Waste Management 38 (2015) 201–209

Contents lists available at ScienceDirect

Waste Management journal homepage: www.elsevier.com/locate/wasman

On the current state of the Hydrologic Evaluation of Landfill Performance (HELP) model Klaus U. Berger Institute of Soil Science, Universität Hamburg, Allende-Platz 2, 20146 Hamburg, Germany

a r t i c l e

i n f o

Article history: Received 10 June 2014 Accepted 12 January 2015 Available online 14 February 2015 Keywords: Landfill liner systems Cover systems HELP model Water balance Validation Simulation

a b s t r a c t The Hydrologic Evaluation of Landfill Performance (HELP) model is the most widely applied model to calculate the water balance of cover and bottom liner systems for landfills. The paper summarizes the 30 year history of the model from HELP version 1 to HELP 3.95 D and includes references to the three current and simultaneously available versions (HELP 3.07, Visual HELP 2.2, and HELP 3.95 D). A sufficient validation is an essential precondition for the use of any model in planning. The paper summarizes validation approaches for HELP 3 focused on cover systems in the literature. Furthermore, measurement results are compared to simulation results of HELP 3.95 D for (1) a test field with a compacted clay liner in the final cover of the landfill Hamburg-Georgswerder from 1988 to 1995 and (2) a test field with a 2.3 m thick socalled water balance layer on the landfill Deetz near Berlin from 2004 to 2011. On the Georgswerder site actual evapotranspiration was well reproduced by HELP on the yearly average as well as in the seasonal course if precipitation data with 10% systematic measurement errors were used. However, the increase of liner leakage due to the deterioration of the clayey soil liner was not considered by the model. On the landfill Deetz HELP overestimated largely the percolation through the water balance layer resulting from an extremely wet summer due to an underestimation of the water storage in the layer and presumably also due to an underestimation of the actual evapotranspiration. Finally based on validation results and requests from the practice, plans for improving the model to a future version HELP 4 D are described. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction The Hydrologic Evaluation of Landfill Performance (HELP) model is presumably the world’s most commonly applied model to calculate the water balance of cover and bottom liner systems for landfills and contaminated sites. The HELP model is a ‘‘quasitwo dimensional’’ layer model which considers many one dimensional hydrologic processes in two directions, vertical (esp. evapotranspiration, infiltration, saturated and unsaturated vertical flow) and lateral (surface runoff, lateral drainage), and combines these processes, but actually does not calculate a two dimensional flow. HELP requires daily weather data for a range of 1–100 calendar years, parameters for calculating evapotranspiration, and soil and design data. Open or closed landfills and bottom liner systems and cover systems may be modeled. The major purpose of the model is the comparison of the hydrologic effectiveness of alternative liner system designs for the climate of the particular site. HELP was developed as a tool for both landfill designers and authorizing agencies (Schroeder et al., 1994a). The HELP model has a history of 30 years and currently three versions exist (see Section 2). The concept and the modeled prohttp://dx.doi.org/10.1016/j.wasman.2015.01.013 0956-053X/Ó 2015 Elsevier Ltd. All rights reserved.

cesses of HELP are described in the model’s documentation (see Section 2) and in the literature on landfill technology (e.g. Koerner and Daniel, 1997). A sufficient validation is essential for the use of any model in the planning practice to allow an acceptable confidence in the modeling results. Section 3 provides information on the validation of the HELP 3 for cover systems from the literature as well as comparisons of measured data and simulation results of HELP 3.95 D from test fields in covers of two locations in Northern Germany. Section 4 looks ahead to enhancements of the model planned for a future version HELP 4 D.

2. History of the HELP model The original version of the HELP model was developed by Paul Schroeder (US Army Waterways Experiment Station, Vicksburg, Mississippi, USA) and several co-developers and was funded by the US Environmental Protection Agency. With its development starting in 1982, HELP version 1 was released in 1984 with an extensive documentation comprising a user’s guide and a technical/engineering documentation describing the modeling approach

202

K.U. Berger / Waste Management 38 (2015) 201–209

in detail (Schroeder et al., 1984a,b). Among others HELP 1 was based on two older models, CREAMS (Chemicals, Runoff, and Erosion from Agricultural Management Systems) and HSSWDS (Hydrologic Simulation on Solid Waste Disposal Sites) (see Schroeder et al., 1984a,b for references). The program was written in Fortran and executable on mainframes; a PC version for the DOS operating system was released in 1986. In 1987 Schroeder and Peyton published two verification studies for HELP 1. Schroeder and Peyton (1987a) was concerned with a comparison of simulation results to measurement results of fieldscale test areas. However, these test areas were designed and operated for other purposes, the layer sequences were mostly those of simple covers and the measurement data of input and output values of the HELP model were more or less incomplete. Therefore, the results of the output comparison were only of limited value. Schroeder and Peyton (1987b) was concerned with the lateral drainage sub-model, comparing simulation results to measurement results of physical models in pilot plant scale and to analytical solutions of the underlying Boussinesq-equation. Taking the verification results of these studies into account, HELP version 2 was released in 1988 by Schroeder and co-authors. HELP 2 included two new sub-models, a weather generator (WGEN, Richardson and Wright, 1984), and a vegetative growth and decay sub-model which was taken from the SWRRB model (Simulator for Water Resources in Rural Basins, Arnold et al., 1990). Furthermore, the modeling of many hydrologic processes was enhanced by improving, for example, the calculation of the unsaturated hydraulic conductivity, and the sub-model concerning lateral drainage which was replaced by an approach of McEnroe and Schroeder (1988). However, a complete documentation of HELP 2 has never been published. The same holds for a verification study of Peyton and Schroeder, based upon the previously mentioned measurement data of field-scale test areas. At the end of 1994 HELP version 3 was released (Schroeder et al., 1994a,b). HELP 3 represents a major enhancement beyond HELP 2 in two respects. Firstly HELP 3 included a menu-driven, though proprietary, user interface for DOS. Weather data from the USA and Canada can be imported from external databases. Weather data can also be imported from ASCII files. Model input and output is not only possible in English (in HELP so called ‘‘customary’’) units but also in metric units. Secondly major enhancements in the modeling approaches were implemented. Flow through geomembranes was added as independent hydrologic process, the sub-model of potential evapotranspiration was replaced by a simplified Penman approach, and the snow melt sub-model was replaced by procedures based upon the SNOW-17 routine of the National Weather Service River Forecast System. Additionally, in the surface runoff sub-model slope and slope length were considered and some new processes were added, although some in a very rudimentary manner (e.g. freezing and thawing of the soil, subsurface inflow, recirculation of lateral drainage). The current original HELP version is HELP 3.07 from November 1997, executable in DOS. The Canadian company Waterloo Hydrogeologic, Inc. (which was purchased by Schlumberger in January 2005) developed a Windows user interface for the HELP 3.07 model which was technically adapted to this user interface. Since about 1997 both are commercially available as Visual HELP stand-alone or as part of the UnSat Suite Plus; the current version is dated November 2004 (Schlumberger Water Services). Major features of Visual HELP are a database for the weather generator containing parameters of more than 2400 locations worldwide, the graphic design of the studied landfill profile, and a report generator for simulation results including graphics of daily, monthly and yearly results. However, the input files of Visual HELP and HELP 3.07 are incompatible because they differ in format.

In the 1990s the German Federal Environmental Protection Agency (‘‘Umweltbundesamt’’) had an interest in the availability of a water balance model for practical applications resulting from the regulations of that period. Therefore a validation study for HELP 3.07 was implemented for German climate conditions, funded by the Federal Ministry of Education and Research (BMBF) (Berger, 1998, see also Berger, 2000, 2002). The objectives of the study were to determine the limits of application of the model in Germany and to adapt HELP 3 for use in Germany. The adapted version HELP 3.07 D was released in early 1999 and contained databases with German soil textures and locations for the weather generator as well as a German-language User’s Guide. The model itself was not modified. Based upon the results of the validation study and further investigation in the following years the author released the enhanced versions HELP 3.50 D (2001), 3.55 D (2002) and 3.80 D (2004). All versions were executable in DOS. In these versions some of the sub-models were enhanced or replaced, including representations of actual evapotranspiration, vegetative growth and decay, frozen soil, and unsaturated/saturated vertical flow in vertical percolation layers. The enhancements are described in more detail in Berger (2002, 2003) and in the supplement to the Engineering Documentation of HELP 3.95 D (Berger and Schroeder, 2013). Two aspects shall be pointed out here:  The sub-models of evapotranspiration and vegetative growth and decay were enhanced in several ways. The interception sub-model was replaced by an adapted empirical model of von Hoyningen-Huene. This model was developed for several crops, includes interception storage from one day to the next and allows calculating the interception of vegetation with larger values of the maximum leaf area index than for grasses. The evapotranspiration sub-model was modified to allow a more realistic relation between soil evaporation and transpiration and a more realistic influence of the vegetation on evapotranspiration. The implementation of the vegetative growth and decay model was completed and the decrease of plant growth due to short air (aeration stress) was added from the model EPIC.  The frozen soil sub-model was pragmatically enhanced. The periods with frozen soil were shortened for German climate conditions and the process representations of soil freezing and thawing were modified to allow a smooth instead of an abrupt freezing and thawing of the soil within the evaporative zone. Consequently the estimation of surface runoff and of the time distribution of lateral drainage in the drainage layer below the topsoil during and after thawing becomes more realistic. Furthermore, since HELP 3.80 D HELP-D allows the user to change vegetation and soil properties within a simulation run to simulate the aging of a landfill profile. For example, it may now account for vegetation succession or the deterioration of lateral drainage layers due to clogging. The usability was enhanced by separate output files for daily, monthly and yearly simulation results which can be imported into a spreadsheet for further processing. To upgrade from the outdated DOS user interfaces, unable to run under the 64-Bit versions of Windows 7 and 8, in 2011 the author released HELP 3.90 D with a basic Windows user interface. It is compatible to the DOS interface of HELP 3. Some import options for weather data of HELP 3.07 were dropped, and an import option for Visual HELP weather files was added. In the current version HELP 3.95 D (Berger and Schroeder, 2013), released in August 2013, the user interface was enhanced by some new features like on-line help texts and the option to simulate with the model versions HELP 3.95 D or HELP 3.07 which was recompiled with technical adaptations to the current Windows versions.

K.U. Berger / Waste Management 38 (2015) 201–209

3. Validation of the HELP model 3.1. Some fundamental methodological remarks An essential precondition for the practical application of any model is a sufficient level of confidence in the models results at least concerning particular processes, system designs or climatic conditions. The methodology to obtain this confidence (validation) is complex and comprises several sub-processes (see Berger, 1998, 2000). Very important and maybe the most interesting sub-process is the operational validation or output comparison of simulation and measurement results of particular systems. Discrepancies between simulation and measurement results may be caused by several reasons or sources of error; see Berger (1998, 2002) for a categorization of possible errors.

3.2. Validation of HELP 3 in the literature There are several validation studies on HELP version 3 published in the literature. The majority of the studies are from the USA and refer to one of the original versions HELP 3.0x. Unfortunately the exact version is not always mentioned, making intercomparisons of model performance difficult. Some studies are concerned with the leachate collected in the bottom liner system and related topics (Lange et al., 1999; Bonaparte et al., 2002 (especially chapter 5.4); Dho et al., 2002; Shariatmadari et al., 2010; Alslaibi et al., 2013). Bonaparte et al. (2004, chapter 4) characterized five models, summarized validation results of HELP 3 and UNSAT-H 2 from the literature, and gave recommendations for the application of water balance models. In the following paragraphs key findings of studies for cover systems in different climate are summarized. Berger compared measured discharges from a test field on the landfill Georgswerder in Hamburg, Northern Germany, for 1988– 1995 to simulation results of HELP 3.06/3.07 (Berger, 1998, 2000) and HELP 3.80 D (Berger, 2008), respectively. Most of the key findings are similar to the results discussed for HELP 3.95 D in Section 3.3 of this paper except for the simulation of frozen soil. Under German climate the periods of frozen soil modeled by HELP 3.07 are too long, leading to an overestimation of surface runoff on frozen soil, and freezing and thawing of the soil in depth is modeled unrealistically fast (one day for the entire evaporative zone), leading to unrealistic high peak values of the average head in the lateral drainage layer below the restoration layer after thawing (Berger, 1998, 2000, 2003). A key finding of a sensitivity analysis of HELP 3.07 for lateral drainage layers was a considerable and unrealistic decrease of actual evapotranspiration and subsequently a considerable increase of lateral drainage if a geosynthetic drainage net instead of a mineral drainage layer was used (Berger, 1998, 2002). This error was caused by the HELP internal number of time steps per day that increased considerably for drainage net properties. However, this error could be shown as operating system or hardware dependent. It was confirmed by the author for Windows 95 and 98 (i.e. MS DOS 7), but did not occur in Windows XP and using the recompiled HELP 3.07 included in HELP 3.95 D in Windows 7. Fleenor and King (1995) compared the vertical percolation approach of HELP 3 (beta) to a finite element solution of the Richards equation for two simple landfill profiles (cover on waste) under three climatic conditions from arid (Phoenix, Arizona) to semi-arid (Brownsville, Texas) up to humid (Cincinnati, Ohio). Barrier layer flux was well estimated for humid climate but overestimated for arid and semi-arid climate. Khire et al. (1997) (see also Benson and Pliska, 1996) compared measurement results from June 1992 to May 1995 to simulation results of HELP 3.01 and

203

UNSAT-H 2 (Fayer and Jones, 1990) for two cover test sections (15 cm silt, vegetated, on 90 cm compacted clay) on a landfill in Atlanta, Georgia, (humid) and in East Wenatchee, Washington, (semi-arid) respectively. Surface runoff was considerably underpredicted for Atlanta and slightly over-predicted for East Wenatchee. Actual evapotranspiration was predicted quite well for Atlanta, but over-predicted for East Wenatchee. Percolation was considerably over-predicted for Atlanta and slightly underpredicted for East Wenatchee. Scanlon et al. (2002) compared simulation results of seven models (among them HELP 3 as the only water balance model, all others are numerical models based on the Richards equation) among each other and to measured data of two different 3 m thick, non-vegetated, engineered covers in semi-arid warm (Texas, cover with a capillary barrier, from October 1997 to September 1998) and cold (Idaho, 21 July 1997 to 31 October 1999) deserts, respectively. Actual evaporation was underestimated by 34% (or 30% of the net precipitation) at the semi-arid warm site and overestimated at the semi-arid cold site. Percolation was over-predicted at both sites, but just slightly at the semi-arid warm site where 28% of the net precipitation went into storage. Scanlon et al. (2002) state the unit gradient approach (missing matric potential) in the unsaturated flow modeling and the missing seepage face option for the lower boundary condition as reasons for over-prediction of the percolation. Within the Alternative Cover Assessment Program (ACAP) of the US Environmental Protection Agency extensive investigations on several models were performed. Albright et al. (2002) compared simulation results of four models, among them HELP 3.07, to measurement results of two lysimeters at an arid site (Hanford, Washington, non-vegetated capillary barrier, from November 1987 to October 1993) and at a humid site (Coshocton, Ohio, lysimeter of the Agricultural Research Service (ARS) with a 1.5 m thick A-horizon of silt loam to loam over 94 cm thick fractured sandstone, probably with a grass vegetation, from 1985 to 1994). Percolation out of the covers was overestimated by a factor of 14 at the Hanford site, attributed to the gravity driven flow model without wicking of water (the measured percolation was 1.3% of the precipitation, the percolation simulated with HELP 3.07 was 18% of the precipitation). At Coshocton percolation was overestimated by 28.5% (no precipitation value mentioned). Albright et al. (2002) criticized the response of calculated percolation to systematically modified input variables (available water capacity, restoration layer thickness) and declared the important HELP concept ‘‘evaporative zone depth’’ as ‘‘fairly nebulous’’. Roesler et al. (2002) compared measurement results from eight ACAP sites to simulation results of HELP 3 (altogether for 17 test sections, at 5 sites for 1 year (2001) and at 3 sites for 2 years (2000–2001)) and of UNSAT-H 2. Surface runoff was estimated quite well for arid and semi-arid sites but over-predicted for humid sites. The percolation of alternative covers was predicted inaccurately but without a general bias. Percolation through composite barriers was calculated well for arid and semi-arid sites, but under-predicted for humid sites. Percolation of compacted clay barriers was extremely underestimated because preferential flow is not modeled in HELP. Lateral drainage was over-predicted in most of the cases. Albright et al. (2013) compared measurement results of final cover test sections with different design and different measurement periods ranging from 3 to 5 years at seven sites across the USA from humid to arid to simulation results of HELP 3. All test sections had a composite barrier (geomembrane over either compacted clay layer or geosynthetic clay liner) and were at least sparsely vegetated. Using as-built soil properties for simulation, surface runoff was mostly extremely over-predicted and actual evapotranspiration was largely under-predicted, lateral drainage was over-predicted and no correspondence between measured and predicted percolation

204

K.U. Berger / Waste Management 38 (2015) 201–209

was found. However, measured percolation usually was less than 1% of the precipitation. Using in-service soil properties measured at the end of the measurement periods, better but still unsatisfactory agreement was obtained for surface runoff and actual evapotranspiration. Dwyer (2003) compared measurement results from six largescale test covers (RCRA Subtitle D and C, four alternative covers) in Albuquerque, New Mexico, from May 1997 to June 2002 to simulation results of HELP 3 and UNSAT-H 2 derived in three kinds of evaluation (design process, as-built conditions and end of monitoring conditions, respectively). Using as-built as well as end of monitoring conditions, surface runoff was significantly under-predicted, measured lateral drainage was near zero and well predicted, percolation was over-predicted in most cases and actual evapotranspiration was under-predicted. Luellen and Brydges (2005) compared percolation rates through a composite liner with a degraded geomembrane calculated by HELP to results of published leakage equations. HELP under-predicted the long-term percolation rates. Hauser et al. (2005) (see also Hauser, 2009, chapter 9) compared measured data from lysimeters of the ARS at Coshocton, Ohio (humid; vegetation: meadow; 1970–1979 and 1987–1993), and Bushland, Texas (semi-arid; vegetation were crops: irrigated alfalfa 1996–1997, and irrigated corn for two seasons in 1989 and 1990) to simulation results of HELP 3.07 and EPIC 8120. At the humid site surface runoff was overestimated, actual evapotranspiration was underestimated and percolation was overestimated. At the semi-arid site actual evapotranspiration was well estimated for alfalfa, but overestimated for corn. Percolation was overestimated for alfalfa and underestimated for corn. Wattendorf (2006) compared measured data of a lysimeter with a 2.1 m thick restoration layer planted with a grove in Leonberg, South-West-Germany, 12 km WNW of Stuttgart, for the period 1 August 2001 to 31 December 2003 to simulation results of HELP 3.80 D. Percolation was underestimated by 19% or 6.7% of the precipitation. The presented validation studies comprise a variety of climate conditions and a variety of cover designs, vegetation and layer properties. The results for particular discharges or water balance values are partly contradictory. Discrepancies of measured and simulated values may be caused by the HELP model, but may be at least partly caused by other sources of error referring to the test sections like nonuniformity of the construction, aging or measurement errors, to the model application like uncertainties or errors in the estimation of the input parameters or lack of model calibration or sometimes even to the computer environment. 3.3. Operational validation for a final cover system with compacted clay liner on the landfill Hamburg-Georgswerder (Germany) The landfill Georgswerder in Hamburg (located at 53.5°N, 10.0°E, and 6 m a.s.l.) was operated from 1948 to 1979 for the disposal of municipal waste including bulky waste and construction waste. However, some hazardous waste was additionally disposed. After detecting dioxin in the landfill leakage in 1983 the City of Hamburg decided to construct a cover system on the landfill. Due to the lack of knowledge on efficiency and long-term performance of cover systems a research project was initiated at the Institute of Soil Science of the University of Hamburg. For this project six test fields (large lysimeters) with different cover systems were constructed within the cover system of the landfill in 1987 with the same materials and the same technology as the cover of the landfill to obtain representative measurement results. The test fields were operated extensively in the research phase from 1988 to 1995 and additionally to a lesser extent until 1998. From 1999 the Environmental Protection Agency of Hamburg has operated

those test fields with the layer design of the landfill as part of the aftercare program. Detailed information on the test fields and the measurement results is given in Melchior (1993) and in Melchior et al. (2010a). Each of the six test fields is 50 m long in slope direction and 10 m wide, three having a slope of 4% and three having a slope of 20%; all are exposed to the north. The layer design is shown in Fig. 1. All test fields have a 75 cm thick restoration layer of loamy sand with grass vegetation. Below is a 25 cm thick lateral drainage layer of a mixture of coarse sand and fine gravel underlain by one out of three different types of liners: (1) compacted soil liner of 60 cm thick glacial marl; (2) composite liners with a geomembrane of high-density polyethylene (HDPE) on a compacted soil liner as in (1), the geomembrane being welded in two test fields and overlapping but not welded in one test field; and (3) an extended capillary barrier. Below each liner there is a lateral drainage layer constructed in a collecting pan of HDPE to measure the discharge below the liner. Discharges of the layers, meteorological values, water contents and matric potentials of the soil, and soil properties were measured and partially are still being measured. Data from the two most intensively investigated test fields from the research period 1988–1995 have been used by the author for the operational validation of HELP since version 3.07 (Berger, 1998, 2000, 2002, 2003). The major validation results of HELP 3.95 D for the test field F1 with 4% slope and a compacted soil liner are presented here. Based on data of the German Weather Service (DWD) of the main meteorological station of Hamburg (Hamburg-Fuhlsbüttel) from 1893 to 2012, on the yearly average the validation period 1988–1995 had 72 mm more precipitation (842 mm/year) than the normal period 1961–1990 (770 mm/year) and 91 mm/year more than the 120-year period 1893–2012 (750 mm/year). There was no very dry year from 1988 to 1995 (the minimum annual precipitation was 668 mm in 1989), but one very wet year (1001 mm, 1993), the third wettest year from 1893 to 2012. The validation period 1988–1995 was not only wetter than the normal period 1961–1990 and the 120-year period 1893–2012, but also warmer. The average yearly temperature 1988–1995 was 9.74 °C, i.e. 1.04 °C warmer than 1961–1990 (8.69 °C) and 0.88 °C warmer than 1893–2012 (8.86 °C). Furthermore, the winters of 1988–1995 were relatively mild with little snow (Berger, 1998). In Table 1 the average yearly measured and simulated discharges of test field F1 from 1988 to 1995 are compared; Fig. 2 shows the corresponding cumulative daily discharges. (To be precise, the period is 05 Jan. 1988–27 Dec. 1995. The precipitation data used for simulation had 10% systematic measurement error which is a level typical for Germany; see Berger (1998) and briefly Berger (2000) for more information on the correction of the systematic measurement errors of precipitation with respect to the simulation results.) The measured as well as the simulated surface runoff is very small. The same holds for the steeper test field S1 (20% slope). The small measured surface runoff is caused by the closed canopy of the grass vegetation, the well-drained restoration layer (no stagnant moisture), and the lack of weather conditions leading to high surface runoff like very heavy rains, and rain or snow melt on frozen soil. The overestimation of surface runoff for the German climate of HELP 3.07 following from the overestimation of the length of periods with frozen soil was fixed in HELP 3.50 D (see Berger, 2003). The runoff curve numbers calculated by HELP from soil texture, stand of grass, and slope conditions are too small for the German climate (Berger, 1998). There are large differences in the yearly average of measured and simulated lateral drainage as well as liner leakage, but the sum of measured lateral drainage and liner leakage is close to the sum of the corresponding simulated values (overestimation

205

K.U. Berger / Waste Management 38 (2015) 201–209

Fig. 1. Layer design of the test fields on the landfill Hamburg-Georgswerder (after Melchior (1993)).

Table 1 Yearly average of measured and simulated water balance components of test field F1 on the landfill Hamburg-Georgswerder from 1988 to 1995.

a

Water balance component

HELP 3.95 D simulated

Measured

Difference HELP – measured

Precipitation (mm)a Surface runoff (mm) Lateral drainage (mm) Liner leakage (mm) Sum lateral drainage + liner leakage (mm) Precipitation – discharges (mm) Actual evapotranspiration (mm) Potential evapotranspiration (mm)

787.0 1.1 324.8 6.2 331.0 454.9 457.2 723.9

(787.0) 2.2 238.8 80.8 319.6 465.2 n.d. n.d.

– 1.1 +86.0 74.6 +11.4 10.3 – –

Precipitation with 10% systematic measurement errors; (): assigned from simulation input; n.d.: not determined; -: not applicable.

Fig. 2. Cumulative measured and simulated daily lateral drainage and liner leakage of test field F1 and daily and quarterly precipitation on the landfill HamburgGeorgswerder from 1988 to 1995.

by HELP 3.95 D for 3.6% or 1.4% of the precipitation; Table 1). Fig. 2 illustrates the reason for this result. The saturated hydraulic conductivity of the compacted soil liner used for simulation was the measured value from the construction of the liner leading to an approximately constant simulated liner leakage with some specific deviations caused by the modeling approach. The measured liner leakage of the compacted soil liner, however, can be divided into three distinct phases. In the first phase from beginning of 1988 to autumn 1989 the liner leakage approximately corresponded to the measured hydraulic conductivity from the construction and the compacted soil liner performed well. In the second phase from

autumn 1989 to autumn 1992 the measured liner leakage increased by a factor of 2.3 as compared to the first phase. In the third phase from autumn 1992 to the end of 1995 liner leakage increased greatly (by a factor of 31.5) as compared to the first phase, indicating deterioration of the liner and lost effectiveness. The increases were caused by a slight desiccation of the compacted soil layer and the formation of interconnected macro pores (verified by an excavation in 1995), presumably especially in the relatively warm and dry summers of 1989 and 1992 (see Melchior, 1993). (During the excavation in 1995 some precipitation of iron oxides and calcium carbonate was found in the upper 10 cm of

206

K.U. Berger / Waste Management 38 (2015) 201–209

the lateral drainage layer, indicating a slight deterioration of this layer.) Because HELP does not model the aging of a landfill profile it cannot reproduce the increasing liner leakage due to the deterioration of the compacted soil liner and hence it cannot reproduce the lateral drainage of the drainage layer above the compacted soil liner. However, HELP 3.95 D reproduces well the sum of lateral drainage and of liner leakage on the yearly average (Table 1) as well as its seasonal course (Fig. 2). Slightly increasing differences occur in the cumulative sums from 1993 to 1995. They might be caused by the relatively wet years 1993 and 1994 and by aging of the test fields, e.g. altering the pore size distribution and the saturated hydraulic conductivity of the restoration layer caused by rooting. The sum of lateral drainage and liner leakage approximately equals the inflow into the lateral drainage layer. This inflow is largely controlled by evapotranspiration and by the vertical percolation within the restoration layer. Because the other water balance components related to the restoration layer are small (surface runoff, change in water content between the beginning and the end of the validation period) or even zero (surface run-on, interflow at the bottom of the restoration layer) the good match of simulated and measured cumulative sums of lateral drainage and liner leakage means also that HELP 3.95 D reproduces the actual evapotranspiration well. 3.4. Operational validation for water balance layers based on test field data of the landfill Deetz near Berlin (Germany) The current German landfill ordinance (DepV, 2009) introduced the concept water balance layer (in German Wasserhaushaltsschicht). A water balance layer is a restoration layer which fulfills very strict specifications, and therefore may be used under specific conditions to drop the second component of a composite liner or even the entire single liner and the lateral drainage layer of a cover system. The basic idea of the water balance layer is to minimize percolation out of the layer by maximizing evapotranspiration; this requires vegetation with high transpiration, especially shrubs or coniferous forest. The requirements of water balance layers are defined more precisely by the Bundeseinheitlicher Qualitätsstandard 7-2 (BQS 7-2, Federal uniform quality standard 7-2; LAGA Adhoc-AG ‘‘Deponietechnik’’, 2012). Roughly summarized the requirements are a minimum thickness of 150 cm, a minimum available field capacity (water between pF 1.8 and 4.2) of the entire layer of 220 mm, an air capacity (pore volume below pF 1.8) of at least 8 vol.% and a boundary value of the percolation that must not be exceeded dependent on the application; usually in the worst five-year period following the initial post-construction five-year development period, the percolation must not exceed 10% of the precipitation, at the maximum 60 mm per year. (Thus in Germany water balance layers may be operated successfully only in relatively dry regions, having an average yearly precipitation of not more than about 650 mm.) The maximum percolation requirement may be proven by in-situ test field measurements or by simulation with an approved hydrological model. Therefore, an operational validation for water balance layers under an appropriate German climate was performed using measurement data of test fields of the landfill Deetz near Berlin. On the landfill Deetz (located at 52.5°N, 12.8°E, 30 m a.s.l.) of the Märkische Entsorgungsanlagen-Betriebsgesellschaft (MEAB) an extensive investigation of discharges and the water balance of different cover systems was carried out by the engineering consultants melchior + wittpohl (Hamburg); Melchior et al. (2010b) give a detailed description. Nine test fields were constructed in 2002, two of them with a water balance layer of different thickness (230 cm and 150 cm, respectively) and shrubs as vegetation. The water balance layers consisted of a 30 cm thick top soil layer

underlain by a so-called storage layer of about 200 cm and 120 cm thickness, respectively. Discharges, precipitation, water content at some dates, and soil physical properties of the materials were measured. Both test fields with the water balance layer were simulated with HELP 3.95 D for the years with completely available measurement data (2004–2011). All input data were input as measured, or as interpolated where appropriate, or as estimated when no measured data were available. No calibration was performed. The five years from 2007 to 2011 represent a load case for the water balance layers as required by BQS 7-2 and DepV (2009), and were therefore used as the assessment period for the percolation criterion (though 2007 is within the initial, post-construction development period and the vegetation was still developing). Based on data of the German Weather Service (DWD) of the meteorological station Potsdam (12 km WSW from Deetz) from 1893 to 2012, on the yearly average the assessment period 2007–2011 had 51 mm more precipitation (641 mm/year) than the normal period 1961–1990 (590 mm/year) and 53 mm more than the entire period 1893–2012 (588 mm/year). Furthermore, the year with the highest precipitation, 2007 (798 mm), occurred in the assessment period. 2007 had a very wet summer (April to September) with 521 mm precipitation. In the test field with 230 cm thick water balance layer in the assessment period 2007 to 2011 HELP 3.95 D overestimated the percolation by 26 mm per year (3.9% of the precipitation; see Table 2). Thus, the design of this test field meets the percolation criterion of not more than 60 mm/year according to the measurement results as well as to the simulation results and HELP 3.95 D gives a conservative estimate of the percolation. The comparison of measured and simulated cumulative daily percolation shows some common characteristics and two major discrepancies (see Fig. 3). The high measured percolation at the beginning of 2004 is not reproduced by HELP because the soil water content at the beginning of the simulation was calculated by HELP as the water content at the end of the so-called initialization year. Usually this is the procedure recommended by the author especially if measured water contents are not available as it is in this case. Alternatively, the soil water content at the beginning could be calibrated based on the percolation data. Due to the water content at the beginning of 2004 the simulated percolation in 2004 and 2005 starts too late. Furthermore, the course of the simulated percolation in 2004 and 2005 is smoother than that of the measured percolation. Presumably this follows from the HELP modeling approach that neither accounts for the matric potential in the hydraulic gradient nor for a preferential flow in coarse pores (one-domain concept of the pore space). However, the lower dynamic of simulated percolation may to some extent also be caused by the soil physical description of the profile. For each of the two layers only one parameter set of pore size distribution and saturated hydraulic conductivity was available from the time of construction in 2002, and from 2006. Data of the 200 cm thick storage layer refer to the upper part; there are no data to differentiate the storage layer in depth. The first major discrepancy between measured and simulated percolation occurs in the first half of 2006 where HELP does not reproduce the measured percolation (altogether 53 mm, mainly occurring between 17 February 2006 and 27 May 2006). The simulated water content in the evaporative zone (output by HELP but not shown here) in the first half of 2006 is above field capacity (at 330 hPa water suction). However, most of the water is in the upper part of the evaporative zone while its lower part and the part of the storage layer below the evaporative zone are relatively dry (a HELP test version by the author also outputs the water content of the model’s internally-computed segments). Therefore, no percolation is simulated. As a potential notable event a 37-day period of frozen soil occurs in the simulated data from 16 January 2006 to 21

207

K.U. Berger / Waste Management 38 (2015) 201–209 Table 2 Yearly averages of measured and simulated water balance components of test field 1 with 230 cm water balance layer on the landfill Deetz from 2007 to 2011. Water balance component

HELP 3.95 D simulated

Measured

Difference HELP – Measured

Precipitation (mm) Surface runoff (mm) Potential evapotranspiration (mm) Actual evapotranspiration (mm) Percolation (mm) Percolation (% of precipitation)

(677.1) 7.4 751.2 596.7 48.6 7.2

677.1 n.d. n.d. n.d. 22.5 3.3

– – – – +26.1 +3.9

(): Input as measured; n.d.: not determined; -: not applicable.

Fig. 3. Cumulative measured and simulated daily percolation of test field 1 with 230 cm thick water balance layer and quarterly precipitation on the landfill Deetz from 2004 to 2011.

February 2006. Therefore, also an actually frozen soil seems to be likely around that time. However, to explain the measured percolation as a reaction of the actual thawing of the soil is speculation and should be reproduced by HELP. Furthermore, no deterioration of the vegetation was observed. In summary this first discrepancy remains unexplained. The second major discrepancy occurs from the second quarter 2007 to the second quarter 2008, and especially in the first half year of 2008. The percolation of this period results from the extremely high precipitation in the summer of 2007 (April to September) when 605 mm precipitation were measured on the landfill Deetz (i.e. about the average yearly precipitation). From 1 April 2007 to 30 June 2008 106 mm percolation was measured, but HELP calculates 217 mm. The simulated water contents of the evaporative zone (not shown here) in this period are considerably above field capacity, and thus the simulated percolation is plausible. The abrupt beginning of the simulated percolation at the end of 2007 and the abrupt end in the middle of 2008 is caused by the modeling approach of HELP 3.95 D. Percolation out of the evaporative zone occurs only at water contents in the bottom segment above field capacity. The large overestimation (111 mm) of this percolation event may be caused by several reasons. First there is no evidence for measurement errors in the percolation and precipitation data. However, as noted above there are uncertainties in the soil physical parameters, especially of the 200-cm thick storage layer. There is a difference between the simulated potential and actual evapotranspiration (ETp–ETa) at the end of the growing season in October 2007 (13.5 mm) and at the beginning of the growing season in April 2008 (28 mm). The lag of ETa behind ETp at the beginning of

the growing season is typical in HELP for German climate conditions, and seems to be caused by a plant growth rate which is too small for the modeled ETp at that time (Berger, 1998). In total, between September 2007 and June 2008 the simulated ETa is 67 mm smaller than the simulated ETp and thus may explain a large portion of the overestimated percolation. Furthermore, another reason of the overestimated percolation may be the approximately unit-gradient approach of the unsaturated vertical water routing. Actually the water storage in the water balance layer may be larger than calculated by HELP. In summary the second discrepancy can be explained only partially.

4. Future prospects of the HELP model In the literature on the validation of HELP version 3 several limitations and unrealistic behavior of the model have been identified which are possible starting-points for improving the model. However, some findings are contradictory (e.g. some studies stated an over-prediction of surface runoff, others an under-prediction). Some weak points of the original HELP 3.07 were fixed in the HELP-D versions up to HELP 3.95 D. Other limitations are largely caused by the maximum time resolution of the model and its input data (one day, daily weather data). Surface runoff largely depends on rainfall intensity. Therefore, any model using daily precipitation data can give only a rough estimate of surface runoff, especially in poorly vegetated regions where large amounts of precipitation occur in heavy storms (arid and semi-arid regions). Furthermore, the curve number method of the Soil Conservation Service underlying the surface runoff sub-model is an empirically derived

208

K.U. Berger / Waste Management 38 (2015) 201–209

procedure that therefore strictly speaking has to be calibrated for every region and time period where it shall be applied. Similarly, although not as distinctive, any model using daily precipitation data cannot reproduce short-term peaks of lateral drainage from lateral drainage layers in final covers. The price to overcome these limitations caused by the time resolution is very high because (1) weather data in the required time resolution in the time scale of minutes or hours are hardly or even not available for many locations, (2) the model has to model diurnal variations of evapotranspiration, and (3) computing time will increase considerably. The author plans the following improvements for a future version HELP 4 D:  The unsaturated vertical flow shall be improved by including the matric potential in the hydraulic gradient. The current unit gradient approach often has been criticized in the literature on validation mentioned in Section 3.2. Furthermore a seepage face option for the lower boundary condition of a vertical percolation layer shall be added. The current approach often leads to an over-prediction and thus to a conservative estimate of percolation.  The model’s internal subdivision of layers into segments, introduced by the concept of the evaporative zone, does not sufficiently consider the thickness of the evaporative zone and the layers. Independent of its depth the evaporative zone always is divided into seven segments. Dependent on the thickness non-barrier layers outside the evaporative zone are divided into one to three segments. Thus segments may become extremely thick e.g. for a several meter thick waste layer. The author plans a standard segment thickness of e.g. 10 cm (4 in.) with refinements where necessary, e.g. near the soil surface. In combination with the modified hydraulic gradient the calculation of the unsaturated vertical flow and of the depth dependent water contents should become more realistic. (Computing time must be kept in view, but is expected to remain acceptable.)  In Germany sometimes other kinds of vegetation than grasses are desired, e.g. shrubs or trees, either for landscaping reasons or to maximize evapotranspiration (see Section 3.4). The evapotranspiration and vegetative growth and decay sub-models shall be enhanced to model these types of vegetation.

Acknowledgements The author thanks Paul Schroeder, Ph.D., from the US Army Corps of Engineers, Waterways Experiment Station in Vicksburg, Mississippi, and all co-developers of the HELP model for their extensive model development and documentation including the release of the source code. The author thanks Dr. habil. Stefan Melchior, at that time at the Institute of Soil Science of the University of Hamburg, Prof. Dr. Günter Miehlich and all co-workers of the Project Georgswerder for the careful planning and realization of the test field measurements on the landfill Hamburg-Georgswerder, and also Dr. Volker Sokollek (retired December 2013) from the Environmental Protection Agency of the City of Hamburg for his long-time and hydrologic sound support of the project. The author’s validation study on HELP 3.07 was funded by the German Ministry of Education and Research (BMBF) under the support code 147 1038. The author thanks the Märkische EntsorgungsanlagenBetriebsgesellschaft, Potsdam, in person Mr. Steffen Raabe, and Dr. habil. Stefan Melchior and Dr. Bernd Steinert from melchior + wittpohl engineering consultants, Hamburg, for the test field data of the landfill Deetz and the review of the corresponding section of this paper. Finally the author thanks Dr. Benjamin Runkle and the anonymous reviewers for valuable comments.

References Albright, W.H., Gee, G.W., Wilson, G.V., Fayer, M.J., 2002. Alternative Cover Assessment Project Phase I Report. Publication No. 41183, Division of Hydrologic Sciences, Desert Research Institute. 195 p. Albright, W.H., Benson, C.H., Apiwantragoon, P., 2013. Field hydrology of landfill final covers with composite barrier layers. J. Geotech. Geoenviron. Eng. 139 (1), 1–12. Alslaibi, T.M., Abustan, I., Mogheir, Y.K., Afifi, S., 2013. Quantification of leachate discharged to groundwater using the water balance method and the Hydrologic Evaluation of Landfill Performance (HELP) model. Waste Manage. Res. 31 (1), 50–59. http://dx.doi.org/10.1177/0734242X12465462. Arnold, J.G., Williams, J.R., Nicks, A.D., Sammons, N.B., 1990. SWRRB, A Basin Scale Simulation Model for Soil and Water Resources Management. Texas A&M University Press, College Station, 142 p., appendixes and 8 floppy discs. Benson, C.H., Pliska, R.J., 1996. Final Covers: HELP Needs Help From the Field. Waste Age, pp. 89–98. Berger, K., 1998. Validierung und Anpassung des Simulationsmodells HELP zur Berechnung des Wasserhaushalts von Deponien für deutsche Verhältnisse. [Validation and adaptation of the simulation model HELP for calculating the water balance of landfills to German conditions.] Umweltbundesamt [Federal Environment Agency of Germany], Fachgebiet III 3.6, Berlin, xii+557 pp. (in German). Berger, K., 2000. Validation of the Hydrologic Evaluation of Landfill Performance (HELP) model for simulating the water balance of cover systems. Environ. Geol. 39 (11), 1261–1274. Berger, K., 2002. Potential and limitations of applying HELP model for surface covers. Pract. Periodical Hazard., Tox., Radioact. Waste Manage. 6 (3), 192–203. Berger, K., 2003. Validation and enhancement of the HELP model to simulate the water balance of surface covers. In: Christensen et al. (Eds.), Sardinia 2003. Ninth International Waste Management and Landfill Symposium, CISA, Italy, 10 p. Berger, K., 2008. Validation and enhancement of Hydrologic Evaluation of Landfill Performance (HELP) Model. In: Global Waste Management Symposium, 7–10 September 2008, Copper Conference Center, Colorado, USA, 10 p. Berger, K., Schroeder, P.R., 2013. The Hydrologic Evaluation of Landfill Performance Model. Version HELP 3.95 D. Software and electronic documents as PDF: User’s Guide. Supplement to the Engineering Documentation of HELP 3.07. Institute of Soil Science, University of Hamburg, Germany, . Bonaparte, R., Daniel, D.E., Koerner, R.M., 2002. Assessment and Recommendations for Improving the Performance of Waste Containment Systems. US Environmental Protection Agency, National Risk Management Research Laboratory, Cincinnati, OH, 1039 p., EPA/600/R-02/099. Bonaparte, R., Gross, B.A., Daniel, D.E., Koerner, R.M., Dwyer, S.F., 2004. Technical Guidance For RCRA/CERCLA Final Covers (Draft). U.S. Environmental Protection Agency, Office of Solid Waste and Emergency Response, Washington DC, EPA 540-R-04-007, 421 p. BQS 7-2: see LAGA Ad-hoc-AG ‘‘Deponietechnik’’. DepV, 2009. Verordnung über Deponien und Langzeitlager (Deponieverordnung – DepV). Bundesgesetzblatt Jahrgang 2009 Teil I Nr. 22, Bonn, 29 April 2009 (in German). Dho, N.Y., Koo, J.K., Lee, S.R., 2002. Prediction of leachate level in Kimpo metropolitan landfill site by total water balance. Environ. Monit. Assess. 73, 207–219. Dwyer, S.F., 2003. Water Balance Measurements and Computer Simulations of Landfill Covers. Ph.D. Thesis, University of New Mexico, Albuquerque, NM, USA, 250 p., (downloaded 30.04.13). Fayer, M.J., Jones, T.L., 1990. UNSAT-H Version 2: Unsaturated Soil Water and Heat Flow Model. PNL-6779, Pacific Northwest Laboratory, Richland, Washington. Fleenor, W.E., King, I.P., 1995. Identifying limitations on use of the HELP model. In: Dunn, R.J., Singh, U.P. (Eds.), Landfill Closures. Environmental Protection and Land Recovery. ASCE, Geotechnical Special Publ. No. 53, pp. 121–138. German Weather Service (DWD): Weather data of Hamburg-Fuhlsbüttel and Potsdam from 1893 to 2012, . Hauser, V.L., 2009. Evapotranspiration Covers for Landfills and Waste Sites. CRC Press, Boca Raton, FL. Hauser, V.L., Gimon, D.M., Bonta, J.V., Howell, T.A., Malone, R.W., Williams, J.R., 2005. Models for hydrologic design of evapotranspiration landfill covers. Environ. Sci. Technol. 39 (18), 7226–7233. Khire, M.V., Benson, C.H., Bosscher, P.J., 1997. Water balance modeling of earthen final covers. J. Geotech. Geoenviron. Eng. 123 (8), 744–754. Koerner, R.M., Daniel, D.E., 1997. Final Covers for Solid Waste Landfills and Abandoned Dumps. ASCE Press, Reston, Virginia, Th. Telford, London. LAGA Ad-hoc-AG ‘‘Deponietechnik’’, 2012. Bundeseinheitlicher Qualitätsstandard 7-2: Wasserhaushaltsschichten in Deponieoberflächenabdichtungssystemen, vom 20.10.2011, published 20.03.12, 33 p., (accessed 06.06.2014, in German). Lange, D.A., Cellier, B.F., Dunchak, T., 1999. A case study of the HELP model: actual versus predicted leachate production rates at a MSW landfill in Northeastern Ohio – 1992 to 1998. In: SWANA 4th Annual Landfill Symposium, pp. 293–313. Luellen, J., Brydges, J., 2005. Long-term hydraulic performance evaluation for a multilayer closure cap. Practice periodical of hazardous, toxic, and radioactive waste management 9, Special Issue Title: Innovative Barrier Systems for Waste Containment, pp. 237–244.

K.U. Berger / Waste Management 38 (2015) 201–209 McEnroe, B.M., Schroeder, P.R., 1988. Leachate collection in landfills: steady case. J. Environ. Eng. 114 (5), 1052–1062. Melchior, S., 1993. Wasserhaushalt und Wirksamkeit mehrschichtiger Abdecksysteme für Deponien und Altlasten. Dissertation Ph.D. thesis, Hamburger Bodenkundliche Arbeiten 22, 330 p. + Appendix (in German). Melchior, S., Sokollek, V., Berger, K., Vielhaber, B., Steinert, B., 2010. Results from 18 years of in situ performance testing of landfill cover systems in Germany. J. Environ. Eng. 136 (8), 815–823. Melchior, S., Steinert, B., Raabe, S., 2010b. Ergebnisse zur Wirksamkeit ausgewählter Oberflächenabdichtungssysteme der Versuchsfelder der MEAB auf der Deponie Deetz. In: 21. Nürnberger Deponieseminar 2010, Tagungsband, pp. 105–116 (in German). Richardson, C.W., Wright, D.A., 1984. WGEN: A Model for Generating Daily Weather Variables. U.S. Department of Agriculture, Agricultural Research Service, ARS-8, 83 p. Roesler, A.C., Benson, C.H., Albright, W.H., 2002. Field hydrology and model predictions for final covers in the Alternative Cover Assessment Program – 2002. Geo Engineering Report 02-08, Geo Engineering Program, University of Wisconsin, Madison, Wisconsin, USA, 227 p. Scanlon, B.R., Christman, M., Reedy, R.C., Porro, I., Simunek, J., Flerchinger, G.N., 2002. Intercode comparisons for simulating water balance of surficial sediments in semiarid regions. Water Resour. Res. 38 (12), 59-1–59-16. Schlumberger Water Services: Visual HELP, (accessed 06.06.14). Schroeder, P.R., Peyton, R.L., 1987a. Verification of the Hydrologic Evaluation of Landfill Performance (HELP) Model Using Field Data. Environmental Protection Agency, Cincinnati, Ohio, 180 p., (EPA/600/2-87/050). Schroeder, P.R., Peyton, R.L., 1987b. Verification of the Lateral Drainage Component of the HELP Model Using Physical Models. Environmental Protection Agency, Cincinnati, Ohio, 117 p., EPA/600/2-87/049.

209

Schroeder, P.R., Gibson, A.C., Smolen, M.D., 1984a. The Hydrologic Evaluation of Landfill Performance (HELP) Model: Volume II. Documentation for Version 1. EPA/530-SW-84-010, Environmental Protection Agency, Cincinnati, Ohio, Draft Report, 256 p. Schroeder, P.R., Morgan, J.M., Walski, T.M., Gibson, A.C., 1984b. The Hydrologic Evaluation of Landfill Performance (HELP) Model, Volume I. User’s Guide for Version 1. EPA/530-SW-84-009, Environmental Protection Agency, Cincinnati, Ohio, 120 p. Schroeder, P.R., N.M. Aziz, C.M. Lloyd, P.A. Zappi, 1994a. The Hydrologic Evaluation of Landfill Performance (HELP) Model. User’s Guide for Version 3. EPA/600/R94/168a. US EPA Risk Reduction Engineering Laboratory. Cincinnati, Ohio, 84 p. and appendix. Schroeder, P.R., Dozier, T.S., Zappi, P.A. McEnroe, B.M., Sjostrom, J.W., Peyton, R.L., 1994b. The Hydrologic Evaluation of Landfill Performance (HELP) Model. Engineering Documentation for Version 3. EPA/600/R-94/168b. US Environmental Protection Agency, Cincinnati, Ohio, 116 p., (accessed 06.06.2014). Shariatmadari, N., Abdoli, M.A., Ghiasinejad, H., Mansouri, A., Alimohammadi, P., 2010. Analysis of help model application in semi-arid areas, study on Tehran test cells. Int. J. Civ. Eng. 8 (2), 174–186. Waterloo Hydrogeologic: see Schlumberger Water Services. Wattendorf, P., 2006: The water balance of recultivation layers – results of measurements and water balance modeling. In: Landfill Workshop Zittau – Liberec 2005, Nov. 10–11, 2005. Conference Proceedings. Wissenschaftliche Berichte 86, No. 2147-2122, University of Applied Science Zittau/Goerlitz, pp. 207–220.

On the current state of the Hydrologic Evaluation of Landfill Performance (HELP) model.

The Hydrologic Evaluation of Landfill Performance (HELP) model is the most widely applied model to calculate the water balance of cover and bottom lin...
822KB Sizes 5 Downloads 8 Views