Abstract

The U.S. Climate Reference Network (USCRN) was deployed between 2001 and 2008 for the purpose of yielding high-quality and temporally stable in situ climate observations in pristine environments over the twenty-first century. Given this mission, USCRN stations are engineered to operate largely autonomously with great reliability and accuracy. A triplicate approach is used to provide redundant measurements of temperature and precipitation at each location, allowing for observations at a specific time to be compared for quality control. This approach has proven to be robust in the most extreme environments, from extreme cold (−49°C) to extreme heat (+52°C), in areas of heavy precipitation (4700 mm yr−1), and in locations impacted by strong winds, freezing rain, and other hazards. In addition to a number of stations enduring extreme winter environments in Alaska and the northern United States, seven of the USCRN stations are located at elevations over 2000 m, including stations on Mauna Loa, Hawaii (3407 m) and on Niwot Ridge above Boulder, Colorado (2996 m). The USCRN temperature instruments and radiation shield have also been installed and run successfully at a station on the Quelccaya Ice Cap in Peru (5670 m). This paper reviews the performance of the USCRN station network during its brief lifetime and the potential utility of its triplicate temperature instrument configuration for measuring climate change at elevation.

1. Introduction

Global climate records show a systematic change in surface air temperatures during the period of instrumental observations since the mid-nineteenth century (Brohan et al. 2006; Lugina et al. 2007; Hansen et al. 2010; NCDC 2010). These changes have been linked to anthropogenic effects on the global earth system, in particular those associated with emissions of greenhouse gases and particles (aerosols) to the atmosphere (Solomon et al. 2007). Because realistic scenarios of future human activity do not promise a quick reversal of the existing tendencies (Nakicenovic et al. 2000), the requirement to accurately and thoroughly monitor the state of the climate (Arndt et al. 2010) as changes occur becomes especially important. Gaps in our ability to monitor climate must be recognized and corrected. One of the most important gaps in our observation networks currently involves land surfaces at high elevation. Existing meteorological networks usually are biased to lower elevations. For example, the mean elevation of the stations located in the western (mountainous) part of the conterminous United States (west of 100°W) is approximately 500 m less than the actual mean elevation of the region. Similar elevation biases exist with observational networks in other mountainous areas of the world. This gap in observing networks raises concerns that fewer observations may be gathered than are needed to detect critical developments that are occurring and/or will occur at high elevations.

There are indications from a number of high-elevation climate records that the amplitude of temperature changes this century may differ from the observed global or hemispheric change record that is dominated by low-elevation observations (Beniston et al. 1997). Steep gradients of temperature and precipitation occur over very short horizontal distances in mountainous regions, forming closely spaced ecosystems separated by ecotones that are highly sensitive to changes in energy and moisture inputs. Sufficiently elevated locations transition to nonvegetated zones dominated by cryospheric activities that are often important for regional hydrology. Despite the relatively small area of land that is elevated, climate change in these locations will have disproportionate impacts on broader regions with extensive populations. In particular, in many dry regions of the tropics and midlatitudes, mountainous regions provide a crucial portion of the water supply (Aizen and Groisman 2009; Barry 2008). Therefore, there is a critical need to monitor climate change more effectively in elevated regions and associate changing climate states with their impacts, both locally and regionally (Diaz and Bradley 1997; Diaz 2003; Diaz et al. 2003; Huber et al. 2005).

To date, climate monitoring in elevated regions has been inconsistent both in network design and in instrumentation. Many individual studies have focused on critical targets, such as high mountain glaciers, but most of these studies are brief in duration and inconsistent in observational methodology. Groups like the Mountain Research Initiative (MRI) and the Coordinated Energy and Water Cycle Observations Project (CEOP) have developed high-elevation initiatives to coordinate international efforts to make in situ measurements in mountainous regions. CEOP-High Elevations (HE) in particular is proposing to link together high-altitude observation stations into a high elevations network (HE-Net) (CEOP-HE 2010). While types of measurements are specified, approaches to measurement and equipment selection are left to the various participants. Minimal standards for observation quality and completeness should be established. It also may be useful for groups representing various measurement technologies/approaches to choose designated HE research stations to collocate their standard automated weather stations and conduct measurement intercomparisons.

There are existing networks that may provide experience and guidance to measurement of climate at high elevations. The U.S. Climate Reference Network (USCRN) was deployed across the United States between 2001 and 2008 for the purpose of yielding high-quality and temporally stable in situ climate observations in pristine environments over the twenty-first century. Given this mission, USCRN stations are engineered to operate largely autonomously with great reliability and accuracy. The primary lesson from USCRN that is very applicable to high-elevation climate observations is the adoption of a triplicate instrument approach for key variables, providing redundant measurements of temperature at one height at each location, for example. This approach not only helps to maintain data contiguity in harsh environments, but also allows multiple observations at a specific time to be compared for quality control purposes. In addition, the USCRN participates in measurement comparisons with the Canadian Reference Climate Station Network, U.S. National Weather Service, World Meteorological Organization precipitation test bed, NASA Soil Moisture Active–Passive mission test bed, and at the USCRN test bed at Marshall Field near Boulder, Colorado. This paper discusses the USCRN approach to gathering climate science quality observations, the performance of the USCRN station network with altitude, and some of the findings at USCRN stations exposed to extreme environments or located at elevation.

2. Triplicate instrument configuration of USCRN

USCRN is designed in a manner analogous to a scientific experiment, where replication is always important to confirming a finding, or, in this case, a measurement. This basic reasoning leads to the reduction of uncertainty in each observation, and, therefore, faster and more confident recognition of trends. The triplicate instrument configuration is in keeping with the 10 climate observing principles expressed in the “Adequacy of Climate Observing Systems” (NRC 1999), an important milestone in the development of better climate observations in the United States and also adopted by the Global Climate Observing System (GCOS 2003). While many more sensors would lead to even less uncertainty, using three sensors provides a reasonable compromise between cost and power consumption, and data continuity and confidence.

Starting with calibration against a National Institute of Standards and Technology (NIST) traceable standard, all instruments seem equal. However, any single instrument, no matter how well calibrated, may be defective or undergo stresses in harsh environments that cause partial or total failure. Of the two, partial failure can lead to reasonable-looking measurements that drift considerably from the true state, yet are very difficult to detect. For example, it could take years to collect enough data to detect a 1°C shift in an instrument by comparing the observations at one station to those from neighboring stations. With three instruments, a pairwise approach can be used in real time to compare measurements and see if any one of the sensors deviates from the other two. For example, the differences

 
formula

can be evaluated relative to a given tolerance level

 
formula

and be determined to be within or outside tolerance. If two pair differences with one instrument in common are outside tolerance, and the third pair difference is within tolerance, then the two instruments in the third pair are producing valid measurements, and the one instrument common to the first two pairs is flagged and not used in calculating the air temperature for that time interval. A simulated example of this can be seen graphically in Fig. 1. On 11 May, sensor 3 quickly changes to a new state 1.5°C cooler than sensors 1 and 2. In a normal temperature time series, this type of deviation would not be detected in real time if only one instrument was available, and may not be detectable in hindsight unless the deviation continued for a long time and the resulting time series could be compared to several nearby stations. However, because three measurements are compared in a pairwise fashion in real time, it is immediately apparent that sensor 3 is problematic, and the problem is solved by replacing the sensor two days later. If this was a situation in which the site was inaccessible, such as can happen during midwinter, the defective sensor may not be quickly replaced, but its data would be flagged to prevent future use.

Fig. 1.

Temperatures observed by sensor 3 deviate from those of 1 and 2 by 1.5°C as indicated by these plots of pairwise differences between sensors. After two days, the fault is corrected by replacing the problematic sensor.

Fig. 1.

Temperatures observed by sensor 3 deviate from those of 1 and 2 by 1.5°C as indicated by these plots of pairwise differences between sensors. After two days, the fault is corrected by replacing the problematic sensor.

The air temperature instruments used in the USCRN are a model of highly accurate platinum resistance thermometers (PRTs), installed within triple-walled radiation shields that are fan aspirated (Fig. 2). Each PRT is rated to have the accuracy of ±0.25°C, but in a laboratory setting, measurements are repeatable to ±0.01°C. With the triplicate instrument configuration, high accuracy can be maintained in the real world. The USCRN pairwise tolerance between two PRTs in normal operating conditions (−50°C to +50°C) is set at ±0.3°C, and field performance is within this tolerance the vast majority of the time. This result is a significant improvement over the Automated Surface Observing System (ASOS), which is designed to have a maximum error of ±1°C and a root-mean-square error of ±0.5°C (NOAA 1998).

Fig. 2.

The USCRN station near Montrose, Colorado. Note the triplicate configuration of shielded and aspirated platinum resistance thermometers on the tower at left, and a weighing bucket gauge with triplicate wires for measuring precipitation amount inside the small double fence intercomparison reference fencing on the right.

Fig. 2.

The USCRN station near Montrose, Colorado. Note the triplicate configuration of shielded and aspirated platinum resistance thermometers on the tower at left, and a weighing bucket gauge with triplicate wires for measuring precipitation amount inside the small double fence intercomparison reference fencing on the right.

When evaluating equipment choices, it is important to choose instruments that are designed to withstand the rigors of elevated observation locations. However, while there may be custom station engineering needed to power systems and to mount and wire instruments in harsh environments, custom instruments themselves are usually not required. For example, all major USCRN instruments and PRT shields have been acquired from commercial sources.

3. USCRN temperature observations in the western United States

There are 40 USCRN stations in the western United States that have at least three years of climate observations. In Fig. 3, the locations of the stations are given with a class number defined by the elevation range in which the station is located, with class 1 stations higher than 2000 m, class 2 between 1500 and 2000 m, class 3 between 1000 and 1500 m, class 4 between 500 and 1000 m, and class 5 below 500 m. For the period 2007–09, USCRN stations in each elevation class were evaluated to determine the percentage of individual PRT observations available, compared to the percentage of cases in which the set of three PRTs at each station were able to yield a valid, quality controlled, calculated 5-min temperature. A valid PRT observation not only passes gross range checks, but also passes a strict pairwise comparison test that flags measurements that are more than 0.3°C different from the other members of the triplicate configuration. In all elevation bands, the value of a triplicate configuration was evident, with there being more cases where air temperature could be calculated than would be expected if a single PRT were deployed. The higher-elevation USCRN stations actually performed better in the west than the lower-altitude stations (Fig. 4), with less than a 0.1% difference between the single PRT reports received versus the percentage of times air temperature was calculated from the set of PRTs. On the other hand, the lower-altitude stations greatly benefited from the triple configuration, yielding 0.5% more calculated air temperatures than if a single PRT was deployed. The lower-elevation stations seem to have more challenges in the west, such as extremely hot deserts where temperatures exceeded 50°C during the three years, or humid coastal environments. In the higher-elevation bands, minimum temperatures plunged below −35°C at times, but did not get as warm as the low lands. The results seem promising for intermediate- to high-elevation deployment of USCRN-like systems for measuring temperatures.

Fig. 3.

USCRN stations in the western United States with a minimum of three years of data. The elevation of the station has been classified into five elevation ranges, with 1 the highest and 5 the lowest elevations.

Fig. 3.

USCRN stations in the western United States with a minimum of three years of data. The elevation of the station has been classified into five elevation ranges, with 1 the highest and 5 the lowest elevations.

Fig. 4.

Western U.S. USCRN air temperature observations from 2007 to 2009, comparing the percentage of possible observations of 5-min air temperature received from the individual PRTs (blue) vs the percentage of times when it was feasible to use the three PRT sets to calculate temperature for a 5-min period.

Fig. 4.

Western U.S. USCRN air temperature observations from 2007 to 2009, comparing the percentage of possible observations of 5-min air temperature received from the individual PRTs (blue) vs the percentage of times when it was feasible to use the three PRT sets to calculate temperature for a 5-min period.

The USCRN network is designed to assess national climate trends on an annual basis. However, for the purposes of discussing station performance at altitude, two groups of stations in the west were separated by elevation to examine the recent western temperature record. Seasonal temperature departures relative to 1971–2000 estimated normals (Sun and Peterson 2005) were calculated for the 13 highest- and 12 lowest-elevation stations in the west and were compared to see if the climate signal varies with altitude for the 2004–08 period. While there is no known elevation bias in the quality of the estimated normals, it should be noted that estimated normals for higher-elevation stations are usually based on long-term stations located at a wider range of altitudes than used in estimating normals for stations at lower elevations. Figure 5 indicates that there are no significant altitude-related differences in seasonal departures during the period for maximum and minimum temperatures. The interstation ranges in temperature departures for each season and altitude group overlap considerably (bars in Fig. 5), so small differences in the seasonal means for the altitude groups (relationships of squares to triangles in Fig. 5) are imposed on considerable spatial variation across the region. The high-elevation stations located above 1800 m display only slight differences in seasonal departure means, seasonal departure standard deviations, and seasonal station ranges compared to the lower-altitude stations located below 800 m elevation (Table 1). A sample of 20 seasons was not sufficient to find statistically significant differences in seasonal mean and standard deviations between the two elevation groups. In fact, it should not be construed from this analysis that the USCRN itself has the ability to clearly detect relationships between elevation and temperature trends, as the density of this network is too coarse to do so and the time series too short. The main point of this analysis is to illustrate that USCRN station design could be deployed to great effect in mountainous terrain, especially with regards to using a triplicate measurement configuration.

Fig. 5.

The seasonal temperature departures from estimated station normals for the western United States, using 13 USCRN stations above 1800-m elevation (brown) and 12 USCRN stations under 800-m elevation (green): (a) maximum temperature departure (°C) and (b) minimum temperature departure (°C). The bars indicate individual station maximum and minimum departures.

Fig. 5.

The seasonal temperature departures from estimated station normals for the western United States, using 13 USCRN stations above 1800-m elevation (brown) and 12 USCRN stations under 800-m elevation (green): (a) maximum temperature departure (°C) and (b) minimum temperature departure (°C). The bars indicate individual station maximum and minimum departures.

Table 1.

USCRN high-elevation (>1800 m) and low-elevation (<800 m) area-averaged maximum and minimum seasonal temperature departures for September–November (SON) 2004 to 2008 in the western United States. The high- and low-altitude stations produce quite similar seasonal area average, seasonal area standard deviation, and seasonal area station range statistics during this brief period of operation.

USCRN high-elevation (>1800 m) and low-elevation (<800 m) area-averaged maximum and minimum seasonal temperature departures for September–November (SON) 2004 to 2008 in the western United States. The high- and low-altitude stations produce quite similar seasonal area average, seasonal area standard deviation, and seasonal area station range statistics during this brief period of operation.
USCRN high-elevation (>1800 m) and low-elevation (<800 m) area-averaged maximum and minimum seasonal temperature departures for September–November (SON) 2004 to 2008 in the western United States. The high- and low-altitude stations produce quite similar seasonal area average, seasonal area standard deviation, and seasonal area station range statistics during this brief period of operation.

Some studies have found more positive observed temperature trends over time at intermediate and high elevations than at lower elevations in the Alps (Durand et al. 2009; Rebetez and Reinhard 2008) and in modeled temperature trends over the Rockies (Fyfe and Flato 1999). Earlier studies using small sets of stations (e.g., Diaz and Bradley 1997; Seidel and Free 2003) seem to have indicated that high-elevation stations were showing more rapid temperature increases in recent decades, but these relationships have not proven to be consistent globally (Pepin and Lundquist 2008; Pepin and Seidel 2005). The temperature trend picture over elevated topography is very complex because of both synoptic-scale and local effects (Daly et al. 2010), although those elevations near the annual 0°C isotherm usually display more positive temperature trends because of snow–albedo–temperature feedbacks as the snow line retreats upward (Pepin and Lundquist 2008).

4. USCRN instrument performance in extreme conditions

The two USCRN stations with the longest exposure to extreme cold conditions such as would typically be faced at high altitude are the experimental stations at Fairbanks and Barrow, Alaska. The station at Barrow is near sea level, but the station at Fairbanks is on a low hill (347 m), so is not subject to the extreme cold of valley bottoms in central Alaska. The network minimum temperature to date was set at Barrow on 3 February 2006, reaching −49.2°C. Both stations have functioned well during the winter since installation in 2002, with the exception of precipitation measurement at Barrow, where the height of the gauge and its shields needed to be adjusted to avoid drifting and allow ground blizzards to move through under the fencing. Because of the impacts of large-scale atmospheric circulation variations, there have been notable differences in winter temperature anomalies between the two sites, especially in 2003 (Fig. 6). The period is too short to be indicative of interdecadal variations or future trends, but the differences found illustrate the need for more station coverage in the Arctic to better define patterns of climate change with high-quality USCRN instrumentation. Deployment of more USCRN stations to Alaska has begun, and an opportunity to place a CRN station in Tiksi, Siberia, is being pursued by the Global Climate Observing System U.S. Office (Diamond 2009).

Fig. 6.

Winter [December–February (DJF)] mean minimum temperatures for the Fairbanks and Barrow USCRN stations, 2003–09 (°C).

Fig. 6.

Winter [December–February (DJF)] mean minimum temperatures for the Fairbanks and Barrow USCRN stations, 2003–09 (°C).

The engineering of USCRN stations has proven to be robust in a variety of circumstances in the continental United States. Stations have been in the path of high-wind events, such as tropical storms crossing the Gulf of Mexico coast, and many hundreds of individual convective or nonconvective events. Very few interruptions have occurred because of wind alone. The USCRN record for observed 10-s wind gust occurred recently at the station near Whitman, Nebraska. During a squall line passage on 24 May 2010, a 32.46-mps (72.6 mph) peak gust was observed at 1.5 m above the ground. The station continued to operate normally after the wind event peak, with solar panels charging batteries and instruments working well. USCRN stations have also survived encounters with hurricanes coming inland, and provided excellent precipitation records while other networks were disabled. Finally, USCRN sites have continued operating during freezing rain events lasting many hours, with a heating collar around the throat of the weighing bucket gauge keeping the precipitation liquid and measureable.

In cases where electrical power is at a premium, alternate deployment approaches have been developed by the USCRN program. While these approaches have not yet been deployed in the USCRN, they have been used in the U.S. Regional Climate Reference Network (USRCRN) pilot, which also was designed by USCRN engineers. Key among these approaches is to place multiple PRTs in one fan-aspirated solar shield with two redundant fans to ensure continuous operation if a fan fails, yet provide three or more independent temperature measurements with the power requirement of only one fan instead of three. This approach has been used at an experimental meteorological station on the Quelccaya Ice Cap of Peru managed by the University of Massachusetts at 5670-m elevation (D. Hardy 2010, personal communication). Approximately two years of measurements from mid-2007 to mid-2009 yielded mean values within 0.0027°C for three PRTs, with a fourth PRT within 0.025°C of the others. For the two most closely matched PRTs, out of approximately 212 000 5-min values, 98.4% of cases agreed within 0.1°C in an extreme high-elevation environment.

In addition to lessons mentioned above, the USCRN has also responded to issues regarding site selection, power generation, and power storage in extreme environments. Where a USCRN station is located at elevation, an effort is usually made to find sites with an open, flat aspect. Therefore, there is limited experience using USCRN approaches on steep sloped elevated environments. In two of our highest-altitude locations, a raised pedestal is used for the precipitation gauge, and it is surrounded with a double Alter shield rather than a wooden Small Double Fence Intercomparison Reference (SDFIR) shield. Power engineering has improved over the years at solar sites, which are now being equipped with improved solar panels and more cold-weather-tolerant batteries. Where extra power generation is needed, wind generators have been added. However, like any scientific measurement endeavor, the USCRN is still learning and improving, mainly through studying situations and locations where equipment failed and revising the engineering of the USCRN on a continuous basis. There are still issues faced by USCRN with regards to power in environments that are both cold and dark during winter, and we are currently examining further engineering changes that will allow for more reliable observing in these situations. Despite these continuing challenges, we are confident in the validity of the triplicate measurement approaches we use for air temperature, precipitation, and soil moisture and temperature observations, and recommend their adoption for high-elevation climate observation networks.

5. Final remarks

  • The U.S. Climate Reference Network will be monitoring continental U.S. and Alaska climate at least for the next 50 years.

  • The USCRN stations have been designed to work reliably in a wide range of environments, including the Arctic and mountainous regions of the country. Analyses of the observations to date show that the stations have performed well in harsh environments during the early network history.

  • The USCRN model or portions of it are being examined elsewhere in the world. In particular, the U.S. Global Climate Observing System (USGCOS) and USCRN are cooperating with the Meteorological Service of Canada via the U.S.–Canada Reference Station Exchange (Sioux Falls, South Dakota and Edgemont, Ontario), participating in WMO precipitation intercomparison studies in Italy, and working with other groups in international collaborations, such as placing USCRN technology at stations in Russia (Tiksi, Yakutia) and Peru (Quelccaya Ice Cap). In addition, USGCOS provided USCRN communication systems for eight South American Andes climate stations.

  • Hourly and daily USCRN data are currently available as ASCII text files for public access at ftp://ftp.ncdc.noaa.gov/pub/data/uscrn/products/hourly02/ and ftp://ftp.ncdc.noaa.gov/pub/data/uscrn/products/daily01/, respectfully.

  • Experience acquired during the decade of USCRN activity shows that the siting criteria and station design characteristics of this network, especially the triplicate measurements of essential climate variables, can be recommended for automated climate monitoring in the harsh environments of elevated regions worldwide. More information can be found on the USCRN Web site at http://www.ncdc.noaa.gov/crn/.

  • As USCRN observations accumulate, this carefully measured dataset will provide confidence that the variations seen are due to actual climate features, not caused by changes in instrumentation, siting, or land use that plague other networks. This will be one of USCRN’s major contributions to the ongoing worldwide climate change monitoring effort.

Acknowledgments

The authors thank Egg Davis of STG, Inc., for performing some of the data analyses in this paper, Doug Hardy of the University of Massachusetts for sharing his findings regarding USCRN temperature sensor performance on the Quelccaya Ice Cap in Peru, and to the internal and external reviewers of the paper for their helpful comments.

REFERENCES

REFERENCES
Aizen
,
V.
, and
P. Ya.
Groisman
,
2009
:
Northern Eurasia High Mountain Ecosystems NEESPI-HE International Workshop 2009
.
Mountain Research Initiative Newsletter, No. 3, MRI, Bern, Switzerland, 44–45
.
Arndt
,
D. S.
,
M. O.
Baringer
, and
M. R.
Johnson
, Eds.,
2010
:
State of the Climate in 2009
.
Bull. Amer. Meteor. Soc.
,
91
,
S1
S224
.
Barry
,
R. G.
,
2008
:
Mountain Weather and Climate
.
3rd ed. Cambridge University Press, 506 pp
.
Beniston
,
M.
,
H. F.
Diaz
, and
R. S.
Bradley
,
1997
:
Climatic change at high elevation sites: An overview
.
Climatic Change
,
36
,
233
251
.
Brohan
,
P.
,
J. J.
Kennedy
,
I.
Harris
,
S. F. B.
Tett
, and
P. D.
Jones
,
2006
:
Uncertainty estimates in regional and global observed temperature changes: A new data set from 1850
.
J. Geophys. Res.
,
111
,
D12106
,
doi:10.1029/2005JD006548
.
CEOP-HE
,
2010
:
Science plan
.
EvK2CNR, SHARE Project, 41 pp
.
Daly
,
C.
,
D. R.
Conklin
, and
M. H.
Unsworth
,
2010
:
Local atmospheric decoupling in complex topography alters climate change impacts
.
Int. J. Climatol.
,
30
,
1857
1864
.
Diamond
,
H. J.
,
2009
:
The U.S. Global Climate Observing System (GCOS) program: An update on continuing efforts to implement reference climate observation sites
.
Extended Abstracts, 25th Conf. on International Interactive Information and Processing Systems (IIPS) for Meteorology, Oceanography, and Hydrology, Phoenix, AZ, Amer. Meteor. Soc., 3B.7. [Available online at http://ams.confex.com/ams/89annual/techprogram/paper_146525.htm.]
Diaz
,
H. F.
, Ed.,
2003
:
Climate Variability and Change in High Elevation Regions: Past, Present and Future
.
Kluwer Academic Publishers, 282 pp
.
Diaz
,
H. F.
, and
R. S.
Bradley
,
1997
:
Temperature variations during the last century at high elevation sites
.
Climatic Change
,
36
,
253
279
.
Diaz
,
H. F.
,
M.
Grosjean
, and
L.
Graumlich
,
2003
:
Climate variability and change in high elevation regions: Past, present and future
.
Climatic Change
,
59
,
1
4
.
Durand
,
Y.
,
M.
Laternser
,
G.
Giraud
,
P.
Etchevers
,
B.
Lesaffre
, and
L.
Merindol
,
2009
:
Reanalysis of 44 yr of climate in the French Alps (1958–2002): Methodology, model validation, climatology, and trends for air temperature and precipitation
.
J. Appl. Meteor. Climatol.
,
48
,
429
449
.
Fyfe
,
J. C.
, and
G. M.
Flato
,
1999
:
Enhanced climate change and its detection over the Rocky Mountains
.
J. Climate
,
12
,
230
243
.
GCOS
,
2003
:
The second report on the adequacy of the global observing systems for climate in support of the UNFCCC
.
WMO Tech. Doc. 1143, 74 pp
.
Hansen
,
J.
,
R.
Ruedy
,
M.
Sato
, and
K.
Lo
,
2010
:
Global surface temperature change
.
Rev. Geophys.
,
48
,
RG4004
,
doi:10.1029/2010RG000345
.
Huber
,
U. M.
,
H. K. M.
Bugmann
, and
M. A.
Reasoner
,
2005
:
Global Change and Mountain Regions: An Overview of Current Knowledge
.
Springer, 650 pp
.
Lugina
,
K. M.
,
P. Ya.
Groisman
,
K. Ya.
Vinnikov
,
V. V.
Koknaeva
, and
N. A.
Speranskaya
, cited
2007
:
Monthly surface air temperature time series area-averaged over the 30-degree latitudinal belts of the globe
.
Nakicenovic
,
N.
, and
Coauthors
,
2000
:
Special Report on Emissions Scenarios
.
Cambridge University Press, 612 pp
.
NCDC
, cited
2010
:
Global Climate at a Glance
.
[Available online at http://www.ncdc.noaa.gov/gcag/.]
NOAA
,
1998
:
Automated Surface Observing System (ASOS) user’s guide
.
National Oceanic and Atmospheric Administration, 61 pp
.
NRC
,
1999
:
Adequacy of Climate Observing Systems
.
National Academy Press, 66 pp
.
Pepin
,
N. C.
, and
D. J.
Seidel
,
2005
:
A global comparison of surface and free-air temperatures at high elevations
.
J. Geophys. Res.
,
110
,
D03104
,
doi:10.1029/2004JD005047
.
Pepin
,
N. C.
, and
J. D.
Lundquist
,
2008
:
Temperature trends at high elevations: Patterns across the globe
.
Geophys. Res. Lett.
,
35
,
L14701
,
doi:10.1029/2008GL034026
.
Rebetez
,
M.
, and
M.
Reinhard
,
2008
:
Monthly air temperature trends in Switzerland 1901–2000 and 1975–2004
.
Theor. Appl. Climatol.
,
91
,
27
34
.
Seidel
,
D. J.
, and
M.
Free
,
2003
:
Comparison of lower-tropospheric temperature climatologies and trends at low and high elevation radiosonde sites
.
Climatic Change
,
59
,
53
74
.
Solomon
,
S.
,
D.
Qin
,
M.
Manning
,
M.
Marquis
,
K.
Averyt
,
M. M. B.
Tignor
,
H. L.
Miller
Jr.
, and
Z.
Chen
, Eds.,
2007
:
Climate Change 2007: The Physical Science Basis
.
Cambridge University Press, 996 pp
.
Sun
,
B.
, and
T. C.
Peterson
,
2005
:
Estimating temperature normals for USCRN stations
.
Int. J. Climatol.
,
25
,
1809
1817
.