Introducing pseudoprecipitation

1
Introducing pseudo-precipitation

• Paul Schultz
• Isidora Jankov
• Huiling Yuan

August 28, 2007
2
Motivation

• Certain methods of statistical post-processing of
  ensemble model output would work better with a
  variable that represents precipitation so that
• It seems continuous in space and time
• It is approximately equal to precipitation where
  positive
• It is something that represents the opposite of
  precipitation (e.g., dryness) where negative
• Straightforward physical interpretation is
  desirable but not required
• We can scale if we must, but we first seek a
  quantity that does not require it
• A way to de-bias ensemble members (useful as
  input to hydro modeling)

3
Formula

• Integrated quantity could also be
• First term on RHS is the vertical integral of
  saturation excess
• Effectively negative-definite
• Highest humidity represented by values
  approaching zero
• Units of depth, like precipitation
• Dry end looks asymptotic

4
sounding

• Negative of the area integral between the red and
  green traces

5
Example

• This is an example based on SREF grids
• Current algorithm should work with any model
  output on isobaric surfaces
• Passes sanity check

6
Histogram

• Units are mm
• The categories for the bottom panel are
  lt-60, -50, -45, -40, -35, -30, -25, -20, -15,
  -10, -7.5, -5, -2.5, 0, 2.5, gt5

7
Features

• NAEFS precipitation is reported in 6-h
  increments, and is the result of processes that
  occurred during the 6 hours of collection
• Saturation deficit is calculated over average
  values of vapor and temperature during the 6 h
• This formulation of PP contains the biggest terms
  of water conservation equation, except
• Surface flux
• Mass of suspended hydrometeors
• Horizontal advection
• Might not matter much on 1 grid
• Conservation properties not important for
  statistical purposes but adds to interpretability

8
Next tests

• Look closely at areas surrounding precipitation
  with a color table that exposes large gradients
  near zero
• Create a loop, check for smooth evolution
• Other model grids
• GFS or NMM
• NAEFS samples