Upload
marjory-henderson
View
215
Download
0
Tags:
Embed Size (px)
Citation preview
Solomon Gebreyohannis GebrehiwotDepartment of Aquatic Sciences and Assessment, SLU, Box 7050, 750 07 Uppsala;
Kevin BishopDepartment of Aquatic Sciences and Assessment, SLU, Box 7050, 750 07 Uppsala; and Department of Earth Sciences,
Uppsala University, Villavägen 16, 752 36 Uppsala
Annmieke GärdenäsDepartment of Soil and Environment, SLU, Box 7014, 750 07 Uppsala;
Jan SeibertDepartment of Geography, University of Zurich – Irchel, Winterthurestrasse 190, CH-8057 Zurich, Switzerland;
Per-Erik MellanderResearch Officer, Agricultural Mini-Catchment Programme, Teagasc, Johnstown Castle Environmental Research
Centre, Co. Wexford, Ireland
23-04-21
INTRODUCTIONHydrological change detection
Statistical analysisChange detection – modeling
Pair watershed approachmodel parameter changes in a single watershed
23-04-21
Study Site – Blue Nile Basin
23-04-21
KogaGilgel Abbay
Birr
Upper-Didesa
Forest change history
Rf ~ 1500 mm
Rf ~ 2000 mmArea = 260 (Koga) –
1900 km2 (Upper-Didesa
Subsistence farming, soil erosion and land use change, seasonal water availability
23-04-21
Snow routine parameters:
Model Application - HBV
23-04-21
Periodic classification
23-04-21
METHODS - Change detection
Parameter comparisonComparison of the distribution of 50 best parameter setsUsing Wilcoxon signed-rank test
Residual comparisonUsing parameter sets from P1 then simulating rainfall in P2
and P3Then calculating residuals for all 3 periods and comparing
themSimulation comparison
Parameter sets from P1, P2 and P3 simulated for the driest and wettest years, and compare the simulations
23-04-21
Parameter comparison
23-04-21
23-04-21
FC LP BETA K1 K2 MAX
Birr
P1 1208.2 0.21 1.15 0.22 0.13 1.96
P2 1386.6 0.17 1.10 0.25 0.12 2.41
P3 1604.6 0.28 1.10 0.21 0.12 2.73
Upper-Didesa P2 773.0 0.31 3.25 0.06 0.08 2.21P3 714.9 0.22 1.54 0.14 0.06 3.04
Gilgel Abbay P1 195.6 0.86 2.40 0.05 IN 2.24P2 227.0 0.94 1.68 0.08 IN 2.54P3 217.1 0.95 1.80 0.09 IN 1.89
Koga
P1 1413.2 0.36 1.15 0.14 0.06 2.19
P2 1637.0 0.44 1.22 0.15 0.08 2.17
P3 1670.5 0.50 1.28 0.11 0.05 3.07
Table. Medians of model parameter values with test results. Groups in a column colored red are significantly different at ρ 0.05 with Wilcoxon signed-rank test, and those do not are non-significantly different, “IN” shows that the parameters respective to each watershed were insensitive .
Model residuals comparison
23-04-21
Annual medians of relative residuals; filled circles showed significant differences and open ones, non significant. “+” indicates the average medians.
Extreme climate simulation comparison
23-04-21
Simulations for the driest and wettest years of the four watersheds with parameter sets from the three periods.
Discussion and conclusionSpecific parameters are significantly changing
23-04-21
Residuals are significantly changing from the reference period – Period 1 (1960-1975)
Discussion and conclusion
23-04-21
Though parameters were said significantly changing, the response of the watersheds to flow remains less/no change
The masking of parameter changes in simulationsScale issueParameter compensation
The impacts of parameter changes might be seen at smaller scale
We recommend analysis of relation of model parameters to specific watershed characteristics in the future
23-04-21
Swedish International Development Agency (SIDA)
23-04-21