To fully utilize the water resources of the basin, several dams were
built in the previous century to control the seasonal and interannual
variability of the Nile flow. The recent conflict over the Nile water
has received significant attention in the past few years after the
decision by Ethiopia to build a large dam on the Blue Nile (the Grand
Ethiopian Renaissance Dam, or GERD) to produce electricity,
mostly for export to neighbouring countries. The dam, currently
under
is relatively
compared
to previous designs
In
the construction,
format provided
by thelarge
authors
and unedited.
for the same location, which raised serious concerns regarding its
effect on water shares of downstream countries (that is, Egypt and
Sudan). If variability of the Nile flow changes in the future, then
water storage capacity in the basin will need to be re-evaluated.
Until recently, attempts to project the future of the Nile flow
yielded inconsistent results. Although several studies examined the
impacts of climate change on the Nile basin using different approaches10–18 , the uncertainty surrounding conclusions from these
studies was high for several reasons. First, none of the previous studies presented observational evidence to support their hypotheses, as
they estimatedS.
theSiam
impactsand
of climate
change
the flow of Nile
Mohamed
Elfatih
A. B.onEltahir
5° N
0−500
500−1,000
1,000−1,500
1,500−2,000
2,000−2,500
SUPPLEMENTARY INFORMATION
LETTERS
0°
DOI: 10.1038/NCLIMATE3273
2,500−3,000
PUBLISHED
ONLINE: 24 APRIL 2017 | DOI: 10.1038/NCLIMATE3273
3,000−3,500
3,500−4,000
0
200 400 600 800
Kilometres
5° S
Climate change enhances interannual variability
of the Nile river flow
*
25° E
30° E
35° E
Figure 1 | Topographic map of Eastern Africa and the Nile sub-basins.
Topographic map of Eastern Africa showing the main tributaries of Nile
basins (Upper Blue Nile, Sobat, Atbara and Bahr el-Jebel) and different
dams in these basins. The rainfall and runoff data analysed in this paper are
averaged over the Upper Blue Nile, Sobat and Atbara basins.
d
Re
The human population living in the Nile basin countries is
Mediterranean Sea
projected
to double
by Department
2050, approaching
one billion1 .Engineering,
The
Ralph M. Parsons
Laboratory,
of Civil and Environmental
Massachusetts Institute of Technology, Cambridge,
increase in water
demand
associated
with this burgeoning
[email protected]
Massachusetts
02139, USA.
*e-mail:
N
population will put significant stress on the available water
resources. Potential changes in the flow of the Nile River as a
30° N
NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange
1
result of climate change may further strain this critical situa2,3
© 2017 Macmillan
Publishers Limited, part of Springer Nature. All rights reserved.
tion . Here, we present empirical evidence
from observations
and consistent projections from climate model simulations
suggesting that the standard deviation describing interannual
variability of total Nile flow could increase by 50% (±35%)
25° N
(multi-model ensemble mean ±1 standard deviation) in the
twenty-first century compared to the twentieth century. We
Aswan Dam
attribute the relatively large change in interannual variability
of the Nile flow to projected increases in future occurrences of
Dongola
El Niño and La Niña events4,5 and to observed teleconnection
Merowe Dam
20° N
between the El Niño–Southern Oscillation and Nile River
Atbara
flow6,7 . Adequacy of current water storage capacity and plans
for additional storage capacity in the basin will need to be
Lower Blue Nile
re-evaluated given the projected enhancement of interannual
Khashm El Girba Dam
variability in the future flow of the Nile river.
Sennar Dam
15° N
The Nile river basin is an ecosystem under severe stress. The
Tekeze Dam
basin is shared by about 400 million people in eleven countries
Roseiras Dam
with economies that depend heavily on agriculture, which employs
1
GERD Dam
the vast majority of the labour force in most of these countries .
Furthermore, almost half of the Nile basin countries are projected to
Upper Blue Nile
10° N
live below the water scarcity level, 1,000 m3 /person/year, by 20308,9 .
Bahr
el-Jebel
Thus, any future changes in the magnitude of the flow volume of
the Nile river can lead to significant impacts on the lives of people
Sobat
Fincha Dam
living within the basin and may increase the already high level of
water stress.
To fully utilize the water resources of the basin, several dams were
5° N
built in the previous century to control the seasonal and interannual
0−500
variability of the Nile flow. The recent conflict over the Nile water
500−1,000
has received significant attention in the past few years after the
1,000−1,500
decision by Ethiopia to build a large dam on the Blue Nile (the Grand
1,500−2,000
Ethiopian Renaissance Dam, or GERD) to produce electricity,
0°
2,000−2,500
mostly for export to neighbouring countries. The dam, currently
2,500−3,000
under construction, is relatively large compared to previous designs
3,000−3,500
for the same location, which raised serious concerns regarding its
3,500−4,000
effect on water shares of downstream countries (that is, Egypt and
Sudan). If variability of the Nile flow changes in the future, then
0
200 400 600 800
5° S
Kilometres
water storage capacity in the basin will need to be re-evaluated.
Until recently, attempts to project the future of the Nile flow
25° E
30° E
35° E
yielded inconsistent results. Although several studies examined the
impacts of climate change on the Nile basin using different ap- Figure 1 | Topographic map of Eastern Africa and the Nile sub-basins.
proaches10–18 , the uncertainty surrounding conclusions from these Topographic map of Eastern Africa showing the main tributaries of Nile
studies was high for several reasons. First, none of the previous stud- basins (Upper Blue Nile, Sobat, Atbara and Bahr el-Jebel) and different
ies presented observational evidence to support their hypotheses, as dams in these basins. The rainfall and runoff data analysed in this paper are
they estimated the impacts of climate change on the flow of Nile averaged over the Upper Blue Nile, Sobat and Atbara basins.
a
Se
Ralph M. Parsons Laboratory, Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge,
NATURE
CLIMATE02139,
CHANGE
| www.nature.com/natureclimatechange
Massachusetts
USA.
*e-mail: [email protected]
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
1
25
26
The Supplementary Information document includes the following sections:
27
28
1- Description of observed flow and rainfall data.
29
2- Selection method of climate models.
30
3- Description of the teleconnection between El Niño Southern Oscillation and the variability in
31
the flow of Nile river.
32
4- Sensitivity of the projected future changes in the flow of the Nile river to the choice of GCMs
33
34
35
36
37
38
39
40
41
42
43
44
2
45
1- Observed flow and rainfall data
46
47
In this paper, we focus on the data of the Eastern Nile sub-basins (Upper Blue Nile, Sobat and
48
Atbara) as they contribute by 80% of the total Nile flow at Dongola, hence, responsible of the
49
seasonality and inter-annual variability in the total flow of the main Nile. Furthermore, an
50
analysis (Supplementary Fig. 12) of the correlation between the flow at Dongola and Blue Nile
51
showed that the Blue Nile explains almost 80% of the variability of the flow at Dongola. The
52
observed flow data for Dongola was available through personal communications with the
53
ministries of water in the countries of the Nile basin and Global River Discharge Database
54
(GRDD)1. The data from the ministries were verified against the flow from the GRDD at Aswan
55
from 1870 till 1960 (before the construction of High Aswan Dam) and at Dongola from 1912
56
till 1984. The data were in good agreement between the different sources. It is also important
57
to note that the flow from 1870 till 1900 was relatively higher compared to the well-documented
58
long-term mean of the Nile flow ~84 km3/year. Thus, we focus in our analysis using the bias
59
correction and sampling approaches on the flow in Dongola starting 1900. The observed flow
60
data for the Upper Blue Nile and Atbara basins was available through personal communications
61
with the ministries of water in the countries of the Nile basin. The authors could not access
62
similar data for the flow of the Sobat basin. Several rainfall stations are also used to estimate
63
the changes in trend of the rainfall over the Eastern Nile basin. However, only a few stations
64
were available over the Upper Blue Nile basin with sufficient data that overlaps with the flow
65
data. These stations are (Addis Ababa, Debre Markos, Dessie, Gondar and Nekemet). The
66
rainfall data were available using the Global Historical Climatology Network (GHCN-V2). The
67
data of the different rainfall stations was averaged by using a multiple linear regression of the
3
68
different rainfall data for each station as the predictors and the observed flow of the Upper Blue
69
Nile basin as predictand. The analysis was repeated using Thiessen polygons approach and the
70
same changes in mean and variability were found.
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
4
91
2- Selection method of climate models.
92
In this study, 18 GCMs that participated in the Coupled Model Inter-comparison Project Phase
93
5 (CMIP5) with the Representative Concentration Pathway (RCP 8.5) for future greenhouse
94
gases projected emissions were selected2 (Supplementary Table 1). The selected models were
95
identified in previous studies by other research groups to be the best in simulating the frequency
96
of El Niño and La Niña events as they have a skewness of rainfall over the Nino3 region greater
97
than 1 and were able to simulate at least one event of extreme El Niño3,4. ENSO is one of the
98
most important phenomena shaping tropical climate. Hence, models used to predict the future
99
climate in any tropical region should have credibility in simulating this phenomenon and also
100
the observed teleconnections with ENSO5. Moreover, we tested the sensitivity of our results
101
using different combinations of these GCMs as presented in Section 4 of this document.
102
103
Supplementary Figure 1 shows the changes in the seasonal cycle of the precipitation and runoff
104
using the 18 GCMs and compared to observed precipitation from TRMM V7-3B43 and CRU
105
TS 3.1, and stream flow for the Upper Blue Nile, Sobat and Atbara basins. Supplementary Table
106
4 summarizes the changes in the mean, standard deviation and coefficient of variation of the
107
runoff from the 18 GCMs.
108
109
110
111
112
113
5
114
115
116
117
Supplementary Table 1 | Summary of Global Climate Models (GCMs) used in this study
Models
AGCM (Lon. x Lat.)
BCC-CSM1-1M
1.12 o × 1.12o
BCC-CSM1-1
CCSM4
CMCC-CMs
CMCC-CM
CNRM-CM5
GFDL-CM3
GFDL-ESM2G
GFDL-ESM2M
MPI-ESM-MR
MRI-CGCM3
NORESM1-ME
NORESM1-M
CMCC-CESM
CESM1-BGC
CESM1-CAM5
2.8 o × 2.8 o
1.25 o × 0.94 o
1.88 o × 1.87 o
0.75 o × 0.75 o
1.4 o × 1.4 o
2.5 o × 2.0 o
2.5 o × 2.0 o
2.5 o × 2.0 o
1.88 o × 1.87 o
1.1 o × 1.1 o
2.5 o × 1.9 o
2.5 o × 1.9 o
3.75 o × 3.71 o
1.2 5 o × 0.94 o
1.25 o × 0.94 o
ACCESS1
1.25 o x1.87 o
FGOALS
2.8 o × 2.8 o
Agency
Beijing Climate Center, China Meteorological
Administration
Beijing Climate Center, China Meteorological
Administration
National Center of Atmospheric Research, USA
Centro Euro-Mediterraneo per I Cambiamenti Climatici
Centro Euro-Mediterraneo per I Cambiamenti Climatici
National Centre of Meteorological Research, France
NOAA Geophysical Fluid Dynamics Laboratory, USA
NOAA Geophysical Fluid Dynamics Laboratory, USA
NOAA Geophysical Fluid Dynamics Laboratory, USA
Max Planck Institute for Meteorology, Germany
Meteorological Research Institute, Japan
Norwegian Climate Center, Norway
Norwegian Climate Center, Norway
Centro Euro-Mediterraneo per I Cambiamenti Climatici
Community Earth System Model Contributors
Community Earth System Model Contributors
Commonwealth Scientific and Industrial Research
Organization/ Bureau of Meteorology (CSIRO-BOM)
LASG, Institute of Atmospheric Physics, Chinese Academy
of Sciences
118
119
120
121
122
123
124
125
126
127
128
6
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
168
169
170
171
172
173
174
175
b
a
Supplementary Figure 1 | Changes in the seasonal cycles of precipitation and runoff. a, b,
Changes in the seasonal cycle of precipitation and runoff over the Eastern Nile basin, where the
blue and red lines are the averages of the periods (1900-2000) and (2000-2100), respectively, while
in (a) the green and brown lines are the average observed precipitation seasonal cycle from TRMM
V7-3B43 for the period (1998-2014) and CRU TS3.1 for the period (1900-2000), respectively.
The green line in (b) is the observed seasonal cycle of the streamflow of the Upper Blue Nile,
Sobat and Atbara basins. In a,b, the error bars are ±1 standard deviation of the rainfall and runoff
for each month. In a,b, the numbers in blue and red are the annual averages for the periods (19002000) and (2000-2100), respectively, and the green and brown numbers are the annual average of
precipitation for the observation from TRMM V7-3B43 and CRU TS 3.1, respectively in (a). The
green number in (b) is the annual average observed streamflow for the Upper Blue Nile, Sobat and
Atbara basins.
154
155
156
157
158
159
160
161
162
163
164
165
166
167
Supplementary Figure 2 | Changes in the mean, standard deviation and coefficient of
variation for the runoff simulated by 18 CMIP5 GCMs over the Eastern Nile basin. Percent
changes in the mean, standard deviation and coefficient of variation for the period (2000-2100)
relative to the period (1900-2000) showing the number of models that fall in each group of these
percent changes.
7
176
3- El Niño Southern Oscillation and the variability in the flow of Nile river.
177
During El Niño events, sea surface temperature (SST) in the tropical Indian Ocean increases in
178
response to the warming in the tropical Eastern Pacific Ocean through oceanic connection and
179
forces a Gill-type circulation, i.e., enhanced westerly, low-level air flow over East Africa and the
180
Indian Ocean. This anomalous low-level flow enhances the flux of air and moisture away from the
181
Nile basin, resulting in a reduction of rainfall and river flow6. This mechanism is somewhat similar
182
to mechanism discussed in previous studies about the westward extension of the warm pool over
183
the Indian ocean leading to extension of the Walker circulation and drying over Eastern Africa7.
184
In comparing the Nile basin to other tropical regions, we note that the correlation between the Nile
185
flow and ENSO indices is among the highest observed correlations between ENSO indices and
186
rainfall/flow over other regions in the world as shown in Supplementary Table 10.
187
188
The longest observational record on the Nile describes the flow at Aswan for the period 1870-2012
189
is presented in Supplementary Figure 3. However, the potential for using this record to study long
190
term trends in the mean flow is limited due to three factors: (a) The rating curve calibration used
191
for generating the flow before 1902 was not accurate resulting in discharges that were about 8%
192
higher8,9; (b) The construction of the Low Aswan Dam in 1902, and the High Aswan Dam in 1960s
193
impacted the location of the station resulting in non-homogeneity of the record, which eventually
194
made it necessary to substitute the flow in Dongola for the flow at Aswan after 1960; and (c) The
195
increasing rate of water abstraction upstream due to expansion of irrigation in Sudan. The current
196
estimates of this abstraction rate range from 12 to 14 cubic kilometer per year10. All these factors
197
taken together limit the potential for using this record to detect long term changes in the mean Nile
198
flow. Instead, we focus mainly on the shorter but homogeneous record presented in Figure 2. We
8
199
assume that the Aswan record can be used to describe approximately the variability of the flow
200
around the slowly varying mean, as we did in Figure 3c, since we use a different mean for assessing
201
variability within a moving window of 30 years.
202
203
The time series of the total annual flow at Dongola (green line) is characterized by two periods
204
(i.e. around 1920 and after 1980) of high inter-annual variability as illustrated by the 30 years
205
moving average of the coefficient of variation (blue line) (Supplementary Figure 3). These periods
206
also coincide with periods of high ENSO activity (shaded in gray) (i.e. high number of El Nino
207
and La Nina events). During these two periods, the number of El Nino and La Nina was more than
208
14 events per 30 years period and the coefficient of variation was around 25%. On the other hand,
209
the period (1940 to 1970) had a low number of El Nino and La Nina events (i.e. around 11 events
210
per 30 years period) and low inter-annual variability as illustrated by a low coefficient of variation
211
less than 15%. The comparison between periods of low and high ENSO activity shows that a slight
212
increase in the number of El Nino and La Nina events can enhance the inter-annual variability of
213
the total Nile flow. Furthermore, the presented analysis based on observations of SST and Nile
214
flow shows that the coefficient of variation of the total Nile flow can increase by approximately
215
67% (from 15% to 25%) with a slight increase of the number of El Nino and La Nina events from
216
around 11 events to 14 events, which has implications of the future inter-annual variability of the
217
flow in the Nile river if the number of future El Nino and La Nina events changes.
218
219
The GCMs simulate the frequency of El Niño, La Niña and neutral events for the twentieth century
220
with some differences but are generally in reasonable agreement with observations
221
(Supplementary Fig. 4a). There is an agreement between the GCM results presented in
9
222
Supplementary Fig. 4a that the frequency of extreme and moderate El Niño and extreme La Niña
223
events may increase in the 21st century. These increases are associated with a decrease in the
224
frequency of neutral years (Supplementary Fig. 4a), which are characterized by average conditions
225
of SST, convective activity and rainfall over the Pacific Ocean3, 4. These changes in the frequencies
226
of El Niño and La Niña events would change the Nile flow patterns through the observed
227
teleconnection between ENSO and the Nile, in which El Niño and La Niña events induce low and
228
high flows in the river, respectively. The teleconnection is evident through direct observations of
229
both SST in the Pacific Ocean and the Nile flow (Supplementary Fig. 4b). The analysis of the
230
GCMs also shows that most of the GCMs can reasonably simulate the teleconnection between the
231
Nile flow and ENSO (Supplementary Table 2 and 9). It is also important to note that some of the
232
GCMs fail to simulate the observed teleconnection between ENSO and the Nile flow. For example,
233
7 out of 18 GCMs have average rainfall during moderate El Nino events that is higher than during
234
neutral years and 5 out of 18 have average rainfall during La Nina events that is lower than neutral
235
years. Thus, it is important to interpret the GCMs results while recognizing their limited accuracy
236
in simulating the Nile-ENSO teleconnection as discussed in Section 4.
237
238
Here, we project future changes in the Nile river flow using a sampling approach (described in
239
Sampling analysis in Methods) that combines GCM projections of changes in the frequency El
240
Niño and La Niña events with empirical relationships describing the teleconnection between
241
ENSO and the Nile flow. The resulting distribution of the projected future flow in the Nile river
242
confirms that the future average flow and inter-annual variability may increase (Supplementary
243
Fig. 4c), which is consistent with the simulated runoff by the GCMs. We also use a standard bias
244
correction approach (Supplementary Fig. 4d) (described in Bias correction using probability
10
245
matching method in Methods) to project the future Nile flows. The main difference between the
246
two approaches is that the sampling approach relies on the frequencies of different El Niño and La
247
Niña events to predict the Nile flows, regardless of the simulated flow by the models. On the other
248
hand, the bias correction approach relies completely on the simulated runoff by the models. The
249
two approaches yield the same conclusions and show similar distributions of the past and future
250
Nile flows (Supplementary Fig. 4c and 4d). Hence, we can conclude that the changes in the
251
variability of the simulated runoff in the Nile by the GCMs is due to the changes in frequencies of
252
El Niño and La Niña events. Furthermore, the results of each approach (i.e. sampling and bias
253
correction) agree on the sign of change in the inter-annual variability in the Nile flow. All the 18
254
GCMs show simulate an increase of the standard deviation using the two approaches. On the other
255
hand, it is also important to note that 5 (4) GCMs out of the 18 GCMs show a decrease in the long-
256
term mean of the Nile flow using the sampling (bias correction) approaches. In addition, two
257
GCMs (i.e. GFDL-ESM2M and ACCESS1) show different direction of the change in the long-
258
term mean (in the two approaches). This discrepancy can be primarily due to the small change in
259
the long-term mean in one of the approaches. For example, the increase in the long-term mean of
260
the GFDL-ESM2M using the bias correction is only +2.3% (from 79.2 km3 to 81.1 km3) and
261
similarly the decline in the long-term mean of the ACCESS1 -0.4% (from 77.4 km3 to 77.1
262
km3).The distribution of the flow in the twentieth century is closely clustered around the mean
263
with few extreme events (i.e., below 70 km3/year and higher than 100 km3/year), while the future
264
flows have fewer normal events (i.e., between 70 km3/year and 100 km3/year) and more high-
265
flow events (i.e., greater than 100 km3/year) (Supplementary Fig. 4c and 4d). These changes in
266
the distribution of the flow are evident in the changes of the 10th and 90th percentiles of the future
267
flow (2000-2100) compared to the past flow (1900-2000) as shown in Supplementary Table 3. In
11
268
addition, the changes in the distribution are tested using Kolmogorov-Smirnov test and 14 out of
269
the 18 GCMs show significant change in the distribution of future Nile flow (2000-2100)
270
compared to the past (1900-2000) as shown in Supplementary Table 5. Similarly, most the GCMs
271
show positive significant trends as shown in Supplementary Table 6 using Mann-Kendall test.
272
Furthermore, the changes in the flow pattern are consistent among the models regardless of which
273
combination of models are chosen to estimate the trends in the mean or variance as discussed in
274
Section 4 of this document (Supplementary Fig. 8, 9 and 10 and Supplementary table 8). It is
275
important to note that the ensemble average runoff during extreme El Nino event for the
276
period (1900‐2000) is 0.58 mm/day, while for the future period (2000‐2100) is 0.65
277
mm/dayandsimilarlyfortheothereventsasshowninsupplementarytable9.Thus,inthe
278
futurethemeanflowmayincreasebecauseoftheincreasedvaluesofflowduringElNino
279
andLaNinaevents,andthevariabilitymayincreasebecauseoftheincreasedfrequencyof
280
ElNinoandLaNinaevents.
281
282
Finally, Supplementary Table 7 summarizes the water storage analysis using the Hurst
283
equationforthe18GCMsandusingthebiascorrectionandsamplingapproaches.Allthe
284
GCMsshowanincreaseofthefuturewaterstoragerequiredtoaccommodatetheenhanced
285
inter‐annualvariabilityinthefutureflowoftheNile.
286
287
288
289
12
290
291
292
293
294
295
296
297
298
299
300
301
302
Supplementary Figure 3 | Time series of total annual Nile flow (green line), and the coefficient
of variation of the total annual Nile flow computed for a 30-years moving window (blue line),
for the period 1870 to 2012. The shaded rectangles in gray are the corresponding periods
from Figure 3c with high number of El Nino and La Nina events (i.e. more than 14 events
per 30 years).
303
304
305
306
307
308
309
310
311
13
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
a
b
c
d
Supplementary Figure 4 | Changes in the future Nile flows and their association with changes
in the frequencies of El Nino and La Nina through the teleconnection between the Nile and
ENSO. a, Multi-model average changes in the future frequencies of El Nino and La Nina events
using 18 CMIP5 GCMs and their comparisons to the observed number of events. b, Relation
between annual Nile flow averaged between June and May and the sea surface temperature over
Nino 3 (5◦N-5◦S, 150◦W-90◦W), Nino 4 (5◦N-5◦S, 160◦E-150◦W) and ENSO index (6N–2◦ N,
170– 90◦ W; 2◦ N–6◦ S, 180oW–90◦ W; and 6oS–10◦ S, 150◦W–110◦ W). c, d, Changes in the
distribution in future Nile flow based on sampling and bias correction approaches respectively.
The change in the long-term average of annual Nile flow for the future period (2000-2100; red
dashed lines) is statically significant above the 95% confidence level using Student t-test compared
the past period (1900-2000; blue solid lines).
14
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
a
b
Supplementary Figure 5 | Changes in the mean, standard deviation and coefficient of
variation for the flow estimated using sampling (a) and bias correction (b) approaches
based on 18 CMIP5 GCMs over the Eastern Nile basin. Percent changes in the mean,
standard deviation and coefficient of variation for the period (2000-2100) relative to the
period (1900-2000) showing the number of models that fall in each group of these percent
changes.
396
397
398
15
399
400
401
402
403
404
405
406
a
b
b
407
408
409
410
411
412
Supplementary Figure 6 | Ranges of the simulated runoff by the GCMs for each type of
413
ENSO events for (a) 1900-2000 period, and (b) 2000-2100. The red line represents the median,
414
the boundaries of the blue box are for the 25% and 75% quantiles of the simulated runoff by the
415
GCMs. The black lines in (b) are the median of the period (1900-2000) for each type of events.
416
Detailed data used in these plots are available in Supplementary Table 2.
417
418
16
419
420
421
422
423
424
425
426
427
428
429
430
a
b
431
432
433
Supplementary Figure 7 | Ranges of the long-term mean (a) and standard deviation (b) of
434
the GCMs runoff using sampling and bias correction approaches for the period (1900-2000)
435
The red line represents the median, the boundaries of the blue box are for the 25% and 75%
436
quantiles of the simulated runoff by the GCMs. The straight black lines are the median of the long-
437
term mean and standard deviation corresponding to the period (1900-2000). Detailed data used in
438
these plots are available in Supplementary Table 4.
17
439
440
441
442
443
444
445
Supplementary Table 2 | Number of different El Nino and La Nina events simulated by
the GCMs and the corresponding runoff simulated over the Eastern Nile basin (ENB).
The classification of the different types of El Nino and La Nina events are as
described in the sampling approach in Methods Section. GCMs highlighted in green are the
models that simulate higher runoff during extreme La Nina events in the past and future
compared to extreme El Nino events in the past and future.
BCCCSM1-1M
BCCCSM1-1
CCSM4
CMCCCMs
CMCC-CM
CNRMCM5
GFDLCM3
GFDLESM2G
GFDLESM2M
MPI-ESMMR
MRICGCM3
# of
Events
ENB
Runoff
# of
Events
ENB
Runoff
# of
Events
ENB
Runoff
# of
Events
ENB
Runoff
# of
Events
ENB
Runoff
# of
Events
ENB
Runoff
# of
Events
ENB
Runoff
# of
Events
ENB
Runoff
# of
Events
ENB
Runoff
# of
Events
ENB
Runoff
# of
Events
Moderate El
Nino
190020002000
2100
Extreme El
Nino
190020002000
2100
Moderate La
Nina
190020002000
2100
Extreme La
Nina
190020002000
2100
Neutral Years
190020002000
2100
29
26
3
4
13
9
6
7
50
53
0.58
0.82
0.59
0.94
0.62
0.72
0.66
1.01
0.62
0.84
29
25
0
1
14
9
5
4
53
59
0.76
0.82
0.00
0.99
0.71
1.08
0.87
0.94
0.80
0.79
9
6
15
23
14
0
5
3
58
65
0.37
0.54
0.41
0.63
0.50
0.00
0.52
0.52
0.46
0.63
9
29
1
9
6
8
7
12
76
39
0.27
0.26
0.24
0.23
0.32
0.30
0.29
0.27
0.29
0.27
13
29
0
3
9
11
7
5
72
48
0.30
0.24
0.00
0.22
0.31
0.25
0.26
0.24
0.24
0.25
5
5
15
25
15
8
4
3
62
51
0.74
0.79
0.80
0.83
0.83
0.94
0.94
0.66
0.84
0.86
2
5
15
33
13
6
4
11
65
38
0.38
0.35
0.39
0.47
0.35
0.47
0.35
0.45
0.35
0.43
11
43
0
2
17
10
2
8
69
26
0.68
0.59
0.00
0.82
0.65
0.67
0.61
0.56
0.69
0.67
5
39
11
10
13
9
4
4
66
26
0.67
0.66
0.62
0.45
0.59
0.64
0.66
0.46
0.57
0.57
16
38
0
2
14
10
4
8
67
40
0.34
0.34
0.00
0.38
0.33
0.83
0.33
0.46
0.33
0.39
18
28
0
5
14
9
3
11
66
43
18
ENB
Runoff
NORESM1
# of
-ME
Events
ENB
Runoff
NORESM1
# of
-M
Events
ENB
Runoff
CMCC# of
CESM
Events
ENB
Runoff
CESM1# of
BGC
Events
ENB
Runoff
CESM1# of
CAM5
Events
ENB
Runoff
# of
ACCES1
Events
ENB
Runoff
# of
FGOALS
Events
ENB
Runoff
Equivalent observed
Nile flow (1900-2000)
446
447
448

0.14
0.21
0.00
0.19
0.17
0.18
0.14
0.21
0.15
0.18
8
11
11
13
10
9
3
1
69
63
0.48
0.54
0.44
0.44
0.48
0.73
0.73
0.87
0.50
0.62
1
14
17
11
9
10
7
2
67
53
0.50
0.46
0.40
0.53
0.54
0.71
0.60
0.55
0.47
0.57
11
22
4
17
13
21
3
4
68
32
0.13
0.13
0.12
0.10
0.14
0.12
0.14
0.17
0.13
0.11
6
6
10
23
13
9
4
3
67
55
0.45
0.49
0.37
0.58
0.49
0.57
0.41
0.75
0.41
0.60
9
17
5
18
14
21
3
7
69
33
0.59
0.75
0.82
0.80
0.69
0.90
0.75
1.18
0.65
0.77
1
0
34
59
13
8
5
5
47
26
0.47
0.00
0.46
0.48
0.40
0.50
0.49
0.43
0.44
0.49
2
6
21
18
14
9
4
6
59
58
0.71
0.89
0.72
0.67
0.93
1.07
0.99
1.10
0.81
1.07
72.6
--
60.4
--
87.1
All the runoff values are in (mm/day).
449
450
451
452
453
454
19
--
95.8
--
83.2
--
455
456
457
458
459
460
Supplementary Table 3 | Statistics of the change in the Nile flow for the different GCMs
estimated using the bias correction and sampling approaches. The statistics include
changes in the 10th and 90th percentiles of the flows for the past period (1900-2000) and future
period (2000-2100).
BCCCSM1-1M
BCCCSM1-1
CCSM4
CMCCCMs
CMCC-CM
CNRMCM5GFDL-CM3
GFDLESM2GGFDLESM2M
MPI-ESMMRMRICGCM3NORESM1ME
NORESM1M
CMCCCESM
CESM1BGC
CESM1CAM5
ACCESS1
FGOALS
Avgerage
All GCMs
10th
Percentile
1900-2000
Sampling
10th
90th
Percentile
Percentile
2000-2100
1900-2000
10th
Percentile
1900-2000
Bias
10th
90th
Percentile
Percentile
2000-2100
1900-2000
90th
Percentile
2000-2100
90th
Percentile
2000-2100
63.9
78.6
105.3
135.8
62.0
80.3
99.2
130.8
63.5
63.9
65.8
81.4
103.6
102.1
117.5
132.4
60.4
64.1
66.3
71.9
98.8
104.2
111.9
160.7
59.0
65.8
54.4
59.3
96.7
96.6
96.2
96.2
60.9
56.9
23.0
51.2
99.7
99.3
98.4
104.9
57.9
62.7
55.7
64.2
99.2
100.3
99.7
122.7
56.9
61.3
67.4
72.6
99.9
100.3
107.5
119.8
67.8
48.9
96.8
94.4
61.6
45.9
101.7
100.3
57.8
52.1
102.1
98.4
56.9
47.1
100.9
113.6
63.5
62.0
97.2
149.0
61.7
66.8
99.6
108.5
63.5
68.0
96.6
142.4
60.9
67.6
96.9
113.5
63.9
63.1
102.1
127.8
62.2
70.6
98.5
112.9
53.7
68.6
102.1
126.1
63.6
71.2
99.4
112.2
57.8
48.4
103.7
95.2
61.4
47.3
99.6
96.1
57.4
83.1
105.3
148.0
61.0
76.0
101.0
149.0
57.9
53.7
52.2
52.3
55.7
53.6
103.6
105.4
103.6
121.8
107.5
160.4
62.1
59.9
61.9
74.5
68.5
67.5
101.0
100.7
100.7
133.5
106.4
137.3
60.4
64.1
100.9
122.9
60.9
64.7
100.0
120.7
461
462
463
464
465
20
466
467
468
469
Supplementary Table 4 | Statistics of the change in the Nile flow for the different GCMs
estimated using the bias correction and sampling approaches. The statistics include changes
in the mean, standard deviation of the flows for the past period (1900-2000) and future period
(2000-2100).
BCCCSM1-1M
BCCCSM1-1
CCSM4
CMCCCMS
CMCC-CM
CNRMCM5GFDL-CM3
GFDLESM2GGFDLESM2M
MPI-ESMMRMRICGCM3NORESM1ME
NORESM1M
CMCCCESM
CESM1BGC
CESM1CAM5
ACCESS1
FGOALS
Avgerage
(All
GCMs)
Average
future
percent
change (All
GCMs)
Std.
deviation
of future
percent
change (All
GCMs)
Mean
1900-2000
Sampling
Std.
Mean
deviation
2000-2100
1900-2000
Std.
deviation
2000-2100
Mean
1900-2000
Bias
Std.
Mean
deviation
2000-2100
1900-2000
Std.
deviation
2000-2100
80.3
112
15.4
22.1
79.9
105.2
15.5
21.5
82.2
79.3
85.7
106.9
16.6
15.1
20.4
17.5
79.9
81.5
92.1
114.5
15.2
17.8
37.1
36.6
80.1
78.3
69.1
83.2
15.4
14.7
30.3
18.3
80.1
79.5
72.2
78.3
15.4
16.7
30.2
18.5
80.1
78.1
82.1
95.1
15.8
17.2
18.7
25.1
79.9
80.3
83.5
96.1
16.2
16.1
18.3
20.0
84.1
76.1
14.0
18.1
80.0
72.3
16.8
21.8
76.6
72.1
19.3
19.1
79.2
81.1
16.9
25.5
81.3
99.8
14.9
29.4
79.8
87.9
14.9
19.6
80.8
103.6
14.1
22.9
78.3
91.5
17.3
22.7
80.4
97.3
13.7
28.5
78.9
92.1
15.3
18.4
79.0
103.1
16.6
24.5
79.9
89.6
15.1
17.2
81.4
67.1
18.7
24.3
80.3
74.6
15.9
26.3
81.0
110.3
19.1
27.8
80.4
105.8
16
28
83.8
77.4
80.4
97.4
77.1
113.9
17.8
19.3
16.5
32.6
24.
29.1
80.2
81.2
80.2
101.9
85.4
99.7
16
16.9
15.8
28
17.6
25.3
80.1
93.4
16.2
25.3
80.2
92.2
16.1
23.8
14.4%
48.4%
13.4%
49.4%
19.1%
32.1%
14.9%
38.2%
21
470
Supplementary Table 5 | The test statistic of the Kolmogorov–Smirnov test for the change in
471
the distribution of the future Nile flow (2000-2100) relative the past (1900-2000) calculated
472
for the runoff simulated by the GCMs (raw data) and using the flows estimated with the bias
473
correction and sampling approaches. At least 14 out 18 GCMs show significant change in the
474
distribution of the future Nile flow compared to the past using three approaches.
475
476
477
478
479
480
481
482
483
484
485
486
487
Bias Correction
Sampling
Raw Data
CNRM-CM5
0.14
0.18
0.14
CESM1-BGC
0.47*
0.38*
0.45*
BCC-CSM1-1M
0.62*
0.72*
0.62*
CCSM4
0.51*
0.45*
0.49*
MRI-CGCM3
0.28*
0.41*
0.27*
FGOALS
0.46*
0.68*
0.44*
CMC-CMS
0.33*
0.31*
0.33*
BCC-CSM1-1
0.29*
0.11
0.28*
CMC-CM
0.09
0.22*
0.09
GFDL-CM3
0.42*
0.37*
0.41*
GFDL-ESM2G
0.21*
0.33*
0.28*
GFDL-ESM2M
0.11
0.2*
0.16
MPI-ESM-MR
0.31*
0.33*
0.34*
NORESM1-ME
0.37*
0.42*
0.36*
NORESM1-M
0.28*
0.31
0.24*
CMCC-CESM
0.25*
0.09
0.27*
CESM1-CAM5
0.41*
0.38*
0.42*
ACCESS1
0.14
0.23*
0.16
*Indicates values that are significant at 5% significance level
22
488
489
490
491
492
493
494
495
496
497
498
Supplementary Table 6 | Mann-Kendall coefficients (tau) for the trends in the time series
of the Nile flow (1900-2100) for the different GCMs, calculated for the simulated runoff
(raw data), and using the flows estimated with the bias correction, and sampling
approaches. The number of GCMs with statistically significant and positive trends are 12, 10,
and 12 for the bias correction, sampling and raw data respectively. The corresponding number
of GCMs with statistically significant and negative trends are 2, 3, and 3 for the bias correction,
sampling and raw data respectively.
Bias Correction
Sampling
Raw Data
CNRM-CM5
0.07
-0.07
0.07
CESM1-BGC
0.42*
0.15*
0.41*
BCC-CSM1-1M
0.48*
0.44*
0.49*
CCSM4
0.42*
0.27*
0.42*
MRI-CGCM3
0.26*
0.2*
0.23*
FGOALS
0.36*
0.34*
0.36*
CMC-CMS
-0.14*
-0.18*
-0.17*
BCC-CSM1-1
0.21*
0.04
0.22*
CMC-CM
-0.03
-0.09
-0.04
GFDL-CM3
0.35*
0.19*
0.35*
GFDL-ESM2G
-0.07
-0.18*
-0.13*
GFDL-ESM2M
0.06
-0.07
0.08
MPI-ESM-MR
0.21*
0.19*
0.24*
NORESM1-ME
0.26*
0.27*
0.27*
NORESM1-M
0.25*
0.16*
0.19*
CMCC-CESM
-0.21*
-0.06
-0.22*
CESM1-CAM5
0.4*
0.14*
0.41*
ACCESS1
0.09
-0.05
0.12*
*Indicates values that are significant at 5% significance level
499
500
501
502
503
23
504
505
506
507
508
Supplementary Table 7 | Changes in the Hurst coefficients and corresponding changes in
the 100-year storage (R100) return period required to accommodate the change in the
future Nile flows using the bias correction and sampling approaches.
Hurst coefficient
Model
Bias Correction
190020002000
2100
R100
Sampling
190020002000
2100
Bias Correction
20001900-2000
2100
Sampling
1900-2000
2000-2100
BCC-CM1-M
BCC-CSM
CCSM4
0.72
0.81
0.86
0.75
0.74
0.74
0.62
0.74
0.72
0.72
0.55
0.67
258.8
359.8
514.4
403.9
502.1
661.4
197.1
270.1
240.1
349.5
196.6
205.2
CMCC-CMS
0.81
0.82
0.67
0.69
366.2
749.9
174.5
239.7
CMCC-CM
0.9
0.72
0.69
0.74
566.7
310.0
221.4
294.2
CNRM-CM5
0.68
0.73
0.65
0.74
237.9
317.3
203.4
287.5
GFDL-CM3
GFDLESM2G
GFDLESM2M
MPI-ESM
0.9
0.8
0.7
0.65
543.7
456.6
250.6
314.2
0.78
0.77
0.81
0.67
355.0
443.9
288.4
221.9
0.75
0.78
0.8
0.75
0.76
0.69
0.68
0.58
316.8
314.6
583.5
368.5
389.5
239.2
266.8
412.5
MRI-CGCM3
NORESM1ME
0.8
0.7
0.69
0.7
394.9
351.0
250.0
445.6
0.65
0.79
0.77
0.66
193.9
403.5
333.1
368.1
NORESM1-M
CMCCCESM
0.86
0.76
0.7
0.57
437.5
335.3
263.8
208.7
0.76
0.7
0.78
0.72
309.9
406.5
432.6
410.2
CESM-BGC
0.75
0.82
0.72
0.72
300.8
693.1
294.1
485.6
CESM-CAM5
ACCESS
FGOALS
Average All
GCMs
Standard
deviation All
GCMs
0.73
0.61
0.76
0.76
0.66
0.78
0.63
0.73
0.75
0.7
0.68
0.64
273.3
183.8
308.0
526.4
231.9
534.5
184.5
290.6
334.7
509.2
340.5
538.5
0.78
0.76
0.71
0.68
346.4
460.1
269.6
338.6
0.08
0.04
0.05
0.06
110.7
147.7
69.2
109.1
509
510
511
512
24
513
4- Sensitivity of the projected future changes in the flow of the Nile river to the choice
514
of GCMs
515
In the following analysis, we test the robustness of the projected changes in the future flow of the
516
Nile river using different combination of GCMs to show that these changes are independent of the
517
choice of GCMs. The 18 GCMs are divided to 5 groups as following:
518
Group 1 includes the GCMs that simulate correctly the teleconnection between Nile flow and
519
ENSO. Each GCM is evaluated based on its ability to simulate the teleconnection between Nile
520
flow and ENSO for four different criteria (i.e., the Nile flow simulated during extreme El Nino
521
years is less than that of extreme La Nina years; the Nile flow simulated during moderate El Nino
522
years is less than that of moderate La Nina years; the Nile flow simulated during moderate El Nino
523
years is less than that of neutral years; and the Nile flow simulated during moderate La Nina years
524
is greater than that of neutral years).
525
Group 2 includes the GCMs that are able to simulate at least three out of four teleconnection
526
between ENSO and the Nile flow as previously described, and the coefficient of determination
527
between the average annual seasonal cycle of the simulated rainfall over the ENB compared to
528
observation from TRMM-V73B43 is at least 50% and the simulated rainfall is within ±20% of the
529
observations.
530
Group 3 combines the GCMs that meet all four teleconnection criteria, and the coefficient of
531
determination between the average annual seasonal cycle of the simulated rainfall over the ENB
532
compared to observations from TRMM-V73B43 is at least 50% and the simulated rainfall is within
533
±20% of the observations.
534
Group 4 includes the GCMs that are the best in simulating the different ENSO feedbacks depicted
535
by Bjerknes index11.
25
536
Group 5 includes the GCMs that have the highest spatial resolution as the models with the highest
537
resolution tend to better represent the rainfall seasonal cycle over the basin12 (i.e., higher than 1.5o
538
x 1.5o), and the coefficient of determination between the average annual seasonal cycle of the
539
simulated rainfall over the ENB compared to observations from TRMM-V73B43 is at least 50%
540
and the simulated rainfall is within ±20% of the observations.
541
542
The results of this sensitivity analysis are presented in Supplementary Figures 8, 9 and 10, and
543
Supplementary Tables 8 and 9. The results are consistent among the different combination of
544
GCMs. Supplementary Figure 8 shows that the 30-year moving average of the runoff mean and
545
standard deviation is increasing with time for the 5 groups. Supplementary Figures 9 and 10
546
show the changes in the distribution of the flows for the period (2000-2100) compared to (1900-
547
2000) for the different combinations and show the increase of extreme wet events in the future
548
and reduction of occurrence of average flow conditions (i.e. between 70 km3/year and 100
549
km3/year). Supplementary Table 8 summarizes the changes in the mean, standard deviation,
550
10th and 90th percentiles of the flow for the different groups of GCMs. The results show that the
551
mean and standard deviation may increase by 15% and 50% respectively in the future. The 10th
552
and 90th percentiles may increase from 60 km3/year to 65 km3/year and from 100 km3/year to
553
120 km3/year respectively.
554
555
556
557
558
559
560
26
561
562
563
564
b
a
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
Supplementary Figure 8 | 30-year moving averages of the mean (a) and standard deviation
(b) of the runoff simulated by the models from 1900 to 2100 for the different combination
of GCMs.
27
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
a
b
c
d
e
Supplementary Figure 9 | Changes in the future Nile flows using different combinations
of GCMs based on bias correction approach. a, b, c, d, e are the flow distributions for the
different combinations of GCMs, which represent Group 1, 2, 3, 4 and 5, respectively. The
future flows for the period (2000-2100) are in red and the past flows for period (1900-2000) are
in blue.
28
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
a
b
c
d
e
Supplementary Figure 10 | Changes in the future Nile flows using different combinations
of GCMs based on a sampling approach. a, b, c, d, e are the flow distributions for the
different combinations of GCMs, which represent Group 1, 2, 3, 4 and 5, respectively. The
future flows for the period (2000-2100) are in red and the past flows for period (1900-2000) are
in blue.
29
684
685
686
687
688
689
Supplementary Table 8 | Statistics of the change in the Nile flow for the different
combinations of GCMs estimated using the bias correction and sampling approaches.
The statistics include changes in the mean, standard deviation and 10th and 90th percentile of
the flows.
Model
Group 1
Group 2
Group 3
Group 4
Group 5
All GMCs
Observation
Period
19002000
20002100
19002000
20002100
19002000
20002100
19002000
20002100
19002000
20002100
19002000
20002100
19002000
Bias Correction
Standard
Mean
deviation
90th
Percentile
Flow Sampling
Standard
Mean
deviation
10th
Percentile
10th
Percentile
90th
Percentile
80.70
16.64
63.05
101.70
79.65
15.90
63.90
103.70
109.84
29.03
76.10
145.75
108.58
17.91
80.00
134.10
80.37
16.48
61.00
101.08
79.46
16.35
60.78
102.98
102.24
26.09
73.90
137.00
101.01
20.19
74.70
128.98
79.96
16.14
61.87
100.00
80.50
15.75
61.07
100.58
101.44
28.91
65.58
136.03
109.61
27.85
73.18
140.80
80.25
16.15
61.34
100.60
78.99
16.59
59.24
101.61
93.30
21.72
68.11
120.90
89.10
21.87
64.14
117.67
80.32
16.1
61.1
100.7
79.95
15.28
59.7
101.9
94.03
24.84
66
124.3
94.57
24.53
65.4
123.8
79.98
15.99
60.9
100.7
80.12
16.2
60.4
100.9
92.16
23.8
64.7
120.7
93.38
25.28
64.1
122.7
80.2
16.1
60.2
100.5
80.2
16.1
60.2
100.5
690
691
692
693
694
695
696
697
698
30
699
700
Supplementary Table 9 | Summary of the runoff simulated over the Eastern Nile basin
during different El Nino and La Nina types for different combinations of GCMs.
Runoff (mm/day)
All GCMs (1900-2000)
Group 1 (1900-2000):
BCC-CSM1-1M, CCSM4,
CMCC-CMS, MRICGCM3, FGOALS
Group 2 (1900-2000):
BCC-CSM1-1M, CCSM4,
CNRM-CM5, CESM1BGC
Extreme El Nino
Std.
Mean
deviation
Moderate El Nino
Std.
Mean
deviation
Neutral
Mean
Std.
deviation
Extreme La Nina
Std.
Mean
deviation
Moderate La
Nina
Std.
Mean
deviation
0.58
0.21
0.54
0.21
0.56
0.24
0.61
0.29
0.58
0.25
0.55
0.22
0.49
0.27
0.54
0.3
0.62
0.39
0.61
0.36
0.54
0.19
0.54
0.16
0.58
0.19
0.64
0.61
0.16
0.23
Group 3 (1900-2000):
BCC-CSM1-1M, CCSM4
Group 4 (1900-2000):
CCSM4, GFDL-CM3,
GFDL-ESM2M,
NOREMS1-ME,
NORESM1-1M, ACCESS1
Group 5 (1900-2000):
BCC-CSM1-1M, CCSM4,
CMCC-CMS, CNRMCM5, MPI-ESM-MR,
CESM1-BGC, CESM1CAM5, ACCESS1
0.5
0.13
0.48
0.15
0.54
0.12
0.59
0.1
0.56
0.09
0.76
0.08
0.75
0.11
0.76
0.79
0.84
0.13
0.77
0.09
0.53
0.22
0.48
0.15
0.5
0.18
0.55
0.22
0.52
0.18
All GCMs (2000-2100)
0.65
0.36
0.64
0.29
0.66
0.39
0.73
0.5
0.71
0.34
0.71
0.53
0.66
0.41
0.77
0.54
0.91
0.8
0.73
0.51
0.74
0.18
Group 1 (2000-2100):
BCC-CSM1-1M, CCSM4,
CMCC-CMS, MRICGCM3, FGOALS
Group 2 (2000-2100):
BCC-CSM1-1M, CCSM4,
CNRM-CM5, CESM1BGC
Group 3 (2000-2100):
BCC-CSM1-1M, CCSM4
Group 4 (2000-2100):
CCSM4, GFDL-CM3,
GFDL-ESM2M,
NOREMS1-ME,
NORESM1-1M, ACCESS1
Group 5 (2000-2100):
BCC-CSM1-1M, CCSM4,
CMCC-CMS, CNRMCM5, MPI-ESM-MR,
CESM1-BGC, CESM1CAM5, ACCESS1
Observed corresponding
Nile Flow (km3/year)
0.17
0.74
0.17
0.14
0.66
0.73
0.21
0.73
0.79
0.2
0.68
0.21
0.73
0.15
0.76
0.35
0.72
NA
0.83
0.07
0.77
0.23
0.87
0.84
0.87
0.17
0.99
0.12
0.61
0.24
0.57.
0.23
0.61
0.21
0.66
0.31
0.68
0.23
60.4
22.2
72.6
17.1
83.2
21.3
95.8
10.6
87.1
19.5
31
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
Additional Supplementary Figures and Tables
Supplementary Figure 11 | Increase in the population with time. Increase in the population
(solid brown line) versus the decrease of the water share per person assuming no changes in the
flow (i.e., 80 km3/year, which is the long-term average between 1900 and 2000, blue solid line),
and the water share per person using the multi-model average time series of the Nile flow
projected using the bias correction (dashed red line). Climate change is likely to increase the
mean flow in the basin. This increase in the flow is a welcome change, and can slightly enhance
the water share per person and thus reduce the water stress in the basin, but this positive impact
will persist for only a few years before being overwhelmed by large decrease in water share per
capita due to projected population growth. Population growth poses a serious challenge to
availability of water in the Nile.
32
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
Supplementary Figure 12 | Relation between annual flows at Dongola and Upper Blue Nile basin from 1965‐2010. Supplementary Table 10 |CorrelationbetweenENSOindicesandRainfall/flowoverdifferent
regionsintheworld.
Basin/Region
CoefficientofCorrelation(R)/
CoefficientofDetermination(R2)
Amazonbasinflow13
‐0.31/9.6%
Congobasinflow13
‐0.31/9.5%
13
Paranabasinflow 0.44/19.3%
13
Nilebasinflow ‐0.50/25%
RainfalloverIndia14
‐0.57/32%
15
RainfalloverSouthernCalifornia 0.41/16.8%
33
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
References
1. Vorosmarty, C.J., Fekete, B.M. and Tucker, B.A. Global River Discharge, 1807–1991,
Version. 1.1 (RivDIS). Data set. Available on-line [http://www. daac. ornl. gov] from
Oak Ridge National Laboratory Distributed Active Archive Center, Oak Ridge,
Tennessee, USA. 1998.
2. Taylor, KE., Stouffer, RJ., Meehl, GA. An overview of CMIP5 and the experiment
design. Bull Am Meteorol Soc 93:485–498. 2015.
3. Cai, W., et al. Increasing frequency of extreme El Niño events due to greenhouse warming.
Nature Climate Change 4, 111–116. 2014.
4. Cai, W. et al., Increased frequency of extreme La Nina events under greenhouse
warming." Nature Climate Change 5, 132-137. 2015.
5. Rowell, D.P. Simulating SST teleconnections to Africa: What is the state of the
art?. Journal of Climate, 26(15), pp.5397-5418. 2013.
6. Siam, M. S., Wang, G., Demory, M. E., Eltahir, E. A. B. Role of the Indian Ocean sea
surface temperature in shaping the natural variability in the flow of Nile River."Climate
dynamics 43, no. 3-4: 1011-1023. 2014.
7. Williams, A.P. and Funk, C. A westward extension of the warm pool leads to a westward
extension of the Walker circulation, drying eastern Africa. Climate Dynamics, 37(11-12),
pp.2417-2435. 2011.
8. Sutcliffe, J.V. and Parks, Y.P. The hydrology of the Nile (No. 5). Wallingford,
Oxfordshire, UK: International Association of Hydrological Sciences. 1999.
9. Hurst, H.E. and Phillips, P. The Nile Basin, Volume IV. Ten-day mean and monthly mean
discharges of the Nile and its tributaries. Cairo: Ministry of Public Works, Physical
Department. 1933.
10. Slaman, S. Sudan Continues Relinquishing a Growing Portion of Nile Water Share. Sudan
Vision, Issue: 3414, December 10th, 2014.
11. Kim, S. T., et al. An., Response of El Niño sea surface temperature variability to
greenhouse warming, Nat. Clim. Change, 4, 786–790. 2014.
12. Siam, M. S., Demory, M. E., Eltahir, E. A. B. Hydrological cycles over the Congo and
upper Blue Nile basins: evaluation of general circulation models simulations and reanalysis
data products. J Clim. doi:10.1175/JCLI-D-12-00404.1. 2013.
34
807
808
809
810
811
812
813
814
815
816
817
818
13. Amarasekera, K.N., Lee, R.F., Williams, E.R. and Eltahir, E.A.B. ENSO and the natural
variability in the flow of tropical rivers. Journal of Hydrology, 200(1), pp.24-39. 1997.
14. Ihara, C., Kushnir, Y., Cane, M.A. and De La Peña, V.H. Indian summer monsoon
rainfall and its link with ENSO and Indian Ocean climate indices. International Journal
of Climatology, 27(2), pp.179-187. 2007.
15. Jong, B.T., Ting, M. and Seager, R. El Niño's impact on California precipitation:
seasonality, regionality, and El Niño intensity. Environmental Research Letters, 11(5),
p.054021. 2016.
35