Available online at www.sciencedirect.com Procedia Environmental Sciences 19 (2013) 113 – 121 Four Decades of Progress in Monitoring and Modeling of Processes in the Soil-PlantAtmosphere System: Applications and Challenges Curve-Number/Green-Ampt mixed procedure for net rainfall estimation: a case study of the Mignone watershed, IT A Petroselli a*, S Grimaldi b,c,d, N Romano e a Department of Sciences and Technologies for Agriculture, Forestry, Nature, and Energy (DAFNE Department), University of Tuscia, Via San Camillo De Lellis snc,01100 Viterbo, Italy b Department for Innovation in Biological, Agro-food and Forest systems (DIBAF), University of Tuscia, Via San Camillo De Lellis snc, 01100 Viterbo, Italy c Honors Center of Italian Universities (H2CU), Sapienza University of Rome, Via Eudossiana 18, 00184 Roma, Italy d Department of Mechanical and Aerospace Engineering, Polytechnic Institute of New York University, Six MetroTech Center, Brooklyn, NY 11201 e Department of Agriculture, Division of Agricultural, Forest and Biosystems Engineering - University of Naples Federico II, Via Università n 100, 80055 Portici, Italy Abstract A mixed procedure, referred to as CN4GA (Curve Number for Green-Ampt), was recently introduced with the aim of distributing in time the net rainfall volume provided at event scale by the Soil Conservation Service - Curve Number (SCS-CN) method The proposed method consists in employing the Green-Ampt infiltration equation and calibrating both the ponding time and the soil hydraulic conductivity using the initial abstraction and the total volume given by the SCS-CN method The procedure is here applied on several rainfall-runoff events observed in an Italian watershed Results confirms the general behavior already noticed in previous studies: the CN4GA procedure provides net rainfall intensities consistent with the runoff observations, the SCS-CN method underestimates peak intensity when applied at sub-daily* resolution, and the differences between the two approaches are relevant when the gross rainfall peak occurs at the beginning of the storm and generally in case of multi-peak events © The Authors Authors Published Publishedby byElsevier ElsevierB.V B.V © 2013 2013 The theScientific ScientificCommittee Committeeofof conference Selection and/or peer-review under responsibility Selection and/or peer-review under responsibility ofofthe thethe conference Keywords: SCS-CN method; Green-Ampt method; net rainfall; infiltration; CN4GA; ungauged basins; rainfall-runoff modeling * * Corresponding author Tel.: +39 0761 357348; fax: +39 0761 357 434 E-mail address: petro@unitus.it 1878-0296 © 2013 The Authors Published by Elsevier B.V Selection and/or peer-review under responsibility of the Scientific Committee of the Conference doi:10.1016/j.proenv.2013.06.013 114 A Petroselli et al / Procedia Environmental Sciences 19 (2013) 113 – 121 Introduction Estimating net rainfall, i.e the difference between gross precipitation and losses, is a crucial task in hydrology and specifically in rainfall-runoff modeling The lack of discharge observations in ungauged basins impels practitioners to resort to empirical approaches for evaluating the rainfall component that contributes to the runoff One of the most used heuristic model is the Soil Conservation Service-Curve Number method (SCS-CN) [1,2], but recent literature [3,4] suggests that this method is not applicable at sub-daily time resolution and it should not be employed for estimating water infiltration into soil, mainly because it is a lumped approach (in space and time) developed in order to define the total direct runoff derived from a rainfall event Recently [5,6] a mixed procedure combining the SCS-CN method and the Green-Ampt (GA) infiltration equation [7,8], named CN4GA (Curve Number for Green Ampt) has been proposed The key concept underlying this procedure is considering the total excess rainfall and the initial abstraction derived from the SCS-CN for a specific rainfall event, and using them to fix the ponding time and to calibrate one key GA parameter, namely the saturated hydraulic conductivity of soil In doing so, the total excess rainfall volume provided by the SCS-CN method can be distributed over time using a physically based infiltration model Previous studies [5,6] described the performances of CN4GA on several observed and synthetic case studies underlining the differences with the standard SCS-CN procedure Conclusions are that the latter method underestimates the rainfall peak intensities mainly when the peak occurs at the beginning of the storm and that the CN4GA parameters are insensitive when applied to extreme rainfall events The aim of the present work is to provide some more insights on the proposed mixed technique, describing an application on a gauged catchment located in central Italy The paper is organized as follows: first the methodologies for estimating the net rainfall are briefly described, then the study area and the proposed analyses are illustrated; finally, the results, discussions and conclusions close the manuscript The net rainfall estimation methods: SCS-CN and CN4GA 2.1 The SCS-CN method The Soil Conservation Service - Curve Number (SCS-CN) method [1,2] is a popular rainfall-runoff model widely used to estimate losses and direct runoff from a given rainfall event It is defined according to the following relations: Q P I a P Ia S , S Đ 1000 Ã 25.4ă 10 CN â , Ia OS (1) where Q is runoff depth (mm), P is gross rainfall depth (mm), Ia is initial abstraction (mm), and S represents potential retention (mm) The only one parameter, named Curve Number (CN), is related to land cover, soil type and antecedent moisture condition, while the constant λ value should be fixed to 0.2 It is noteworthy that literature suggests to estimate different λ values when rainfall-runoff observations are available: basically a change in λ value causes a shift in the start of net rainfall, and hence an optimal value can be selected by imposing the same start of net discharge 115 A Petroselli et al / Procedia Environmental Sciences 19 (2013) 113 – 121 2.2 The GA method and the CN4GA mixed procedure The Green and Ampt [7,8] equation is a physically-based, although simplified, model to describe an infiltration process in soil It is expressed according to: q ° ° ® t °q0 r ê 'T'h ằ K s ô1 ô I t ằ ơ ẳ for t (2) for t t where q0 is infiltration rate; r is rainfall rate; is ponding time; Δh=(hsurf − hf), with hsurf being the depth of water on the soil surface and hf a constant matric pressure-head at the moving wetting front I(t) is cumulative infiltration, and Δθ=(θs–θi) represents the change in soil-water content with θi being the initial value and θs the field saturated soil-water content Ks is the saturated hydraulic conductivity, which should not be the value of the laboratory-measured saturated hydraulic conductivity, but has to be estimated by in-situ experiments It is more convenient to refer to an effective value of the saturated hydraulic conductivity, Keff, obtained by the optimization procedure The CN4GA mixed procedure consists in an the GA infiltration equation according the following steps: (a) estimation of the total rainfall excess and ponding time with the SCS-CN method (b) Estimation of the Keff parameter enforcing the GA cumulative infiltration, IGA, which must be equal to the corresponding SCS-CN cumulative infiltration, ISCS-CN: average values for the GA parameters reported in the literature are first assigned based on the soil type; the estimated IGA is then compared to ISCS-CN If IGA is higher (lower) than ISCS-CN, then GA is run again using a lower (higher) Keff value The parameter Keff is iteratively decreased (increased) until IGA becomes equal to the corresponding ISCS-CN At the end of this iterative procedure, an optimal value for the effective saturated hydraulic conductivity (Keff-opt) is obtained According to the previous step, the net rainfall is maintained at zero until ponding time is reached Consequently, by applying the GA with the estimated Keff-opt and given the literature values of other GA parameters, the CN4GA rainfall excess is computed The CN4GA net storm has the same cumulative rainfall excess value and the same ponding time of the net storm derived by the SCS-CN method Case study area and description of the performed analyses The Mignone watershed is located in Central Italy and selected as case study for this paper Its drainage area is 440 km2, elevations range from m to 618 m, maximum distance between divide and outlet is 59.1 km and basin average slope is 7.8% Figure shows the watershed DEM (75 m cell-size) and the extracted drainage network Rainfall and discharge observations are available, at a time resolution of 1-hour and from 1998 to 2010 For the purpose of this work, 12 extreme events (i.e the ones having the annual maximum gross discharge) are selected and their main properties are summarized in Table The data of this case study are used to estimated the net rainfall applying both the SCS-CN method and the CN4GA method, and to evaluate the performances comparing the two resulting hydrographs to the observed one The estimation of the net discharge is preliminary and mandatory to such analyses and has been performed starting from the available gross discharge data by implementing a recursive filter application [9,10]; knowing the net discharge total volume, the SCS-CN procedure has been applied calibrating the CN and λ parameters on each event, comparing the cumulated net rainfall volume to the total direct runoff volume An example of such procedure is shown in Figure 2a for the 2004 event 116 A Petroselli et al / Procedia Environmental Sciences 19 (2013) 113 – 121 After having determined the net rainfall scenario provided by the SCS-CN method, the application of the CN4GA procedure leads to second net rainfall scenario: the two scenarios are characterized by the same ponding time and the same cumulative net rainfall volume The last step useful for hydrograph estimation is the rainfall-runoff modeling: advanced terrain analysis techniques for DEM pre and post processing [11,12] are here applied using the WFIUH-1par geomorphological unit hydrograph [13,14] employing the time of concentration as estimated by the NRCS formula [15] For the selected case study area the concentration time is equal to 17 hours and the corresponding WFIUH is shown in Figure 2b After having determined the basin WFIUH, the convolution with the two net rainfall scenarios is possible, in doing so obtaining two modeled hydrographs that can be compared with the observed net discharge data and hence allowing to investigate on the performances of the mixed procedure Fig (left) case study localization; (right) DEM and the extracted drainage network Table Rainfall-runoff events main properties: year, gross peak discharge (GQ), net discharge (NQ), cumulative net runoff volume (NV), cumulated gross rainfall (GR), gross rainfall peak intensity (Grip), total number of net runoff peaks (PN) year 1998 1999 2000 2001 2003 2004 2005 2006 2007 2008 2009 2010 GQ NQ NV 3 (m /s) 68.3 132.8 211.7 128.9 137.6 304.7 186.1 215.1 128.9 304.0 156.4 277.5 (m /s) 60.2 100.1 182.7 113.8 116.2 278.2 153.6 158.1 87.4 261.1 131.1 217.9 (10 m ) 2.4 1.9 4.7 5.1 4.2 20.6 7.3 4.2 1.3 24.5 6.2 7.8 GR GRip PN (mm) 47.3 44.0 47.4 46.2 53.5 86.5 36.3 34.8 24.0 125.4 159.6 141.5 (mm/h) 7.5 6.5 5.7 5.4 10.9 9.2 3.6 8.6 5.4 9.9 13.8 32.1 (-) 2 1 3 117 A Petroselli et al / Procedia Environmental Sciences 19 (2013) 113 – 121 400 0.15 (2004) 300 0.10 10 IUH 200 p(t) [mm/h] q(t) [m /s] 0.05 100 15 0 20 40 t [h] 60 80 100 0.00 10 15 20 t [h] Figure (a) Example of net discharge and net rainfall determination in the study catchment Primary x-axis: observed gross discharge (continuous black line) and baseflow (discontinuous black line) Secondary x-axis: observed gross rainfall (thin light grey) and SCS-CN net rainfall (filled gray) balancing the total net discharge volume (b) The WFIUH-1par obtained enforcing the concentration time value (17 hours) in the selected basin Results and discussion In Figure 3, the hyetographs estimated with the two methods and the hydrographs observed and resulting from the rainfall-runoff modeling are reported for the selected maximum annual events A visual inspection of this plot supports the following comments First, results obtained in previous studies [5,6] are confirmed: the SCS-CN approach should be avoided at sub daily time resolution, because Ia is not related to the infiltration properties of the soil, and the consequence is an overestimation (underestimation) of losses at the beginning (end) of the event This can be easily observed in years 2004, 2008 and 2009 It is confirmed that applying the SCS-CN equation over time the first gross rainfall peaks are entirely lost independently of their intensity, while, conversely, this circumstance does not occur when applying the CN4GA scheme, since this last method is related to the infiltration properties of the soil and it is able to preserve the main peaks of the storm This can be supported by the same panels of years 2004, 2008 and 2009, where the CN4GA hydrographs, both in timing both in peak values, are similar to the observed ones; this CN4GA characteristic can be generalized to all the panels, where CN4GA performs better than SCS-CN, except in 2005 and 2006 where the two approaches provide the same performances Moreover, looking at the events depicted in Figure 3, additional comments can be drawn about the overall behavior of the SCS-CN and CN4GA approaches Actually, a first set of events (years 1999, 2001, 2003, 2004, 2008, 2009) is characterized by multiple gross rainfall peaks determining multiple peak discharges In such circumstances, the SCS-CN equation seems not able to follow the gross rainfall shape, determining a hydrograph particularly different from the recorded one, both at the beginning and at the end of the event This particular behavior seems to be less pronounced in the case of events having only one peak of gross rainfall; in particular for the years 1998, 2005, 2006 and 2010 the results of two rainfall-runoff modeling approaches are similar and no particular differences are observed in terms of modeled peak discharges 118 A Petroselli et al / Procedia Environmental Sciences 19 (2013) 113 – 121 75 150 (1998) q(t) [m /s] 15 20 t [h] 40 50 15 60 0 250 NQ Q CN4GA Q SCS-CN GR 10 R SCS-CN R CN4GA 20 40 t [h] 60 80 (2001) 3 50 10 0 20 40 t [h] 60 50 10 80 150 NQ Q CN4GA Q SCS-CN GR R SCS-CN R CN4GA 0 20 40 t [h] 60 80 (2004) 30 60 t [h] 90 120 q(t) [m /s] NQ Q CN4GA Q SCS-CN 10 GR R SCS-CN R CN4GA 100 15 0 50 t [h] 100 150 p(t) [mm/h] 30 p(t) [mm/h] q(t) [m /s] 200 10 NQ Q CN4GA Q SCS-CN GR 20 R SCS-CN R CN4GA 50 100 300 (2003) 100 p(t) [mm/h] 100 p(t) [mm/h] NQ Q CN4GA Q SCS-CN GR R SCS-CN R CN4GA q(t) [m /s] 100 150 100 150 (2000) 200 p(t) [mm/h] 25 q(t) [m /s] 100 NQ Q CN4GA Q SCS-CN GR 10 R SCS-CN R CN4GA p(t) [mm/h] q(t) [m /s] 50 (1999) 119 A Petroselli et al / Procedia Environmental Sciences 19 (2013) 113 – 121 200 200 100 q(t) [m /s] NQ Q CN4GA Q SCS-CN GR R SCS-CN R CN4GA 10 0 20 40 60 t [h] 100 15 80 0 100 NQ Q CN4GA Q SCS-CN GR 10 R SCS-CN R CN4GA 15 30 45 t [h] (2008) q(t) [m /s] 20 t [h] 40 150 15 60 20 40 60 t [h] 80 100 300 200 NQ Q CN4GA Q SCS-CN GR 10 R SCS-CN R CN4GA (2010) 200 30 t [h] 60 90 q(t) [m /s] q(t) [m /s] p(t) [mm/h] 30 15 NQ Q CN4GA Q SCS-CN GR 30 R SCS-CN R CN4GA 100 p(t) [mm/h] 10 NQ Q CN4GA Q SCS-CN GR 20 R SCS-CN R CN4GA 120 (2009) 100 p(t) [mm/h] 300 p(t) [mm/h] q(t) [m /s] NQ Q CN4GA Q SCS-CN GR R SCS-CN R CN4GA 60 450 (2007) 50 p(t) [mm/h] (2006) p(t) [mm/h] q(t) [m /s] (2005) 45 0 20 40 t [h] 60 80 Figure Rainfall–runoff extreme events observed in the study catchment Primary x-axis: observed net discharge (NQ), SCS-CN net discharge (Q SCS-CN), CN4GA net discharge (Q CN4GA) Secondary x-axis: observed gross rainfall (GR), SCS-CN net rainfall (R SCS-CN), CN4GA net rainfall (R CN4GA) 120 A Petroselli et al / Procedia Environmental Sciences 19 (2013) 113 – 121 Conversely, there are some events, like the ones occurring in 2000 and 2007, characterized by only one rainfall peak but where SCS-CN is not able, again, to determine an hydrograph similar to the recorded one; the circumstance is due to the presence of SCS-CN net rainfall at the end of the event, consequence of (1); a long sequence of net rainfall is hence generated applying the SCS-CN method, and this does not occur with the CN4GA application neither is expected by looking the observed hydrographs Conclusions A mixed rainfall excess method, named CN4GA, which combines the SCS-CN method with the Green-Ampt infiltration equation, was developed and tested [5,6] This method consists in applying the SCS-CN approach to quantify the cumulative net rainfall amount and using the Green-Ampt equation to distribute in time this volume In the present work the mixed procedure was tested on a small basin located in central Italy and that is characterized by the availability of rainfall and discharge observations at hour time step Twelve extreme events have been analyzed and previous literature results are confirmed: the SCS-CN underestimates the net rainfall at the beginning of the event and overestimates it at the end Several events show similar results and this occurrence is mainly due to the position of the gross rainfall peak inside the event The CN4GA solution, that has a sound physical basis, can yield significantly different outcomes in terms of hydrograph shape, both in timing and in peak values, with respect to those provided by the classic SCS-CN scheme if the latter is applied at a sub-daily time resolution The discrepancies can be particularly relevant especially when the rainfall peak position is located at the beginning of the storm, or in the cases of events characterized by multiple peak rainfall or peak discharges Acknowledgements Prof F Napolitano and Dr V Montesarchio of Sapienza University of Rome are acknowledged for having provided the rainfall-runoff data References [1] Soil Conservation Service (SCS) National Engineering Handbook, Section 4, Hydrology; U.S Department of Agriculture, Washington D.C.; 1972 [2] Natural Resources Conservation Service (NRCS) National Engineering Handbook part 630, Hydrology; U.S Department of Agriculture, Washington D.C.; 2008 [3] Garen DC, Moore DS Curve number hydrology in water quality modeling: Uses, abuses, and future directions Journal of the American Water Resources Association 2005;41:377-388 [4] Woodward DE, Hoeft CC, Hawkins RH, Van Mullem J, Ward TJ Discussion of "Modifications to SCS-CN Method for LongTerm Hydrologic Simulation" by K Geetha, S K Mishra, T I Eldho, A K Rastogi, and R P Pandey Journal of Irrigation and Drainage Engineering 2010;136:444-446 [5] Grimaldi S, Petroselli A, Romano N Green-Ampt Curve Number mixed procedure as an empirical tool for rainfall-runoff modelling in small and ungauged basins Hydrological Processes 2013; doi:10.1002/hyp.9303 (in press) [6] Grimaldi S, Petroselli A, Romano N Curve-Number/Green-Ampt mixed procedure for streamflow predictions in ungauged basins: Parameter sensitivity analysis Hydrological Processes 2013 (accepted) [7] Green WH, Ampt GA Studies on soil physics J Agric Sci 1911;4:1-24 [8] Mein RG, Larson CL Modeling Infiltration during a Steady Rain Water Resources Research 1973;9:384-394 A Petroselli et al / Procedia Environmental Sciences 19 (2013) 113 – 121 [9] De Martino G, De Paola F, Fontana N, Marini G, Ranucci A Preliminary design of combined sewer overflows and stormwater tanks in southern Italy Irrigation and Drainage 2011; 60: 544-555 (Published online 10 November 2010 in Wiley Online Library (wileyonlinelibrary.com) doi: 10.1002/ird.591) [10] De Martino G, De Paola F, Fontana N, Marini G, Ranucci A Pollution reduction in receivers: storm-water tanks, Journal of Urban Planning and Development, ASCE 2011; 137(1): 29–38, ISSN 0733-9488/2011/1 [11] De Paola F and Ranucci A Analysis of spatial variability for stormwater capture tanks assessment Irrigation and Drainage Article first published online: 20 SEP 2012 doi: 10.1002/ird.1675 [12] Lyne V, & Hollick M Stochastic time-variable rainfall-runoff modelling Institute of Engineers Australia National Conference Publ 79/10; 1979, p 89-93 [13] Serinaldi, F and Grimaldi, S Synthetic design hydrographs based on distribution functions with finite support Journal of Hydrologic Engineering 2011;16:434–446 [14] Grimaldi S., Teles V., Bras R.L., Sensitivity of a physically-based method of terrain interpolation to variance of input data and to the constraint of an enforced surface drainage direction Earth Surface Processes and Landforms 2004;29:587-597 [15] Nardi F., Grimaldi S., Santini M., Petroselli A., Ubertini L., Hydrogeomorphic properties of simulated drainage patterns using digital elevation models: the flat area issue, Hydrological Science Journal 2008;53:1176-1193 [16] Grimaldi S., Petroselli A., Nardi F., & Alonso G Flow Time estimation with variable hillslope velocity in ungauged basins Advances in Water Resources 2010;33:1216-1223 [17] Grimaldi S, Petroselli A, Nardi F A parsimonious geomorphological unit hydrograph for rainfall runoff modeling in small ungauged basins Hydrological Science Journal 2012;57:73-83 [18] Natural Resources Conservation Service (NRCS) Ponds-Planning, Design, Construction Agriculture Handbook No 590 U.S Natural Resources Conservation Service, Washington D.C.; 1997 121 ... that literature suggests to estimate different λ values when rainfall- runoff observations are available: basically a change in λ value causes a shift in the start of net rainfall, and hence an optimal... (i.e the ones having the annual maximum gross discharge) are selected and their main properties are summarized in Table The data of this case study are used to estimated the net rainfall applying... net discharge data and hence allowing to investigate on the performances of the mixed procedure Fig (left) case study localization; (right) DEM and the extracted drainage network Table Rainfall- runoff