Streamflow timing trends across western North American mountains have shifted towards earlier in the water year, with changes up to one month earlier. These shifts are correlated with increasing spring temperatures and have continued through 2008. While models suggest acceleration towards earlier timing for some snow-dominated regimes, the results are not statistically significant when considering spatial and temporal correlations in the datasets. Some regions have experienced regime shifts towards greater rain domination since the late 1980s and late 1990s.
Is there acceleration in streamflow timing trends across western North American mountains? [Iris Stewart]
1. Is there acceleration in
streamflow timing trends
across the western North
American mountains?
Iris Stewart (Santa Clara University)
Holger Fritze (Universitaet Muenster)
Edzer Pebesma (Universitaet Muenster)
2. Western North America: Mountain
snow storage = key for water supply
• Precipitation
low and
seasonal
• Southern
and western
areas
projected to
become
drier
3. Determining
streamflow
timing
measures
for a) snow
b) rain
c) mixed
regimes
All Gauges at: http://webpages.scu.edu/ftp/streamflowtiming/
4. Snowmelt Domination Categories
(SDCs)
• SDC1: clearly rain dominated
– Snowmelt pulse in < 30% of years
• SDC2: mostly rain dominated
– Snowmelt pulse in >= 30 & < 50% of years
• SDC3: mostly snow dominated
– Snowmelt pulse in >= 50 & < 70% of years
• SDC4: clearly snow dominated
– Snowmelt pulse in > 70% of years
12. Have changes in streamflow
timing accelerated?
What regime shifts are taking
place?
13. Two linear
regression
models:
1) Second order OLS
- Acceleration when 2nd order
term negative
2) Piecewise linear
- 2 connected straight lines, is
there a change in slope?
14. Models appear to suggest
acceleration for SDCs 3 and 4
• β2 negative =>
acceleration
towards earlier
present
• β’2 negative =>
change in slope
towards earlier
BUT: if we incorporate spatial-temporal
covariance model to account for
autocorrelations, results no longer
statistically significant
17. Summary
• Shifts towards earlier snowmelt runoff timing
continued through 2008
– Regionally coherent
– Mostly connected to warmer spring temps
• No statistically significant acceleration, when
considering spatial and temporal correlation
– Earlier timing indicated
– Short time series with high interannual variability
• Regime shifts taking place in several regions with
high vulnerability to warmer temps
– Most shifts towards greater rain domination