Temperature Prediction
ATMS 452
Temperature Measurements
• 2-m
• In shade
• Over a natural vegetated surface
Stevenson Screen
AKA: Cotton Region Shelter
The Traditional Approach
ASOS Temperature/Humidity
Sensor
Standard NWS/FAA instrumentation now
Why Not Simply Use Model
Output Directly and Go Home?
• Model surface elevation may be very
different than actual surface elevation due
to limited model resolution.
– Differences can be in the hundreds of meters or
more.
– Example UW 36 km MM5 lowest level over
Boeing field is 256 m. Actual 5 m.
– Improves as resolution increases
Why Not Simply Use Model
Output Directly and Go Home?
• Model resolution is often inadequate for
properly simulating important features that
influence temperature.
– Examples include the Columbia River Gorge
and drainage flows in valleys
– Coastal temperature contrasts
MM5 12-km
MM5 4km
Near Sea Level Gap
On Border of WA and OR
Columbia Gorge
Flow Simulation
36 km grid spacing
12 km grid spacing
Portland
The Dalles
Portland
The Dalles
Pass Height = 700 m
Pass Height = 600 m
4 km grid spacing
Portland
Cascade
Locks
The
Dalles
Pass Height = 400 m
12 km grid spacing
Portland
The Dalles
Pass Height = 600 m
1.33 km grid spacing, Pass Height = 150 m
•
Portland
Cascade Locks
Troutdale
The Dalles
444.4 m grid spacing, Pass Height = 100 m
Portland
Troutdale
Cascade Locks
• Model temperatures may be seriously in error due
to poor model physical parameterizations, such as
for the planetary boundary layer, radiation, surface
energy fluxes, cloud and precipitation processes.
– Example: overmixing in PBL can result in the inability
to maintain shallow cold air masses. Result: warm bias
– Example 2: improper soil moisture (too wet or dry) can
greatly influence temperatures. (too wet produces too
cool temps, vice versa)
– Example 3: poor snow distribution (very frequent)
• As a result of such errors, models can have serious
systemic temperature biases.
00 UTC
12 UTC
Temperature Biases
• Large temperature biases can occur
– When there is a shallow layer of cold air (few
hundred m deep) that is mixed out by model
incorrectly.
– During transition seasons (particularly spring,
when land surface conditions are problematic)
– During summer during warm periods.
Example:
Shallow Fog…Nov 19, 2005
• Held in at low levels for days
• MM5 held in the deep inversion…generally
without the shallow mixed layer of cold air
a few hundred m deep
• MM5 could not maintain the moisture at
low levels
Why you can’t simply take the
model temps
• Models can also have synoptic and timing
errors that can greatly undermine
temperature forecasts.
Downslope Warming
• Large warming during downslope flow
• Often large over Cascade foothills (North Bend),
but apparent all over the world, including to the
lee (east) of the Rockies--the Chinook Wind.
• In Europe called the Foehn Wind.
• Uusually, air comes from mid-levels where
potential temperature is higher than at the surface.
A drop in dewpoint usually accompanies it.
Adiabatic Cooling with Upslope
Flow
• Can be harder to see because:
– Clouds and precipitation obscure the signal
– Cooling leads to saturation, resulting in smaller
cooling rate with lifting (saturated adiabatic)
Diabatic Effects
Diabatics
• Radiation:
– direct radiational heating and cooling of the air is
relatively small
– indirectly, very large influence on air temp
through modulation of ground temperature and
communication into the boundary layer by
turbulence. Surface heating by the solar flux and
nighttime cooling by IR flux dominate surface
temperatures.
– Clouds have a major effect on radiational
heating/cooling.
Clouds, Radiation and Surface
Temps
• Clouds lessen warming during day by
reducing solar radiation
• Clouds decrease cooling at night by
intercepting IR radiation and reradiating IR
back to the surface.
• Thin cirrus…only minor influence (1-2F)
• Thick altostratus--5-10F influence or more.
• Bottom line: clouds reduce diurnal variation
One can see association between
changes in clouds and
temperature at short-time scales
IR Opacity of Air Can Influence
Surface Temps
• IR opacity is proportional to humidity.
• Increased water vapor content increases
opacity--better absorption and emission in
IR.
• Less water vapor results in better IR cooling
under clear conditions. That is why deserts
cool very rapidly at night. Why
Washington DC stays hot all night.
• Reduces diurnal temp. range
Conduction and Turbulence
Effects in the BL.
Fans in orchards and vineyards
Phase Changes of Water
• Can have a large impact on temperature and
type of precipitation.
Evaporation
SE Everett
• Snohomish, WA
• April 4, 2005
IMPACTS OF LOCAL
DIURNAL CIRCULATIONS ON
TEMPERATURE
Hoquiam, WA: July 15, 2002
18
70
Wind Speed (kts)
Temperature (ЎF)
16
65
12
60
10
55
8
6
50
4
45
2
0
40
1
2 3
4 5 6
7
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Hour (LDT)
Temperature (ЎF)
Wind Speed (knots)
14
Comparison of the daily temperature cycle between
observations (blue dashed) and the model (red triangles) for a
coastal location (Boston, MA) on 10 August 1997 impacted by
the passage of a sea breeze front.
2-m air temperature and 10-m wind
velocity in the afternoon of a sea breeze
event on April 18, 2005. The land-sea
temperature contrast drives the onshore
sea breeze. Local spatial variations in
sea surface temperature (SST) can
leave a signature on the sea breeze,
especially when complex coastlines
create multiple sea breeze fronts (e.g.,
the sea breeze penetrating from Long
Island Sound in the image above).
In our multiply-nested COAMPS
simulations, nest 4 (1.33 km resolution)
and nest 5 (0.44 km resolution) were
enhanced to account for urban effects
and to include time-varying (hourly)
high-resolution SST's from Alan
Blumberg's New York Harbor Observing
and Prediction system (NYHOPS). The
diurnal heating of the ocean surface is
thus represented in our simulations.
We are in the process of quantifying the
impact of these effects on the sea
breeze circulation for several distinct
events, and evaluating the forecast skill
of the system.
Near Sea Level Gap
On Border of WA and OR
Cold Pools
Cold air can collected in mesoscale
basins…such as eastern Washington,
Colorado.
Near Sea Level Gap
On Border of WA and OR
Portland wind distribution for all
days by direction and speed
• Wintry precipitation at PDX
– Almost exclusively E or SE
– Tendency towards E for snow and
SE for freezing rain
Wind
distribution
for days with
snowfall
Wind
distribution
for days with
Freezing
Rain
December 11-15, 2000 Case Study
• Strong winds for nearly four days
– Gale force a times
• Wintry mix in Western Gorge
and exit area.
– Little snow accumulation but
significant icing.
• Selected because:
– Data availability was good (esp.
ACARS)
– Model Initialization ok
Domain
Definition
36 km grid spacing
12 km grid spacing
Portland
The Dalles
Portland
The Dalles
Pass Height = 700 m
Pass Height = 600 m
4 km grid spacing
Portland
Cascade
Locks
The
Dalles
Pass Height = 400 m
12 km grid spacing
Portland
The Dalles
Pass Height = 600 m
1.33 km grid spacing, Pass Height = 150 m
•
Portland
Cascade Locks
Troutdale
The Dalles
444.4 m grid spacing, Pass Height = 100 m
Portland
Troutdale
Cascade Locks
T on
150 m
Surface
Portland
Troutdale
Southerly Buster…SE Australia…just like the push
MOS Compared to the NWS: Temperature
Cool Season Mi. Temp – 12 UTC Cycle
Average Over 80 US stations
7
6
5
4
3
2
'66-67 '71-72 '76-77 '81-82 '86-87 '91-92 '96-97 '01-02
Cool Season
24-h GUID.
24-h LCL
48-h GUID.
48-h LCL
Temperature
MAE (F) for the seven forecast types for all stations,
all time periods, 1 August 2003 – 1 August 2004.
Large one-day temp changes
MAE for each forecast type during periods of large temperature
change (10F over 24-hr), 1 August 2003 – 1 August 2004.
Includes data for all stations.
MAE for each forecast type during periods of large
departure (20F) from daily climatological values,
1 August 2003 – 1 August 2004.
Number of days
each forecast is the
most accurate, all
stations.
In (a), tie situations
are counted only
when the most
accurate
temperatures are
exactly equivalent.
In (b), tie situations
are cases when the
most accurate
temperatures are
within 2F of each
other.
Looser Tie Definition
Number of days
each forecast is
the least
accurate, all
stations.
In (a), tie situations are
counted only when the
least accurate
temperatures are exactly
equivalent. In (b), tie
situations are cases when
the least accurate
temperatures are within
2F of each other.
Looser Tie Definition
Highly correlated time series
Time series of MAE of MAX-T for period one for all stations, 1 August
2003 – 1 August 2004. The mean temperature over all stations is
shown with a dotted line. 3-day smoothing is performed on the data.
Cold spell
Time series of bias in MAX-T for period one for all stations, 1
August 2003 – 1 August 2004. Mean temperature over all
stations is shown with a dotted line. 3-day smoothing is
performed on the data.
MAE for all stations, 1 August 2003 – 1 August
2004, sorted by geographic region.
MOS Seems to have the most problems at high elevation stations.
Bias for all stations, 1 August 2003 – 1 August 2004,
sorted by geographic region.
Precipitation Comparisons
Brier Scores for Precipitation for all stations for the entire
study period.
Brier Score for all stations, 1 August 2003 – 1 August
2004. 3-day smoothing is performed on the data.
Prob. Of Precip.– Cool Season
(0000/1200 UTC Cycles Combined)
0.7
Brier Score Improvement over Climate
0.6
Guid POPS 24 hr
Local POPS 24 hr
Guid POPS 48 hr
Local POPS 48 hr
0.5
0.4
0.3
0.2
0.1
0
1966
1969
1972
1975
1978
1981
1984
Year
1987
1990
1993
1996
1999
2002