Transcript Document

The Diurnal Cycle of Diabatic Heating and TRMM Precipitation Estimates in West Africa
Adam J. Davis
Department of Atmospheric Science, Colorado State University, Fort Collins, CO
Abstract
Numerous investigations have examined the diurnal cycle of convective activity in West Africa based
exclusively on satellite observations. However, a unique opportunity exists to study this problem using
combined in situ and satellite data thanks to the African Monsoon Multidisciplinary Analysis (AMMA)/NASA
AMMA (NAMMA) field campaign that took place in 2006. In particular, a network of radiosonde launch sites
was set up from June through September 2006, with the most intensive observations collected over parts of
Niger, Nigeria, Benin, Togo, and Ghana. In the present study, composite vertical profiles of diabatic heating
through the diurnal cycle are computed within this region of West Africa, based on the AMMA sounding data.
Then, these heating profiles are placed within the context of precipitation estimates derived from several
TRMM products for the same time period. In particular, the structures and timing of the heating profiles are
compared with precipitation feature information provided by the TRMM database of the University of Utah
Tropical Meteorology Group. This dataset includes precipitation features based on applying thresholds to data
from several different instruments, including the PR, TMI, and VIRS. Differences in the composite diurnal
timing of rainfall as detected by these various types of precipitation features are explored and compared with
the signatures of convective and stratiform precipitation suggested by the observed diabatic heating. Alignment
between the timing of the diabatic profiles and more-processed satellite products, such as 3B42 rain estimates,
is also assessed.
Diabatic Heating Profiles
TRMM 3B42 Precipitation Estimates
• At midnight, mid to upper level heating and
drying overlaying low level cooling are
suggestive of a predominance of stratiform
precipitation systems from 7o to 15o N, with
general cooling over the ocean to the south and
the semi-arid and desert areas to the north
• By 6 AM, the stratiform signal persists, with
some northward expansion, while a more
convective profile develops near the coast and a
shallow convection signature appears over the
ocean
• At noon, a deep convection-type profile
predominates, with the greatest magnitudes near
the coast (around 7o N) and between 10o and 13o
N, corresponding with the high terrain of the Jos
Plateau
• 3B42 precipitation
estimates look to have
a spatially broader rain
distribution and lower
peak magnitudes than
the PR-based PF data
• This could be
associated with less
precise discernment of
rainfall by using IR
data as a proxy,
compared to the PR
• The coarser spatial
resolution (1o vs. 0.25o)
of the PF data and the
grouping of an entire
feature’s rainfall into its centroid’s grid box could also account for the sharper, more intense rainfall peaks in the
PF datasets
• By 6 PM, a convective profile persists near the coast, albeit weaker than at noon, while cooling begins over the ocean and a more
stratiform structure, with northward expansion of heating, emerges farther inland
Dataset Descriptions
Heating Profiles
• Q1 and Q2 profiles were computed using a dataset
derived from 4x-daily radiosonde observations collected
in the summer of 2006, as part of the AMMA/NAMMA
(African Monsoon Multidisciplinary Analyses / NASA
AMMA) field campaign
• This 1o resolution analysis is based solely on the
sounding data in areas of good coverage, supplemented by
the ECMWF special AMMA reanalysis in regions of
sparse observations
• The map at left shows the sounding stations, with the
degree of fill of the circle indicating the completeness of
that station’s record
• The present study focuses on longitudes from 0o to 7o E,
which includes the core of the radiosonde network with
the best spatial and temporal coverage, and the time
period of July 1 through September 30, 2006
TRMM Precipitation
• Precipitation analyses are carried out using the TRMM
precipitation feature database provided by the University of
Utah Tropical Meteorology Group
• Full descriptions of the database can be found in Liu et al.
(2008) and online at trmm.chpc.utah.edu
• As shown in the schematic from Liu et al. (2008),
observations from several TRMM instruments are
collocated, precipitation features (PFs) are defined using
various criteria based on the collocated data, and statistics of
the PFs are computed monthly on 1o grids at 8 time periods
through the diurnal cycle
• This study examines Radar PF rainfall data, based on areas
of surface rain as detected by the TRMM precipitation radar
(PR), and Radar Projection PF rainfall data, based on the
ground projection area of PR reflectivity greater than 20
dBZ, thereby also including thick anvils aloft
• Additionally, TRMM 3B42 precipitation rate data from NASA
Goddard are analyzed
• The 3B42 product uses monthly TRMM calibration
parameters to yield adjusted precipitation estimates from
merged infrared satellite observations 8 times daily at 0.25o
resolution
• All
• Many of the features evident in the
heating profiles above correspond with
the precipitation estimates from PR data
• Around midnight, low to moderate rain
rates exist over much of the 7o to 15o N
band, suggesting stratiform rainfall may
indeed predominate
• A rainfall peak develops later in the
night around 11o N, likely associated with
westward-propagating MCSs that cross
this domain overnight
• Rainfall intensifies near the coast (4o to
6o N) in the morning, with an offshore
peak in late morning, along with
continued rainfall inland
• Precipitation is concentrated between 7o
and 13o N during the afternoon, matching
the convective heating structure at that
• A caveat to consider in interpreting the radar PF rainfall data is that all rainfall associated time, with a notable late afternoon peak
with a single feature is assigned to the grid box where the PF centroid lies
near the coast
• Thus, the actual rainfall distribution may be spatially broader than portrayed by the
• Later in the evening, the greatest rainfall
relatively sharp peaks shown here
occurs from 10o to 16o N, as MCSs begin
to enter the Sahel portion of the domain
• Additionally, differences in comparison with the heating profiles above may be
from the east
associated with the offset in the time periods used by each dataset, as well as the noncontinuous availability of PR data (only available when TRMM satellite passes overhead)
Radar Precipitation Features
Radar Projection Precipitation Features
Liu et al. (2008), Figure 1
heating profile and precipitation plots represent data averaged over longitude
0o to 7o E and the time period from July 1 through September 30, 2006
• Radar projection PF rainfall generally
corresponds well with radar PF values,
suggesting most 20 dBZ echoes aloft in this
region are associated with surface
precipitation
• Radar projection PFs do appear spatially
shifted relative to radar PFs in a few cases,
and radar projection rain estimates are
sometimes greater (e.g. 12o N 12-15 UTC)
• This suggests areas of reflectivity aloft that
do not match with surface rainfall, due to
non-precipitating anvils or low-level
evaporation of precipitation
• The semi-arid Sahel location of the
mentioned point and the heating profile
structure in that area are consistent with
low-level evaporation of falling rain
• The broader 3B42 rainfall distribution is consistent with Liu et al. (2007), who found that only 35-57% of the
area of their VIRS cold cloud features corresponded with PR rainfall, and noted that land areas of central Africa
had a relatively larger proportion of non-raining cold clouds than ocean regions they examined
• The diurnal cycle of the 3B42 estimates is generally consistent with the PF datasets, although the 3B42 rainfall
shows a greater peak at the coast and nearby ocean around noon, the afternoon peak in 3B42 rainfall magnitude
occurs later than in the PF analyses, and the 3B42 data has less of a peak in the Sahel region in the evening than
the PF data
• The later timing of some peaks in the 3B42 estimates compared to the PF rainfall may be associated with the
delay between the maximum intensity of precipitation in convective systems and the subsequent generation and
spreading of the large, cold anvil that is used as a proxy for rainfall by IR methods
• This possible mechanism was suggested by the work of Liu and Zipser (2008), who found that cold clouds
reached a maximum 2 to 3 hours later than surface rainfall over tropical land regions in their study
Discussion and Conclusions
• The correspondence between heating profiles derived from AMMA radiosonde data and the diurnal cycle of
rainfall as diagnosed from TRMM PR precipitation features is generally good, despite caveats with the PF
analysis making interpretation of specific details challenging
• Rainfall estimates from radar projection PFs generally differ little from estimates from radar PFs in this domain,
though some spatial shifting is evident, and a few points in the Sahel region are suggestive of low-level
evaporation of rainfall or non-precipitating anvils, consistent with the heating profiles in that area
• The IR-based TRMM 3B42 rainfall estimates show a broader and lower-magnitude precipitation distribution
compared to the PR-based feature analysis
• While this result agrees with previous findings comparing PR rain area to cold cloud area, differences in
data resolution and methodology preclude a definitive interpretation
• A delay in the phase of peak rainfall exists in the TRMM 3B42 estimates relative to the PR data
• This outcome matches with the determinations of previous studies as well, and is an important factor to
consider when using IR-estimated precipitation data to examine the diurnal cycle
References and Acknowledgments
• Liu, C., E. J. Zipser, and S. W. Nesbitt, 2007: Global distribution of tropical deep convection: Different
perspectives from TRMM infrared and radar data. J. Climate, 20, 489-503.
• Liu, C. and E. J. Zipser, 2008: Diurnal cycles of precipitation, clouds, and lightning in the tropics from 9 years
of TRMM observations. Geophys. Res. Lett., 35, L04819.
• Liu, C., E. J. Zipser, D. J. Cecil, S. W. Nesbitt, and S. Sherwood, 2008: A cloud and precipitation feature
database from nine years of TRMM observations. J. Appl. Meteorol., 47, 2712-2728.
Radar Projection PF Precip. Rate
(Radar Projection PF - Radar PF) Precip. Rate Difference
• The author would like to thank the members of the University of Utah Tropical Meteorology Group for their
work in designing and constructing the TRMM precipitation feature database, as well as for graciously making
this data openly available in the furtherance of scientific inquiry
• This work has been made possible by NASA PMM Grant NNX10AG81G (PI: Richard H. Johnson), a NASA
CEAS graduate fellowship awarded to Adam J. Davis, and the support of Dr. Ramesh Kakar, Dr. Richard Johnson,
Dr. Scott Braun, and Paul Ciesielski.