estimation and prediction of fresh water runoff based on atmospheric data: preliminary results cody...
TRANSCRIPT
Estimation and Prediction of Fresh Water Runoff Based on
Atmospheric Data:Preliminary results
Cody D. S. Sipkema4th Year Environmental Engineering Co-op Student
Dalhousie University and
J. Chassé, BIO.
Project Goals
• Develop an algorithm to estimate local river discharge rates by routing fresh water through watershed systems
• Estimate historical freshwater discharge rates where no data is available
• Predict the effect of climate change on local river discharge rates by connecting the routing model to GCMs
• Incorporate this algorithm into the CRCM to provide data for regional atmosphere-ocean downscalling systems
Context: Freshwater is important in coastal water
• Freshwater maintains estuarine circulation
• Activate exchange rate with the ocean
• Affect stratification
• Affect Ice formation
Methods: Data Used• USGS HydroSHEDS datasets
• Domain: 145W-52W, 25N-60N (Spatial)– DEM (Digital Elevation Model) at 30 arc seconds
– Flow Accumulation Raster at 30 arc seconds
– Watershed Polygons
• NCEP Reanalysis I datasets• Domain: January 1, 1948- June 27, 2007 (Temporal)
Same spatial domain, data every 6 hrs.– Surface Temperature at 2m
– Precipitation Rate
– Latent Heat Flux
Methods: Watersheds
• Watershed Polygons extracted from shape file
• Only those which drain into the Atlantic basin and are of or above 200 sqkm area are deemed “significant watersheds”
• Significant watersheds are given an ID number
• Solved problem of small (often single cell) coastal watersheds
Methods: Watersheds
• Large number of small and single celled coastal watersheds
• These smaller watersheds are insignificant and should be ignored
•The area covered by these smaller watersheds however cannot
• Outlets were found for each significant watershed using the flow accumulation raster
• The cell within a polygon with the highest flow accumulation value was deemed the outlet
• Outlet coordinates and ID numbers were stored and carried into the “New Mesh”
Methods: Watersheds
Methods: New Mesh
• New mesh constructed for calculations and modeling
• 0.25 Decimal Degree mesh used
• 30 times coarser than the original DEM data
• Cell values were assigned using significant watershed polygon list and the DEM
• Cell values were assigned based on whether their center was found to be…
– Within a significant watershed polygon (Cell given ID value of basin)– Within the ocean or water body (Cell given no data flag -9999)
• If the cell was not found to be within either of these possibilities, they were given the ID value of their closest significant watershed neighbour (aggregation and assimilation of insignificant watershed areas)
• This resulted in a “Routing Raster” that indicates where received precipitation in each cell would travel as runoff
Methods: New Mesh
• NCEP data provided in coarser T62 Gaussian grid
• For each time step (6hrs) across the spatial domain values of precipitation (P), temperature (T) and latent heat flux (LHF) were calculated (in cells containing IDs) by the bilinear interpolation method
Methods: Fresh Water Routing
• Evaporative losses (E) were approximated using interpolated LHF values and the Latent Heat of Vaporization (LHV) at the interpolated temperature.
• LHV of fresh water at a given temperature based on a cubic regression relationship
• LHV(T) = −0.0000614342 (T^3) + 0.00158927(T^2) − 2.36418(T) + 2500.79
• Density of water was assumed constant• Ρwater =1000kg/(m^3)
Methods: Fresh Water Routing
• Snow pack and melt was incorporated using a temperature indexing approach on a cell by cell basis
• Snow is accumulated whenever the temperature falls below the threshold value (0 degrees Celcius)
• When temperatures rise above this threshold the snow melts at a given rate constant proportional to the difference in temperature
Melt= Cmelt * (Tcell-TThreshold) Where: Cmelt: Empirical coefficient ( m^3/(s*(degC)) Tcell: Interpolated cell value for temperature TThreshold: The temperature at which snow begins to melt.
Methods: Fresh Water Routing
• Flow Delay is incorporated to approximate the travel time of water
• This is done assuming a constant flow velocity for all rivers and assuming a straight path from the cell to the outlet
• Thus distant cells runoff will not contribute to the flow at the outlet until a set future time dictated by the delay.
Methods: Fresh Water Routing
Preliminary Results
Parameters
Ef = 0.65 (Evaporation Coefficient)
Cmelt =1E-8 m^3/s*C
V= 1.3 m/s (River velocity)
Saint John River Basin Watershed ID: 65675
Significant Watershed ID: 323
Outlet Coordinates: 71.1958, 46.8125
Area:54,927.2sqkm
Saint John River Basin
• Saint John River (Annual Means)• Correlation Coefficient = 0.736–0.81(Missing Data)
– Note: Historical Data from Pokiok and Mactaquac gauging stations.
Annual Discharge
0
200
400
600
800
1000
1200
1400
1945 1950 1955 1960 1965 1970 1975 1980 1985 1990
Time (Year)
Dis
ch
arg
e (
m^
3/s
)
Historical Data
Model Discharge
• Saint John River (Monthly Means)• Correlation Coefficient = 0.8097
Saint John River Basin
Saint John River Monthly Means
-1000
0
1000
2000
3000
4000
5000
1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969
Time
Dis
char
ge (m
^3/s
)
Historical Data
Discharge (Cmelt 1E-8)
Churchill River Basin Watershed ID: 30370
Significant Watershed ID: 107
Outlet Coordinates: -60.1875, 53.3208
Area: 92,698sqkm
Churchill River Basin
• Saint John River (Annual Means)• Correlation = 0.57• Note: Churchill Hydro Began operations in December 1971
Project started in July 1967
Controlled flow and diverted waterways (Affects results)
Annual Discharge of Churchill Basin
0
500
1000
1500
2000
2500
1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986
Time
Dis
ch
arg
e (
m^
3/s
)
Model Discharge
Historical Data