Credit refers to the quantity of stormwater or pollutant reduction achieved either by an individual BMP or cumulatively with multiple BMPs. Stormwater credits are a tool for local stormwater authorities who are interested in
This page provides a discussion of how bioretention practices can achieve stormwater credits. Bioretention systems with and without underdrains are both discussed, with separate sections for each type of system as appropriate. In this discussion, bioretention systems with an underdrain are called biofiltration systems, while bioretention systems with no underdrain are called bioinfiltration systems.
Bioretention is a terrestrial-based (up-land as opposed to wetland) water quality and water quantity control process. Bioretention consists of an engineered soil layer designed to treat stormwater runoff via filtration through plant and soil media, and evapotranspiration from plants. Pretreatment is REQUIRED for all biofiltration facilities to settle particulates before entering the BMP. Bioretention practices may be built with or without an underdrain. Other common components of bioretention systems may include a stone aggregate layer to allow for increased retention storage and an impermeable liner on the bottom or sides of the facility if located near buildings, subgrade utilities, or in karst formations. Bioretention is a versatile stormwater treatment method applicable to all types of settings such as landscaping islands, cul-de-sacs, parking lot margins, commercial setbacks, open space, rooftop drainage, and streetscapes.
Systems with no underdrain are called bioinfiltration, while those with an underdrain are called biofiltration. Biofiltration, commonly termed bioretention with underdrains, is primarily a stormwater quality control practice. Some water quantity reduction can be achieved through infiltration below the underdrain, particularly if the underdrain is raised above the bottom of the Best Management Practice (BMP), and through evapotranspiration. Biofiltration includes an underdrain layer to collect the filtered runoff for downstream discharge.
See Bioretention terminology for a discussion of different types of bioretention systems.
Biofiltration has one of the highest nutrient and pollutant removal efficiencies of any BMP (Mid-America Regional Council and American Public Works Association Manual of Best Management Practice BMPs for Stormwater Quality, 2012). Biofiltration pollutant removal primarily occurs through filtering by the engineered soil media and vegetation. Biofiltration also provides pollutant removal and volume reduction through evaporation, infiltration, transpiration, biological and microbiological uptake, and soil adsorption; the extent of these benefits is highly dependent on site specific conditions and design. In addition to phosphorus and total suspended solids (TSS), which are discussed in greater detail below, biofiltration treats a wide variety of other pollutants.
Removal of phosphorus is dependent on the engineered media. Media mixes with high organic matter content typically leach phosphorus and can therefore contribute to water quality degradation. The Manual provides a detailed discussion of media mixes, including information on phosphorus retention.
Stormwater treatment trains are multiple Best Management Practice (BMPs) that work together to minimize the volume of stormwater runoff, remove pollutants, and reduce the rate of stormwater runoff being discharged to Minnesota wetlands, lakes and streams. Under the Treatment Train approach, stormwater management begins with simple methods that prevent pollution from accumulating on the land surface, followed by methods that minimize the volume of runoff generated, and is completed by BMPs that reduce the pollutant concentration and/or volume of stormwater runoff. Biofiltration facilities are typically located in upland areas of the stormwater treatment train, controlling stormwater runoff close to the source.
This section describes the basic concepts and equations used to calculate credits for volume, Total Suspended Solids (TSS) and Total Phosphorus (TP). Specific methods for calculating credits are discussed later in this article.
Biofiltration generates credits for Total Suspended Solids (TSS) and Total Phosphorus (TP). Biofiltration does not substantially reduce the volume of runoff but may qualify for a partial volume credit as a result of evapotranspiration, infiltration occurring through the sidewalls above the underdrain, and infiltration below the underdrain piping. Biofiltration is also effective at reducing concentrations of other pollutants including nitrogen, metals, bacteria, and hydrocarbons. This article does not provide information on calculating credits for pollutants other than TSS and TP, but references are provided that may be useful for calculating credits for other pollutants.
In developing the credit calculations, it is assumed the bioretention practice is properly designed, constructed, and maintained in accordance with the Minnesota Stormwater Manual. If any of these assumptions is not valid, the BMP may not qualify for credits or credits should be reduced based on reduced ability of the BMP to achieve volume or pollutant reductions. For guidance on design, construction, and maintenance, see the appropriate article within the bioretention section of the Manual.
In the following discussion, the water quality volume (WQV) is delivered instantaneously to the BMP. The WQV is stored as water ponded above the filter media and below the overflow point in the BMP. The WQV can vary depending on the stormwater management objective(s). For construction stormwater, the water quality volume is 1 inch off new impervious surface. For MIDS, the WQV is 1.1 inches.
In reality, some water will infiltrate through the bottom and sidewalls of the BMP as a rain event proceeds. The instantaneous volume method therefore may underestimate actual volume and pollutant losses.
Volume credits for biofiltration are available only if the BMP permanently removes a portion of the stormwater runoff via infiltration through sidewalls or beneath the underdrain piping, or through evapotranspiration. These credits are assumed to be instantaneous values based on the design capacity of the BMP for a specific storm event. Instantaneous volume reduction, also termed event based volume reduction, can be converted to annual volume reduction percentages using the MIDS calculator or other appropriate modeling tools.
Volume credits for biofiltration basins with underdrains are calculated by a combination of infiltration through the unlined sides and bottom of the basin, the volume loss through evapotranspiration (ET), and the retention volume below the underdrain, if applicable (this is based on the assumption that this stored water will infiltrate into the underlying soil). The main design variables impacting the volume credits include whether the underdrain is elevated above the native soils and if an impermeable liner on the sides or bottom of the basin is used. Other design variables include surface area at overflow, media top surface area, underdrain location, and basin bottom locations, total depth of media, soil water holding capacity and media porosity, and infiltration rate of underlying soils.
The volume credit (V) for biofiltration basins with underdrains, in cubic feet, is given by
\( V = V_{inf_b} + V_{inf_s} + V_{ET} + V_U \)
where:
Volume credits for infiltration through the bottom of the basin (Vinfb) are accounted for only if the bottom of the basin is not lined. As long as water continues to draw down, some infiltration will occur through the bottom of the BMP. However, it is assumed that when an underdrain is included in the installation, the majority of water will be filtered through the media and exit through the underdrain. Because of this, the drawdown time is likely to be short. Volume credit for infiltration through the bottom of the basin is given by
\( V_{inf_B} = A_B\ DDT\ I_R/12 \)
where
The drawdown time is typically a maximum of 48 hours, which is designed to be protective of plants grown in the media. The Construction Stormwater permit requires drawdown within 48 hours and recommends 24 hours when discharges are to a trout stream. With a properly functioning underdrain, the drawdown time is likely to be considerably less than 48 hours.
Volume credit for infiltration through the sides of the basin is accounted for only if the sides of the basin are not lined with an impermeable liner. Volume credit for infiltration through the sides of the basin is given by
\( V_{inf_s} = (A_O - A_U)\ DDT\ I_R/12 \)
where
This equation assumes water will infiltrate through the entire sideslope area during the period when water is being drawn down. This is not the case, however, since the water level will decline in the BMP. The MIDS calculator assumes a linear drop in water level and thus divides the right hand term in the above equation by 2.
Volume credit for media storage capacity below the underdrain (VU) is accounted for only if the underdrain is elevated above the native soils. Volume credit for media storage capacity below the underdrain is given by
\( V_U = n\ D_U\ (A_U + A_B)/2 \)
where
This is an instantaneous volume. This will somewhat overestimate actual storage when the majority of water is being captured by the underdrains. This equation also assumes this stored water will infiltrate into the underlying soil. Thus, using the entire porosity will overestimate the actual drainage. The MIDS calculator assumes water between the soil porosity and field capacity will infiltrate into the underlying soil.
The volume of water lost through ET is assumed to be the smaller of two calculated values: potential ET and measured ET. Potential ET (ETpot) is equal to the amount of water stored in the basin between field capacity and the wilting point. Measured ET (ETmea) is the amount of water lost to ET as measured using available data and is assumed to be 0.2 inches/day. ETmea is converted to ET by multiplying by a factor of 0.5. ET is considered to occur over a period equal to the drawdown time of the basin. Volume credit for evapotranspiration is given by the lesser of
\( ET_{mea} = (0.2/12)\ A\ 0.5\ t \) \( ET_{pot} = D\ A\ C_S \)
where
Provided soil water content is greater than the wilting point, ET will continually occur during the non-frozen period. However, because the above volume calculations are event based, t will be equal to the time between rain events. In the MIDS calculator, a value of 3 days is used because this is the average number of days between precipitation events. ET will occur over the entire media depth. D may therefore be set equal to the media depth (DM). In this case, the value for A would be the average area through the entire depth of the media. The MIDS calculator limits ET to the area above the underdrain. If infiltration is being computed through the bottom and sidewalls of the basin, then CS would be field capacity minus the wilting point of soils (cubic feet per cubic foot) since water above the field capacity would infiltrate (or go to an underdrain).
The volume of water passing through underdrains can be determined by subtracting the volume loss (V) from the volume of water instantaneously captured by the BMP. No volume reduction credit is given for filtered stormwater that exits through the underdrain, but the volume of filtered water can be used in the calculation of pollutant removal credits through filtration.
The volume reduction credit (V) can be converted to an annual volume if desired. This conversion can be generated using the MIDS calculator or other appropriate modeling techniques. The MIDS calculator obtains the percentage annual volume reduction through performance curves developed from multiple modeling scenarios using the volume reduction capacity for biofiltration, the infiltration rate of the underlying soils, and the contributing watershed size and imperviousness.
TSS reduction credits correspond with volume reduction through infiltration and filtration of water captured by the biofiltration basin and are given by
\( M_{TSS} = M_{TSS_i} + M_{TSS_f} \)
where
Pollutant removal for infiltrated water is assumed to be 100 percent. The mass of pollutant removed through infiltration, in pounds, is given by
\( M_{TSS_i} = 0.0000624\ (V_{inf_b} + V_{inf_s} + V_U)\ EMC_{TSS} \)
where
The EMCTSS entering the BMP is a function of the contributing land use and treatment by upstream tributary BMPs.
Removal for the filtered portion is less than 100 percent. The Stormwater Manual provides a recommended value of 85 percent removal for filtered water, while the MIDS calculator provides a value of 65 percent. Alternate justified percentages for TSS removal can be used if proven to be applicable to the BMP design.
The mass of pollutant removed through filtration, in pounds, is given by
\( M_{TSS_f} = 0.0000624\ (V_{total} - (V_{inf_b} + V_{inf_s} + V_U))\ EMC_{TSS}\ R_{TSS} \)
where
The above calculations may be applied on an event or annual basis using the appropriate units.
Total phosphorus (TP) reduction credits correspond with volume reduction through infiltration and filtration of water captured by the biofiltration basin and are given by
\( M_{TP} = M_{TP_i} + M_{TP_f} \)
where
Pollutant removal for infiltrated water is assumed to be 100 percent. The mass of pollutant removed through infiltration, in pounds, is given by
\( M_{TP_i} = 0.0000624\ (V_{inf_b} + V_{inf_s} + V_U)\ EMC_{TP} \)
where
The EMCTP entering the BMP is a function of the contributing land use and treatment by upstream tributary BMPs.
The filtration credit for TP in biofiltration with underdrains assumes removal rates based on the soil media mix used and the presence or absence of amendments. Soil mixes with more than 30 mg/kg phosphorus (P) content are likely to leach phosphorus and do not qualify for a water quality credit. If the soil phosphorus concentration is less than 30 mg/kg, the mass of phosphorus removed through filtration, in pounds, is given by
\( M_{TP_f} = 0.0000624\ (V_{total} - (V_{inf_b} + V_{inf_s} + V_U))\ EMC_{TP}\ R_{TP} \)
Again, assuming the phosphorus content in the media is less than 30 milligrams per kilogram, the removal efficiency (RTP) provided in the Stormwater Manual is a function of the fraction of phosphorus that is in particulate or dissolved form, the depth of the media, and the presence or absence of soil amendments. For the purpose of calculating credits it can be assumed that TP in storm water runoff consists of 55 percent particulate phosphorus (PP) and 45 percent dissolved phosphorus (DP). The removal efficiency for particulate phosphorus is 80 percent. The removal efficiency for dissolved phosphorus is 20 percent if the media depth is 2 feet or greater. The efficiency decreases by 1 percent for each 0.1 foot decrease in media thickness below 2 feet. If a soil amendment is added to the BMP design, an additional 40 percent credit is applied to dissolved phosphorus. Thus, the overall removal efficiency, (RTP), expressed as a percent removal of total phosphorus, is given by
\( R_{TP} = [0.8 * 0.55] + [0.45 * ((0.2 * (D_{MU_{max=2}})/2) + 0.40_{if amendment is used})] * 100 \)
where
This section provides specific information on generating and calculating credits from biofiltration for volume, Total Suspended Solids (TSS) and Total Phosphorus (TP). Stormwater runoff volume and pollution reductions (“credits”) may be calculated using one of the following methods:
Users may opt to use a water quality model or calculator to compute volume, TSS and/or TP pollutant removal for the purpose of determining credits for biofiltration. The available models described below are commonly used by water resource professionals, but are not explicitly endorsed or required by the Minnesota Pollution Control Agency. Furthermore, many of the models listed below cannot be used to determine compliance with the Construction Stormwater General permit since the permit requires the water quality volume to be calculated as an instantaneous volume.
Use of models or calculators for the purpose of computing pollutant removal credits should be supported by detailed documentation, including:
The following table lists water quantity and water quality models that are commonly used by water resource professionals to predict the hydrologic, hydraulic, and/or pollutant removal capabilities of a single or multiple stormwater BMPs. The table can be used to guide a user in selecting the most appropriate model for computing volume, TSS, and/or TP removal for biofiltration BMPs.
Comparison of stormwater models and calculators. Additional information and descriptions for some of the models listed in this table can be found at this link. Note that the Construction Stormwater General Permit requires the water quality volume to be calculated as an instantaneous volume, meaning several of these models cannot be used to determine compliance with the permit.
Link to this table
Access this table as a Microsoft Word document: File:Stormwater Model and Calculator Comparisons table.docx.
Model name | BMP Category | Assess TP removal? | Assess TSS removal? | Assess volume reduction? | Comments | |||||
---|---|---|---|---|---|---|---|---|---|---|
Constructed basin BMPs | Filter BMPs | Infiltrator BMPs | Swale or strip BMPs | Reuse | Manu- factured devices |
|||||
Center for Neighborhood Technology Green Values National Stormwater Management Calculator | X | X | X | X | No | No | Yes | Does not compute volume reduction for some BMPs, including cisterns and tree trenches. | ||
CivilStorm | Yes | Yes | Yes | CivilStorm has an engineering library with many different types of BMPs to choose from. This list changes as new information becomes available. | ||||||
EPA National Stormwater Calculator | X | X | X | No | No | Yes | Primary purpose is to assess reductions in stormwater volume. | |||
EPA SWMM | X | X | X | Yes | Yes | Yes | User defines parameter that can be used to simulate generalized constituents. | |||
HydroCAD | X | X | X | No | No | Yes | Will assess hydraulics, volumes, and pollutant loading, but not pollutant reduction. | |||
infoSWMM | X | X | X | Yes | Yes | Yes | User defines parameter that can be used to simulate generalized constituents. | |||
infoWorks ICM | X | X | X | X | Yes | Yes | Yes | |||
i-Tree-Hydro | X | No | No | Yes | Includes simple calculator for rain gardens. | |||||
i-Tree-Streets | No | No | Yes | Computes volume reduction for trees, only. | ||||||
LSPC | X | X | X | Yes | Yes | Yes | Though developed for HSPF, the USEPA BMP Web Toolkit can be used with LSPC to model structural BMPs such as detention basins, or infiltration BMPs that represent source control facilities, which capture runoff from small impervious areas (e.g., parking lots or rooftops). | |||
MapShed | X | X | X | X | Yes | Yes | Yes | Region-specific input data not available for Minnesota but user can create this data for any region. | ||
MCWD/MWMO Stormwater Reuse Calculator | X | Yes | No | Yes | Computes storage volume for stormwater reuse systems | |||||
Metropolitan Council Stormwater Reuse Guide Excel Spreadsheet | X | No | No | Yes | Computes storage volume for stormwater reuse systems. Uses 30-year precipitation data specific to Twin Cites region of Minnesota. | |||||
MIDS Calculator | X | X | X | X | X | X | Yes | Yes | Yes | Includes user-defined feature that can be used for manufactured devices and other BMPs. |
MIKE URBAN (SWMM or MOUSE) | X | X | X | Yes | Yes | Yes | User defines parameter that can be used to simulate generalized constituents. | |||
P8 | X | X | X | X | Yes | Yes | Yes | |||
PCSWMM | X | X | X | Yes | Yes | Yes | User defines parameter that can be used to simulate generalized constituents. | |||
PLOAD | X | X | X | X | X | Yes | Yes | No | User-defined practices with user-specified removal percentages. | |
PondNet | X | Yes | No | Yes | Flow and phosphorus routing in pond networks. | |||||
PondPack | X | [ | No | No | Yes | PondPack can calculate first-flush volume, but does not model pollutants. It can be used to calculate pond infiltration. | ||||
RECARGA | X | No | No | Yes | ||||||
SHSAM | X | No | Yes | No | Several flow-through structures including standard sumps, and proprietary systems such as CDS, Stormceptors, and Vortechs systems | |||||
SUSTAIN | X | X | X | X | X | Yes | Yes | Yes | Categorizes BMPs into Point BMPs, Linear BMPs, and Area BMPs | |
SWAT | X | X | X | Yes | Yes | Yes | Model offers many agricultural BMPs and practices, but limited urban BMPs at this time. | |||
Virginia Runoff Reduction Method | X | X | X | X | X | X | Yes | No | Yes | Users input Event Mean Concentration (EMC) pollutant removal percentages for manufactured devices. |
WARMF | X | X | Yes | Yes | Yes | Includes agriculture BMP assessment tools. Compatible with USEPA Basins | ||||
WinHSPF | X | X | X | Yes | Yes | Yes | USEPA BMP Web Toolkit available to assist with implementing structural BMPs such as detention basins, or infiltration BMPs that represent source control facilities, which capture runoff from small impervious areas (e.g., parking lots or rooftops). | |||
WinSLAMM | X | X | X | X | Yes | Yes | Yes | |||
XPSWMM | X | X | X | Yes | Yes | Yes | User defines parameter that can be used to simulate generalized constituents. |
The Simple Method is a technique used for estimating storm pollutant export delivered from urban development sites. Pollutant loads are estimated as the product of mean pollutant concentrations and runoff depths over specified periods of time (usually annual or seasonal). The method was developed to provide an easy yet reasonably accurate means of predicting the change in pollutant loadings in response to development. Ohrel (2000) states: "In general, the Simple Method is most appropriate for small watersheds (<640 acres) and when quick and reasonable stormwater pollutant load estimates are required". Rainfall data, land use (runoff coefficients), land area, and pollutant concentration are needed to use the Simple Method. For more information on the Simple Method, see The Simple method to Calculate Urban Stormwater Loads or The Simple Method for estimating phosphorus export.
Some simple stormwater calculators utilize the Simple Method (STEPL, Watershed Treatment Model). The MPCA developed a simple calculator for estimating load reductions for TSS, total phosphorus, and bacteria. Called the MPCA Estimator, this tool was developed specifically for complying with the MS4 General Permit TMDL annual reporting requirement. The MPCA Estimator provides default values for pollutant concentration, runoff coefficients for different land uses, and precipitation, although the user can modify these and is encouraged to do so when local data exist. The user is required to enter area for different land uses and area treated by BMPs within each of the land uses. BMPs include infiltrators (e.g. bioinfiltration, infiltration basin, tree trench, permeable pavement, etc.), filters (biofiltration, sand filter, green roof), constructed ponds and wetlands, and swales/filters. The MPCA Estimator includes standard removal efficiencies for these BMPs, but the user can modify those values if better data are available. Output from the calculator is given as a load reduction (percent, mass, or number of bacteria) from the original estimated load.
Because the MPCA Estimator does not consider BMPs in series, makes simplifying assumptions about runoff and pollutant removal processes, and uses generalized default information, it should only be used for estimating pollutant reductions from an estimated load. It is not intended as a decision-making tool.
Download MPCA Estimator here: File:MPCA Estimator.xlsx
A quick guide for the estimator is available Quick Guide: MPCA Estimator tab.
The Minimal Impact Design Standards (MIDS) best management practice (BMP) calculator is a tool used to determine stormwater runoff volume and pollutant reduction capabilities of various low impact development (LID) BMPs. The MIDS calculator estimates the stormwater runoff volume reductions for various BMPs and annual pollutant load reductions for total phosphorus (including a breakdown between particulate and dissolved phosphorus) and total suspended solids (TSS). The calculator was intended for use on individual development sites, though capable modelers could modify its use for larger applications.
The MIDS calculator is designed in Microsoft Excel with a graphical user interface (GUI), packaged as a windows application, used to organize input parameters. The Excel spreadsheet conducts the calculations and stores parameters, while the GUI provides a platform that allows the user to enter data and presents results in a user-friendly manner.
Detailed guidance has been developed for all BMPs in the calculator, including biofiltration. An overview of individual input parameters and workflows is presented in the MIDS Calculator User Documentation.
A simplified approach to computing a credit would be to apply a reduction value found in literature to the pollutant mass load or concentration (EMC) of the biofiltration device. Concentration reductions resulting from treatment can be converted to mass reductions if the volume of stormwater treated is known.
Designers may use the pollutant reduction values reported in this manual or may research values from other databases and published literature. Designers who opt for this approach should
The following references summarize pollutant reduction values from multiple studies or sources that could be used to determine credits. Users should note that there is a wide range of monitored pollutant removal effectiveness in the literature. Before selecting a literature value, users should compare the characteristics of the monitored site in the literature against the characteristics of the proposed biofiltration device, considering such conditions as watershed characteristics, biofiltration sizing, soil infiltration rates, and climate factors.
Field monitoring may be used to calculate stormwater credits in lieu of desktop calculations or models/calculators as described. Careful planning is HIGHLY RECOMMENDED before commencing a program to monitor the performance of a BMP. The general steps involved in planning and implementing BMP monitoring include the following.
The following guidance manuals have been developed to assist BMP owners and operators on how to plan and implement BMP performance monitoring.
Geosyntec Consultants and Wright Water Engineers prepared this guide in 2009 with support from the USEPA, Water Environment Research Foundation, Federal Highway Administration, and the Environment and Water Resource Institute of the American Society of Civil Engineers. This guide was developed to improve and standardize the protocols for all BMP monitoring and to provide additional guidance for Low Impact Development (LID) BMP monitoring. Highlighted chapters in this manual include:
AASHTO (American Association of State Highway and Transportation Officials) and the FHWA (Federal Highway Administration) sponsored this 2006 research report, which was authored by Oregon State University, Geosyntec Consultants, the University of Florida, and the Low Impact Development Center. The primary purpose of this report is to advise on the selection and design of BMPs that are best suited for highway runoff. The document includes the following chapters on performance monitoring that may be a useful reference for BMP performance monitoring, especially for the performance assessment of a highway BMP:
In 2014 the Water Environment Federation released this White Paper that investigates the feasibility of a national program for the testing of stormwater products and practices. The information contained in this White Paper would be of use to those considering the monitoring of a manufactured BMP. The report does not include any specific guidance on the monitoring of a BMP, but it does include a summary of the existing technical evaluation programs that could be consulted for testing results for specific products (see Table 1 on page 8).
The most current version of this manual was released by the State of California, Department of Transportation in November 2013. As with the other monitoring manuals described, this manual does include guidance on planning a stormwater monitoring program. However, this manual is among the most thorough for field activities. Relevant chapters include:
This online manual was developed in 2010 by Andrew Erickson, Peter Weiss, and John Gulliver from the University of Minnesota and St. Anthony Falls Hydraulic Laboratory with funding provided by the Minnesota Pollution Control Agency. The manual advises on a four-level process to assess the performance of a Best Management Practice, involving:
Use these links to obtain detailed information on the following topics related to BMP performance monitoring:
In addition to TSS and phosphorus, bioretention BMPs can reduce loading of other pollutants. According to the International Stormwater Database, studies have shown that biofiltration BMPs are effective at reducing concentrations of pollutants, including metals, and bacteria. A compilation of the pollutant removal capabilities from a review of literature are summarized below.
Relative pollutant reduction from bioretention systems for metals, nitrogen, bacteria, and organics.
Link to this table
Pollutant | Constituent | Treatment capabilities1 |
---|---|---|
Metals2 | Cadmium, Chromium, Copper, Zinc, Lead | High |
Nitrogen2 | Total nitrogen, Total Kjeldahl nitrogen | Low/medium |
Bacteria2 | Fecal coliform, e. coli | High |
Organics | Petroleum hydrocarbons3, Oil/grease4 | High |
1 Low: < 30%; Medium: 30 to 65%; High: >65%
2 International Stormwater Database, (2012)
3 LeFevre et al., (2012)
4 Hsieh and Davis (2005).
See Reference list