Soil Database (version 1.1) . 2009, FAO, Rome, Italy and 281
IIASA, Laxembourg, Austria. 282
13. Jung, H.C.; Jasinski, M.F. Sensitivity of a Floodplain Hydrodynamic Model to Satellite -Based DEM Scale 283
and Accuracy: Case Study —The Atchafalaya Basin. Remote Sens. 2015 , 7, 7938 -7958. 284
14. Diaconu , C.; Serban , P. Sinteze si regionalizari hidrologice , Editura HGA : Bucuresti , Romania , 1994 ; pp. 151– 285
185, 973-98530 -8-0. 286
15. Yu, B.; Rose, W.C.; Ciesiolka, C.A.; Cakurs, U. The relationship betwee n runoff rate and lag time and the 287
effects of surface treatments at the plot scale. Hydrol.Scienc . 2000 , 45(5), 709 -726. 288
16. Sarpe, C. A.; Haidu, I. Temporal sampling conditions in numerical integration of hydrological systems 289
time series. Geogr. Tech. 2017 , 12, Issue 1, 2017, 82 – 94. 290
17. United States Department of Agriculture, Soil Conservation Service (USDA -SCS). Estimation of Direct 291
Runoff from Storm Rainfall. In National Engineering Handbook; USDA -SCS: Washington, DC, USA, 2004. 292
18. Xu, J.; Zhao, Y.; Zhong , K.; Ruan, H.; Liu, X. Coupling Modified Linear Spectral Mixture Analysis and Soil 293
Conservation Service Curve Number (SCS -CN) Models to Simulate Surface Runoff: Application to the 294
Main Urban Area of Guangzhou, China. Water 2016 , 8, 550. 295
19. Miță, P.; Muscanu , M. Small river basins runoff coefficients , Studii și cercetări de hidrologie : Bucuresti, 296
Romania, 1986, pp. 45 -58. 53. 297
20. Kim, U.; Kaluarachchi, J.J. Application of parameter estimation and regionalization methodologies to 298
ungauged basins of the Upper Blu e Nile River Basin, Ethiopia. Journ. Hydrol. 2008 , 362, 39–56. 299
21. Young, D.F.; Carleton, J.N. Implementation of a probabilistic curve number method in the PRZM runoff 300
model. Environ. Modell. Softw. 2006 , 21, 1172–1179. 301
22. Kowalik, T.; Walega , A. Estimation of CN Parameter for Small Agricultural Watersheds Using Asymptotic 302
Functions. Water 2015 , 7, 939 -955. 303
23. Posner, A.; Georgakakos, K.; Shamir, E. MODIS Inundation Estimate Assimilation into Soil Moisture 304
Accounting Hydrologic Model: A Case Stu dy in Southeast Asia. Remote Sens. 2014 , 6, 10835 -10859. 305
24. Jeon, J. -H.; Lim, K.J.; Engel, B.A. Regional Calibration of SCS -CN L -THIA Model: Application for 306
Ungauged Basins. Water 2014 , 6, 1339 -1359. 307
© 2017 by the authors. Submitted for possible open access publication under the 308
terms and conditions of the Creative Commons Attribution ( CC BY ) license 309
(http://creativecommons.org/licenses/by/4.0/). 310