the new england cold-air damming experiment (cadex)stmiller/stmiller... · 3/27/2014 · the new...

The New England Cold-Air Damming Experiment (CADEX) 1

THE NEW ENGLAND COLD-AIR DAMMING EXPERIMENT (CADEX)

Sam Miller

Eric Hoffman

Eric Kelsey

Jason Cordeira

Judd Gregg Meteorology Institute

Department of Atmospheric Science and Chemistry

Plymouth State University

27 March 2014

ABSTRACT

Cold air damming (CAD) has been extensively studied in the U.S. Midatlantic region,

east of the Appalachian Mountains, but not in the northern New England states of New

Hampshire and Maine. CAD can develop through synoptically-forced cold-air

advection, smaller-scale evaporative processes, or some combination of both. Regardless

of the formation process, CAD can create hazardous conditions, and presents problems

in practical forecasting, including prediction of daily temperature ranges, precipitation

type (especially in winter), surface wind direction and speed, and aviation hazards such

as wind shear, turbulence, and icy runway conditions. This paper includes the

formulation of a new CAD index, a preliminary climatology based on a two-year dataset,

and two case studies of significant CAD events in northern New England that occurred

over the 2013-2014 winter season. We conclude with a proposal for a staggered five-year

research project, taken in stages, involving Plymouth State University faculty and

students, and several elements of the U.S. National Weather Service, that will (1) create a

thorough climatology of CAD events in New England; (2) conduct case studies of

selected events utilizing both observational (operationally-available and special field)

data and WRF model runs; (3) evaluate the relative effectiveness of NCEP numerical

weather models at predicting the development, evolution, consequences, and decay of

CAD events; and (4) work with NWS offices to develop new numerical tools for

forecasting CAD in an operational setting.

1. LITERATURE REVIEW.

An excellent summary of Cold Air Damming (CAD) can be found in Chapter 8 of Lackmann

(2012). A number of previous studies have examined the synoptic and mesoscale evolution of cold-air

damming. Many of these studies have investigated CAD east of the Appalachians and particular

attention has been given to this phenomenon in the mid-Atlantic region of the U.S. The seminal paper on

this subject, Bell and Bosart (1988), developed the first climatological study of CAD in this region. Bell


and Bosart (1988) showed that CAD develops when cold stable air is located to the east of a north-south

mountain barrier. If the geostrophic flow is easterly but the cold stable air is unable to flow over the

barrier, it becomes blocked. High pressure develops east of the barrier hydrostatically. Simultaneously,

warm air above the low-level inversion is able to flow over the mountain and down slope on the western

side. This results in lee troughing to the west of the barrier. The combination of these two features results

in a characteristic trough/ridge couplet in the sea-level pressure field across the mountain barrier.

Another feature of a CAD event identified by Bell and Bosart (1988) is the development of a northerly

low-level jet to the east of the barrier. This jet develops in the immediate vicinity of the mountain where

the low-level flow is blocked, and is caused by a balance between the pressure gradient force directed

from north-to-south, and friction. Near the surface, friction keeps wind speeds relatively low, but just

above the surface, and below the inversion layer, a low-level wind maximum on the order of 15 m s-1

develops (Bell and Bosart, 1988).

Bailey et al. (2003) developed a 10-year climatology and synoptic classification scheme for CAD

events in the mid-Atlantic region. In this study the authors described a spectrum of CAD events divided

into three classifications: classic, hybrid, and in situ. Classic CAD events are strongly forced by synoptic

scale flows and feature a strong parent anticyclone (central sea-level pressure typically > 1030 hPa)

centered to the northeast of the CAD region. This anticyclone provides the source of the low-level cold air

that initiates the CAD, and the role of diabatic processes in producing the low-level cold air is small

relative to the synoptic scale cold air advection. On the other end of the spectrum, in situ events develop

when diabatic cooling processes are dominant in developing the low-level cold air to the east of the

mountains. In these cases a relatively dry airmass exists east of the mountains with relatively weak high

pressure to east or northeast. As a synoptic scale low-pressure system approaches from the west,

isentropic lift in the region of warm-air advection ahead of the cyclone allows precipitation to begin east

of the mountains. Evaporation and/or melting cool the low-level cold air in situ. Sea-level pressures rise

or remain steady east of the mountains, producing the characteristic trough/ridge couplet. The last

category according to Bailey et al. (2003) are hybrid cases in which both synoptic scale low-level cold air

advection and diabatic processes contribute to the development of cold air and high pressure east of the

mountains.

Bailey et al. (2003) found that during the winter months the monthly frequency was 2-3 CAD

events with classic and hybrid CAD events accounting for 1-2 events. CAD was found to occur during

the warm season and most of those events were found to in-situ events.

Once the cold air damming has been established, a significant forecasting issue involves

understanding and predicting when the cold air near the surface will erode. Lackmann (2012) suggests

several possible mechanisms for CAD erosion including: cold-air advection aloft, solar heating, near-

surface divergence, shear-induced mixing, and frontal advance. It appears that no studies have tried to

understand which of these erosion mechanisms is most frequent, and if there is any relationship between

the CAD event type (classic, hybrid, in situ) and the erosion mechanism. In addition, most of the studies

of CAD have focused on the mid-Atlantic region, east of the highest ridges of the Appalachian mountain

range. Few studies have addressed CAD in New England specifically. Therefore among the goals of this

research will be the identification and classification of CAD events in New England (specifically New

Hampshire), observing the small-scale structures associated with CAD in the complex terrain, and

observing, identifying, and classifying the CAD erosion mechanisms.

Cold air erosion processes. Cold air erosion is one of the biggest forecasting challenges of CAD events due

to the mesoscale and diabatic processes often involved. Relatively small errors in timing of CAD erosion


(e.g., 1-6 hours) can result in huge social and economic disruptions resulting from increased

accumulations of snow, sleet and/or freezing rain.

Dammed cold air may erode via one or more processes (Lackman 2012):

1. Sufficiently strong wind shear in combination with a sufficiently weak inversion (i.e., small bulk

Richardson number) can generate turbulence and vertical mixing down through the inversion.

Increased synoptic scale pressure gradients and elevated convection are processes that can

increase wind shear and turbulence through the inversion.

2. Cold air advection above the inversion, usually associated with the passage of the cold front, acts

to weaken the inversion (i.e., decrease the bulk Richardson number) and create a near dry-

adiabatic vertical temperature profile that allows strong vertical mixing down to the surface.

Cold air advection aloft also promotes subsidence and drying of the atmosphere, which allows

insolation to enhance the cold air erosion.

3. Insolation can promote surface-up warming and vertical mixing where thick cloud cover does

not exist. New England CAD events, however, are typically associated with a thick stratus layer

at the height of the inversion.

4. When the surface anticyclone moves far enough to the east and north/northeast, low-level flow

ceases to reinforce the cold air, and divergence within the cold air dome may occur. This

divergence thins the depth of the cold air, which leads to subsidence, drying, and during the

daytime, increased insolation.

5. Lastly, if the cold air is saturated, precipitation formed above the inversion (where the air is

warmer) will transport its heat down into the cold air. This warming is in contrast to the

evaporative cooling and CAD reinforcement that occurs when the cold air is not saturated.

Once the dammed cold air has eroded enough vertically to expose the higher mountain peaks to

warmer air aloft, the peaks will perturb the flow of warm air and enhance vertical erosion of the cold

dome via orographically-induced turbulence. The characteristic and intensity of turbulence generated is

described by the Froude number (Stull 1988):

𝐹𝑟 = 𝜋 �̅�

𝑁𝐵𝑉𝑊𝑇

(1)

where �̅� is the magnitude of the mean wind speed of the warm air [ms-1], 𝑵𝑩𝑽 is the stability and vertical

wind shear within the warm air (Brunt–Väisälä frequency; a unitless quantity), and 𝑾𝑻 is the length of

the mountain peak exposed to the warm air [m]. Squaring �̅� gives the kinetic energy per unit mass

available for conversion to turbulence kinetic energy and cold air erosion. Overall, Fr is a unitless

quantity. The Froude number is typically used to characterize the motion of air as it flows over an isolated

hill or mountain surrounded by a flat Earth surface. Here it is applied in the situation where the top of

the cold air dome is a fixed horizontal surface (encompassing the lower portion of a hill or mountain)


unless exposed to the downward momentum flux of turbulent eddies. Fr ~ 0.1 results in air flowing

predominantly around the mountain top with little if any generated turbulence. At Fr ~ 1.0, the effective

wavelength of the exposed mountain top (2𝑊𝑇) is equal to the natural wavelength of the warm air

flowing over the mountain and mountain waves will develop downstream of the mountain. As Fr

increases beyond 1.0, the wavelength of the warm air is greater than the effective wavelength of the

mountain top and increasing turbulence is generated downstream of the mountain.

2. FORMULATION AND CLIMATOLOGY OF THE COLD-AIR DAMMING INDEX (CADINX).

In the context of this study, we define a cold-air damming (CAD) event in central New

Hampshire as one that produces colder (denser) air near the geographic center of the state, with warmer

(lighter) air to the north, west and south (see Fig. 1). Virtual potential temperature (v) was used to

evaluate variations in the density of the low-level airmass across the region. Time series of v were

computed from base hourly observations for the stations shown in Table 1 and Fig. 1, and then used

(according to the schema outlined in Fig. 2) to compute a Cold-Air Damming Index (CADINX) that

identifies periods and intensities of CAD in central New Hampshire. The details of the CADINX

calculations are discussed below.

Table 1: Stations used to compute CADINX.

Station ICAO Latitude [°N] Longitude [°E] Elevation [m ASL]

Glens Falls, NY KGFL 43.34 -73.61 100

Plymouth, NH K1P1 43.78 -71.75 157

Providence, RI KPVD 41.72 -71.43 16

Sherbrooke, QC CYSC 45.40 -71.83 241

Fig. 1: Map of stations used to compute CADINX. (Image courtesy Google Earth, 2014.)

100 KM

NY

VT

QC

NH

ME

MA

CT RI


Fig. 2: Schematic of CADINX stations showing relevant distances.

Quality-controlled time series created from hourly surface weather observations were used to

compute hourly values of virtual temperature (Tv) for each station by:

𝑇𝑣 = 𝑇 [1 +

𝑤𝜀

1 + 𝑤] (2)

where Tv has units of Kelvins, T is the reported in situ air temperature [K], w is the water vapor mixing

ratio [kgvapor/kgair], computed from the reported dew point, and is the ratio of the molar masses of

vapor and dry air (0.622) (Rogers and Yau, 1989). Virtual potential temperature (v) was then computed

using Poisson’s Equation, i.e.:

Ѳ𝑣 = 𝑇𝑣 (1000

𝑆𝐿𝑃 −𝑆𝑇𝑁𝐸𝐿

10

)

0.2857

(3)

where SLP is the station’s reported sea-level pressure [hPa], and STNEL is the station elevation [m ASL].

Dividing the station elevation by 10 provides an estimate of the difference between sea-level pressure and

station pressure, which is necessary because the latter is not routinely reported in standard surface

weather observations (Rogers and Yau, 1989; NASA, 1966).

Sherbrooke, QC (CYSC) data were used to represent conditions in the northern part of the study

area; Plymouth, NH (K1P1) in the center of the study area; Glens Falls, NY (KGFL) in the western part of

the study area; and, Providence, RI (KPVD) in the southern part of the study area. If CYSC was cooler

than K1P1 ( Ѳ𝑣𝐶𝑌𝑆𝐶 < Ѳ𝑣

𝐾1𝑃1), K1P1 was warmer than KPVD ( Ѳ𝑣𝐾1𝑃1 > Ѳ𝑣

𝐾𝑃𝑉𝐷), or K1P1 was warmer than

KGFL ( Ѳ𝑣𝐾1𝑃1 > Ѳ𝑣

𝐾𝐺𝐹𝐿), the conditions meeting our definition of CAD were not met, and the CADINX

CYSC

KPVD

KGFL

Y1 = 180.3 km

Y2 = 230.1 km

X = 157.6 km K1P1


was set to zero. Otherwise, finite differences approximating the north-south and east-west v gradients

were computed by:

𝐴 = Ѳ𝑣

𝐶𝑌𝑆𝐶 − Ѳ𝑣𝐾1𝑃1

∆Y1

(4.1)

𝐵 = Ѳ𝑣

𝐾1𝑃1 − Ѳ𝑣𝐾𝑃𝑉𝐷

∆Y2

(4.2)

𝐶 = Ѳ𝑣

𝐾1𝑃1 − Ѳ𝑣𝐾𝐺𝐹𝐿

∆X (4.3)

The index was then computed by:

𝐶𝐴𝐷𝐼𝑁𝑋 = 𝑚𝑒𝑎𝑛([|𝐴|, |𝐵|, |𝐶|])𝑥 100 (5)

which has units of °C (100 km)-1. The vertical bars around each term indicate that the absolute value of

each finite difference was used.

To avoid lengthy gaps in available data, and adjust for the relatively recent installation of the

AWOS at K1P1 (in 2005), a two-year hourly time series (17,544 hours) of the CADINX was computed.

The time series begins on 01 April 2006, at 0000 UTC, and ends on 31 March 2008, and 2300 UTC (Fig. 3).

Bulk statistics for the entire two-year period of CADINX intensity are shown in Table 2. The mode

indicates that the most common value of intensity is 0 °C (100 km)-1, meaning, most of the time, cold-air

damming is not occurring.

Fig. 3: Time series of CADINX. Datapoints correspond to hours since 01 April 2006 at 0000 UTC.


Table 2: Bulk statistics of CADINX intensity for full two-year period. All values shown have units of °C (100

km)-1

.

Max 9.54

Min 0

Mean 0.43

Median 0

Mode 0

Standard Deviation 0.95

Contoured charts of in situ surface temperature were spot checked against corresponding

CADINX values. For example, the extreme CADINX value of 9.54 0 °C (100 km)-1 (visible at the right end

of the time series plot shown in Fig. 3) occurred on 03 March 2008, at 0800 UTC. The surface isotherms

corresponding to that time are shown in Fig. 4. A dome of cold, dense air was centered over central New

Hampshire, with warmer air to the north, west, and south. These results indicate that the CADINX

correctly identifies those conditions we have defined as cold-air damming in central New Hampshire.

Fig. 4: Surface (2 m) isotherms for 04 March 2008, 0800 UTC. Contours are drawn at 1 °C intervals. The

temperature field shown corresponds to a CADINX value of 9.54 °C (100 km)-1

.

Table 3 shows the relative frequency of different hourly CADINX intensities, grouped by

intervals of 1 °C (100 km)-1. Hourly intensities of zero were most frequent during the two-year period,

occurring for 14,296 hours, or 81.5 percent of the time. The most frequent non-zero intensity is between 1

and 2, which occurred for 1730 hours, or 9.9 percent of the total time period of 17,544 hours. Using these

results, CADINX intensities were grouped into three classes: Weak (> 0 to ≤ 3 °C (100 km)-1), Moderate (>

3 to ≤ 6 °C (100 km)-1), and strong (> 6 °C (100 km)-1). Table 4 shows the result of this grouping, and

indicates that weak cold-air damming occurred for 3,339 hours, or 19.0 percent of the total time; moderate

for 531 hours, or 3.0 percent of the time; and strong for 20 hours, or 0.1 percent of the time.


Table 3: Number of hours with CADINX values in different intensity ranges, in intervals of 1 [°C (100 km)-1

.

Percentages were computed using the number of hours shown, and a total time series length of 17,544 hours.

Intensity Range

[°C (100 km)-1

]

No. of Hours Percentage of total hours

0 14296 81.5

> 0 to ≤ 1 642 3.7

> 1 to ≤ 2 1730 9.9

> 2 to ≤ 3 967 5.5

> 3 to ≤ 4 366 2.1

> 4 to ≤ 5 131 0.7

> 5 to ≤ 6 34 0.2

> 6 to ≤ 7 12 0.07

> 7 to ≤ 8 3 0.02

> 8 to ≤ 9 2 0.01

> 9 3 0.02

Table 4: Number of hours with CADINX values in different intensity classes of weak, moderate, and strong.

Percentages were computed using the number of hours shown, and a total time series length of 17,544 hours.

Intensity Class Intensity Range

[°C (100 km)-1

]

No. of Hours Percentage of total hours

Weak > 0 to ≤ 3 3339 19.0

Moderate > 3 to ≤ 6 531 3.0

Strong > 6 20 0.1

Monthly bulk statistics of CADINX intensity were computed, and are shown in Table 5. Similar

statistics by meteorological season are shown in Table 6, and by hour of the day (UTC) in Table 7. A

corresponding set of monthly statistics describing the duration of events are shown in Table 8. Events are

defined as a contiguous run of at least one hour when the CADINX had intensities of greater than zero.

Combined, these results indicate that the strongest cold-air damming intensities occurred during the

meteorological winter and spring, with maximum intensities exceeding 6.00 °C (100 km)-1 December

through March. The most extreme case occurred in March. They also show that cold-air damming can

occur at any time of year (consistent with the findings reported by Bell and Bosart 1988), although

CADINX intensities remain moderate or weaker during the meteorological summer and autumn. While

cold-air damming occurred at all hours of the day, strong cold-air damming did not occur from afternoon

through evening. Maximum CADINX intensities of more than 6.00 occurred between 0400 UTC (2300

LST) through 1800 UTC (1300 LST). Finally, while the greater number of contiguous events occurred in

July, these were short-lived, with a maximum duration of 17 hours. Events of 48 hours or more occurred

in November, December, January and February.


Table 5: Bulk statistics of hourly CADINX intensity by month. All values shown have units of °C (100 km)-1

. In

all cases, the minimum, median, and modal values were zero.

Month Max Mean

Jan 6.07 0.55

Feb 6.55 0.37

Mar 9.54 0.39

Apr 5.37 0.37

May 4.50 0.51

Jun 4.35 0.41

Jul 4.37 0.41

Aug 4.34 0.40

Sep 4.32 0.52

Oct 5.15 0.37

Nov 5.05 0.43

Dec 6.91 0.43

Table 6: Bulk statistics of hourly CADINX intensity by meteorological season. All values shown have units of

°C (100 km)-1

. In all cases, the minimum, median, and modal values were zero.

Month Max Mean

Winter (DJF) 6.91 0.45

Spring (MAM) 9.54 0.42

Summer (JJA) 4.37 0.41

Autumn (SON) 5.15 0.44

Table 7: Bulk statistics of hourly CADINX intensity by UTC hour. All values shown have units of °C (100 km)-1

.

In all cases, the minimum, median, and modal values were zero. Correction to local standard time is UTC-5.

UTC Hour Max Mean

00 4.96 0.34

02 5.22 0.39

04 7.79 0.45

06 8.26 0.51

08 9.54 0.54

10 8.34 0.57

12 6.30 0.71

14 6.07 0.54

16 6.43 0.33

18 6.55 0.27

20 4.15 0.24

22 4.79 0.31


Table 8: Bulk statistics of hourly cold-air damming duration by month. Events are defined as a contiguous run

of at least one hour when the CADINX intensities of greater than zero.

Month Number of Events

Max Duration [hours]

Min Duration [hours]

Mean Duration [hours]

Median Duration [hours]

Modal Duration [hours]

Standard Deviation of Duration [hours]

J 42 56.0 1.0 7.5 5.0 1.0 9.5

F 30 37.0 1.0 6.0 4.5 1.0 7.3

M 27 27.0 1.0 6.9 3.0 2.0 7.5

A 65 35.0 1.0 4.9 2.0 1.0 6.4

M 92 34.0 1.0 4.5 3.0 1.0 4.9

J 93 33.0 1.0 3.9 2.0 1.0 4.6

J 117 17.0 1.0 3.3 2.0 1.0 3.4

A 90 35.0 1.0 4.1 2.0 1.0 5.1

S 93 17.0 1.0 4.4 3.0 1.0 4.2

O 80 15.0 1.0 3.7 2.0 1.0 3.7

N 54 48.0 1.0 5.7 3.0 1.0 8.5

D 39 56.0 1.0 7.2 4.0 1.0 10.0

Bulk statistics of CADINX intensity were also parsed by the surface (10 m) wind direction

reported at K1P1, and the 700 hPa wind direction reported by the Portland, ME (KPWM) radiosonde

stations. This procedure was suggested by Bell and Bosart (1988). The comparison to the 700 hPa wind

required reducing the available hourly values of CADINX intensity to those computed for the synoptic

hours 0000 and 1200 UTC. Because of a large gap in available data for KPWM, it also required reducing

the period of the comparison, with the new time interval beginning on 01 April 2006 at 0000 UTC, and

ending on 18 September 2007, at 0000 UTC.

The bulk statistics of the comparison to the hourly surface winds at K1P1 are shown in Table 9.

This result indicates that the strongest maximum and mean CADINX intensities occurred with calm

surface winds (associated with a direction of “0”), and the second highest maximum and mean values

were associated with a surface wind from the east-northeast. The comparison to the 700 hPa wind was

done by means of cross-correlation, with the wind taken as the forcing, in ten-compass degree

increments, and the CADINX taken as the response. Fig. 5 shows the result. While the maximum

positive correlation was as high as about 15 percent (with 95 percent confidence at about 1.5 percent), the

correlation indicated a relatively weak relationship. In general, a northerly 700 hPa wind was associated

with a maximum positive correlation, at a time lag of about 36 hours. A southerly 700 hPa wind was also

associated with maximum correlation, but with a negative associated lag, meaning the increase in

CADINX preceded the southerly wind by about 60 hours.


Table 9: Bulk statistics of hourly CADINX intensity by surface wind sector. All values shown have units of °C

(100 km)-1

. In all cases, the minimum, median, and modal values were zero.

Wind Sector Max Mean

Calm 9.54 0.76

> 0 to ≤ 30 2.47 0.18

> 30 to ≤ 60 3.15 0.39

> 60 to ≤ 90 6.29 0.67

> 90 to ≤ 120 4.66 0.44

> 120 to ≤ 150 5.15 0.50

> 150 to ≤ 180 5.04 0.28

> 180 to ≤ 210 2.83 0.08

> 210 to ≤ 240 3.24 0.09

> 240 to ≤ 370 3.31 0.04

> 270 to ≤ 300 3.36 0.04

> 300 to ≤ 330 3.76 0.05

> 330 to ≤ 360 2.37 0.12

Fig. 5: Cross-correlation between 700 hPa wind, and CADINX. The former is the forcing, and the latter is the

response. Upper panel shows correlation coefficient with respect to meteorological-convention wind direction; lower

panel shows the associated lag (in hours).

The CADINX was combined with time series of surface (2 m) temperature recorded at Concord,

New Hampshire (KCON), and the surface (10 m) wind direction recorded at KCON, to examine the

relationship to precipitation type observed at KCON. (KCON was used instead of K1P1, because the

latter is not equipped with a freezing precipitation sensor. Concord is located approximately 65 km south

of Plymouth.) The result of this comparison is shown in Fig. 6, and indicates that, for some wind

directions, the CADINX and surface temperature at KCON was sufficient (in the mean) for distinguishing

freezing precipitation from solid and ordinary liquid precipitation.


Fig. 6: Parametric plot showing precipitation type at KCON as a function of three variables. Lines indicate

mean values of indicated parameters associated with different observed precipitation phases. Green is liquid

precipitation (rain and drizzle); red is freezing precipitation (freezing rain and freezing drizzle); blue is frozen

precipitation (snow and snow grains).

3. CASE STUDY: 05-06 DECEMBER 2013.

A preliminary field study was conducted during a CAD event that occurred on the fifth and sixth

of December, 2013. Fig. 7 shows the hourly CADINX intensities during the event and for the entire

month. The event began shortly after 05 December 2013 at 1400 UTC (0900 LST; LST = GMT - 5), taking

about 8 hours to reach a quasi-steady CADINX intensity of slightly more than 5 °C (100 km)-1, making

this a “moderate” intensity event. The peak CADINX intensity of 5.25 °C (100 km)-1 occurred on 06

December 2013 at 0600 UTC, after which the event quickly disintegrated. The CADINX dropped back to

zero by 06 December 2013 at 1000 UTC, about four hours after the event peaked. The lower panel of Fig.

7 also indicates that about 18 more CAD events occurred during the month, making this result consistent

with the results noted in Table 8 (which noted 39 total events during December of 2006 and 2007, or

between 19 and 20 events each). This event lasted a total of 20 hours, which is within one standard

deviation of the mean duration in December. The peak intensity (5 °C (100 km)-1) was less than the


extreme intensity noted in Table 5, as was a stronger event later in the month that peaked near 6 °C (100

km)-1.

Fig. 7: CADINX for cold-air damming event on 05 and 06 December 2013. Upper panel is CADINX during the

event; lower panel shows CADINX for the entire month of December 2013. Triangles on lower panel indicate detail

area in upper plot.

Synoptic analysis of the event. The synoptic scale evolution associated with the CAD event was similar to

those classified as “Hybrid” by Bailey et al. (2003). During the day on December 4th, a weak ridge of high

pressure (~1020 hPa) associated with a cold and dry continental polar (cP) air mass extended northwest

to southeast across New England. Clear skies over much of northern New England during the early

evening hours allowed temperatures to cool into the 30’s Fahrenheit. A significant low pressure center

developed over the central Plains and moved northeastward toward the Great Lakes. After midnight a

band of light to moderate precipitation associated with warm advection at 850 hPa moved across New

England, and evaporative cooling in the dry air below the cloud base likely contributed to development

of the cold air east of the White Mountains.

By 1200 UTC on 5 December, the center of the low (994 hPa) was located north of Lake Superior,

and a characteristic weak surface ridge associated with the cold air damming was already evident

extending from Downeast Maine to eastern New York (Fig. 8). The warm front associated with the

cyclone was well to the southwest of New England and the band of precipitation moved north of New

England (not shown). The surface potential temperature analysis (Fig. 9), shows the characteristic cold-

air damming thermal trough extending southeastward from central Maine to northern New Jersey.

Winds in the cold air were calm, or light from the northeast. Fig. 9 shows that southerly flow and a

thermal ridge west of the mountains extended from western Pennsylvania to northern NY and into

northern Vermont.


Fig. 8: NCEP surface analysis from 1200 UTC 5 December 2013. Sea level pressure isobars (4 hPa interval),

fronts, and surface observations (standard station model) are shown.


Fig. 9: Surface potential temperatures (°C; red and blue contours), and standard surface station model plots

for winds, sky cover, and obstructions to visibility with potential temperatures plotted (°C) in the upper right.

At 850 hPa (Fig. 10), weak warm air advection ahead of the deepening cyclone pushed a thermal

ridge northward from West Virginia across western New York and into southeastern Quebec, but the

warmest air had not moved across New England. In the upper troposphere, a negatively tilted

shortwave trough associated with a closed low at 500 hPa (Fig. 11) and the surface cyclone was located in

western Ontario. A significant jet streak in the polar jet stream (300 hPA, Fig. 12) developed in the

southwesterly flow extending from the southwestern U.S. to the Great Lakes. Over New England, the

upper-level ridge axis (Figs. 11,12) moved into eastern Canada and winds became westerly.


Fig. 10: 850 hPa, 1200 UTC 5 December 2013. NAM-NMM 00-hr forecast, 850 hPa Heights (dm; black,

interval=3), Temperatures (°C; color shaded, interval=5) and observations (station model).


Fig. 11: 500 hPa, 1200 UTC 5 December 2013. NAM-NMM 00-hr forecast, 500 hPa Heights (dm; black, interval=6),

Temperatures (°C; color shaded, interval=5) and observations (station model).


Fig. 12: 300 hPa, 1200 UTC 5 December 2013. NAM-NMM 00-hr forecast, 300 hPa Heights (dm, black,

interval=12), Temperatures (°C, color shaded, interval=5) and observations (station model).

During the day on 5 December, the surface and upper-level lows continued to move

northeastward into eastern Canada. By 1800 UTC on the 5th (Fig. 13), the surface low deepened to 986 hPa

and was located near James Bay, but the warm front had not moved significantly, and extending from

Ontario southeastward to northern New Jersey. The weak surface ridge associated with the cold air

damming over New England and eastern NY was still evident. The surface potential temperatures at

1700 UTC (Fig. 14) continue to show the thermal trough east of the mountains and ridge to the west

across northern New England. Even though the contours indicate warming between 1200 UTC (Fig. 9)

and 1700 UTC (Fig. 14), the observations show that potential temperatures had not changed in central

New Hampshire and Maine. The Barnes objective analysis scheme used in these analyses was unable to

accurately correctly capture the enhanced thermal gradients on the northern and southern boundaries of

the cold air dam.


Fig. 13: As in Fig. 8, except 1800 UTC 5 December 2013.


Fig. 14: As in Fig. 9, except for 1700 UTC 5 December 2013.

By 0000 UTC on the 6th, the surface low was nearly stationary over James Bay (Fig. 15). However,

NCEP analyses show that the surface cold frontal trough had moved east into central New York while the

warm front remained stalled across eastern New York and southern New England. The weak surface

ridge axis associated with the CAD was not evident. Yet, the 0000 UTC surface potential analysis (Fig. 16)

continued to show the thermal trough southeast of the White Mountains and the thermal ridge extending

northward into northern Vermont and New Hampshire. Potential temperatures in central New

Hampshire and central Maine remained nearly unchanged from their 1800 UTC values. The moderate

potential temperature gradient and westerly flow associated with the approaching cold front were

evident across western New York and Pennsylvania.


Fig. 15: As in Fig. 8, except 0000 UTC 6 December 2013.


Fig. 16. As in Fig. 9, except for 0000 UTC 6 December 2013.

Above the low-level cold air in New England, the main synoptic features were warm air

advection and southwesterly flow. The 850 hPa analysis at 0000 UTC on the 6th (Fig. 17) showed a

thermal ridge in strong southwesterly flow along with the enhanced thermal gradient associated with the

surface cold front in the western Great Lakes. The low over James Bay was nearly vertically stacked with

closed circulation evident at 850 hPA (Fig. 17) and 500 hPa (Fig. 18). At 300 hPa (Fig. 19), a significant jet

streak developed in the southwesterly flow extending from Wisconsin into western Quebec.


The cold air damming event ended during the early morning hours on the 6th. The surface

potential temperature analyses continued to show evidence of the thermal trough and low-level cold air

in central New Hampshire until nearly 0500 UTC on the 6th (not shown). The NCEP surface analyses

indicate that the cold front passed through central New Hampshire at 0900 UTC on the 6th and the Maine

coast by 1200 UTC (not shown). During the brief 4-hr period between 0500 and 0900 UTC, winds in

central New Hampshire became westerly and the surface potential temperatures increased from 1-2 C to

~4-5 C. However, the low-level cold air remained firmly in place in central Maine until the cold front

passed.

Hourly analyses of surface (2 m) air temperature are shown in the Appendix, illustrating the

evolution of the regional temperature field over the duration of the event. A review of these figures

indicates an inverted ridge of cold air originating in western Maine, with relatively warm air to the north,

west, and south of Plymouth, New Hampshire. As the CADINX reaches the plateau intensity above 5 °C

(100 km)-1 at 2200 UTC on the 5th, the inverted cold ridge grows southwestward into central New

Hampshire. Over the next eight hours, the fine structure of the inverted ridge varies somewhat, but one

low-temperature core remains centered of central New Hampshire. As the CADINX begins its rapid

decline beginning after 0600 UTC on the 6th, the cold core over New Hampshire recedes to the east. This

is accompanied by cooling to the north (over southern Quebec) and west (in Vermont and New York

State).

Local analysis of the event. Research personnel from the Judd Gregg Meteorology Institute deployed

several small temperature sensors, equipped with dataloggers, in the region around Plymouth. These

devices were placed several hours before the beginning 05-06 December 2013 event, and set to log in situ

temperature at 1-min intervals. Figs. 20 and 21 show the placement of the temperature sensors, and Fig.

22 shows time series of the data produced by the sensors. The time series begin on 05 December 2013 at

1300 UTC, and end 06 December 2013 at 2100 UTC, which is approximately 2 hours before the beginning

of the cold-air damming event (as indicated by the CADINX) to about 12 hours after it ended. The total

period recorded is 32 hrs 1 min long.

The temperature sensors were placed in several clusters (Fig. 21). Three of these clusters were in

the form of transects up the slope of hills. One of these transects, located in Rumney, New Hampshire

(northwest of Plymouth), consisted of four sensors ranging in elevation from 514 to 1140 feet above sea

level. Another consisted of three sensors, located in Ashland, New Hampshire (located south of

Plymouth), that ranged from 463 to 546 feet sea level. A third consisted of two sensors, located in

Campton, New Hampshire (located north of Plymouth). The lower sensor was at 603 feet of sea level,

and the upper sensor was at 1060 feet above sea level. Table 10 shows the sensor locations and ID

numbers. Fig. 22 shows the 32-hr temperature time series recorded by each sensor.


Fig. 20: Locations of field sensors placing during cold-air damming event on 05 and 06 December 2013. All

sensors were placed in central New Hampshire. Box in the center of the map indicates detail area shown in Fig. 21.

(Background map from Google Maps, 2014.)

50 KM

NH

VT

ME


Fig. 21: Detail map showing locations of temperature sensors deployed during CAD event of 05-06

December 2013. Insets show details of sensors clusters, labeled with sensor elevations (feet ASL). (Background

maps from Google Map, 2014.)


Table 10: Locations and IDs of temperature sensors deployed in preliminary cold-air damming study. Sensor

locations are shown in Figs. 8 and 9.

Sensor ID Latitude [°N]` Longitude [°E] Elevation [ft ASL]

HOBO 68 45.756793 -71.690297 653

HOBO 69 43.891346 -71.586772 975

HOBO 70 43.732463 -71.661257 546

HOBO 71 43.873025 -71.708534 1060

HOBO 72 43.854400 -71.666800 603

HOBO 73 43.800317 -71.663202 720

HOBO 74 43.802495 -71.771396 903

HOBO 75 43.804646 -71.770965 1140

HOBO 77 43.791381 -71.786997 514

HOBO 78 43.730567 -71.663824 463

HOBO 79 43.731515 -71.662729 467

HOBO 80 43.732329 -71.661522 521

Fig. 22: Time series of in situ temperatures recorded by sensors shown in Figs. 8 and 9. Time series begin 05

December 2013 at 1300 UTC, and end 06 December 2013 at 2100 UTC. Total period shown is 32 hrs 1 min.

All transect clusters show the buildup and erosion of the cold-air damming event, but the

Rumney cluster (which includes HOBOs 73-77), because of its relatively fine vertical resolution, provides


the most detailed picture. By differencing the time series recorded by the sensors in this cluster, the

evolution of the event can be described. Fig. 23 shows the time series of the upper sensor (HOBO 75)

minus the lower sensor (HOBO 77). The vertical elevation difference between these two locations is 626

feet (191 meters).

The time series shows that the temperature at the upper station started out the 32 hr period 1.4 °F

degrees colder than the temperature at the lower station (resulting in negative difference values).

This corresponds to an environmental temperature lapse rate (e) of +4.1 °C km-1, which is less

than the moist adiabatic lapse rate (m). The lower atmosphere was absolutely stable at this time.

The CADINX (Fig. 7) indicates that the cold-air damming event began shortly after 0900 LST

(1400 UTC) on 05 December 2013. Fig. 23 indicates that the upper station was 2.8 °F cooler than

the lower station, corresponding to e of 8.0 °C km-1, which is conditionally stable. The stability

continued to decrease until 0950 LST, when e reached about 10 °C km-1, which corresponds to

the dry adiabatic lapse rate (d). After this the trend reversed, and the stability began increasing.

By 1000 LST (1500 UTC), the CADINX began its rapid increase, and cold-air damming began to

develop in central New Hampshire. This is reflected in the vertical temperature differences

shown in Fig. 11. By 1100 LST (1600 UTC), the vertical temperature difference had decreased to

about 2 °F, corresponding to a e of +5.8 °C km-1, which is approximately the same as m. This

trend toward greater stabilization continued for the next several hours.

At about 1830 LST (2330 UTC), roughly the same time the CADINX reached its plateau value

above °C (100 km)-1), the low-level e became negative, with the upper station indicating warmer

temperatures than the lower station. This condition grew increasingly extreme, reaching a peak

of about 7.4 °F at about 0350 EST (0850 UTC) on 6 December 2013, corresponding to a e of -21.5

°C km-1 - an extremely stable condition. After this, the stability began to rapidly erode. Note that

the peak stability occurred almost three hours after the CADINX indicated the beginning of the

erosion of the event. This indicates that the pool of cold air associated with the damming event

was deeper than the highest sensor in the Rumney transect cluster, and that the strongest positive

vertical temperature gradient (negative e) began to register as the cold pool eroded to the point

where its upper surface corresponded to the height of the upper sensor in the cluster.


Fig. 23: Difference time series of HOBO 75 – HOBO 77, occurring over a vertical elevation distance of 626

feet (191 meters). Lower panel shows whole time series, beginning 05 December 2013 at 1300 UTC, and ending 06

December 2013 at 2100 UTC. Upper panel shows detail of period indicated by blue triangles in the lower panel.

Positive values occurred when temperatures at the upper sensor (HOBO 75) were warmer than temperatures at the

lower sensor (HOBO 77).

Additional insight is gained by differencing the temperature sensors placed at 1140 and 903 ft

ASL (HOBOs 75 and 74, respectively). This temperature difference time series is shown in Fig. 24. As

with the previous difference time series, the vertical temperature profile began with the lower sensor

warmer than the upper sensor. The vertical temperature lapse rate then reversed as cold-air damming

built into the region. However, in this case, the temperature difference briefly disappeared (highlighted

with an arrow in Fig. 24) before climbing to a peak value. This occurred at 0200 EST (0700 UTC) on 6

December 2013, which is one hour after the CADINX (Fig. 7) indicates the cold-air damming event began

to erode. An examination of the two base temperature series indicates that the vertical temperature

difference disappeared because the temperature at the upper sensor decreased, that is, the depth of the cold

pool of air temporarily increased. After about 10 minutes, the temperature at the upper sensor began to

rapidly increase (the depth of the cold air pool began to decrease again), and the vertical temperature

gradient reached a maximum value about two hours later, at 0350 LST (0850 UTC) – about one hour

before the CADINX indicates the cold-air damming event ended. Note that the climb toward the peak

vertical temperature difference value appears to have occurred in four steps, each characterized by its

own smaller peak, followed by a brief decrease in the vertical temperature difference. The final decrease

in the vertical temperature difference resulted when the lower temperature sensor rapidly warmed

(indicating a rapid decrease in the depth of the cold pool of air). By about 0740 LST (1240 UTC) on 6

December 2013, the lower sensor became warmer than the upper sensor.


Fig. 24: Difference time series of HOBO 75 – HOBO 74, occurring over a vertical elevation distance of 237

feet (72 meters). Lower panel shows whole time series, beginning 05 December 2013 at 1300 UTC (0800 LST), and

ending 06 December 2013 at 2100 UTC (1600 LST). Upper panel shows detail of period indicated by blue triangles

in the lower panel. Positive values occurred when temperatures at the upper sensor (HOBO 75) were warmer than

temperatures at the lower sensor (HOBO 74).

As discussed in Section 1, the onset, intensity, and demise of CAD events may be influenced by

precipitation. Figure 25 illustrates the temporal evolution of the KGYX radar reflectivity alongside the

CADINX for the duration of the 5–6 December CAD event. The reflectivity data indicate two periods of

precipitation (rain) over central New Hampshire: One at approximately 1700 LST (2200 UTC) on 5

December 2013 (label “B”) with reflectivity values of ~15 dBZ, and another from 0400 to 0900 LST (0900-

1400 UTC) on 6 December 2013 (label “D”) with reflectivity values of 20–30 dBZ. The corresponding

reflectivity images from the WSR-88D at KGYX at 1700 LST on 5 December 2013 and 0500 LST on 6

December 2013 are shown in the right portion of the image. The former time period indicates scattered

rain showers across central New Hampshire, while the latter time period indicates broader synoptic-scale

precipitation. The lower portion of the image is a reproduction of the CADINX from Fig 7a for

comparison.

The initial increase in the CADINX from 0.0 to >2.0°C (100 km)–1 between 1400 and 1700 LST on 5

December 2013 (labels “A” and “B”) occurred prior to the observed reflectivity values of ~15 dBZ

over HOBO sensor 75 and 77.

The period of scattered rain showers at 1700 LST on 5 December 2013 and the time period

immediately thereafter is associated with the subsequent increase in the CADINX from >2.0°C

(100 km)–1 to >5.0°C (100 km)–1. The increase in the CADINX is concurrent with the observed

increase in the temperature difference between HOBO-75 and HOBO-77 from <0.0°C to >0.0°C at

~1800 LST (2300 UTC) on 5 December 2013, indicating that the upper station (HOBO-75) was

now warmer than the lower station (HOBO-77) as shown previously in Fig. 23. Investigation of

the air temperature and dew point temperature at Plymouth (K1P1), New Hampshire suggests

that this period of rain was associated with evaporative cooling that allowed near-surface


temperatures to cool relative to higher-altitude temperatures (not shown). This process would

lead to an increase in the CADINX value as observed.

Further comparison between the KGYX reflectivity data and the CADINX indicate that the CAD

cold pool began to erode prior to the onset of synoptic scale precipitation between 0100 and 0400

LST (0600 and 0900 UTC) on 6 December 2013 (label “C”); however, the full erosion of the cold

pool by 0500 LST (1000 UTC) 6 December 2013 (label “D”) occurred in conjunction with observed

reflectivity values >20 dBZ beginning at ~0400 LST (0900 UTC) on 6 December 2013.

While the initial period of rain showers may have served to enhance the low-level cold air via

evaporative cooling across the interior valleys of central New Hampshire on 5 December 2013,

the later period of synoptic-scale rain may have served to mix warmer air at higher levels down

to the surface and fully erode the low-level cold pool on 6 December 2013.

Fig. 25: KGYX base radar reflectivity (dBZ; 0.5 elevation scan) in 10-minute intervals from 0900 LST (1400 UTC)

5 December 2013 to 0900 LST 6 December 2013 at the location of HOBO-75 (solid) and HOBO-77 (dashed; top

left), plan-view reflectivity at 1700 LST (2200 UTC) 5 December 2013 and 0500 LST (1000 UTC) 6 December

2013 (upper right thumbnails), and the CADINX for 0900 LST 5 December 2013–1400 LST 6 December 2013.

Four time periods discussed in the text are labeled “A”, “B”, “C”, and “D”.


The pattern described above is consistent with the passage of several waves on the upper surface

of the cold pool of air. We hypothesize that as the cold dome vertically erodes to expose mountain tops,

turbulence generated downstream of the mountains accelerates cold air erosion downstream of the

mountain tops. During the initial exposure of mountain tops, 𝑊𝑇 is small, Fr is >>1 and turbulence is

generated. This initial turbulence will be small spatially and with low TKE. As the top of the cold air

dome drops in elevation (e.g., via divergence, or vertical shear turbulent mixing), the spatial extent of the

turbulence perpendicular to the flow increases proportionately with the width of the exposed mountain

top perpendicular to the flow, and TKE increases. We speculate that Fr remains >1 for most of the cold

air erosion, and smaller topographic features gradually become exposed and increase the number and

spatial coverage of orographically enhanced cold air erosion.

During the 4-6 December 2013 CAD event, analysis of the nearest upstream sounding (from

Albany, NY; KALB) at 0000 UTC on 6 December (Fig. 26) reveals a well-mixed layer of warm air (876-747

hPa) with Fr = 3.38, assuming an exposed mountain top width of 1 km. This Fr value suggests turbulent

downstream flow that accelerates cold air erosion.

Fig. 26: Radiosonde sounding profile from KALB (Albany, New York) for 0000 UTC on 6 December 2013.

4. CASE STUDY: 09-12 JANUARY 2014.

Another notable CAD event occurred between the eighth and 12th of January, 2014. Fig. 27 shows

the hourly CADINX intensities during the event and for the entire month. The event in question

occurred in two phases. The first phase lasted about 12 hours, and was followed by a 10-hr period when

the CADINX indicates cold-air damming was no longer operative. The second phase lasted about two

and a half days, and was associated with a freezing rain event localized in central New Hampshire.


Fig. 27: CADINX for cold-air damming event from 08 through 12 January 2014. Upper panel is CADINX during

the event; lower panel shows CADINX for the entire month of January 2014. Triangles on lower panel indicate detail

area in upper plot.

The first phase began at 2300 UTC (1800 EST) on 8 January, and reached moderate intensity

(above 4 °C (100 km)-1) within four hours. This phase of the event remained active for about ten hours,

then rapidly eroded. Fig. 28 shows surface (2-m) isotherms for the region at 1200 UTC, when the event

peaked. A “finger” of cold air extends from western Maine into central New Hampshire, showing the

characteristic signature of regional CAD found during the December, 2013 event (See Appendix). For the

next ten hours, the CADINX went back to zero, indicating that cold-air damming was no longer

operating in central New Hampshire.

The United States Surface Analysis (Fig. 29) for the period during and after this phase of the

event indicates the presence of a ridge of high pressure, with the center of the surface anticyclone moving

across the Midwest and into southern New England. The hourly surface observations (Fig. 30) from

Plymouth Municipal Airport (K1P1), in central New Hampshire, indicate light wind conditions with clear

skies for most of the night, and the plotted radiosonde observation (Fig. 31) for 1200 UTC on 9 January,

from Gray, Maine (KGYX), indicates the presence of a weak, surface-based radiation inversion. From this

evidence it seems likely that the first phase of this CAD event was initiated by the advection of cold, dry

continental polar air from interior Canada into central New Hampshire, which was then further

stabilized by radiational cooling under clear skies and light winds. After sunrise on 9 January, the

unshielded cold pool was warmed by insolation, causing the cold-air damming event to break down.


Fig. 28: Surface (2 m) isotherms for 09 January 2014 at 1200 UTC.

Fig. 29: United States Surface Analysis for 09 January 2014 at 1200 UTC.


Fig. 30: 24-hr Meteogram for Plymouth Municipal Airport (K1P1), for 09 January 2014.

Fig. 31: Plotted radiosonde observation from Gray, Maine (KGYX), for 1200 UTC on 9 January 2014.

The second phase of the event began at 0200 UTC on 10 January 2014, which corresponds to 2100

EST on 9 January. By 0800 UTC (0300 EST), the CADINX (Fig. 27) reached moderate levels (> 3 °C (100

km)-1) similar to those it had achieved the previous night. After about 1100 UTC, the CADINX began to

drop again, and fell to about 1 °C (100 km)-1 by 2200 UTC. The difference between the behavior of the

CADINX during the early morning hours of the 9th and the 10th may have been caused by the continued

eastward movement of the surface anticyclone.

By the morning of the 10th, the surface high center had moved offshore of the U.S. Midatlantic

region, and southerly flow on the west side of the high, along with an upper-level disturbance, were

bringing an area of cloud cover and precipitation into New England. Light precipitation moved through


the region during the early part of the day, exiting to the east by about 2000 UTC. Had the air column

been unsaturated, evaporation (or sublimation) of the falling precipitation might have enhanced the low-

level cold pool via latent heat removal. The KGYX radiosonde observation for 1200 UTC on 10 January

indicates that the low-level column of air was becoming saturated, and the radiosonde at 1100 UTC on 11

January indicates it was in a near-saturation state from the surface to just above 700 hPa. Under these

conditions, very little evaporative cooling would have occurred, so the addition of precipitation would

not have further enhanced the low-level cold pool via latent heat removal.

Cloud cover remained in place from 0000 through 1200 UTC on 11 January, and the CADINX

indicates that the cold-pool remained essentially stable during this period. CADINX values hovered

between 2 and 3 °C (100 km)-1, indicating weak damming. At 1200 UTC, a large precipitation shield

associated with a major winter storm began to move into the region (Fig. 32). Warm-air advection into

the region, associated with this midlatitude cyclone, produced a strong CAD situation: The areas north,

west, and south of central New Hampshire warmed up substantially, while temperatures in central New

Hampshire remained near freezing. Fig. 33 shows surface (2-m) isotherms at 1800 UTC (1300 EST).

The combination of the approaching midlatitude cyclone, and the trapped cold pool produced a

freezing rain event in the central part of New Hampshire. Figs. 34 and 35 show meteograms for Concord

(KCON) and Manchester (KMHT), NH on 11 January 2014. KCON is shown because K1P1 does not have

a freezing precipitation sensor, and KCON is the closest observing site that does have one. KCON and

KMHT are shown to illustrate the intensity of the horizontal temperature gradient across the southern

part of New Hampshire. Freezing rain occurred in KCON (and through most of the rest of state) between

0800 and 1500 UTC. The southern edge of the cold pool had a very sharp transition zone, with

temperatures changing by more than 20 °F (11°C) over a horizontal distance of about 30 km (19 mi). The

transition zone moved north KMHT at 2000 UTC (1500 EST), but remained south of KCON until 0300

UTC (2200 EST).

Fig. 32: United States Surface Analysis for 11 January 2014 at 1200 UTC.


Fig. 33: Surface (2 m) isotherms for 11 January 2014 at 1800 UTC.

Fig. 34: 24-hr Meteogram for Concord (KCON), for 11 January 2014.


Fig. 35: 24-hr Meteogram for Manchester (KMHT), for 11 January 2014.

5. SCIENTIFIC QUESTIONS AND PROPOSED METHODS.

We propose to address the following eight questions:

1. When are New Hampshire CAD events most likely (time of day, time of year)? 2. What is the average duration of a CAD event in central New Hampshire? 3. What is the influence of CAD events on precipitation type? What portion of freezing precipitation

events in central New Hampshire can be attributed to CAD?

4. What are the physical processes associated with the development and decay of a New Hampshire CAD event?

5. What role does topography play in the development and decay of New Hampshire CAD events? 6. What is the vertical depth of the cold pool associated with CAD events? 7. How, and to what extent, do vertical variations in wind, and the propagation of waves along the

upper surface of cold pools occurring with CAD events, create hazards to aviation, such as wind shear and turbulence?

8. How often are CAD events eroded prior to vs. during the passage of a subsequent cold front and its associated precipitation?

9. How well are maximum temperatures forecasted by numerical weather prediction models during CAD events?

We propose to answer these questions in the following order:

Questions 1 – 3 (Beginning Year 1): The climatology of CAD events over central New Hampshire already

accomplished was developed using a CAD index (CADINX), computed from observational data at four

surface stations, but represented only a two-year period from 1 April 2006 to 31 March 2008. We propose

to extend the length of the CAD event climatology using observational data and derived quantities (such


as the CADINX) to better understand when CAD events are most likely (e.g. time-of-day and time-of-

year), the average duration of CAD events over New England, and the historical relationship between

freezing precipitation and CAD. This climatology will also involve the use of NARR composite analyses

to quantify the mean synoptic and mesoscale conditions associated with synoptic, hybrid and in situ CAD

events in New England.

Resources: Three faculty members (one with partial support; the others donating time as

matching), one graduate student, and one member of the National Weather Service.

Duration: 18 months.

Outcome: One master’s thesis; American Meteorological Society (AMS) and/or North East Storm

Conference (NESC) presentation; one paper submitted to peer-reviewed journal.

Questions 4 – 8 (Beginning Year 2): In-depth case studies, in addition to and in more detail than those

presented above, will be conducted on select events identified in the climatology, to better understand the

variations around the mean conditions quantified with the NARR ensembles, and the physical processes

associated with the development, depth, intensity, and decay of CAD events over New England. These

case studies will utilize both high-resolution NARR reanalysis and Weather Research and Forecast (WRF)

simulations of the selected CAD events. WRF model simulations will also be used to assess the role of

topography and boundary-layer processes in the development, depth, intensity, and decay of CAD

events as hypothesized in the two case studies discussed above.

The case studies of past CAD events will be complemented by high-resolution in situ

observational data collected during Intensive Observation Periods (IOPs) occurring in the period of

award by Plymouth State University meteorology students and faculty. Observational data will be

collected from (a) HOBO temperature data loggers, (b) portable AWOS systems, (c) a portable

rawinsonde system co-located with an atmospheric profiler, and (d) tethered dirigible-based weather

instrumentation, all of which will contribute to a complete description of the horizontal and vertical

structure of CAD events and their evolution. The HOBO temperature data loggers and portable AWOS

systems will be necessary to identify the horizontal characteristics of CAD events, while the portable

rawinsonde, profiler, and dirigible weather instrumentation will be crucial to identify the high-resolution

vertical thermodynamic and kinematic structure of CAD events. A proposed implementation of

observing systems during a typical CAD IOP in central New Hampshire is presented in Fig. 36.

Resources: Four faculty members, four graduate students, and several undergraduate students.

o One faculty member and one graduate student will focus on NARR analysis.

o One faculty member and one graduate student will focus on WRF model simulations.

o One faculty member and one graduate student will focus on vertical profiler data and

aviation hazards.

o One faculty member, one graduate student, and the undergraduates will conduct field

studies.

Duration: 18 – 24 months.

Outcome: Four master’s thesis and several senior research papers; 3 – 5 AMS and/or NESC

presentations; 3 – 4 papers submitted to peer-reviewed journals.


Fig. 36: Deployment strategy for hypothetical IOP involving low pressure passing to the northwest of New

England and CAD developing in central New Hampshire. The red line represents a subjective analysis of the

location of the warm front (see inset) that is highly dependent on the depth of the cold air relative to local topography.

In this hypothetical analysis, the warm front approximately parallels the 500 m above sea level isopleth to the south of

the White Mountain National Forest.

Question 9 (Beginning Year 3): Observational evidence suggests that the scale (horizontal and vertical)

and dynamics of CAD events may not be accurately represented by current numerical weather prediction

(NWP) model forecasts. Therefore, we propose to assess the accuracy and skill of NWP models in

forecasts of precipitation type and temperature prior to, during, and after CAD events over New

England.

The assessment of forecast skill will be accomplished by a series of case studies, and be

complemented by developing and testing decision forecast support services and tools, such as a forecast

CADINX for use in the National Weather Service’s Interactive Forecast Preparation System, derived from

daily forecast data provided by the National Centers for Environmental Prediction (NCEP) and locally

run high-resolution WRF model simulations. Case studies will be used evaluate the relative skill of

NCEP models and a locally-run WRF model, for all CAD types, and by CAD classification (synoptic, in

Vermont

New Hampshire

Maine

Surface fronts

Plymouth State

University

Plymouth Regional

Airport (K1P1)

Mount Washington

Observatory

Observing platforms:

HOBO/AWOS

Tether-sonde

Rawinsonde

Target locations for HOBO

or portable AWOS transect

Tether-sonde deployment:

North South

Tenney Mtn

Stinson Mtn

~500 m


situ, and hybrid). We will also examine whether forecast CADINX values become more representative as

an actual CAD event becomes progressively more immediate.

We envision this third and final component of the CADEX project will be completed in close

collaboration with the National Weather Service forecast office in Gray, Maine, or the office in Caribou,

Maine.

Resources: Two faculty members, two graduate students, and one or more members of the

National Weather Service.

Duration: 18 – 24 months.

Outcome: Two master’s theses; 1 – 2 AMS and/or NESC presentations; two papers submitted to

peer-reviewed journal; operational (numerical) tools for use in NWS forecast offices.


REFERENCES CITED

Bailey, C.M., G. Hartfield, G.M. Lackman, K. Keeler, and S. Sharp, 2003: An objective climatology,

classification scheme, and assessment of sensible weather impacts for Appalachian cold-air damming,

Weather and Forecasting, 18 (4), 641 – 661.

Bell, G.D., and Bosart, L.F., 1988: Appalachian cold-air damming, Monthly Weather Review, 116 (1), 137-

161.

Lackmann, G., 2012: Mid-latitude Synoptic Meteorology: Dynamics, Analysis, and Forecasting, 1st edition,

American Meteorological Society, 345 pp.

National Aeronautics and Space Administration (NASA), 1966: U.S. Standard Atmosphere Supplement

1966, U.S. Government Printing Office, 300 pp.

Rogers, R.R., and M.K. Tau, 1989: A Short Course in Cloud Physics, 3rd Ed., Butterworth-Heinemann, 290

pp.

Stull, R.B., 1988: An Introduction to Boundary Layer Meteorology, Kluwer Academic Publishers, 670 pp.


ADDITIONAL COLD-AIR DAMMING REFERENCES

Brennan, M.J., 2004: Cold air damming, Bulletin of the American Meteorological Society, 85 (5), 659 – 660.

Forbes, G.S., R.A. Anthes and D.W. Thomson, 1987: Synoptic and mesoscale aspects of an Appalachian

ice storm associated with cold-air damming, Monthly Weather Review, 115 (2), 564 – 591.

Fritch, J.M., J. Kapolka, and P.A. Hirschberg, 1992: The effects of subcloud-layer diabatic processes on

cold air damming, Journal of the Atmospheric Sciences, 49 (1), 49 – 70.

Lackmann, G. M., and W. M. Stanton, 2004: Cold-air damming erosion: Physical mechanisms, synoptic

settings, and model representation. Preprints, 20th Conf. on Weather Analysis and Forecasting, Seattle, WA,

Amer. Meteor. Soc., CD-ROM, 8.3.

Raman, S., N.C. Reddy, and D.S. Niyogi, 1998: Mesoscale analysis of the Carolina coastal front, Boundary-

Layer Meteorology, 86 (1), 125 – 145.

Rauber, R.M., L.S. Olthoff, M.K. Ramamurthy, D. Miller and K.E. Kunkel, 2001: A synoptic weather

pattern and sounding-based climatology of freezing precipitation in the United States east of the Rocky

Mountains, Journal of Applied Meteorology, 40 (10), 1724 – 1747.

Xu, Q., 1990: A theoretical-study of cold air damming, Journal of the Atmospheric Sciences, 47 (23), 2969 –

2985.

Xu, Q., and S.T. Gao, 1995: An analytic model of cold-air damming and its applications, Journal of the

Atmospheric Sciences, 52 (3), 353 – 366.

Xu, Q., S.T. Gao, and B.H. Fiedler, 1996: A theoretical study of cold air damming with upstream cold air

flow, Journal of the Atmospheric Sciences, 53 (2), 312 – 326.


APPENDIX

Fig. A1: Surface (2 m) isotherms for 05 December 2013 at 1400 UTC.


the new england cold-air damming experiment (cadex)stmiller/stmiller... · 3/27/2014 · the new...

Documents