Solar Effects on Rainfall at Lake Victoria, East Africa

SOLAR EFFECTS ON RAINFALL AT
LAKE VICTORIA, EAST AFRICA.
J. Curt Stager, Natural Resources, Paul Smith's College
LAKE VICTORIA
World's second largest lake
(ca. 69,000 sq.km)
Relatively shallow (ca. 80 m)
Outflow into the Nile.
Most inputs from rainfall,
most losses by evaporation:
like a huge rain gauge.
"LONG RAINS" (MAM) & "SHORT RAINS" (OND).
SUNSPOTS AND THE
LEVELS OF LAKE
VICTORIA
From 1896 to 1927 AD, a strong
(0.87) correlation existed
between SSN and the levels of
Lake Victoria.
Victoria became a classic
example of sun-rainfall effects.
After 1927, SSN-lake
correlations decayed or
reversed.
This is now most often
described as a classic
case of
false correlations.
A common explanation
is that the record
examined was originally
too short...
AN UPDATED RECORD
1. Since 1967, TSI-lake CC's = 0.53. TSI-lake CC's were
significant (0.62) for most of the century (same for SSN).
2. Every sunspot peak coincided with a lake level rise; but
the pulse timing fits > relative magnitude.
NOT JUST LIMITED TO LAKE VICTORIA
Other lakes here have similar decade-scale pulses.
10-13 yr lacustrine pulses in Holocene varves (Kenya)
and in the Nile flood minimum record (lake outflow).
"Accepted" rain-triggers for this region:
ENSO = heavier OND "short" rains...but not always (0.35).
Indian Ocean Dipole = some with ENSO, better fit (0.58).
DOES INSTABILITY NEGATE AN ASSOCIATION?
Linkages between tropical rainfall, ENSO, and IOD are
unstable, shutting on & off; as does the "solar association."
http://www.jamstec.go.jp/frsgc/jp/press/yamagata/011129/eng/index_e.html
POSSIBLE MECHANISMS
1. If higher TSI increases tropical SST's:
a. Western Indian Ocean warming is associated with
enhanced East African rainfall
(Black et al., Nicholson et al., Mistry & Conway, Linthicum et al.).
b. Tropical Indian Ocean SST's track solar cycles
(W. White et al., JGR 1997)
2. Warm-wet, cool-dry pattern dominated late Quaternary.
WARM = WET,
and
COOL = DRY
A dense paleosol
underlies early
Holocene sediments,
even in mid-lake.
This core from Damba
Channel, Uganda,
contains a shell deposit
at the paleosol contact.
(Stager, 1984)
DIATOMS
Calendar years before 2000 A.D.
Climate cycles present
in the GISP2 ice core
related to NH insolation
and solar variability.
8000
9000
10000 11000 12000 13000 14000 15000 16000 17000 18000 19000 20000
0
a
20
40
(ppb)
GISP2 sodium
b
60
21,000 yr
11,400 yr
2200 yr
GISP2 bpc's
1460 yr
stronger
insolation?
950 yr
c
higher
GISP2 CT series
OD
d
OLD
weaker
insolation?
H-1
relative lake level
They also occur in
monsoon records from
the Arabian Sea.
YD
Victoria CA2
When added together,
they produce irregular
pulses that track the
levels of Lake Victoria.
lower
dry
8000
9000
dry?
10000 11000 12000 13000 14000 15000 16000 17000 18000 19000 20000
Calendar years before 2000 A.D.
(Stager et al., 2002)
If higher TSI increases rainfall, then RVF outbreaks
might be predictable years in advance, simply by
tracking the sunspot cycle...
TOO GOOD TO BE TRUE?
TAKING A CLOSER LOOK:
How does the 20th century sun-lake association relate
to sun-RAINFALL relationships?
Rainfall pulses are abrupt, shut on/off with ITCZ passage.
Lake levels change gradually, smoothing and lagging
rainfall by ca. 1 year.
Rainfall-lake relations are imperfect (0.4 overall):
a. Rain pulses 1910-1924 not reflected in lake levels
b. Lake rise after 1960 >> rain increase
c. Long rains have > consistent lake signal than short rains
Early century: highest SSN-lake CC's (0.8).
0 lag: RAIN v. LEVEL = 0.16; RAIN vs. SSN = 0.21
-1 yr lag: RAIN v. LEVEL = 0.50; and RAIN vs. SSN = 0.34
= SIG.DIFF.
A SEASONAL SIGNAL:
Linked to ITCZ system?
LONG RAINS
slightly wetter in rising
phases of solar cycle.
SHORT RAINS
slightly wetter in closing
phases of solar cycle.
No significant correlation between annual rainfall and
sunspots/TSI, with 0 or +1 yr lags.
With 3-yr smoothing, the highest CC is 1927-67 (0.25).
CC's rise to levels
resembling those
with ENSO when
annual rainfall
precedes TSI by -1
year.
1901-1930 = 0.36
(NINO4 = 0.32)
African rains cause sunspots?
1901-1952 = 0.21
Fortuitous?
(NINO4 = 0.27)
OR...Rainfall increases most
during early phases of solar cycles?
1964-1998 = 0.25
(NINO4 = 0.20)
Rising phase of each sunspot peak coincided with a
large rainfall pulse.
Closing phases also had rain pulses in mid-century;
relative sizes gradually reversed, cancelling CC's then.
Lake pulses fit solar peaks better than the rain series does.
SSN series shifted back 1-2 years:
SSN peaks overlay rain pulses, as with lake series.
Alternating 5-6 yr rain pulses weaken the CC's with SSN.
Is the "solar signal" just hidden by ENSO, or coincidental,
or does the solar cycle entrain the ENSO rhythm?
Lake level (5,8,9,12,16 yrs)
Smoothed rainfall (5,6.5,8,9,12,16 yrs)
Monthly SSN (5,8,10-11 yrs)
NINO4 index (5.5,9-10,13 yrs)
LINGERING QUESTIONS:
1. Why are sun-rainfall and ENSO/IOD-rainfall
associations so unstable and non-linear?
2. Does the solar cycle interact with ENSO/IOD in
setting the pace of tropical rainfall pulses?
3. Why a 12-yr pulse in rains and lake levels?
(Solar = 10-11 yrs, ENSO/IOD = 13 yrs)
and
4. Why the recent rise in Victoria levels?
TSI, SST's, and rains increased in late 20th century.
Smoothed rain vs. SSN = 0.92, with 8 year lag.
WHAT CAN WE CONCLUDE?
1. Victoria lake-sun linkages are visually appealing,
statistically significant, and plausible.
2. They are also unstable (as are links between lake
levels and ENSO/IOD, as well as levels and rainfall
series).
3. Most major ENSO/IOD-driven rain pulses occur on
either side of SSN maxima, not on the peaks.
5. The 12-year periodicity in rain/lake is close to but not
identical to SSN or ENSO/IOD periodicities.
CONFUSING ENOUGH YET?
Pilkington Bay core P2K-1; shallow-water diatoms
(SWD), amoeba remains, and inferred conductivity.
WARM SS PEAKS NOW:
WET
WARM E. HOLOCENE:
WET
COOL GLACIAL:
DRY
COOL LIA:
WET!?
Can there be more than one
mode of sun-rainfall
interactions here?
HYPOTHESIS
Could a weaker sun reduce
the range over which the
ITCZ migrates?
This might increase the
duration of rainy seasons over
East Africa (COOL = WET).
http://www.cla.sc.edu/geog/faculty/carbone/modules/mods4car/africa-itcz
/
SOLAR VARIABILITY vs. ORBITAL CYCLES
If this sun-rainfall linkage is real, then the LIA
differed fundamentally from "true" ice ages, as
well as from today.
"Real" ice ages are related to long-term cooling of
alternating hemispheres (by orbital cycles).
Wouldn't solar weakening during the LIA encourage
symmetrical cooling of both hemispheres?
BUT if minima caused
wetting in LIA, why
would rainfall increase
during solar maxima
NOW?
Teleconnections between
East Africa and higher
latitudes switched phasing
at the close of the LIA.
If these sun-rain links are
real, then they are
localized and UNSTABLE!
ACKNOWLEDGEMENTS
National Science Foundation
R. Anyah, D. Conway, J. Lean, P. Mayewski,
L.D. Meeker, R. Sutcliffe
Paul Smith's College
Climate Change Institute, U. Maine, Orono