Advances in Macroscale Hydrology Modeling for the Arctic Drainage Basin Dennis P. Lettenmaier Department of Civil and Environmental Engineering University.

Advances in Macroscale Hydrology Modeling for the Arctic Drainage Basin

Dennis P. LettenmaierDepartment of Civil and Environmental Engineering

University of Washington

53rd Arctic Science ConferenceUniversity of Alaska Fairbanks

September 19, 2002

Thermohaline Circulation

G. Holloway, Institute of Ocean Sciences, Sidney, BC

Arctic drainage basin

Ob

Mackenzie

Lena

Yenesei

0

20

40

60

80

100

45 55 65 75

Latitude (degrees)

Ba

sin

Are

a (

%)

gauged area

ungauged area

Mackenzie River basin early version VIC snow season length results

RS

T Index

Energy Balance

Ob River basin early version VIC snow season length results

RS

T Index

Energy Balance

• 21 participating land surface models (typically land surface representations in coupled land-atmosphere models, representing surface energy and water balances

• Study site: Torne-Kalix River basin (Sweden and Finland), ~58,000 km2

• Each model provided ~10 years of gridded (1/4 degree) surface radiative and meteorological forcings

• Streamflow, snow extent, and surface water balance observed or inferred from observations

PILPS (Project for Intercomparison of Land surface Parameterization Schemes) Experiment 2e

Figure 1. Location of the Torne and Kalix Rivers (red) within the BALTEX domain (white)

Mean annual snowfall apportionment to melt and sublimation

Predicted annual average latent heat flux (1989 – 1998) and estimate from basin water balance

Predicted average last day of snow cover (1989 – 1998) and satellite estimate

PILPS-2e Conclusions• Inter-model variations in mean annual runoff were primarily related to

winter snow sublimation, even though summer ET was much higher.

• Storage of snowmelt runoff in the soil column primarily influenced the timing of peak runoff, rather than volume.

• Models with high sublimation generally lost their snow pack too early and underpredicted annual runoff. Differences in snow sublimation were largely a result of differences in snow surface roughness.

• The greatest among-model differences in energy and moisture fluxes occurred during the spring snowmelt period.

• Differences in net radiation were governed by differences in the surface temperature during winter, and by differences in surface albedo during snowmelt, but were minor when snow was absent

• The formulation of aerodynamic resistance and stability corrections in areas of no overstory were at least as important as the sensitivity to representation of canopy interception in explaining intermodel differences in winter evaporation.

Lakes and wetlands

Source: San Diego State University Global Change Research Group

Landcover from Landsat MSS images (Muller et al. 1999).

Putuligayuk River

Snowmelt water balance

Snow Water Equivalent +87 +124 +89Surface Runoff -56 -87 -56

Evaporation/Condensation

-6 -7 +4

Change in SurfaceStorage

+25 +30 +37

1999 2000 2001

Saturated extent 1999 and 2000

0

100

200

300

400

6/10 6/30 7/20 8/9 8/29Inu

nd

ate

d a

rea

(km

2 )

19992000

2000

= wet = dry

a.

b. c. d. e.

Predicting the effects of lakes and wetlands

• Lake energy balance based on:– Hostetler and Bartlein

(1990)

– Hostetler (1991)

• Assumptions:– One “effective” lake for

each grid cell;

– Laterally-averaged temperatures; and

Lake energy balance

Lake surface energy balance

Mean daily values, June-August 2000

Mean diurnal values, June-August 2000‘Lake 1’, Arctic

Coastal Plain, Alaska

Observed

Simulated

Mean temperature profile (1993-1997)Toolik Lake, Alaska

Lake ice formation and break-upTorne River, Sweden

ice formationice break-up

= area > 20 km2 = area < 20 km2

Wetland Algorithm

soilsaturated

land surface runoff enters

lake

evaporation depletes soil

moisture

lake recharges

soil moisture

Simulated saturated extentPutuligayuk River, Alaska

Simulated mean annual evaporation

with lake algorithm without lake algorithm

• Simulated annual evaporation increases by 60%

Blowing Snow

Günter Eisenhardt 3.31.2002, Iceland

Distribution of terrain slopes

Trail Valley Creek, NWT Imnavait Creek, Alaska

Sub-grid variability in wind speed

• Wind speeds assumed to follow a Laplace (double exponential) distribution

• Requires the standard deviation of wind speed, proportional to: – grid cell mean wind speed – standard deviation of terrain slope– autocorrelation of terrain slope

• Total sublimation flux found by summing sublimation for the average wind speed of ten equally-probable intervals

Non-equilibrium Transport

average fetch, f

transport = 0

transport = Qt(x= f)

snow

Estimating average fetch

vegetation type terrain slope terrain st. dev

Simulated annual sublimation from blowing snow

Sensitivity to fetch

SWE and active layer depth