Skip to main content

Lecture15

Forest cover doesn't just affect small and medium but also larger floods; The larger the flood, the larger the forest cover and said flood. How large is large? Younes says large is any larger than a 10 year flood.

 

The argument of the new paradigm is of the most fundamental nature. It targets the previously held hypothesis, controls, etc.

  • The previous paradigm was caused by "being fooled by randomness".
  •  

Younes' work argues against previous hydrology works stating that they're uncontrolled; they're using the science method wrong and are unable to establish causation. The wrong paradigm has guided forest hydrology to the current position and has influences every single aspect. Every single research project has used the same methodology. 

 

Past 100 years of forest hydrology has not looked at frequency

 

 

Grand forks, first flood  of high level 1948; second time 2018; flood level has repeatedly come close to the 2018 levels over the past 4 years.

Scientific paradigms

"A collection of beliefs and agreements, shared by scientists, that guides the research efforts of scientific communities, and by which a field of inquiry is most clearly identified as a science"

  • Subscribers of old paradigms will fight for their old paradigm to "save their pride"
  • Science must be objective; scientists are humans and therefore cannot be objective. They are subjective by nature.

 

Everything that we have learned is shaped by a paradigm. 

  • Forest hydrology is embroiled in controversies that never get resolved
    • Forest reduce flood RISK but do not reduce
    • Forest cover eliminates some flood events
    • Disagreement among scientists have led to an every increasing schism between science, public perception, and management policies.

 

Anonymous peer reviewers and journal-editors make it difficult for Younes and lab to release research.

 

 

"Forest hydrology has a sad history of being embroiled in controversies that never seem to get resolved" (consensus virtually does not exist in any forest science field, if we have real consensus it has nothing to do with the establishing causation).

 

Disagreements among scientists are common in many fields of studies:

  • Wouldn't be "nice to know why and how highly competent scientists can repeatedly come up with completely contradictory results?"

 

 

Old paradigm

Non-causal deterministic reductionist approach.

Unable to draw any connections; has been misused to say stuff that can't be said. As there is randomness, looking at any single event provides misleading info.

As example, it's like trying to determine if cigarettes cause cancer by taking 1 smoker, 1 non-smoker and seeing if either of them develop cancer; basing the answer solely on this one test.

If smoker develops cancer; cigarettes cause 100% cancer rates (untrue). If smoker does not develop cancer; cigarettes don't cause any cancer (also untrue).

If non-smoker develops cancer but smoker does not, cigarettes prevent cancer (untrue). If non-smoker develops cancer and smoker develops cancer (untrue); smoking does not affect cancer rates (untrue).

Current paradigm with forest effects on floods is a dogma / belief. 

 

Reductionist (it reasons one storm at a time) and deterministic (it doesn't invoke the stochastic aspect of floods).

 

Stochastic: randomly determined; having a random probability distribution or pattern that may be analyzed statistically but may not be predicted precisely.

 

 

Floods are stochastic and have multiple causes, a cause CANNOT be determined because they are not testable or controllable.

Paired Watershed studies:

  • Difference in flood magnitude in two watersheds, when both are subject to the same storm
  • Log peak flow of control vs treatment
  • Regression = correlation
  • Every point = pair of two peak flows when both watersheds are subject to the same storm
  • Two regression lines converge at greatest magnitude.

 

Explain why the frequency pairing graph diverges and then later becomes parallel at the end?

  • Flood frequency graphs do not converge and there is a reason for it.
  • Anytime the flood frequency graph moves up, frequency increases

 

  • Reasoning:
    • If both watersheds are subject to the same storm, the angle of the regression line will be proportional on both axis (45 degrees) meaning that there is no difference between the two.
    • If forest cover does not impact flooding, then when the control vs treatment is paired by storm size, the points will be distributed in a way that approaches a 45 degree angle.
      • I.e., same point on each axis.
    • If points are above 45 degree angle, the flood magnitude is bigger in the treatment than the control.
    • If points are below 45 degree angle, the flood magnitude is smaller in the treatment than the control.

Anytime mean increases by 55%, variability around the mean increases by 30%. When variability increases the post harvest frequency curve steepens, which is why they diverge before becoming parallel. 

 

 

History: Sponge theory

The role of forest in flood generation ahs been conceptualized around a century-old "sponge theory"

 

Forest is imagined to soak up water during storm and releases it later on over time (Pinchot, 1905)

 

New paradigm

Frequency pairing.

 

An increase in the mean of the flood frequency distribution after logging is a function of the amount of moisture available for run off. 

 

With trees:

  • Logging → supress ET → More moisture available → Soils are more wet → greater run-off → decreased buffer → higher variability

 

Compaction from machinery and paving → less infiltration of precipitation → water does not infiltrate the storage → higher run-off rate → increase efficiency in which water reaches the stream → increase in peak flow variability.

 

 

Stream network is not static, it expands / shrinks based on seasonal wetness. At the beginning of a storm, the stream may be less developed than what it is at the end of the storm, with stream elements developing as it gets wet.

 

If the watershed is subdued, snow melt is synchronized under the canopy.

Watershed characteristics:

Uniform topography → flashy watershed.

Mountainous topography with high peaks → gradient of different melt rates at different elevations

 Different aspects → desynchronization of melt → flood risk mitigation.

logging → melt synchronization → increased flood risk → high peak flow variability

 

Why does the frequency pairing graph become parallel? 

Floods are generated by the combination of storm and soil moisture 

Forested → threshold return period → after, the soil is saturated → slope changes and becomes steeper

Logged -> smaller threshold return period because soil is wetter → condition at which peak flows saturate soil is faster 

Before: Subsurface flow dominates → after, overland flow dominates 

After saturation is reached → Slope is controlled by the variability of precipitation

 

An increase in the mean of the flood frequency distribution after logging is a function of the amount of moisture available for runoff.

 

With trees → take out trees → suppress ET → wetter soil → more moisture available

Compaction from machinery → less precipitation infiltration → higher rate of runoff.

Important stuff to know

Calculating return period based on exceedance.

 

The return period is defined as 1/P, e.g. an annual exceedance probability P of 0.1 (10%) implies a return period T of ten years.