Canadian Hydrological Drought: Processes and Modelling

1
Canadian Hydrological Drought Processes and
Modelling

• John Pomeroy, Robert Armstrong, Kevin Shook,
  Logan Fang, Tom Brown, Lawrence Martz
• Centre for Hydrology, University of
  Saskatchewan,
• Saskatoon

2
Prairie Hydrology - Reality
Smith Creek, Saskatchewan
3
Overview

• OBJECTIVE
• To better understand, describe and model the
  development of hydrological drought on the
  Prairies
• FOCUS evaluation and drought sensitivity of
• Processes
• Snow Redistribution, Accumulation and Melt
• Runoff Generation/Wetland Recharge
• Areal Evaporation
• Modelling
• Prairie Hydrological Modelling CRHM platform to
  create physically based hydrological models of
  soil moisture, evaporation, snow accumulation,
  small prairie stream runoff and wetland recharge

4
Calculating Prairie Snowmelt Runoff
5

• Spatially distributed blowing snow model
• 262,144 grids
• On each grid calculation of fluxes based on

St Denis, Saskatchewan
6
Spatially Distributed Blowing Snow Accumulation
- Feb
7
Spatially Distributed Blowing Snow Accumulation
End of March
8
Blowing Snow in Operational Drought Modelling

• For hydrological and agricultural water balance
  applications, need landscape type specific
  calculations, aggregated approach
• To calculate mass balance for landscape unit
  (TILE), need inputs from upwind tiles (source to
  sink)
• Possible to calculate transport from one tile to
  another
• Calculation order based on tile aerodynamic
  sequence (smooth to rough high elevation to low
  elevation). Transport out from one tile is
  transport in to next tile.
• Important to preserve continuity at multiple
  scales

Fallow Field
9
Spatially Aggregated Blowing Snow Accumulation, 7
tiles
10
Distributed vs Aggregated Blowing Snow Modelling
Areal average SWE from two resolutions of blowing
snow model and snow surveys distributed 111 mm,
aggregated 90 mm, and observed 97 mm.
11
Prairie Evaporation

• Actual Evaporation critical component of drought
• Uncertainty in estimating Evaporation
• Various theoretical relationships with differing
  sets of parameters (a, zo, d, vegetation, water),
  variables (K?, L?, u, T, q) and state variables
  (?, Ts)
• Highly spatial variability)
• subgrid variability
• Advection to ponds
• Aggregation in LSS.
• Tiles
• Problem of changing tile area during drought
• Continuity
• All models limit water for evaporation by
  tracking supply
• Prairie plants dont care and send roots to
  available water (3 m)

12
Field Observation NECESSARY
St Denis National Wildlife Area, Saskatchewan
13

• St Denis, SK, summer 2006, dry but no drought
• 3 physically based methods (Granger GD,
  Penman-Monteith, Dalton Bulktransfer LSS-like
  compared to best observation sets from eddy
  correlation.
• Possible to set soil moisture for resistance
  and continuity aspects of CRHMfrom field
  measurements of soil water (no model calibration).

14
Lethbridge Ameriflux Site (2001)
EnteredDrought as summer 2001 Progressed Severe
Decline in Soil water content anddaily
actual evaporation
15

• Modeling evaporation under drought conditions
  requires soil moisture accounting
• Influence of canopy resistance term increases
  as season progresses
• Uncertainty in reference minimum for resistance
  PM, BT
• Not possible to set physically realistic
  parameters for Penman-Monteith and Dalton Bulk
  Transfer resistance schemes,
• Granger GD method in CRHM performed well in
  severe drought

16
Spatial Variability of Prairie Evaporation

• Important for hydrology
• Wetland recharge and dessication
• Streamflow generation, contributing area for
  runoff
• Two eddy correlation systems, 2007, pond and
  dryland

17
Thermal Image of St. Denis NWA (2007)Taken from
an infrared imager from an airplane
Provides basis for spatial distribution of net
radiation in Granger GD evaporation method
18
Distributed Daily Evaporation St Denis
Distributed -Outgoing longwave -Outgoing
shortwave -Aerodynamic roughness Granger GD
Modelwith commonatmospheric feedback, T, RH,
19
Spatial Frequency Distribution of Actual
Evaporation (one day)
wetland
dryland
mm daily actual evaporation
20
Cold Regions Hydrological Model Process Modules

• Developed from research at University of
  Saskatchewan and EC over several decades
• Radiation (slopes, estimation procedures)
• Blowing snow (snow transport sublimation)
• Interception (rain and snow)
• Snowmelt (open forest, advection, energy
  balance)
• Infiltration (frozen and unfrozen soils)
• Evaporation (Granger or Penman-Monteith)
• Soil moisture balance (with groundwater
  interaction)
• Routing (hillslopes, sub-surface and streamflow)

21
CRHM Use for DRI

• Hydrological evolution and feedbacks in drought
• Hydrological Drought Indices based on small basin
  soil moisture, streamflow and wetland levels
• Scaling methodology and process test bed
• Evaluate prairie land surface parameterisations
  and aggregation for MESH
• Develop prairie hydrology routing for MESH
• Provide drought hydrology tool for users

22
1999-2004/05 Drought Impact at St. Denis,
cumulative effect on the hydrological
processesand wetland water levelmodelled with
CRHM
23
CRHM Test at Wetland 109, St Denis
24
CRHM Application to Prairies

• Apply to two representative types of basins (RB)
• Well drained small prairie stream
• Wetland basin with much surface storage
• Create prairie drought surface of basin state
  variables
• Need standard atmospheric observations or
  reanalysis data (U,T, RH, Precip)
• Needs radiation (sparse observations!!!)
• Calculates soil moisture, streamflow, water
  storage, snowpack as state variables

25
NARR Daily Qsi
26
NARR gt CRHM Simulated Hourly Qsi
27
Creating Frequency Distributions of Wetlands for
Hydrological Modelling of Prairie Wetland
Representative Basin

• Need to have characteristic frequency
  distribution of wetlands this changes during
  drought.
• Test at St Denis where excellent data exists
• Simply route water excess along surface
  topography from one storage area to the next

28
0.1 m water added to DEM
29
0.3 m water added to DEM
30
St. Denis Slough Simulation
Runoff from spatially-constant precip -
spatially-constant Evap.
Need spatially variableevaporation
31
Conclusions

• Successful physically-based prairie hydrological
  modelling for small basins using CRHM no
  calibration
• Spatial scale for blowing snow accumulation and
  spring runoff calculation determined tiled
  approach adequate
• Suppression of blowing snow transport and
  enhancement of frozen soil infiltration
  responsible for much of wetland desiccation in
  drought
• Evaluation of evaporation models and observations
  suggests that soil moisture should be a product
  rather than a driver of evaporation calculations.
  Possible to distribute Granger method.
• Spatial distribution of evaporation, pond storage
  and runoff provides basis for upscaling
  atmospheric feedbacks and calculating hydrology
  in drought.
• CRHM ready for application to develop
  Prairie-wide hydrological drought products,
  Representative Basin soil water, runoff, water
  storage