Post on 24-Feb-2016
description
An Accuracy Assessment of the Polar MM5 Snow Accumulation Model
• Jared Carse• Mentors: Dr. David Braaten, Dr. Claude Laird• Graduate Mentors: Aaron Gilbreath, Mitch
Oswald
Polar MM5 Model• Fifth Generation Mesoscale Model
modified for polar climates• Developed by Burgess et al.
– “A spatially calibrated model of annual accumulation rate on the Greenland Ice Sheet (1958–2007)”
• Calibrated using firn cores and meteorological station data
• Spans year 1958-2007• Raster data set
3 of XX
Radar Traverse• 375 kilometer
traverse from NGRIP to NEEM
• Snow Accumulation Radar
• Layers traced in MatLab
4 of XX
NGRIP
NEEM
5 of XX
Converting Radar data• Extract the thickness
of between annual traced layers
• Convert the water equivalent units using ice core density profiles
• Density interpolated between NGRIP and NEEM density profiles
6 of XX
1994195019061862181817741730168616420
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
NGRIP Density Profile
firn density
YearFi
rn D
ensi
ty (g
/cm
3)
Import Radar Data into ArcGIS• Each layer extracted
from MatLab has Lon/Lat coordinates
• Projected into the same coordinate system that the Model raster data uses
7 of XX
Convert Radar Data to Raster• Same spatial
resolution is needed to accurately compare between radar and model– Mean of points that lie
in each pixel• Raster Calculator
used to form error assessment
8 of XX
Model Error and Bias
9 of XX
1958
1961
1964
1967
1970
1973
1976
1979
1982
1985
1988
1991
1994
1997
2000
2003
2006
-0.06-0.04-0.02
00.020.040.060.080.1
0.12 Model Bias
bias
Wat
er e
quiv
alen
ce (m
eter
)
Average Bias =23.403 mm
1958
1961
1964
1967
1970
1973
1976
1979
1982
1985
1988
1991
1994
1997
2000
2003
2006
0
0.02
0.04
0.06
0.08
0.1
0.12RMSE
RMSE
Wat
er e
quiv
alen
ce (m
eter
)
Average RMSE = 40.598 mm
10 of XX
1958
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
0.09
0.11
0.13
0.15
0.17
0.19
0.21
0.23
0.25
Radar vs. Model Average Accumulation
Radar MeanModel Mean
Year
Accu
mul
atio
n - (
w.e
. - m
eter
)
Model st. dev. = 21.834 mm
Radar st. dev. = 26.211 mm
11 of XX
0
0.05
0.1
0.15
0.2
0.25
Southern Third (NGRIP side)
ModelRadar
0
0.05
0.1
0.15
0.2
0.25
Middle Third
ModelRadar
1958
1962
1966
1970
1974
1978
1982
1986
1990
1994
1998
2002
2006
0
0.05
0.1
0.15
0.2
0.25
0.3
Northern Third (NEEM side)
ModelRadar
12 of XX
1994
1991
1988
1985
1982
1979
1976
1973
1970
1967
1964
1961
1958
0
0.05
0.1
0.15
0.2
0.25
0.3
NGRIP - Model vs. Ice Core (1958 - 1994)
ModelIce Core
Year
Accu
mul
atio
n w
ater
equ
ival
ence
(met
ers)
2003
1999
1995
1991
1987
1983
1979
1975
1971
1967
1963
1959
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
NEEM – Model vs. Ice Core (1958 - 2003)
ModelIce Core
Year
Accu
mul
atio
n –
wat
er e
quiv
alen
ce (m
eter
)
NAO - Radar
13 of XX
0.04
0.06
0.08 0.1 0.1
20.1
40.1
60.1
8 0.2 0.22
-4
-2
0
2
4
6
Southern Third – NGRIP side
Southern MedianLinear (Southern Median)
Accumulation
NAO
Inde
x
0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22-4-3-2-1012345
Middle Third
Middle MediansLinear (Middle Medians)
Accumulation
NAO
Inde
x
Correlation = .17264
0.08 0.1 0.120.140.160.18 0.2 0.220.240.260.28-4-3-2-1012345
Northern Third – NEEM side
Northern Me...
Accumulation
NAO
Inde
xCorrelation = .090376
Correlation = .081642
Photo extracted from /www.ldeo.columbia.edu/res/pi/NAO/
NAO – Ice Cores
14 of XX
0.1 0.12 0.14 0.16 0.18 0.2 0.22 0.24 0.26-4
-3
-2
-1
0
1
2
3
4
5 NGRIP Ice Core
NGRIP Ice Core
Accumulation - water equivalent (meter)
NAO
Inde
x
0.09 0.13 0.17 0.21 0.25 0.29-4
-3
-2
-1
0
1
2
3
4 NEEM Ice Core
NEEM Ice Core
Accumulation w.e (meter)
NAO
Inde
x
correlation -0.17534
Correlation0.22803
Summary of NAO• From ice core data
– Negative NAO year produce higher accumulation at NEEM– Positive NAO years produce higher accumulation at NGRIP
• To be statistically significant– At alpha = 0.10– Correlation = .243– The largest correlation occurs at NGRIP ice core with
correlation = .22803– Therefore the relationship between NAO and accumulation is not
significant. A larger sample size is needed, i.e. more years need to be measured
15 of XX
How well does the model perform?
• Ice core bias = 17.545 mm
• Radar bias = 34.355 mm
• Model compared to both radar and ice cores, consistently over-predicts
16 of XX
1994
1991
1988
1985
1982
1979
1976
1973
1970
1967
1964
1961
1958
-0.1
-0.05
0
0.05
0.1
0.15
0.2NGRIP - Bias Values
Model - Core Bias
Model - Radar Bias
Wat
er e
quiv
alen
ce (m
eter
)
Future uses of model• Could be used as tool
to help trace layers– Model corresponds
with ice cores fairly well
– Large-scale coverage rather than point sources that ice cores give us
17 of XX
Caveats • The model accumulation is set annually,
January 1 – December 31• Radar layers can be variable
– Large storms could produce layers that appear to be annual
• Model appears to be less variable than accumulation detected through radar
18 of XX
19 of XX