Using fixed process-oriented model structure as a starting point for hydrological researchUsing fixed process-oriented model structure as a starting point for hydrological researchOlga Semenova Olga Semenova

Hydrograph Model Research Group, St. Petersburg, RUSSIA, www.hydrograph-model.ruHydrograph Model Research Group, St. Petersburg, RUSSIA, www.hydrograph-model.ru

In this poster the general line of the Hydrograph model applications in different environments started at EGU 2009 is followed. The main idea is to show that a single process-based model with fixed distributed structure may be used to investigate the processes in very different environments. It seems reasonable in each case to adjust existing, widely tested model structure which is consistent with the scales of hydrological processes and the availability of parameter information.

As usual, some general thoughts on how we can advance hydrological modelling from its current state are presented as well:•The hydrological community should find the courage to leave some old, discredited theories and ideas in the past; instead of continuing to constructing model structures from non-physically-based blocks and conceptual simplifications; maintaining an understanding of the general processes combined with high-quality observational data may give a rise to new insights. •The convenience of some mathematical approaches should not take priority over the understanding of the processes. •New model structures should be carefully verified by independent data on similar watersheds (even if only runoff is available).•Model structure, where this is flexible, should not fully rely on calibration techniques to find “optimal structures”. Rational intellectual calibration of model parameters and decisions on appropriate model structures based on hydrological principles can be used if supported by an understanding of the underlying processes.•Reasonably developed, process-oriented, robust, fixed model structures can be an excellent starting point for hydrological research and will provoke serious consideration of the hydrological system and deficiencies in its description. For instance, the limitations of fixed model structures may point out new directions in observations that are necessary to more properly implement the model and form the basis of enhanced collaborations between modellers and experimentalists.

1. Background1. Background

2. Hydrograph Model2. Hydrograph Model

R • Single model structure for

watersheds of any scale

• Adequacy to natural

processes while looking

for the simplest solutions

• Minimum of manual

calibration

Forcing data: precipitation,

temperature, relative

humidity

Output results: runoff, soil

and snow state variables,

full water balance

Slope transformationof surface flow

Initial surfacelosses

Infiltration andsurface flow

Heat dynamicsin soil

Snow coverformation

Heat energy

Interception

Heat dynamicsin snow

Snow melt andwater yield

EvaporationWater dynamics in soil

Channel transformation

Runoff at basin outlet

Underground flow

Transformation of underground flow

PrecipitationRain Snow

3. Red River of the North, MN, USA - 3. Red River of the North, MN, USA - South Branch of Buffalo River (749 kmSouth Branch of Buffalo River (749 km22))

4. Dry Creek, Idaho, USA4. Dry Creek, Idaho, USA

5. California, USA – San Joaquin River 5. California, USA – San Joaquin River as an example (4341 kmas an example (4341 km22))

6. Summary6. Summary

AcknowledgementsAcknowledgements• Dry Creek project was made possible with funding from NOAA Grant NA08NWS4620047. The author Dry Creek project was made possible with funding from NOAA Grant NA08NWS4620047. The author

acknowledges the invaluable assistance from James McNamara and Pam Aishlin at Boise State acknowledges the invaluable assistance from James McNamara and Pam Aishlin at Boise State University, and Pedro Restrepo at the Office of Hydrologic Development of the National Weather University, and Pedro Restrepo at the Office of Hydrologic Development of the National Weather Service, USA.Service, USA.

• Red River study was made possible with contract from NOAA NWS OHD. The author acknowledges Red River study was made possible with contract from NOAA NWS OHD. The author acknowledges the assistance of Pedro Restrepo at the OHD NWS, the support of Mike DeWeese and Steve Buan the assistance of Pedro Restrepo at the OHD NWS, the support of Mike DeWeese and Steve Buan from the NCRFC, Carrie Ohlheiser from the NOHRSC, and Victor Koren from OHD. The administrative from the NCRFC, Carrie Ohlheiser from the NOHRSC, and Victor Koren from OHD. The administrative and management support of the contract by Robert Brown of Len Technologies is highly and management support of the contract by Robert Brown of Len Technologies is highly appreciated.appreciated.

• California project was made possible with funding from NOAA Grant NA09NWS4620044. The author California project was made possible with funding from NOAA Grant NA09NWS4620044. The author acknowledges the assistance from Xiogang Gao at the Regents of University of California, and Pedro acknowledges the assistance from Xiogang Gao at the Regents of University of California, and Pedro Restrepo at the Restrepo at the OHD NWSOHD NWS, USA., USA.

• Attendance to EGU 2011 was made possible with the support of German-Russian Laboratory for Attendance to EGU 2011 was made possible with the support of German-Russian Laboratory for Polar and Marine Research which is highly appreciated.Polar and Marine Research which is highly appreciated.

Catchment Area: 28 km2

Elevation Range: 1030-2130 m

Grasses, shrubs, and conifer forests vary with aspect and elevation

Low Elevation Grass

Mid Elevation Shrub

High Elevation Forest

Bare ground

Grass

Shrubs

Trees

Runoff formation complexes

01.200910.200807.200804.200801.200810.200707.200704.200701.2007

Te

mp

era

ture

, d

eg

ree

C

35

30

25

20

15

10

5

0

-5

vo

lum

e w

ate

r c

on

ten

t

0.25

0.20

0.15

0.10

0.05

0.00

observed simulated observed simulated

Lower Weather station (1151 m), Lower Weather station (1151 m), soil, 5 and cm depth, 2007-2008soil, 5 and cm depth, 2007-2008

01.200910.200807.200804.200801.200810.200707.200704.200701.2007

Te

mp

era

ture

, d

eg

ree

C

25

20

15

10

5

0

vo

lum

e w

ate

r c

on

ten

t

0.25

0.20

0.15

0.10

0.05

observed simulated observed simulated

observed simulated

05.200303.200301.200311.200209.200207.200205.200203.200201.200211.2001

sn

ow

de

pth

, m

1.0

0 .8

0 .6

0 .4

0 .2

0 .0

2002-2003

05.200902.200911.200808.200805.200802.2008

sn

ow

de

pth

, m

1 .4

1 .2

1 .0

0 .8

0 .6

0 .4

0 .2

0 .0

2008-2009

Tree Line station (1651 m), Tree Line station (1651 m), snow depth (m)snow depth (m)

Map showing location Map showing location of the watershedsof the watersheds

observedsimulated - even soi l depthsimulated - di str ibuted soi l depth

08.200906.200904.200902.200912.200810.2008

m3

/s

1. 2

1.1

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

Observed and simulated flow for even Observed and simulated flow for even and distributed soil depth casesand distributed soil depth cases

observedno losses, no cal ibrationno cal ibration, l osses implementedcal ibrated, losses implemented

08.200807.200806.200805.200804.200803.200802.2008

m3

/s

2. 1

1.8

1.5

1.2

0.9

0.6

0.3

0.0

Results of sequential calibration of Results of sequential calibration of groundwater parametersgroundwater parameters

observed simulated

01.200412.200311.200310.200309.200308.200307.200306.200305.200304.200303.200302.200301.2003

m3

/s

0. 9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

observed simulated

01.200512.200411.200410.200409.200408.200407.200406.200405.200404.200403.200402.200401.2004

m3

/s

1 . 7

1.6

1.5

1.4

1.3

1.2

1.1

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

Observed and simulated runoff, LowerGageObserved and simulated runoff, LowerGageNS (2000 – 2009) = 0.43 NS (2000 – 2009) = 0.43

observed simulated

12.1010.1008.1006.1004.1002.1012.0910.0908.09

So

il m

ois

ture

, V

C

0 .20

0 .15

0 .10

0 .05

0 .00

observed simulated

01.1111.1009.1007.1005.1003.1001.1011.0909.0907.09

Te

mp

era

ture

, d

eg

ree

C

2 4 .0

22 .0

20 .0

18 .0

16 .0

14 .0

12 .0

10 .0

8 .0

6 .0

4 .0

2 .0

0 .0

-2 .0

-4 .0

-6 .0

-8 .0

observed Hydrograph Sacramento

01.200910.200807.200804.200801.200810.200707.200704.200701.2007

m3

/s

70

60

50

40

30

20

10

0

observed Hydrograph Sacramento

12.200910.200908.200906.200904.2009

m3

/s

300

280

260

240

220

200

180

160

140

120

100

80

60

40

20

0

The hydrology of the study area is characterized by:

•snowcover modified by wind redistribution;

•high surface runoff from the major spring snowmelt event as a result of frozen soils;

•most rainfall occurring in spring and early summer from large frontal systems and the most intense rainfall in summer from convective storms over small areas;

•poorly-drained natural stream networks;•large amount of artificial agricultural

tiles influencing the redistribution of runoff into the streamflow network;

•small post-glacial depressions which may not drain to any natural external drainage system and have large storage capacity.

The study area is located at the south-east part of the Red River of the North watershed. The main landscape is agricultural fields. Annual precipitation amounts to 600 – 750 mm.

Observed and simulated soil moisture (50 cm depth) and Observed and simulated soil moisture (50 cm depth) and temperature at (10 cm depth) at Crescent Lake stationtemperature at (10 cm depth) at Crescent Lake station

Observed and simulated runoff by Observed and simulated runoff by Hydrograph and Sacramento modelsHydrograph and Sacramento models

Accounting for Depression StorageThe depth of depression storage is the part of surface runoff depth expended for filling micro-relief depressions of the watershed surface. Potential opportunities to retain a definite part of the surface runoff are caused by the free capacity of depression storage, i. e. by the difference between the maximum possible capacity (DM) and the amount of water in the depressions at the moment of commencing surface runoff (HP). Then, the depth of surface runoff retention (D) with known surface runoff depth (Hq) can be defined by the equation

M

qPM D

HHDD exp1)(

Accounting for artificial drainage systemThe winter soil frost creates conditions for surface flow when frozen soil infiltration rates are very low and prevent water from infiltrating deep into the soil. In those cases, surface flow occurs and goes directly to the agricultural drainage systems loosing some water to the depression storages on the way. In such a way, this type of flow can not be considered typical surface flow in the sense it is used in the Hydrograph model. This type of flow has a more deferred contribution into the river network. Given the available information about the river network and drainage system, which is more generic rather than specific, accounting for those processes was made possible by a change in the surface runoff elements parameters. In this way, the values of the conditional constants a* (see [1] for details), were changed from 1000 to 10 for a surface type of flow and from 100 to 5 – for soil type of flow.

Used precipitation dataUsed precipitation data

1) Station data2) Snow Data Assimilation

System (SNODAS)3) MAPX – mean areal

precipitation calculated using the radar- and gauge-based information.

California

Feather River

American River

San Joaquin River

Ü Ü4237

119

Ü15 - 23

23 - 33

33 - 43

43 - 51

51 - 63

ÜEvergreen forest land

Lakes

Mixed forest land

Residential

Shrub and brush tundra

Ü

s1061

s1062

s1063

s1065

s1066

s1067

s1068

s1070

s1073

s1074

s1075

s1117

s1118

s740

s746

s747

s748

s751

s752

s818

s8369

s845

#0

#0

#0

#0

#0

#0

#0

#0

#0#0

#0

#0

#0

#0#0

#0

#0

#0

#0

#0

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

%2

FLR

MBR

TMR

PSR

PCK

OKH NTP

NFR

MHP

KSP

HNT

GTM

GRV

GRM

FRT

CRR

CHM

BGC

BDM

AGW

MTF

BTT

NFR

TMR

PSR

KSP

GTM

GRV

GRM

FRT

VLC

UBC

STR

RCK

GEM

FCH

AGW

Ü

San Joaquin River watershed

#0 precipitation

!( temperature

%2 relative humidity

San Joaquin River San Joaquin River watershed elevation watershed elevation

schemeschemeSan Joaquin River San Joaquin River

precipitation distributionprecipitation distribution

San Joaquin River soil San Joaquin River soil type distributiontype distribution

San Joaquin River land San Joaquin River land use type distributionuse type distribution

San Joaquin River San Joaquin River meteorological stations meteorological stations

distributiondistribution

simulated observed

10.200508.200506.200504.200502.200512.2004

m3

/s

850

800

750

700

650

600

550

500

450

400

350

300

250

200

150

100

50

0

simulated observed

10.200608.200606.200604.200602.200612.2005

m3

/s

850

800

750

700

650

600

550

500

450

400

350

300

250

200

150

100

50

0

simulated observed

12.200710.200708.200706.200704.200702.200712.2006

m3

/s

200

150

100

50

0

simulated observed

10.200808.200806.200804.200802.200812.2007

m3

/s

300

250

200

150

100

50

0

Observed and simulated runoff for San Joaquin River (2005 – 2008)Observed and simulated runoff for San Joaquin River (2005 – 2008)No calibration is implementedNo calibration is implemented

•The Hydrograph model performs well in different environments•Most parameters are estimated from physical characteristics and do

not require further calibration•Some parameters of conceptual components do require calibration•The correct way to approve or reject a model structure is by testing it

multiple times over basins in different conditions•Reasonably developed, process-oriented, robust, fixed model structure

can be an excellent starting point for any hydrological research provoking

serious consideration of the hydrological system and deficiencies in its

description

We do not have resources for developing a specific model

for every single watershed… We do not have resources to

play “modelling exercises” in the sense of making

thousands of runs aimlessly searching for some hidden

model structure or parameter meaning….

Can YOU afford them?