chuck o’hara * vladimir j. alarcon * william mcanally ** james martin ** jairo diaz **

22
GeoResources Institute AN RPC EVALUATION OF THE WATERSHED MODELING PROGRAM HSPF TO NASA EXISTING DATA, SIMULATED FUTURE DATA STREAMS, AND MODEL (LIS) DATA PRODUCTS Chuck O’Hara* Vladimir J. Alarcon* William McAnally** James Martin** Jairo Diaz** Zhiyong Duan** * GeoResources Institute, Mississippi State University ** Civil Engineering Department, Mississippi State University

Upload: glain

Post on 14-Jan-2016

25 views

Category:

Documents


0 download

DESCRIPTION

AN RPC EVALUATION OF THE WATERSHED MODELING PROGRAM HSPF TO NASA EXISTING DATA, SIMULATED FUTURE DATA STREAMS, AND MODEL (LIS) DATA PRODUCTS. Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz ** Zhiyong Duan **. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute

AN RPC EVALUATION OF THE WATERSHED MODELING PROGRAM HSPF TO NASA EXISTING DATA, SIMULATED FUTURE

DATA STREAMS, AND MODEL (LIS) DATA PRODUCTS

Chuck O’Hara* Vladimir J. Alarcon*William McAnally**

James Martin**Jairo Diaz**

Zhiyong Duan**

* GeoResources Institute, Mississippi State University

** Civil Engineering Department, Mississippi State University

Page 2: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute Introduction

• The modeling of watershed hydrology requires input from several types of databases.

• Digital elevation data are used at the beginning of the watershed modeling process to abstract the watershed physiographic features into a manageable number of parameters.

• Many other datasets that include land cover, soils, precipitation, and others are employed in the models to characterize the land, hydrology, and forcings for model event-driven simulations.

• Data values for layers in the model are frequently aggregated by their majority value for each sub-watershed area.

• Sensitivity analysis can be used to assess the influence of variables/parameters on the state of a modeled system.

• The analysis of sensitivity of state variables has been widely used in water resources modeling to quantify the reliability of the output or during model calibration processes.

Page 3: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute

Hydrologic Modeling – A Data Driven Process

• Effects of the quality of elevation data in hydrological simulations are substantial.• Soil moisture conditions, precipitation, land cover, soil characteristics, roughness,

and other factors also play significant roles in providing input data for watershed modeling.

• Primary tasks that include watershed delineation and delineation of sub-watershed model-response units are largely dictated by the quality and information content of the DEM data employed.

• Digital Elevation Model’s grid size, scale affect significantly the calculation of topographic descriptors of

catchments slope, catchment area, topographic index, etc.

• Topographic parameters are used by hydrological models to estimate runoff, stream flow, base flow and other hydrological indicators.

• Land and soil characteristics are significant to describe the land processes and hydrologic conditions within the watershed.

• Weather information and forecasts provide critical “forcings” information to modeling efforts that predict low-flow conditions, flood conditions, or water quality characteristics.

Page 4: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute

RPC HSPF Experiment Components

Existing NASA Data Sources that offer potential enhancements to current BASINS-HSPF Implementations:

MODIS land coverMODIS vegetation indices Shuttle Radar Topography Mission (SRTM-30m)

Future NASA Data Sources that will be employed as RPC Simulated Data Sources to Evaluate:

Synthetic VIIRS Land CoverSynthetic VIIRS Vegetation IndicesLIS-based precipitation (4-km)LIS-based ET (1-km)LIS-based solar radiation

Page 5: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute Study area

• 2 catchments in Saint Louis Bay Watershed Wolf River

Catchment area: 983 sq. km Average flow: 20.1 cms

Jourdan River: Largest contributor of flow to

the Saint Louis Bay Catchment area: 882 sq. km Average flow: 24.5 cms

Jourdan River Catchment

Page 6: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute RPC Evaluation Activities

• Implement baseline HSPF model for selected watersheds and employ a series of DEM layers to conduct watershed delineation, subdelinetion, and characterization of watershed parameters. Elevation:

USGS: NED 30-meter SRTM: 30-meter (NASA data set)

Landuse and Vegetation Condition – MODIS and * Synthetic VIIRS

Precipitation, Solar Radiation, & Soil moisture – * NASA LIS

• Impact on hydrological simulation Flow Water quality

* RPC Synthetic Data and Model Data Evaluation Components

Page 7: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute

HSPF RPC EVALUATION OBJECTIVES

The purpose of this research is to identify how sensitive are HSPF model outputs including flowrate estimations are to perturbations in topographical parameters and to substitution of data sets that characterize the land as well as weather processes.

The baseline model will be constructed for the Jourdan and Wolf Rive catchments in coastal Mississippi and will be extended to consider areas being modeled along with data substitutions being implemented by the team at Goddard Space Flight Center.

X1 Flow

Topographical parameters

HSPFX2

X3

& RPC substitution of NASA data

sources (existing and futures simulated)

Page 8: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources InstituteMethodology

Comparison• BASINS summarizes the topographic information per sub-basin

and per stream in two tables:• These tables are used to do a comparison (per sub-basin)

between the resulting delineations from the different elevation datasets for each of the watersheds under study.

AttributesStreams

HSPFAttributes

Sub-basins

DELINEATION TABLES

BASINS

Page 9: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources InstituteMethodology

Comparison tables

BASINS : Attributes of Sub-basins

Sub-basin area

AREA

SCHEMATIC

AREA FACTOR

Sub-basin slope

SLO1

Stream depth

DEP1

PWAT_PARM2

SLSUR

F-TABLES

Used as a reference depth to calculate

other F-table depth values

Stream width

WID1

F-TABLES

Used to calculate mean wet area with

depth and length

BASINS : Attributes of Streams

Maximum/minimum Elevation

MaxEl/MinEl

RCHRES-HYDR-PARM2

Used to calculate DELTH

Stream length

LEN2

RCHRES-HYDR-PARM2

LEN

HSPF

HSPF

A

B

Page 10: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources InstitutePrevious Research, Model Calibration, and

Sensitivity Analysis Methodology

• The Jourdan River catchment was delineated using elevation data from the National Elevation Dataset (NED) in BASINS: 30 Meter Resolution (1 arc-second)

• An HSPF application for Jourdan River was generated from within BASINS and calibrated for flow rate. USGS station 02481570 at Santa Rosa for the period 1962-1966

• To isolate the effects of topography-related parameters during sensitivity analysis, the calibration was performed using parameters that are not related to topography: LZSN: Lower zone nominal soil moisture storage INFILT: Index to the infiltration capacity of the soil UZSN: Upper zone nominal soil moisture storage

• Parameters that are mildly dependent on topography were also used INTFW: Interflow inflow parameter

Page 11: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute

• Sensitivity of flow rate estimations to perturbations in the following variables: Length of the overland flow plane: LSUR Slope of the overland flow plane: SLSUR F-tables

Stream width (WID1), Stream length (LEN2), and

• Strategy: One-variable-at-a-time approach.

Percent perturbations to the variables included in this study were increased/decreased in: ±100%, ±50%, ±10% and ±1% from the base values.

The estimations of flow for the calibrated case were considered the base case.

The combination of small and big perturbations allowed identifying non-linear sensitivities.

Normalized sensitivity values were calculated for each perturbation.

Methodology

Sensitivity analysis (cont’d)

Depth

Area Volume Outflow

F-TABLE

Page 12: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute

NSUR

DEP1

Stream Length L

EN2

SUB-BASIN AREA

SLSUR

LSUR

WID1

Max Elev

Min Elev

NSUR

DEP1

SUB-BASIN AREA

SLSUR

LSUR

WID1

HSPF variables Used in sensitivity experiments

Max Elev

Min Elev

Page 13: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute Conclusions

• HSPF-estimated flow is not very sensitive to LSUR (length of the overland flow plane) and SLSUR (slope of the overland flow plane). Relative changes in flow estimations to perturbations to these

variables are lower than 1% from the base values. • The variables Stream Length and Stream Width produce significant

percent changes on simulated flow when small changes (1%) are made to base values of those variables.

• 20% of those changes are at least greater than 1.5% reaching up to 16% change from the base values.

• Since stream length and stream width are highly dependant from the size and shape of the corresponding sub-catchment, the delineation of the watershed will drive the value of those variables.

Page 14: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute Methodology

Watershed delineation• Two elevation datasets were used to delineate the Saint Louis Bay

watersheds. EPA-USGS DEM: 300 Meter Resolution, 1-Degree Digital Elevation Models

(DEM) that corresponds to 3 arc-second (or 1:250,000-scale) USGS topographic map series.

EPA-NED: USGS 30 Meter Resolution, One-Sixtieth Degree National Elevation Dataset.

Current studies include 30-m-SRTM and 5-m-IFSAR data. Results will be presented in future reports.

• The watersheds under study were delineated using the automatic delineation option available in BASINS.

• To compare results, all delineations were performed with: no-flow towards inner cells, 38 sq km threshold area, 31 outlets (1 outlet was manually placed at the location of the USGS 02481510

Station at Landon). • The National Hydrographic Dataset (NHD) for streams was used in all

delineation procedures.

Page 15: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute Results• Jourdan and Wolf Rivers catchments in Saint Louis Bay

A B USGS-DEM (250K, 300 m)

NED (24K, 30 m)

Page 16: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute ResultsPercent differences in topographical indicators for Jourdan River

PERCENT DIFFERENCES

Basin Sub-basin name Area Slope1 Wid1 Dep1 Length2 Slo2 Min El Max El

9 Hickory Creek 0.51 209.69 0.30 0.20 10.02 77.31 -46.17 10.33

10 White Cypress Creek 0.35 295.51 0.21 0.14 27.20 17.86 -46.28 -14.22

11 Catahoula Creek -1.25 190.76 -0.75 -0.50 4.08 10.18 -63.29 -9.04

12 Crane Pond Branch -9.36 209.95 -5.72 -3.84 11.76 8.90 -68.00 -30.63

14 Jourdan River -16.81 1322.55 -10.46 -7.12 7.26 245.15 -55.00 -8.50

13 Crabgrass Creek -3.50 238.11 -2.11 -1.42 5.17 218.06 -57.17 1.20

17 7.39 344.73 4.37 2.91 3.32 39.84 -70.63 -32.25

18 Dead Tiger Creek -11.94 295.71 -7.35 -4.95 -70.46 806.53 -70.88 -44.33

20 Jourdan River -42.66 508.70 -28.38 -19.96 16.65 -16.31 -75.33 -29.75

• 300m-250K-USGS-DEM-calculated overland flow plane slopes (SLO1) are up to 14 times bigger than SLO1 values calculated using 30m-24K-NED.

• Stream Lengths (LEN2) are slightly bigger• Minimum Elevation (Min El) values are slightly smaller

Page 17: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute Results• Percent differences in topographical indicators for Wolf River

PERCENT DIFFERENCES

BasinSub-basin name Area Slo1 Wid1 Dep1 Len2 Slo2 MinEl MaxEl

1 Wolf River 1.73 -59.57 1.03 0.69 -5.00 -5.06 13.51 2.81

2Alligator Creek -0.34 -66.70 -0.20 -0.14 -60.96 -42.87 6.98 -11.10

3 Wolf River 1.09 -67.56 0.65 0.43 -18.87 -22.76 30.04 0.55

4 Murder Creek 0.62 -61.92 0.37 0.25 -3.88 -31.08 28.33 -0.58

5 Crane Creek 6.45 -67.52 3.82 2.53 -9.56 -2.87 -3.45 -8.29

6 Wolf River -3.99 -65.56 -2.42 -1.61 -18.69 29.84 -0.99 2.12

23 Wolf River (*) -1.38 -64.20 -38.21 -27.46 -12.46 -28.25 0.22 -43.65

• 300m-250K-USGS-DEM-calculated overland flow plane slopes (SLO1) are half smaller than SLO1 values calculated using 30m-24K-NED.

• Stream Lengths (LEN2) are slightly smaller• Minimum Elevation (Min El) values are slightly bigger

Page 18: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute Conclusions• Resolution of elevation data affects watershed delineation by providing

more sub-basins when using coarser datasets. • Higher-resolution datasets allow better delineation of flat areas.• For flat areas (Jourdan)

overland flow plane slope values estimated using the USGS-DEM dataset are bigger than slope values estimated using the NED elevation data.

Length of streams are slightly bigger when using USGS-DEM Minimum and maximum elevations values also present noticeable

percent differences.• For Rougher areas: Luxapallila and Wolf:

Overland flow slope values resulting of using the NED dataset are also different (50% in average) than those values calculated using the USGS-EPA dataset.

NED-generated sub-basin slope values are bigger than the USGS-EPA generated slopes (for Jourdan this was reversed).

• This seems to suggest that coarser datasets overestimate sub-basin slopes in flat watersheds and underestimate slopes in roughed terrain.

Page 19: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute RPC Evaluation Components

• Future delineation studies using other elevation data SRTM: 30-meter

• Impact on delineation Sub-basins Stream characterization

Longitudinal (stream length and slope) Cross sectional (F-tables)

Page 20: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute RPC Evaluation Components

• Integration of NASA-LIS parameters for parameters including, but not limited to the following:

Precipitation, Solar Radiation, and Soil moisture

Page 21: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute RPC Evaluation Components

Standard implementations of HSPF do not typically draw upon remote sensing inputs for land characterization. The baseline configuration of the model will be adapted to integrate NASA MODIS data products for inputs including the following:

– Land Cover – Vegetation Condition and Change– Vegetation Indices

In the evaluation, after the MODIS data is successfully integrated into the HSPF model, synthetic VIIRS data will be created from MODIS data streams and results will be tested against results generated through the use of the MODIS products within the configured model.

Page 22: Chuck O’Hara * Vladimir J. Alarcon * William McAnally ** James Martin ** Jairo Diaz **

GeoResources Institute PDR Questions and Discussion Items

RPC Experimental Design:

Baseline and Future Data Assimilation Plan:

Strength of RPC Team:

Adequacy of Field Data Campaign and Local Knowledge Expertise:

Identification of Pathway to ISS: