Brief introduction to DRIFT
Downstream Response to Imposed Flow Transformations
History• Development began in 1998
• Examples of applications:– 1998-2002: Lesotho Highlands Water Project
– 1998-2014: 19 river system in South Africa
– 2006: Zambezi Delta (Mozambique), funded by the International Crane Foundation and the Carr Foundation
– 2007: Mzingwane River (Zimbabwe), funded by IUCN
– 2007: Phuthiatsana River (Lesotho), funded by Lesotho Department of Water Affairs
– 2006-9: Pangani Basin (Tanzania), funded by IUCN and the Tanzanian government
– 2008-9: Nile River (Sudan), funded by the Sudanese Dams Implementation Unit
– 2008-10: Okavango Basin (Angola, Namibia, Botswana), funded by GEF/UNDP
– 2009-10: Lower Zambezi River (Mozambique), funded by Riversdale Mining
– 2009-10: Kunene River (Namibia and Angola), funded by the Angolan and Namibian Governments
– 2010-11: Neelum-Jhelum River (Pakistan), funded by the Government of Pakistan
– 2011: Kagera River (Nile Basin), funded by the Nile Basin Initiative
– 2011: Huaura River (Peru), funded by SN Power
– 2013-14: Poonch River (Pakistan), funded by Mira Power
– 2014: middle Zambezi River (Zambia and Zimbabwe), funded by the Zambezi River Authority
– 2015: Lower Mekong River (Laos, Thailand, Cambodia, Vietnam), funded by the Mekong River Commission
– 2016: Elephant Marshes, Shire River (Malawi), funded by Malawi Ministry of Water Development and Irrigation.
• Recognised as a good practice methodology by the World Bank, ADB, IFC, IUCN, OKACOM, and the South African, Tanzanian and Pakistan governments
Pangani River
Zambezi River
Okavango RiverCunene River
Nile River
Huaura River
Pongola River
Mekong River
N-J Basin
Location of applications
c. 50 projects in total
Cuanza River
Trishuli River
Kouilou-Niari Rver
Kagera River
Orange/Senqu Basin
Olifants-Doorn Basin
Berg Basin
Pungwe River
Poonch River
Shire River
For a project of the magnitude of the KishengangaHydropower Project, the Court is of the view that an in-depth assessment of the type that Pakistan has attempted for these
proceedings is an appropriate tool for estimating potential changes in the downstream environment.
Verbatim: In the matter of the Indus Waters Kishenganga arbitration
The International Court of Arbitration constituted in accordance with the Indus Waters Treaty 1960December 2013
The Hague: Permanent Court of Arbitration (December 2013)
Step 1: Select scenarios
Baseline
Scenarios
Generic steps in DRIFT
Step 3: Model hydrology, hydraulics
Step 5: Assign Baseline Status and trends
Step 6: Knowledge captureSet up DRIFT all sites
Create response curves
Step 7: Calibration
Step 8: AnalysisRun DRIFT for all scenarios and generate prediction of change
Step 4: Indicators
Step 2: Select focus areas
Step 1: Scenario selection
• Scenarios are a means of exploring possible pathways into the future
• Describe a range of potential development of the river (design, location and operation of infrastructure/abstractions)
• Dedicate process for scenario selection
• Informs site and indicator selection
Step 2: Site selection
• The sites are the focus for the DRIFT predictions of ecosystem change
Step 3: External modelling
• Time-series of simulations for baseline and each scenario, at each site:
– Hydrology
– Hydraulics
– estimated sediments
• Imported into DRIFT
Step 4: Status and Trends Assessment
• Status and Trends:
– describe the ecological status of the rivers at the time of the study;
– describe the past ecological status of the rivers, and possibly;
– describe the future ecological status rivers with and without the water-resource developments included in the scenarios
• Status of the rivers at the time of the study is most frequently used as the baseline from which to describe change
– Relative change is predicted – no absolutes
Step 5: Indicator selection
• DRIFT indicators are inputs to the DRIFT model
• Each indicator must have a describable relationship to the flow or sediment regime
• Indicators describe:
– the flow regime of the river, e.g., duration of
the dry season
– ecosystem attributes; e.g.; abundance of white fish
– river-linked social attributes, e.g., riverbank gardens
• DRIFT will predict how each indicator will change from baseline
Step 6: Mapping of indicator links
• Map links between driving and responding indicators
Step 7: Knowledge Capture
Each mapped link requires a response curve
Construct the response curves
-5
-4
-3
-2
-1
0
1
2
0 100 200 300 400
Dry season duration (days)S
ev
erit
y r
atin
g in
dic
ati
ng c
ha
nge
in a
bun
da
nce
Response Curves
• Means of capturing information and understanding:– from in-depth scientific data, international
knowledge, national knowledge or local wisdom. – created by EF specialists with a working
knowledge of the river ecosystem and its users– graphic and explicit with supporting explanations– allow qualitative as well as quantitative
knowledge to be captured;– amenable to adjustment as knowledge
increases.
Severity
rating
Severity
change
Equivalent Loss or gain
5 Very large 501-∞ (to pest
proportions)
4 Large 251-500
3 Moderate 68-250
2 Low 26-67
1 Negligible 1-25
0 None No change
-1 Negligible 0-20
-2 Low 20-40
-3 Moderate 40-60
-4 Large 60-80
-5 Very large 100-80
Example: Response Curve
Fish Guild A
-5
-4
-3
-2
-1
0
1
2
3
4
5
0 10 20 30 40 50 60 70 80
Dry Season Minumum Discharge
Sev
eri
ty o
f c
han
ge
rela
tiv
e t
o P
D M
edia
n
specialists draw curve
Response of one ecosystem indicator (Fish Guild A) to minimum dry-
season flows in a year
Specialists construct curve
150 000 t/day > median
c. 25% increase
150 000 t/day < median
c. 20% decrease
230 000 t/day > median
c. 45% increase
The higher the wet season average sediment load, the more river energy
that is expended carrying sediment, and the lower likelihood of erosion.
10 weeks later median
c. 15% decrease
6 weeks earlier median
c. 15% increase
2 weeks later median
No change
The onset of the dry season represents a time rhithron species are able to
migrate to shallower areas with suitable substrate for spawning, earlier onset
allows the fish greater time to migrate but late onset can distrupt spawning
migration and maturation. Also if dry season starts earlier, it beneficial as fish
can mature in less stressful conditions prior to spawning.
Step 7: Calibration
• Refine Response Curves using the historical flow record, so that DRIFT outputs reflect current understanding and monitoring data
• Use a series of calibration sequences to further refine the response curves, e.g.:– Sequence of wet years– Sequence of dry years– Sequence of several dry years, followed by several
wet years, repeated
• Calibration increasingly stabilises the model
Step 8: Analysis
• For each site under each scenario:– Enter input time-series into DRIFT (hydrology,
hydraulics, sediments)– Define other concerns for scenarios (barriers,
management, etc.)– Calculate the values for all input indicators– Run DRIFT to pass input indicators through the
ecosystem response curves– Generate predictions of change for each indicator and
overall ecosystem condition
• Post-process results into report format
Each responding indicatorModelled time series
Transformed into time series of
driving indicators
Fish Guild A
-5
-4
-3
-2
-1
0
1
2
3
4
5
0 10 20 30 40 50 60 70 80
Dry Season Minumum Discharge
Sev
eri
ty o
f c
han
ge
rela
tiv
e t
o P
D M
edia
n
Scenario: Dry season minimum discharge for each year
External modelled time series
Transformed into time series of
driving indicators
DRIFT prediction of
change for each year
30 years of record = 30
values
Each responding indicator
Curves combined using multi-criteria decision analysis procedures
Resp
on
se c
urv
es f
or
ea
ch
lin
ke
d ind
icato
r
Time-series of change
in indicator A
VegetationHydraulics
Macroinvertebrates
Geomorphology
Modelled hydrology time-series
Linked indicators for Indicator A
Depth
Velocity
Temperature
Macroinvert spp.
Marginal vegetation
Sandy banks
Modelled sediment time-series
Modelled water quality time-series
Indicators and Linked Indicators
Examples of DRIFT outputs
Time-series of change for one indicator at one site under fifteen
scenarios
Change in indicator abundance under each scenario, colour-coded to show
severity of changeEcosystem Indicators
KMax
K3.9
4
K10
K20
K40
K60
K100
K7E3
K9E1
Geom
orph
olog
y
Secondary channels, backwater -49.60 -39.43 -29.78 -24.07 -10.75 -9.81 -9.00 -30.39 -41.13
Cobble and boulder bars 13.96 12.62 10.87 5.12 1.52 2.74 1.96 10.83 12.87
Sand and gravel bars -20.19 -21.18 -22.15 -26.52 -24.45 -22.26 -21.59 -22.25 -21.05
Bed sediment size 10.21 11.81 13.56 16.25 20.06 18.74 17.38 13.54 11.52
Active channel width -9.31 -9.31 -9.31 -9.31 -6.72 -4.71 -4.15 -9.34 -9.32
Depth of pools 3.29 3.29 3.29 3.32 4.90 5.52 5.19 3.29 3.28
Wat
er
qual
ity
Dilution of pollution loads -30.89 -23.37 -12.55 -3.26 -2.97 -2.86 -2.69 -20.16 -27.24
Temperature -47.05 -33.00 -2.66 0.42 0.34 0.34 0.36 -23.67 -42.74
Rive
rine
vege
tatio
n Algae 5.28 7.08 8.53 4.43 2.03 2.43 2.43 7.97 6.23
Marginal vegetation 4.50 4.51 4.50 4.51 1.67 -0.79 -0.78 4.52 4.51
Natural terrace veg. -3.57 -3.57 -3.57 -3.60 -3.62 -3.62 -3.62 -3.61 -3.57
Mac
ro-
inve
rteb
Simuliidae 4.79 6.48 9.06 4.35 5.28 5.00 4.79 7.43 5.24
Other flies & midges 21.92 21.62 20.42 10.62 9.71 9.94 10.24 21.29 21.67
EPT abundance -42.70 -29.13 -10.00 -2.90 5.64 5.38 5.77 -22.41 -35.66
Fish
Brown Trout -76.92 -50.55 -1.14 5.51 9.54 14.45 19.15 -36.66 -66.64
Tibetan Snow Trout -74.68 -59.15 -38.12 -29.24 -19.09 -18.50 -15.08 -51.86 -67.11
Alwan Snow Trout -64.15 -44.76 -22.04 -13.10 -13.23 -14.06 -12.17 -37.57 -53.57
High Altitude Loach -79.06 -72.82 -49.03 -34.98 -20.48 -20.09 -16.96 -65.54 -76.79
K. Hillstream Loach -78.84 -72.33 -50.85 -41.91 -25.43 -24.46 -21.85 -64.93 -76.49
Himalayan Cat Fish -36.56 -23.63 -9.02 -8.00 0.10 0.19 0.61 -18.51 -29.28
Change in ecosystem condition (integrity) at a site under scenarios
-4
-3.5
-3
-2.5
-2
-1.5
-1
-0.5
0
PD
MA
MD
HW
K5
N9
K1
0N
9
K2
0N
9
K4
0N
9
K6
0N
9
K8
0N
9
K1
00N
9
KH
5E
5N
9
KH
7E
3N
9
KH
9E
1N
9
KM
DN
9
KH
WN
9
K3
94N
9
KM
D3N
9
KM
D4N
9
Ove
rall in
teg
rity
sco
re w
ith
Min
an
d M
ax (
LO
C)
Scenarios
Integrity A to B B to C C to D D to E E to F
A
B
D
E
C
Present Day Medium High
A
D
BC
ENot assessed
Low
Snapshot of basin-wide ecosystem condition under different scenarios