running testing and understanding the hydrological simulation program fortran hspf wang duru...

Upload: vstojnic

Post on 08-Jan-2016

227 views

Category:

Documents


0 download

DESCRIPTION

HSPF Running Testing and understanding the Hydrological Simulation Program Fortran HSPF Wang Duru Hjelmfelt

TRANSCRIPT

  • RUNNING, TESTING, AND UNDERSTANDING

    THE HYDROLOGICAL SIMULATION PROGRAM - FORTRAN (HSPF)

    Menghua Wang, J. Obiukwu Duru, and Allen T. Hjelmfelt, Jr.

    ABSTRACT

    This report consists of an introduction to the Hydrological Simulation Program-

    FORTRAN (HSPF) and two other related programs: HSPEXP and ANNIE. HSPF is a

    comprehensive, conceptual, continuous watershed model designed for simulation of

    watershed hydrology and water quality; HSPEXP is an expert system for calibration of

    HSPF; and ANNIE uses a binary, direct - access file called a Watershed Data

    Management (WDM) file to store and manipulate the input/output data for HSPF. The

    relations among the three programs and how they work together for a complete watershed

    simulation are explained in this report. By using the sample data sets provided, these

    programs were tested to verify that the programs were correctly installed and to examine

    how they work. General descriptions of the input data structures for these programs,

    model parameter evaluation, and the storage processes for runoff simulation are given.

    For a lucid clarification, a real working example for simulation of runoff from a research

    field in Goodwater Creek Watershed, located in central Missouri, is used to explain how

    to prepare data for these programs, to estimate model parameters, to run these programs

    and calibrate HSPF, and to explore simulation results. The possible problems for using

    HSPF to simulate the Goodwater Creek watershed which has a thin soil layer underlain

    by a claypan are also explained.

    1

  • HSPF, HSPEXP, and ANNIE all run in DOS environment. To prepare input data

    and explore simulation results are time-consuming. The BASINS (Better Assessment

    Science Integrating Point and Nopoint Sources) model makes watershed and water

    quality studies easier by bringing key data and assessment models together "under one

    roof'. One component of BASINS, NPSM, combines a windows-based interface with

    HSPF model, and is linked to ArcView. This can make HSPF more easily be used. In

    the report, however, this aspect is not elaborated on.

    The watershed analysis system with the core of HSPF can be used to make a

    broad-based evaluation of a watershed's hydrologic balance and chemical loading for

    watershed planning.

    2

  • RUNNING, TESTING, AND UNDERSTANDING

    THE HYDROLOGICAL SIMULATION PROGRAM - FORTRAN (HSPF)

    PROGRAM INTRODUCTION

    The hydrological Simulation Program - FORTRAN (HSPF) is a comprehensive

    model of watershed hydrology and water quality that allows the integrated simulation of

    land and soil contaminant runoff processes with stream hydraulic, water temperature,

    sediment transport, nutrient, and sediment-chemical interactions. For simulation with

    HSPF, the basin has to be represented in terms of land segments and reaches. A segment

    of land that has the capacity to allow enough infiltration to influence the water budget is

    considered pervious. Otherwise, it is considered impervious. A reach indicates a stream,

    a channel, or a completely mixed lake/reservoir. The HSPF has three modules:

    PERLND, IMPLND, and RCHRES for simulations of pervious lands, impervious lands,

    and free-flow reaches/mixed reservoirs, respectively. The operation functions for each

    module are described in PART E: FUNCTION DESCRIPTION in HSPF User's Manual

    (Bicknell et al., 1996).

    In pervious land simulation, HSPF models the movement of water along three

    paths: overland flow, interflow, and groundwater flow, each of which is an outflow from

    a surface storage, an interflow storage, and a groundwater storage, respectively. In

    addition to the three storage zones, two others, upper zone and lower zone, are also used.

    These storage zones are utilized to represent the storage processes that occur on the land

  • surface and in the soil horizon. Figure 1 depicts these zones. The storage depiction is

    modified from Fig. 2.2 in ARM User's Manual (Donigian et al., 1978). The upper zone

    is designed to simulate the diversion of overland flows into depression storage, soil

    fissures, and disturbed or dry surface soil (Crawford and Linsley, 1966). The lower zone

    is the root zone.

    SURFACE FLOW

    INTERFLOW

    Fig. 1 Model soil layers in HSPF

    In impervious land segments where there is no infiltration, overland flow,

    evaporation, and water quality constituent movement are simulated. Reaches are used to

    simulate the hydraulic and water quality processes that occur in the river channel network

    or completely mixed lakes/reservoirs. Runoff routing in a reach is done using a modified

    version of the kinematic wave equation.

    4

  • For modeling purposes, water, sediment, and water quality constituents leaving

    one segment or reach move laterally to a downslope segment or to a reach. In Fig. 2, the

    watershed consists of two pervious lands, one impervious land, and one reach. Water,

    soils, and pollutants from PERLND 1 and IMPLND 1 flow into RCHRES 1. The total

    outflow of materials from the watershed is the sum of outflows from PERLND 2 and

    RCHRES 1.

    Fig. 2. A watershed with two pervious lands, one impervious

    land, and one reach.

    The HSPF needs at least two input files: a Watershed Data Management (WDM)

    file and a User's Control Input (UCI) file. The WDM file is a binary, direct-access file

    that is used to store watershed characteristic data and time-series data, such as

    5

  • precipitation and runoff, in data sets. A WDM file can contain several data sets

    belonging to different watersheds. The relations among the WDM file, data sets, and

    detailed data (watershed characteristic data and time-series data) are shown in Fig. 3.

    File

    Data Sets

    Detail Data Watershed characteristic data, such as drainage area

    Time-series data, such as precipitation records

    Fig. 3. The relations among a WDM file, Data sets, and detail data

    The WDM file provides the user with a common data base, thus eliminating the need to

    reformat data from one application to another. The UCI file contains HSPF model

    parameters and function flags that control how the HSPF performs.

    ANNIE is a program for management of a Watershed Data Management (WDM)

    file, which is required as an input file for HSPF. The ANNIE program mainly has the

    following functions:

    Create a WDM file

    Import data into a WDM file

    Interactively update a WDM file

    Retrieve and plot data from a WDM file

    6

  • Export data from a WDM file.

    The Expert System for Calibration of HSPF (HSPEXP) allows the user to

    interactively edit model parameters in UCI file, run HSPF, plot HSPF outputs, compute

    error statistics for a simulation, and provide the user with expert advice on which

    parameters should be changed up or down to improve the calibration. The expert system

    uses over 35 rules involving over 80 conditions to recommended parameters. The rules

    are based on the experience of experts in the use of HSPF. The relations among HSPF,

    ANNIE, HSPEXP, and two data files of WDM and UCI are shown in Fig. 3. The HSPF

    takes input data from the two files of WDM and UCI and can run independently or

    through HSPEXP. The outputs from HSPF can be written into the WDM file or ASCII

    files.

    7

  • HSPEXP

    Fig. 3 The relations among HSPF, ANNIE, HSPEXP, and two data

    files of WDM and UCI.

    PROGRAM TESTING

    Program testing is used to verify that programs are correctly installed and to

    examine how these programs work. Each of the three programs: ANNIE, HSPF, and

    HSPEXP provides several tests which are described in the program README files. The

    summaries of those tests are given in the following.

    1. TESTS FOR ANNIE

    There are three general tests and six graphic tests for ANNIE. The general tests

    show how to create a WDM file, to import data into the WDM file, to generate new data

    sets using existing data, and to export data from the WDM file. The graphic tests are

    8

  • used to show the capability of ANNIE for plotting time-series data. For understanding

    those tests, TEST 1 is examined in details.

    TEST 1 was designed to create a WDM file and import watershed characteristic

    data and time-series data into the WDM file. There are three input files TEST1.LOG,

    CANE.EXP, and ILEXP, and one output file TEST.WDM. The file TESTI .LOG

    contains a sequence of keystrokes, each of which represents a command. Keystrokes of

    non-printing keys, such as the backspace and function keys, are represented by special

    codes. Each code has a # with three digits and its definition is listed in Table B.4 in

    ANNIE User's Manual (Flynn et al., 1995). The contents of TESTI .LOG is listed and

    explained in Table 1. The files CANE.EXP and 1L.EXP contain data for updating the file

    TEST.WDM. To examine the data structure of the two EXP files, parts of CANE.EXP

    are listed in Table2.

    9

  • Table 1. Description of TESTI .LOG

    Line Contents Explanation

    1 fbtest.wdm This line represents three commands: 1) In the Opening screen menu of ANNIE, press "f" key for File option; 2) in the File (F) screen, press "b" key for Build option to build a new WDM file; and 3) in the Build (B) screen, type "test.wdm" as the name of the new WDM file. The keystrokes from the three commands form the content of fbtest.wchn in the first line.

    2 #402 Accept, function F2, is executed. This causes the screen back to File (F). 3 rai..\data\cane.exp This line represents four commands: 1) Return from the File (F) screen;

    2) select Archive option in the Opening screen; 3) select Import option in the Archive (A) screen; and 4) in the Import (I) screen, type "..\data\cane.exp" for importing data into the file test.wdm.

    4 #402 F2 function key. 5 #402 F2 function key.

    6 #402 F2 function key.

    7 i..\data\il.exp 1) In Archive screen, select Import(I) option; and 2) in the Import(I) screen, type "..\data\il.exp" for importing the data in il.exp into test.wdm.

    8 #402 F2 function key.

    9 #402 F2 function key.

    10 #402 F2 function key. 11 rfs 1) Return from the Archive (A) screen; 2) select File option in the

    opening screen: and 3) in the File(F) screen, select Summarize option for summarizing the contents of test.wdm.

    12 #402 F2 function key.

    13 rr 1) Return from the File (F) screen; and 2) select Return option in the Opening screen.

    10

  • Table 2. Partial listing of input file CANE.EXP.

    1s4). Contents

    1 DATE 2 WDMSFL 3 SYSTEM 4 COMMENT Cane Branch, KY daily: precip, disch, evap, min/max temp 5 unit: precip, disch

    ,

    6 END COMMENT . 7 DSN 5 TYPE TIME NDN 1 NUP 1 NSA 35 SP 70 NDP 150 8 LABEL 9 PARMCD 45 10 STATCD 6 11 TSBYR 1956 12 TSSTEP 1

    TCODE 4 13 TGROUP 6 14 TSFORM 1 15 COMPFG 1 16 VBTIME 1 17 DAREA 0.67 18 TSFILL-9.9999E+4 19 STAID 365205084265702 20 TSTYPE PREC 21 STANAM Cane Branch, KYprecipitation

    END LABEL 22 DATA STARTS: 1956 1 1 0 0 0 ENDS: 1957 12 31 24 0 0 23 1955 12 31 24 0 0 4 1 0 46 1 0.0000 24 1956 2 15 24 0 0 4 1 0 4 0 0.2700 2.710

    0.5700 0.5800 1956 2 19 24 0 0 4 1 0 4 1 0.0000 1957 9 23 24 0 0 4 1 0 4 1 0.0000

    1957 9 27 24 0 0 4 1 0 2 0 0.2100 0.3700

    1957 9 29 24 0 0 4 1 0 93 1 0.0000 END DATA

    END DSN

    11

  • The file CANE.EXP contains 7 data sets numbered from 5 to 11 for daily data of

    precipitation, discharge, evaporation, minimum and maximum temperatures, as well as

    the precipitation and discharge in 30 minutes time intervals. Table 2 lists the heads of an

    EXP file and parts of data set (DSN) 5. Data-set attributes used in the archive file,

    CANE.EXP, can be found in APPENDIX A in the ANNIE User's Manual (Flynn et al.,

    1995). The data structures of CANE.EXP are explained as the following.

    In Table 2, lines 1-6 list the heads of CANE.EXP. Data sets begin from line 7.

    Line 7 indicates that the data set number (DSN) is 5 and the data set type (TYPE) is time-

    series (TIME). The time-series type (TSTYPE) is given as precipitation (PREC) in line

    19. The precipitation station ID (STAID) and name (STANAM) are listed in lines 18 and

    20, respectively. The number 4 for TCODE in line 13 indicates that this is daily data and

    the value of 1 for COMPFG in line 15 means the data can be in compressed format. The

    recorded precipitation data in line 23 is in compressed format and the data in line 24 is in

    uncompressed format, which are indicated by 1 and 0 in column 11, 1 for compression

    and 0 for uncompression. When data is in compressed format, one value can be used to

    represent several individual data. In line 23, the value of 0.000, which is given in the 12 th

    column, is used to represent 46 individual data (Jan. 1 - Feb. 15, 1956), where the number

    46 is given in the tenth column. When data is in uncompressed format, the values of each

    data must be given individually. In line 24, the data values are given as 0.2700, 2.710,

    0.5700, and 0.5800 for 4 individual data (Feb. 16 - 19). In addition, the number 4 in

    column 7 of lines 23 and 24 needs to be explained. This number must match the value of

    TCODE, which indicates time interval. That is 4 for data set 5.

    12

  • The file IL.EXP contains 31 data sets numbered from 16 to 46. The data sets

    contained in ILEXP, however, do not have detailed time-series data. After the data in the

    files CANE.EXP and IL.EXP are imported into the file TEST.WDM; the WDM file

    contains 38 data sets.

    TEST 1 can be automatically run following the instructions given in the

    README file of ANNIE. It is helpful, however, to run this test step by step based on the

    keystroke descriptions for the file TEST 1 .LOG, which is listed in Table 1, for better

    understanding this test.

    2. TESTS FOR HSPF

    There are 13 test runs for HSPF. The first run is to use ANNIE to create two new

    WDM files named TEST.WDM and TEST2.WDM, and to use the data contained in the

    files TESTALEXP and TESTA2.EXP to update the two WDM files, respectively. The

    HSPF can be used to retrieve and import data for a WDM file. It cannot, however, be

    used to create a new WDM file that must be created using ANNIE or other programs.

    Tests 2 to 4 were designed for using HSPF to update the file TESTMDM and

    export data from the WDM file. Tests 5 to 13 are used to test various features of HSPF

    for simulations of snow melt, runoff, sediment, pesticides, and nutrients. The detailed

    descriptions for each test are given in the README file of HSPF.

    To run HSPF, at least two input files, a WDM file and a user control input (UCI)

    file, are needed. WDM files contain watershed characteristic data and time-series data,

    and UCI files contains HSPF model parameters and function control data. The data

    13

  • formats in UCI files are described in PART F: FORMAT FOR THE USER'S CONTROL

    INPUT in HSPF User's Manual (Bicknell et al., 1996).

    A UCI file is the core-input file for running HSPF. The data in this file controls

    how HSPF functions, how different segments link together, and how simulation results

    are presented. The HSPF model parameters are also contained in this file. To

    understand UCI files, the data structure of the UCI file for TEST 9, which is listed in

    APPENDIX I, is explained.

    This test was designed to simulate snow and runoff for one year in 1976. This is

    indicated in GLOBAL block. The test has one pervious land (PERLND) named

    PERLND 1 and four reaches (RCHRES) named RCHRES 1 to 4, respectively, in OPN

    SEQUENCE block. The following is the OPN SEQUENCE block in TEST 9. OPN SEQUENCE

    INGRP INDELT 01:00 PEELED 1 RCHRES 1 RCHRES 2 RCHRES 3 RCHRES 4 DISPLY 1 DISPLY 2 DISPLY 3 PLTGEN 1

    END INGRP END OPN SEQUENCE

    The linkages among the pervious land and the four reaches are described in

    NETWORK block. The data in the NETWORK block in TEST 9 is shown in the

    following:

    14

  • NETWORK Tran ***

    # # #strg # # # # *** PERLND 1 PWATER PERO 500. RCHRES 1 EXTNL IVOL RCHRES 1 HYDR OVOL 1 1. RCHRES 2 EXTNL IVOL RCHRES 1 HYDR OVOL 1 12.1 DISPLY 1 INPUT TIMSER RCHRES 1 HYDR OVOL 1 12.1 PLTGEN 1 INPUT MEAN 2 RCHRES 1 HYDR OVOL 2 1. RCHRES 3 EXTNL IVOL RCHRES 1 HYDR OVOL 2 12.1 DISPLY 2 INPUT TIMSER RCHRES 1 HYDR OVOL 2 12.1 PLTGEN 1 INPUT MEAN 1 RCHRES 2 HYDR ROVOL 1. RCHRES 4 EXTNL IVOL RCHRES 3 HYDR ROVOL 1. RCHRES 4 EXTNL IVOL RCHRES 4 HYDR ROVOL 12.1 DISPLY 3 INPUT TIMSER RCHRES 4 HYDR ROVOL 12.1 PLTGEN 1 INPUT MEAN 3 END NETWORK

    PERLND 1 has a drainage area of 500 acres, which is indicated in the column

    labeled as . RECHES 1 takes the total outflow from PERLND 1 (PERO) times

    500 as its external inflow (IVOL). The outflow from RECHES 1 (OVOL) is diverged

    into RECHES 2 and 3 as their inflows (IVOL). Finally the total outflows from RECHES

    2 and 3 (ROVOL) merge as the inflow to RECHES 4. Based on the linkages, a possible

    drainage network is shown in Fig. 4. The variables used in the NETWORK block, such

    as PERO, OVOL, WOL, and ROVOL, can be found in Time Series Catalog section of

    PART F in HSPF User's Manual (Bicknell et al., 1996). The other information in

    NETWORK block is for displaying and plotting simulation results from the pervious land

    and the reaches using DISPLAY and PLTGEN modules. The DISPLY module

    summarizes a time series and presents the results in neatly formatted tables and the

    PLTGEN module prepares one or more time series for plotting. Instead of NETWORK

    block, another way for describing segment linkages is to use SCHEMATIC and MASS-

    LINK blocks.

    15

  • PERLND

    RECHES 1

    RECHES 3

    RECHES 4

    RECHES 2

    Fig. 4 The possible drainage links for TEST 9

    There are three DISPLYs and one PLTGEN, which are named as DISPLY 1 to 3

    and PLTGEN 1 in OPN SEQUENCE block. The NETWORK block indicates that the

    simulation results from reaches will go to the DISPLAYs and the PLTGEN. These data

    are written using FORTRAN I/O unit numbers 65 and 93, respectively. This is indicated

    in the following DISPLY and PLTGEN blocks.

    16

  • DISPLY DISPLY-INFO1

    *** # _ # Title > TRAN PIVL DIG1 FIL1 PYR DIG2 FIL2 YRND 1 OUTLET INFLOW (CFS) AVER 1 2 65 12 2 SPILLWAY INFLOW (CFS) AVER 1 2 65 12 3 KITTLE CREEK FLOW (CFS) AVER 1 2 65 12

    END DISPLY-INFO1 END DISPLY

    PLTGEN PLOTINFO # - # FILE NPT NMN LABL PYR PIVL *** 1 93 3 24

    END PLOTINFO

    where the I10 unit numbers 65 and 93 are marked using red color. The files

    corresponding to the I/O unit numbers 65 and 93 are defmed as TEST09.D65 and

    TEST09.P93 in FILES block. The following gives the FILES block in TEST 9.

    FILES *** Unit-systems Printer *** # - # t-series Engl Metr ***

    in out *** 1 BICKNELL FARM 1 1 1 0

    END GEM-INFO

    and

    RCHRES

    17

  • GEN- INFO RCHRES< Name >Nexit Unit Systems Printer # - # t-series Engl Metr LKFG

    in out 1 MEIER POND 2 1 1 1 0 1 2 OUTLET 1 1 1 1 0 3 SPILLWAY 1 1 1 1 0 4 KITTLE CREEK 1 1 1 1 0

    END GEN-INFO

    where the printer I10 unit number 1 is marked using red color.

    The definition of each individual model parameter in UCI files is not described

    in this report. All model parameter defmitions can be found in PART F: FORMAT FOR

    THE USER'S CONTROL INPUT in HSPF User's Manual (Bicknell et al., 1996). The

    parameter evaluations for runoff simulation are discussed in the section of RUNOFF

    SIMULATION and in details through a working example.

    3. TESTS FOR HSPEXP

    There are two test runs for HSPEXP. The first test is for daily simulation and the

    second is for hourly simulation. The two tests were designed to show how to

    interactively modify model parameters based on the expert rules provided by HSPEXP, to

    run HSPF simulations, and to statistically analyze and present results. The simulation

    results can be saved into a file, displayed on the screen, and plotted.

    Each HSPEXP test has a keystroke file. They are TEST1.LOG and TEST2.LOG,

    respectively. The keystroke files for HSPEXP are similar to the keystroke files for

    ANNIE. So the files TEST1.LOG and TEST2.LOG are not further explained here.

    In addition to the UCI file and the WDM file required by HSPF, HSPEXP needs

    another file with EXS as its extension. They are TEST1.EXS and TEST2.EXS for the

    18

  • two HSPEXP tests. The EXS files contain the information about storm events. These

    two files can be created using HSPEXP.

    Following the instructions described in the README file of HSPEXP, the two

    HSPEXP tests can be run using their LOG files. To understand these tests, however, it is

    better to run them step by step following the keystrokes contained in their LOG files.

    RUNOFF SIMULATION

    Simulation of runoff using HSPF will be described in details in this section. .

    Pervious land simulation will be emphasized. HSPF uses six storages to simulate water

    budgets. They are interception, surface zone, upper zone, interflow zone, lower zone,

    and groundwater zone. The last five zones are displayed in Fig. 1.

    All incoming precipitation enters interception storage until a pre-assigned

    interception storage capacity (CEPSC) is filled. Available moisture that is not infiltrated

    directly will increase surface detention storage. The increment to surface detention will

    either contribute to overland flow and interflow or enter upper zone storage. Depression

    storage and the storage in highly permeable surface soils are modeled by the upper zone.

    Hence, the upper zone inflow rates are independent of rainfall intensity but upper zone

    storage capacity is low. Moisture is lost from the upper zone by evaporation and

    percolation to the lower zone and groundwater storage (Crawford and Linsley, 1966).

    The baseflow is from groundwater storage. Lower zone is the root zone and the moisture

    loss from this zone is through evapotranspiration. The functions for simulations of these

    storage behaviors are described in PART F: FUNCTION DESCRIPTION in HSPF User's

    19

  • Manual (Bicknell et al., 1996). Understanding how HSPF works is helpful for estimating

    model parameters.

    HSPF has several model parameters for runoff simulation. These parameters are

    defined in PART F: FORMAT FOR THE USER'S CONTROL INPUT in HSPF User's

    Manual (Bickenll et al., 1996). How to evaluate these model parameters can be found in

    SECTION 5: MODEL PARAMETERS AND PARAMETER EVALUATION in ARM

    User's Manual (Donigian et al., 1978 ) and in CHAPTER V: OPERATION OF THE

    MODEL in Stanford Watershed Model User's Manual (Crawford and Linsley, 1966).

    Both of these manuals contain evaluation guidelines on a parameter-by-parameter basis.

    The guidelines for hydrologic calibration can also be found in the two user's manuals and

    in Application Guide for HSPF (Donigian et al., 1984). The two models, ARM and

    Stanford Watershed Model, and NPS (Donigian et al., 1984) are predecessors of HSPF.

    Many of their model parameters are shared in common. To expedite the use of the

    valuable information contained in predecessor user's manuals, Donigian et al (1984) gave

    an equivalency table for selected HSPF and ARM/NSP model parameter names in

    APPENDIX C in Application Guide for HSPF.

    Among these model parameters for runoff simulation, some of them may cause

    confusion for understanding or problems in evaluation. These parameters are explained

    here.

    UZS and LZS are the current soil moisture storage in the upper zone and in the

    lower zone, respectively. UZSN and LZSN are nominal storage levels that are

    approximately equal to the median values of the corresponding storage for upper zone

    and lower zone. An approximate mapping of LZSN values for the United States is given

    20

  • in ARM User's Manual (Donigian et al, 1978, pp. 56). This map indicates the value of

    7.0 in. for Missouri. The nominal storage in the upper zone (UZSN) is generally related

    to LZSN and watershed topography. The value of UZSN relative to LZSN is defined as a

    function of watershed topography and cover. An estimate of UZSN relative to LZSN can

    be found in ARM User's Manual (Donigian et al., 1978, pp.55) and in Stanford

    Watershed Model User's Manual (Crawford and Linsley, 1966, pp.75). Another major

    model parameter for water budget simulation is INFILT, an index to the mean infiltration

    rate on the watershed. The values of INFILT and LZS/LZNS greatly affect infiltration.

    The guideline for INFILT evaluation can be found in ARM User's Manual (Donigian et

    al., 1978, pp.61).

    Logically, upper zone storage (depression), lower zone storage, and the rate of

    infiltration all interact. So the parameters UZSN, LZSN, and INFILT cannot be

    independent (Crawford and Linselt, 1966). The fmal values of these three parameters

    must be determined through model calibration.

    A WORKING EXAMPLE FOR RUNOFF SIMULATION

    To examine the ability of HSPF for runoff simulation, a real working example

    was developed. The input files for the working example are listed in the attached

    appendixes and explained in this report. The simulation results from HSPF are also listed

    and explained.

    21

  • 1. Site Description

    HSPF was used to simulate one year (1995) of runoff from a research field in

    Goodwater Creek watershed. The watershed is located in central Missouri and

    established as a research experiment by U.S. Department of Agriculture, Agricultural

    Research Service in 1971. The watershed has a level to gently slopping surface underlain

    by a layer of loess. This layer of loess contains a claypan located about 30 cm below the

    surface. The claypan is often treated as impermeable.

    The research field is used for extensive water quality studies. The field has a

    drainage area of 36 ha and the major crop in this field is corn. The drainage boundary and

    network of the field is displayed in Fig. 5

    22

  • Fig. 5 The drainage boundary and network of the study field.

    23

  • 2. Data Preparation

    There is no obvious stream flow in the study field. To use HSPF for runoff

    simulation, the field is simplified as a pervious land. In this example, several input data

    files need to be prepared. They are FIELD1.EXP, FIELD1.UCI, FIELD LEXS, and

    FIELD1.WDM. The files FIELD1.EXP, FIELD1.UCI, and FIELD1.EXS are three

    original data input files, which are listed in APPENDIX II to IV, respectively. The file

    FIELD1.EXS is a simple file containing storm event information, which can be

    interactively prepared using HSPEXP. The file FIELD LWDM is created from

    FIELDLEXP using ANNIE. Here, the files FIELDLEXP and FIELD1.UCI are briefly

    described in the following.

    1) FIELD1.EXP

    In this example, snowmelt is not considered. For simulating runoff from the study

    field in 1995, three daily data series, precipitation, evaporation, and runoff, are prepared.

    The runoff data is used for comparison of simulation results.

    In the file FIELD LEXP, the time-series data of precipitation, evaporation, and

    runoff are organized as three data sets with Data-Set-Numbers (DSN) 39, 41, and 136,

    respectively. The three data sets are in compressed formats. In data set for the time-

    series precipitation or runoff, the data is compressed only if the precipitation or runoff

    values in several consecutive days are equal to zero. In data set for the time-series of

    evaporation, the data is compressed if the evaporation values in several consecutive days

    are same.

    Following the three data sets of observed precipitation, evaporation, and runoff,

    there are nine data set templates in the file FlELDLEXP. These data set templates are

    24

  • used to contain simulation results from HSPF. For example, the simulated total runoff

    time series will be put into the data set with DSN 420 and the simulated surface runoff

    into the data set with DSN 421. The data set with DSN 430 is used to contain the unit-

    converted runoff in cube feet per second. The original daily runoff data is in inches and is

    contained in the data set with DSN 136. The file FIELD1.EXP is utilized to create

    FIELD1.WDM by using ANNIE. FIELD1.WDM is an input file for HSPF.

    2) FIELD1.UCI

    In this example, FIELD1.UCI is the user control input file to run HSPF. The data

    in the file is explained section by section as the following.

    GLOBAL block: The time period of 1995 for the one-year runoff simulation is

    given in this data block.

    GLOBAL Field1, Goodwater Creek watershed, MO START 1995 1 1 0 0 END 1995 12 31 24 0 RUN INTERP OUTPUT LEVEL 3 RESUME 0 RUN 1 TSSFL 0 WDMSFL 0 UNITS 1

    END GLOBAL

    FILES block: There are three files in the file block. HSPF reads data from the

    file FlELD1.WDM and writes simulation results back to the file using FORTRAN I10

    unit 26. HSPF will write operation message to FILED1.ECH using I10 unit 25 and the

    simulation result summary to FIELD LOUT using I10 unit 90.

    FILES .**< fname MESSU 25 field1.ech WDM 26 fie101.wdm

    90 field1.out END FILES

    25

  • OPN SEQUENCE block: There is only one pervious land named as PERLND

    307 for this simulation. COPY 100 is used for transferring simulation results into the file

    FIELD 1 .WDM. The input data is daily time-series. To get better results, the calculation

    time step used for the runoff simulation is 30 minutes, which is indicated using the

    variable INDELT.

    OPN SEQUENCE INGRP

    PERLND 307 COPY 100

    END INGRP END OPN SEQUENCE

    INDELT 00:30

    PERLND block: There are several data sections in this block, which are

    described in the details:

    ACTIVITY section - In this working example, only water budget is simulated.

    The flag of PWAT is set to be 1 for active.

    ACTIVITY Active Sections *** x - x ATMP SNOW PWAT SED PST PWG PQAL MSTL PEST NITR PROS TRAC ***

    307 0 0 1 0 0 0 0 0 0 0 0 0 ND ACTIVITY

    PRINT-INFO section - The values in this section control printout level for the file

    FIELD LOUT. The default values for all flags are 4. The value of 5 under PWAT means

    the printout for water balance simulation is in yearly level.

    PRINT-INFO ********************* p rint -flags ************************* pivi, pyR x - x ATMP SNOW PWAT SED PST PWG PQAL MSTL PEST NITA PHOS TRAC *********

    307 4 4 5 4 4 4 4 4 4 4 4 4 1 12 END PRINT-INFO

    26

  • 27

    Name

    307 Corn END GEN-INFO

    GEN- INFO x - X

    MOLES Unit-systems Printer*** t-series Engl Metr***

    in out *** 1 1 90 0

    GEN-INFO section - This section gives general information for PERLND 307 and

    input/output data units. The value of 1 for unit flag means using English unit. The value

    of 90 for I10 unit means that the yearly level printout will be written into FILE LOUT.

    PWAT-PARM1 section - The flags beginning with "v" indicate whether or not

    certain parameters will be assumed to vary throughout the year on the month basis: 1

    means they do vary, 0 means they do not. In this sample, the interception storage

    capacity, upper zone nominal storage, Manning's roughness coefficient, and lower zone

    evapotranspiration are varied monthly. The flags corresponding to these variables are

    VCS, VUZ, VNN, and VLE, respectively and their values are set to 1.

    PWAT-PARM1 *** Flags *** x - x CSNO ATOP UZFG VCS VUZ VNN VIFW VIRC VLE 307 0 1 1 1 1 1 0 0 1 END PWAT-PARM1

    PWAT-PARM2 section - The values for overland flow length (LSUR) and land

    slope (SLSUR) are determined from the topography of the study field. The parameters

    KVARY and AGWRC are related to ground water recession. For small watersheds or

    fields, the two parameter values are generally close to 0.0 (ARM User's Manual -

    Donigian et al., 1978, pp.63). In this study, these two parameters are assumed to have

    their minimum values. LZSN is lower zone nominal storage and LNFILT is an index to

    the mean infiltration rate on the watershed. The two parameters mainly control

    infiltration and baseflow. The initial values of the two parameters can be assumed based

  • on the parameter evaluation guidelines described in ARM user's Manual (Donigian et al.,

    1978). Their fmal values, however, must be determined through model calibration. The

    calibrated values for LZSN and INFILT are 7.0 in and 0.001 in/hr.

    PWAT-PARM2

    *** FOREST LZSN INFILT LSUR SLSUR KVARY AGWRC

    x - x (in) (in/hr) (ft) (1/in) (1/day) 307 0.0 7.000 0.001 750 0.0062 0.000 0.001 END PWAT-PARM2

    PWAT-PARM3 section - The values for the first four parameters PETMAX,

    PETMIN, INFEXP and INFILD in this section are the same as the values used in the Iowa

    River study, which was taken as a sample case study in Application Guide for HSPF

    (Donigian et al., 1984). These values are also accepted for general runoff simulation

    using HSPF. The other parameters related to deep percolation (DEEPFR), baseflow,

    evapotranspiration (BASETP), and groundwater evapotranspiration (AGWETP) are set to

    be 0.0 because of the underlying claypan soil in the study field. The claypan makes

    baseflow and the groundwater storage in the study field to be ignored.

    PWAT-PARM3

    *** PETMAX PETMIN INFEXP INFILD DEEPFR BASETP AGWETP

    *** x - x (deg F) (deg F) 307 40.0 35.0 2.0 2.0 0.000 0.000 0.000 END PWAT-PARM3

    PWAT-PARM4 section - The parameters INTFW and IRC are related to interflow

    recession. The parameter values are given based on the case study used in Application

    Guide for HSPF (Donigian et al, 1984) and adjusted through model calibration. The

    parameter values listed for interception storage (CEPSC), upper zone nominal storage

    (UZSN), Manning's n, and lower zone evapotranspiration are not the values used for

    HSPF simulation because their flags indicated that these parameter values are given

    monthly. The monthly varying data are listed following this section.

    28

  • PWAT-PARM4 ***

    CEPSC UZSN NSUR INTFW IRC LZETP *** x - x

    (in) (in) (1/day) 307 0.0 1.0 0.1 1.0 0.6 0.0 END PWAT-PARM4

    The monthly varying interception storage (MON-INTERCEP), upper zone

    nominal storage (MON-UZSN), and lower zone evapotranspiration (MON-LZETPARM)

    for PERLND 8 in Iowa River Study given in the Application Guide for FISPF (Donigian

    et al., 1984) are used here. The weather and land conditions in Iowa and Missouri are

    similar.

    NON-INTERCEP *** Interception storage capacity at start of each month (in) *** x - x JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

    307 0.03 0.03 0.03 0.03 0.01 0.03 0.10 0.16 0.18 0.14 0.03 0.03 END NON-INTERCEP

    MON-UZSN *** Upper zone nominal staorage start of each month (in) *** x - x JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

    307 0.4 0.4 0.4 0.4 1.4 1.0 1.0 1.2 1.2 1.2 1.0 0.9 END MON-UZSN

    MON-LZETPARM *** Lower zone evapotransp parm at start of each month *** x - x JAN FEB MAR. APR MAY JUN JUL AUG SEP OCT NOV DEC

    307 0.20 0.20 0.20 0.23 0.23 0.25 0.60 0.80 0.75 0.50 0.30 0.20 END MON-LZETPARM

    MON-MANNING section - The monthly values of Manning's n are assumed to

    not vary in this study. The value of 0.1 is used for all of the months for the study field.

    MON-MANNING *** Manning's n for overland flow at start of each month (in) *** x - x JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

    307 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 END MON-MANNING

    PWAT-STATE1 section - this section gives the initial values for several

    parameters. These values are assumed based on the field condition and adjusted during

    model calibration.

    PWAT-STATE1 *** PWATER state variables (in) *** x - x CEPS SURS UZS IFWS LZS AGWS GWVS

    29

  • 307 0.0 END PWAT-STATE1

    0.0 0.1 0.0 3.5 0.0 0.0

    COPY block: This block indicates that there are eight mean-value time-series

    data being transferred using the COPY block.

    COPY

    TIMESERIES Copy-opn***

    *** x - x NPT NMN 100 0 8 END TIMESERIES

    END COPY

    EXT SOURCES block: The precipitation data and evaporation data from data

    sets with DSN 39 and DSN 41 in the file FIELDLWDM are used for the study field

    (PERLND 307) simulation. The evaporation data in DSN 41 were observed from

    evaporation pan. The data times 0.8 is taken for land use. This block also indicates that

    the runoff data from the data set with DSN 136 in the file FIELD1.WDM will be read

    into COPY 100 as its time-series 1, which is used for unit conversion from inch to cube

    feet per second.

    EXT SOURCES SsysSgapTran x x tern strgstrg x x x x WDM1 39 PREC 10 ENGL 1.0 PERLND 307 0 EXTNL PREC 1 1 WDM1 41 EVAP 10 ENGL 0.8 PERLND 307 0 EXTNL PETINP 1 1 WDM1 136 FLOW 10 ENGL 1.0 COPY 100 0 INPUT MEAN 1 1

    END EXT SOURCES

    EXT TARGETS block: The first command line is for putting the total simulated

    runoff from the study field (PERLND 307) into the file FIELD1.WDM as data set 420.

    The second command line is for converting runoff data in inches from COPY 100 into the

    corresponding data using cubic feet per second as unit, and then put these data into the

    file FIELD1.WDM as data set 430. The value of 3.7355556 in this line is the unit

    30

  • conversion factor calculated based on the drainage area of 36 ha. The other command

    lines in this block are used for transferring data from COPY 100 into the file

    FIELD1.WDM. The data in COPY 100 are defined through the two blocks

    SCHEMATIC and MASS-LINK.

    EXT TARGETS Tran Tsys Aggr Amd *** x x xstrg x gf tern strg strg*** PERLND 307 PWATER PERO 1 1 1.0 WDM1 420 QDEP 1 ENGL AGGR REPL COPY 100 OUTPUT MEAN 1 1 3.7355556 WDM1 430 FLOW 1 ENGL AGGR REPL COPY 100 OUTPUT MEAN 2 1 1.0 WDM1 421 SURO 1 ENGL AGGR REPL COPY 100 OUTPUT MEAN 3 1 1.0 WDM1 422 IFWO 1 ENGL AGGR REPL COPY 100 OUTPUT MEAN 4 1 1.0 WDM1 423 AGWO 1 ENGL AGGR REPL COPY 100 OUTPUT MEAN 5 1 1.0 WDM1 425 PETX 1 ENGL AGGR REPL COPY 100 OUTPUT MEAN 6 1 1.0 WDM1 426 SAET 1 ENGL AGGR REPL COPY 100 OUTPUT MEAN 7 1 1.0 WDM1 427 UZSX 1 ENGL AGGR REPL COPY 100 OUTPUT MEAN 8 1 1.0 WDM1 428 LZSX 1 ENGL AGGR REPL END EXT TARGETS

    SCHEMATIC block: This block indicates that the simulation result from

    PERLND 307 will go to COPY 100 using MASS-LINK 1.

    SCHEMATIC x PERLND 307 END SCHEMATIC

  • MASS-LINK

    MASS-LINK 1

    Tran x xstrg

    x x lc *

    PERLND PWATER SURO COPY INPUT MEAN 2 PERLND PWATER IFWO COPY INPUT MEAN 3 PERLND PWATER AGWO COPY INPUT MEAN 4 PERLND PWATER PET COPY INPUT MEAN 5 PERLND PWATER TAET COPY INPUT MEAN 6 PERLND PWATER UZS COPY INPUT MEAN 7 PERLND PWATER LZS COPY INPUT MEAN 8

    END MASS-LINK 1

    END MASS-LINK

    3. MODEL RUN AND CALIBRATION

    The first step is to run ANNIE for creating the file FIELDLWDM using the file

    FIELD1.EXP. To run HSPF, the two files FIELD1.UCI and FIELD1.WDM are needed.

    The program HSPF can be run independently or through the Expert System for

    Calibration of HSPF (HSPEXP). To run HSPEXP, additional file FIELDLEXS is

    needed. The file contains the information about storm events.

    HSPEXP allows the user to interactively edit the input UCI file for the

    hydrological Simulation Program (HSPF), simulate with HSPF, produce plots of HSPF

    output compared to observed values, compute error statistics for a simulation, and

    provide the user with expert advice on which parameter should be changed up or down

    (HSPEXP README file). In the example for runoff simulation, the major calibrated

    parameter is the index of infiltration rate (INFILT). The parameter evaluation guidelines

    in ARM User's Manual (Donigian et al., 1978, pp.61) give a range from 0.01 to 1.0 in/hr

    depending on the cohesiveness and permeability of the soil. INFILT can range from 0.01

    to 0.05 in/hr for SCS Hydrologic Soil Group D, high runoff potential soil lands. Our

    study field is in this category. In this example, an initial value of 0.01 in/hr for INFILT is

    used. This value, however, gives over 30 percent of total runoff as baseflow. The

    32

  • simulation result does not match the field observation that baseflow can be ignored

    because of the underlying claypan in the study field. To eliminate the baseflow over-

    estimation problem, the final value of INFILT is calibrated to be 0.001 in/hr that is much

    smaller than the initial value. HSPF was designed for simulating infiltraion in uniform

    soils. To deal with layered soils, especially the case of a thin surface layer underlain by a

    claypan in our field study, INFILT must be calibrated. The statistical summary of

    calibration results is shown in Fig. 6 and the general report for the calibration results is

    given in the file FIELD LOUT, which is listed in APPENDIX V.

    Fig.6 Statistics summary of calibration results.

    Output from most recent HSPF calibration run:

    Simulated Observed Total annual runoff, in inches 15.010 15.006 Total of highest 10% flows, in inches 13.360 14.597 Total of lowest 50% flows, in inches 0.110 0.000

    Evapotranspiration, in inches

    Simulated Potential

    28.180 29.920

    Total storm volume, in inches Average of storm peaks, in cfs Basef low recession rate

    Simulated 11.100 3.686 0.830

    Observed 13.142 4.269

    -547.900

    Total simulated storm interflow, in inches 1.180 Total simulated storm surface runoff, in inches > 12.850

    Simulated Observed Summer flow volume, in inches 4.550 4.327 Winter flow volume, in inches 3.460 3.084 Summer storm volume, in inches 2.270 2.728

    Figure 6 indicates that the simulation results match the observed values well. The

    simulated total runoff for the year of 1995 is 15.010 in and the observed value is 15.006

    33

  • in. The simulated total storm volume for the seven selected storm events, which are

    listed in the file FIELD1.EXS, is 11.100 in and the corresponding observed value is

    13.142 in. The simulated and observed average peak discharges for the seven storm

    events are 3.686 and 4.269, respectively. The general report given in the file

    FIELD LOUT lists individual values for total moisture supply - precipitation (SUPLY),

    total runoff, surface runoff, interflow, and baseflow. They are 44.820, 15.007, 12.850,

    1.179, and 0.978 inches, respectively. The results indicate that only a very small percent

    of total runoff is interflow and baseflow. This is consistent with the field observations.

    In the example for runoff simulation, daily precipitation data is used. More

    detailed precipitation data, such as hourly data, is expected to be used for Goodwater

    Creek watershed simulations.

    CONCLUSION

    HSPF is a comprehensive, conceptual, and continuous watershed model designed

    for simulation of watershed hydrology and water quality; HSPEXP is an expert system for

    calibration of HSPF; and ANNIE uses a binary, direct-access file called a Watershed Data

    Management (WDM) file to store and manipulate the input/output data for HSPF.

    Various features of these three programs were tested and the input/output data structures

    are described. The details of how to use ANNIE to store data into a VVDM file for HSPH

    use, how to prepare the User Control Input (UCI) data, and how to use HSPF and

    HSPEXP to simulate runoff are explained through a working example - simulating runoff

    from a research field in Goodwater Creek watershed located in Central Missouri. This

    34

  • watershed was established as a research experiment by U.S. Department of Agriculture,

    Agricultural Research Service in 1971.

    The tests conducted in this study indicate that HSPF can be used to simulate

    snowmelt, runoff, sediment, nutrient, and pesticide. The example for runoff simulation

    shows a good match between simulation results and observed data.

  • gYZ-5 5'

    t7 6

    REFERENCES:

    Bicknell, B. R., J. C. Imhoff, and A. S. Donigian, 1996. User's Manual for Hydrological

    Simulation Program-FORTRAN. U.S. Environmental Protection Agency, Athens, GA.

    Crawford, N. H and R. K. Linsley, 1966. Digital simulation in hydrology: Stanford

    Watershed Model N. Technical Report No. 39, Stanford University, Stanford, CA.

    Donigian, A. S. and H. H. Davis, 1978. User's Manual for Agricultural Runoff

    Management (ARM) Model. U.S. Environmental Protection Agency, Athens, GA.

    Donigian, A. S. and N. H. Crawford, 1979. User's Manual for the Nonpoint Source

    (NPS) Model. U.S. Environmental Protection Agency, Athens, GA.

    Flynn, K. M., P. R. Hummel, A. M. Lumb, and J. L. Kittle, 1996. User's Manual for

    ANNIE, Version2, a computer program for interactive hydrologic data management.

    Water-Resources Investigation Report 95-4085, U.S. Geological Survey, Reston, VA.

    Lumb, A. M., R. B. McCammon, and J. L. Kittle, 1994. User's Manual for an Expert

    System (HSPEXP) for Calibration of the Hydrological Simulation Program-FORTRAN,

    Water-Resources Investigations Report 94-4168, Reston, VA.

    36

    Page 1Page 2Page 3Page 4Page 5Page 6Page 7Page 8Page 9Page 10Page 11Page 12Page 13Page 14Page 15Page 16Page 17Page 18Page 19Page 20Page 21Page 22Page 23Page 24Page 25Page 26Page 27Page 28Page 29Page 30Page 31Page 32Page 33Page 34Page 35Page 36