Climate Downscaling Techniques

1
Climate Downscaling
Techniques

• Marina Timofeyeva
• UCAR/NWS/NOAA

2
TALK OUTLINE

• What is Downscaling
• NWS NCEP CPC climate outlooks and regional
  climate
• Downscaling methods
• Application of CPC methods in Developing Local
  Climate Products
• Messages to take home

3
DOWNSCALING
DOWNSCALING is the transformation from a LARGE
SCALE feature to a SMALL SCALE one (not
necessarily of the same kind). DOWNSCALING
implies increases resolution of output.

4
NWS NCEP CPC weather climate outlooks
5
NWS NCEP CPC climate outlooks
This is a map of 344 climate divisions currently
in use over the U.S. Note the changing size as
one goes from east to west, as well as from one
state to another.
CPC uses 102 mega or forecast divisions in their
forecasts. The divisions in the West closely
correspond to NCDC climate divisions.
6
NWS NCEP CPC climate outlooks
7
NWS NCEP CPC climate outlooks
Precipitation Climatology 1971 - 2000
National precipitation map based on
high-resolution PRISM data (left) and on Climate
Divisions (right). Note the large gradients and
fine-scale variability in the Western U.S. that
is not reproduced in the right map.
Slide courtesy of Klaus Wolter, CDC
8
NWS NCEP CPC climate outlooks and Regional
Climate
Official CDs for Colorado (left) and Experimental
CDs (right) based on multivariate statistical
analysis of climate data that also include SNOTEL
data. Such new CDs are being derived for the
entire U.S.A.
Slide courtesy of Klaus Wolter, CDC
9
NWS NCEP CPC climate outlooks and Regional
Climate
of stations that did not reject the test H0
If climate at station is the same as at climate
division, then mean and variance at station and
climate divisions should be the same. THEY ARE
NOT!
10
NWS NCEP CPC climate outlooks and Regional
Climate
NUMBER OF STATIONS (OUT OF 9) THAT HAVE SQUARED
CORRELATION WITH CD gt0.8
11
DYNAMICAL DOWNSCALING
12
DYNAMICAL DOWNSCALING
13
DYNAMICAL DOWNSCALING

• ETA and AVN are examples on meteorological
  scale
• Climate applications
• Regional Spectral Model (RSM) and Seasonal
  Forecast Model (SFM)
• Nested in T62 and T40 NCEP coupled AOGCM using
  50, 30 and 20 km resolution grids
• tested for 1997 winter El Nino
• results
• RSM shows improvement in temperature forecast in
  comparison with AOGCM
• 50-km RSM is unable to forecast anomalies over
  high mountains
• 20-km RSM provides realistic distribution of
  precipitation, but overestimate its maxima

14
STATISTICAL DOWNSCALING WEATHER GENERATORS
Global Circulation Pattern observed
Climate observations
Calibration
Statistics
Class
Frequency analysis
Statistics
Global Circulation Pattern predicted
Prediction
Climate predictions
15
STATISTICAL DOWNSCALING CORRELATION MODELS
Climate Observations
GCM fields
Calibration
Statistics
Statistics
Statistical relationship
Prediction
Modeled climate Statistics
16
APPLICATION OF CPC METHODS IN DEVELOPING LOCAL
CLIMATE PRODUCTS

• EXAMPLE PARTNERSHIP PROJECT
• Western Region HQ (Andrea Bair)
• NWS OCWWS CSD (Marina Timofeyeva)
• NWS NCEP CPC (David Unger)

17
Utilized CPC Methods of Downscaling
Use of composites in the forecast of local
climate using climate variability modes
Translation of probability of exeedance (POE)
outlooks from climate divisions to local station
temperature forecasts
18
Utilized CPC Methods of Downscaling

Temperature Outlook Precipitation Outlook
Degree Day Outlook
Temperature Outlook Precipitation Outlook
Degree Day Outlook
19
Utilized CPC Methods of Downscaling - POE
Translation of forecast division POE to station
forecast was developed in CPC by Barnston, Unger,
and He. The POE outlooks became available in
December 1994, and the translations to station
temperature in 2000.
Observed T
POF ()
20
Utilized CPC Methods of Downscaling - POE
http//www.cpc.ncep.noaa.gov/pacdir/NFORdir/citydi
r/cpcctytd.dat
21
Utilized CPC Methods of Downscaling - POE
Step 1 Defining stations where the station
outlooks are used
22
Utilized CPC Methods of Downscaling - POE
Step 2 Developing regression equations for those
stations
In most cases there is a strong relationship
between temperature at station (y axis) and
climate division (x axis).
23
Utilized CPC Methods of Downscaling - POE
Step 2 Developing regression equations for those
stations Regression coefficients are adjusted for
the trend
Difference Tstation Tcd (ºF)
years
24
Utilized CPC Methods of Downscaling - POE
Step 3 Test of regression equations stability -
Cross validation
Calibration data set Verification Year
1971 1971
1972 1972
1973 1973
1974 1974
1975 1975

1997 1997
1998 1998
1999 1999
2000 2000
Cross validation allows expansion of the test
sample and protects against over fitting
25
Utilized CPC Methods of Downscaling - POE
Step 3 Test of regression equations stability -
Cross validation
Calibration data set Verification Year
1971 1971
1972 1972
1973 1973
1974 1974
1975 1975

1997 1997
1998 1998
1999 1999
2000 2000
26
Methods for Climate Forecast Verification
Ranked Probability Score and Continuous Ranked
Probability Score
RPS is not sensitive for forecast spread.
Tobs
The Rank Probability Score (RPS), graphically
represents the performance of the probability
forecast (Wilks, 1995).
Continuous Rank Probability Score (CRPS) takes
into account spread of the forecasted
distribution.
CRPS Skill Score (CRPSS) is the final measure of
the forecast performance.
27
Progress Report - CRPSS
28
Methods for Climate Forecast Verification
Reliability Diagrams
Reliability Diagrams exhibit the correspondence
between the observed and forecasted percentiles.
Reliability Diagrams allow verification of each
POE for each station. The analysis is done for a
forecast and compared with climatology.
Under-forecasting
Over-forecasting
29
Progress Report Reliability Diagrams
30
Methods for Climate Forecast Verification
Bias Analysis
The bias is computed for CPC forecasts and for
climatology using the equation shown on the right
for climate divisions and stations for each
forecast month and each lead season. The bias
shows range and sign of deviations between
forecasts and observations. Mean Square Error
(MSE) is other common accuracy measure of climate
forecast leading to skill score (SS) estimates.
Expected PDF of difference between forecasts and
observation is normal distribution with mean that
is not significantly different from 0.
31
Utilized CPC Methods of Downscaling - POE
When this method fails
32
Utilized CPC Methods of Downscaling - POE
When this method fails
?
Station Forecast Spread
R Measure of Confidence in Downscaling
33
Utilized CPC Methods of Downscaling - POE
When this method fails
Temperature is a normally distributed variable,
therefore the downscaling method based on
regression can provide good estimates
Precipitation (right chart) is too skewed for
normal distribution. The regression would
require a transformation of this variable.
Compositing can be used for Precipitation
forecasts because it does not employ regression
analysis.
Mean 0.30 St. Dev. 0.38 Median 0.19 Mode
0.01 Skewness 3.11 Kurtosis 14.67
NOT a good fit
34
Utilized CPC Methods of Downscaling COMPOSITES

• Levels of sophistication in use of composites
• Composite means
• Raw Composite distribution
• Smooth resampled Composite Distribution - boot
  strapping techniques
• All of the above with trend and some other mode
  of climate variability taken into account using
  new approach developed by Higgins, Unger and Kim

35
Utilized CPC Methods of Downscaling COMPOSITES
36
Utilized CPC Methods of Downscaling COMPOSITES
level 1
37
Utilized CPC Methods of Downscaling COMPOSITES
level 2
38
Utilized CPC Methods of Downscaling COMPOSITES
level 2
B N A
Jan 1 3 5
Feb 1 4 4
Mar 5 3 1
JFM 0 7 2
FORECAST Given El Nino, Denver Tmean has a shift
in Tmean toward above normal for Jan, below
normal for Mar, and near normal for JFM
39
Utilized CPC Methods of Downscaling COMPOSITES
level 2
JFM composites
Nino3.4 Term Warm Neutral Cold
Above 2/90.22 9/320.28 2/90.22
Near 7/90.78 12/320.38 4/90.45
Below 0/90 11/320.34 3/90.33
For each Nino 3.4. event we compute probability
of the climate variable to be in Below/Near/Above
normal category.
40
Utilized CPC Methods of Downscaling COMPOSITES
level 2
CPC CURRENT ENSO FORECAST http//www.cpc.ncep.noa
a.gov/products/precip/CWlink/ENSO/total.html
Nino3.4 Term Warm Neutral Cold
Above 2/90.22 9/320.28 2/90.22
Near 7/90.78 12/320.38 4/90.45
Below 0/90 11/320.34 3/90.33
NINO 3.4 INITIAL TIME 3 2003 PROJECTION
FRACTION BELOW NORMAL ABOVE AMJ 1 0.039
0.372 0.589 MJJ 2 0.066 0.446 0.488 JJA 3
0.126 0.494 0.380 JFM 9 0.343 0.434 0.243 FM
A 10 0.318 0.438 0.244
FORECAST USING CURRENT CPC Nino 3.4
Example ElNino with 8.5 month lead (forecast
for JFM 2004)
41
Utilized CPC Methods of Downscaling COMPOSITES
level 3
To obtain a smooth distribution we can use
resampling, which allows 93 729 different
combinations of the months in each season. To
better represent extremes in the distribution we
sample with replacement using boot strapping
technique.
El Nino Years January February March
1958
1966
1969
1973
1983
1987
1992
1995
1998
X
X
X
X
X
X
X
X
X
X
X
X
42
Utilized CPC Methods of Downscaling COMPOSITES
level 4
DJF North Dakota Temp (C)
The trend should not be removed arbitrarily.
Hinge, for example, explains climate changes
during the last decades. 11-year Moving Average
(MA) explains decadal variations in climate.
43
Research Products Versus Operational Products
Method Research Output Operational Products
POE Regression equation coefficients Method performance evaluation based on hind cast applications (1995-2003) Seasonal forecasts issued monthly for 13 seasonal leads using equations and CPC CD forecasts Verification issued monthly for forecast made in preceding month
Composites All composite statistics mean, variance, each category probabilities with and without trend, and POE. Method performance evaluation based on hind cast applications (1982 2003) Seasonal forecasts issued monthly for 16 seasonal leads using station statistics and CPC Nino 3.4. forecasts Verification issued monthly for forecast made in preceding month
44
WR NEW LOCAL CLIMATE PRODUCTS Where do we go
EXPECTED OPERATIONAL FUNCTIONS REGIONAL CSPM 1.
POE downscaling 1.1. Developing the
translation equations for 87 sites (completed)
1.2. Tests of the equations (2003-2004)
1.3. Monthly coordination of local product
release (starting 2005) 1.4. Annual update
of the equations (starting 2005) 2. Composites
2.1. Coordination of Composites products
existing or released in the local offices
2.2. Developing Composites for 87 sites
(2003-2004) 2.3. Hind cast test of the
composites (2003-2004) 2.4. Monthly
coordination of local product release (starting
2005) 2.5. Annual update of the composites
(starting 2005)
45
WR NEW LOCAL CLIMATE PRODUCTS Where do we go
EXPECTED OPERATIONAL FUNCTIONS WFO CSFP 1. POE
downscaling 1.1. Selection of sites within
WFO CWA for downscaling (completed) 1.2.
Delivering on monthly basis seasonal outlooks for
selected sites within WFO CWA (starting in
FY05) 1.3. Verification of the previous
month forecasts (starting in FY05) 2.
Composites 2.1. Keeping self updated on
climate variability mode status (starting in
October 2003) 2.2. Developing composites
local studies (optional) 2.3. Release of
monthly and seasonal outlooks for selected
sites within each WFO CWA (starting in 2005)
2.4. Verification of the Composites forecasts
(starting in 2005) 3. Public outreach on these
new products
46
MESSAGES TO TAKE HOME

• THERE ARE NO NWS CONSISTENT LOCAL CLIMATE
  PRODUCTS AVAILABLE NOW
• DOWSCALING CAN BE USED AS A TOOL FOR LOCAL
  CLIMATE PRODUCTS
• THE LOCAL CLIMATE PRODUCT SHOULD BE CONSISTENT
  WITH THE NATIONAL WEATHER SERVICE PRODUCTS
  (CPC)
• CPC METHODS COULD BE USED IN DEVELOPING SUCH
  LOCAL CLIMATE PRODUCTS
• IN DEVELOPING OF DOWNSCALING AT LEAST THREE NWS
  ENTITIES SHOULD BE INVOLVED REGIONAL OFFICE,
  CSD, AND CPC