Comparison of Surface Mountain Climate with Equivalent Free-Air
Parameters Extracted from NCEP/NCAR Reanalysis, Mount Kilimanjaro,
Tanzania Nick Pepin 1 Doug Hardy 2 William Duane 3 Mark Losleben 4
1.Department of Geography, University of Portsmouth, U.K.,
[email protected]@port.ac.uk 2.Glacial
Systems Center, University of Massachusetts, Amherst, MA, U.S.A.,
[email protected]@geo.umass.edu 3.Department of Geography,
University Brunei Darussalam, Brunei,
[email protected]@fass.ubd.edu.bn 4. National Phenology
Network Office, University of Arizona,
[email protected]@email.arizona.edu Introduction
Predicting and understanding future changes in climate in
mountainous regions requires an understanding of how the mountain
surface may become decoupled from the free atmosphere, along with
surface feedback effects (Bradley et al. 2006). Most GCMs predict
future changes in climate based on simulation of the physics of the
free atmospheric response to changes in greenhouse gases.
Unfortunately surface complexities in mountain regions, and poor
grid resolution make it difficult to downscale such predictions in
mountain areas, and temperature trends in mountain areas are
different to those in the free atmosphere (Pepin & Seidel
2005). This study examines the case study of Kilimanjaro, in East
Africa, the largest free-standing mountain in the world. Surface
temperatures and vapour pressures, as measured by loggers installed
on the south-west slope of the mountain (Duane et al. 2008) are
compared with free-air climate measured at the same elevation.
Unsurprisingly the mountain modifies the climate considerably,
creating its own distinctive atmosphere, and responding strongly to
the intense radiation input at this equatorial location. This means
that the surface regime becomes decoupled from the free atmosphere,
especially above treeline. Some consequences for the mass balance
of the summit ice fields are discussed. Surface Temperature and
Humidity Data Surface data consists of hourly observations recorded
from September 2004 until October 2007 at ten locations on the
south-western slope of Kilimanjaro (Table 1). Ten Hobo loggers were
programmed to record air temperature and relative humidity every
hour, installed in home made radiation shields on tripods, 1.5
metres above ground level. Shields were orientated horizontally,
with the long-axis orientated north-south, to minimize solar
heating. Conclusions and Implications for Ice Sheet Mass Balance
The main findings of our study are listed below: a)Below 2500 m,
despite the existence of tropical rainforest, at least some of the
variability in surface and free-air temperature anomalies is
correlated, and a common response to synoptic forcing. Above
treeline however, strong localised patterns of surface heating and
cooling decouple Kilimanjaro mountain climate from that of the free
atmosphere, almost every day and irrespective of synoptic scale
forcing. b)There is a strong thermally-induced circulation on
Kilimanjaro, importing moisture upslope during daylight hours, well
above that available in the free atmosphere. Thus daytime
temperature differences of over 5 deg C (compared with the free
atmosphere) and vapor pressure excesses in excess of 5 mb are
reported. Although these excesses decline with elevation, they are
still present at the summit of the mountain. c)Fluctuations in
upper air flow strength and direction (at 500 mb) appear to have
limited influence on the strength of surface heating and upslope
moisture advection, which is dominated by the diurnal radiative
cycle rather than inter-diurnal synoptic controls. Taken together
these findings suggest that local changes in land-use such as
deforestation (Hemp 2005) could have a strong influence on the
mountain climate and summit ice of Kilimanjaro, making it somewhat
decoupled from larger scale advective changes. In addition, local
observations are crucial since interpolation from free atmospheric
models is problematic. References Bradley, R.S., Vuille, M., Diaz,
H.F. & Vergara, W. (2006), Science 312, 1755-1756. Cullen,
N.J., Mlg, T., Kaser, G., Hussein, K., Steffen, K. & Hardy,
D.R., (2006). Geophys. Res. Lett., 33, L16502,
doi:10.1029/2006GL027084 Duane, W.J., Pepin, N.C., Losleben. M.L.
& Hardy, D. (2008). AAAR, In press. Hemp, A. (2005). Global
Change Biology 11, 1013-1023. Kaser, G., Hardy, D.R., Mlg, T.,
Bradley, R.S. & Hyera, T.M. (2004). Int. Jnl Clim. 24, 329-339.
Kistler, R. et al. (2001). Bull. Amer. Met. Soc. 82, 247-268.
Pepin, N.C. & Seidel, D.J. (2005). J. Geophys. Res.110, D03104:
doi:10.1029/2004JD005047. Mlg, T., Hardy, D.R. & Kaser, G.
(2003). J. Geophys. Res.108, D23,4731: doi:10.1029/2003JD003546.
Mlg, T. & Hardy, D. R. (2004). J. Geophys. Res. 109, D16104,
doi:10.1029/2003JD004338. Thompson, L. et al. (2002), Science 298,
591-593. Troll, C. & K. Wien, (1949). Geog. Ann. 31, 257-274.
NCEP/NCAR Reanalysis (Free-Air) Six hourly air temperatures,
relative humidities, and u and v wind components were interpolated
from the NCEP/NCAR reanalysis version 1 available at CDC
(ftp.cdc.noaa.gov) (Kistler et al. 2001) to exactly the same
locations as the sensors in the field. Since surface data are not
in the reanalysis model, this represents free air conditions. The
reanalysis is derived four times daily, corresponding to 0300,
0900, 1500 and 2100 LST (or East African Time). Graphs below show
the time series of temperature, humidity and u (zonal) and v
(meridional) wind components interpolated to site 1 (1890 m) (left
hand panel) and site 10 (5803 m) (right hand panel) respectively,
using daily data at 1500 EAT for 2004 -2007.ftp.cdc.noaa.gov Table
1: Elevation and Site Characteristics of the 10 stations Figure 1.
Distribution of Hobo Logger Stations, SW slope of Kilimanjaro The
climate sensors range in elevation from 1890 m to 5803 m above
sea-level and are representative of a range of ecological
environments. The lower slopes (sensors 1-3) are covered by
rainforest, with relatively high precipitation and humidity. The
trees give way to giant heathers around the treeline (~3500 m) as
represented by site 4. The heathland becomes drier and more
sparsely vegetated as elevation increases, becoming a semi-desert
landscape by site 7 (~4570 m). Above 5000 m there are areas of
permanent snow and ice cover and the summit site (5803m) in the
crater of the volcano is installed on the northern ice field.
Logger No.ElevationSite description 11890mDense montane rainforest
22340mDense montane rainforest 32760mSparse montane rainforest
43170mTransitional zone between rainforest and sub-alpine heathland
53630mSub-alpine heathland 64050mAlpine with limited vegetation
74570mAlpine with limited vegetation 84970mBare rock 95470mBare
rock 105800mIce field Despite some missing data, due to vandalism
and/or logger malfunctions, we retrieved some data at all
locations, and continuous records (nearly 12,000 hourly
observations) have been obtained at sites 4, 8 and 10 for the 3
year period (see Duane et al. 2008 for a discussion of the surface
data). There are clear differences in the free air behaviour
between the top and the bottom of the mountain. In particular
seasonal signatures change with elevation. In the following
discussion, because Kilimanjaro is in the southern hemisphere,
summer refers to January/February and winter July/August.
Temperatures show the most obvious contrast, and are universally
warmer in summer, although the annual cycle is subdued at higher
elevations, meaning steeper lapse rates during this season.
Relative humidities and moisture availability decrease with
elevation, particularly in winter, when extremely dry air is common
at high elevations. Changes in mean zonal and meridional wind are
complex. On average free-air flow has an easterly component, but
this becomes stronger at low elevations in summer, and high
elevations in winter. Meridional flow on the other hand has a
strong seasonal cycle at low elevations, indicative of the
northerly (winter) and southerly (summer) monsoons. This becomes
much less distinct at high elevations. The graphs above show annual
variation (January-December) of mean surface/free-air temperature
differences and thus indicate the extent to which the mountain acts
as a heat source or heat sink in comparison with the free
atmosphere. Because of the tropical latitude, there is a strong
diurnal contrast, with mostly positive differences (surface warmer
than the free atmosphere) during the day (1500 EAT) and negative
differences (surface colder) at night (0300 EAT). However daytime
heating is concentrated above site 3. Up to this elevation the
forest absorbs most of the solar energy, preventing the surface
from heating. Negative differences remain during the day,
especially in the southern winter (July/August) when low cloud is
common. In contrast above treeline, afternoon heating is dramatic,
particularly strong at mid-elevations (sites 4 and 5). This is
expected to be where montane thermal circulation (upslope flow) is
strongest. The seasonal contrast in heating, with largest
temperature excesses during the south-eastern monsoon in
July/August (at all sites from 4 to 9) is marked. At first glance
this suggests that the thermal circulation maybe more developed
during this season, but consideration of moisture transport (see
below) paints a different picture. Heating is very small at site
10, over the ice field, because of the increased latent heat flux
at this location. At night all sites show a cooling below the free
atmosphere but again this effect is subdued at the lowest sites,
protected from radiative loss by the forest canopy, and perhaps
surprisingly at sites 8 and 9, although sky view factors are
reduced somewhat here on the side of the steep volcanic cone. Such
variations in the temperature contrast show that the mountain
creates its own thermal circulation pattern (Troll & Wien
1949). This has implications for vapour pressure excesses reported
at the surface (graphs to the right). At all elevations there is on
average greater vapour pressure at the mountain surface than in the
free atmosphere, with surface vegetation providing an importance
source through evapotranspiration. Thus mean vapour pressure
excesses decrease with elevation. The northern ice field (site 10)
has very little influence on atmospheric moisture, the largest
excesses being only 1 mb. There is also a large diurnal increase in
vapour pressure excess at all sites, illustrating the importance of
the thermal circulation in transporting moisture to the upper
mountain. This is particularly effective in the wet seasons but
strangely less so in July/August despite the largest mountain
temperature excesses being reported then. In summary much of the
atmospheric moisture at the highest elevations is likely to have
originated from the lower forest zone and not the free atmosphere.
Surface/Free-Air Temperature Contrasts Surface/Free-Air Vapour
Pressure Contrasts Influence of Upper Level Flow Regime The summit
ice fields of Kilimanjaro are undergoing rapid shrinkage (Cullen et
al. 2006) and if current rates of decrease continue (a big
assumption) the summit snows may disappear by 2020 (Thompson et al.
2002). There is debate as to the relative contributions of large
scale global forcing and local/regional drying (which may or may
not be related to deforestation on the lower slopes) (Hemp 2005,
Kaser et al. 2004, Molg & Hardy 2004, Molg et al. 2003). Our
comparison of surface climate with free atmospheric properties
shows that the mountain creates its own atmosphere, with enhanced
moisture availability and thermal heating by day giving way to
cooling at night. This circulation transports moisture from the
lower slopes to the vicinity of the glacier, ultimately controlling
ice accumulation in the form of precipitation. Surface versus
Free-Air Coupling The relationships between surface/free-air
temperature and moisture contrasts at the top of the mountain (site
10) and the free-air (synoptic scale) forcing are weak. Although
large excess surface heating and vapour pressure excesses cannot
occur with high gradient wind speeds (see top two graphs), low
speeds do not necessarily mean such differences develop. Perhaps
surprisingly, upper air flow direction (bottom two graphs) has
relatively little influence on the extent to which the mountain
decouples itself from the free- atmosphere. The graphs above show
the correlations between simultaneous surface and free-air
temperatures at sites 1, 4, 7 and 10. As elevation increases
decoupling intensifies, with the mountain regime being increasingly
independent of that in the free- atmosphere. The major change takes
place above treeline where local heating/cooling is induced more by
the strong equatorial sun and less by advection.