cosmo pp qpf workshop
DESCRIPTION
COSMO PP QPF workshop. 8 March 2006 Langen Marco Arpagaus. Hello & thanks!. Hello to everybody … … and thanks for coming! Special thanks to Silke for the organisation!. Motivation. Improve QPF of LM! – Or at least try … How? Understand the problem! Suggest changes. Is it any better? - PowerPoint PPT PresentationTRANSCRIPT
COSMO PP QPF workshop
8 March 2006Langen
Marco Arpagaus
Hello & thanks! Hello to everybody … … and thanks for coming! Special thanks to Silke for the organisation!
MotivationImprove QPF of LM! – Or at least try …
How? Understand the problem! Suggest changes. Is it any better? Would things improve in an ,ideal’ world?
Motivation (2) Task 1: Consolidate forecast failure reports and
verification findings. Task 2: Provide standardized set of model
changes to be used for sensitivity studies. Task 3: Run the sensitivity experiments and draw
conclusions concerning possible improvements of the LM QPF performance.
Task 4: Run the same sensitivity experiments for moist benchmark cases.
Motivation (3) Task 1.1: Consolidate QPF related problem
reports from all COSMO member states. Task 1.2: Consolidate QPF related verification
results from all COSMO member states and provide prototype cases reflecting the observed QPF problems.
Task 1.3: Condense the lists provided by tasks 1.1 and 1.2 by selecting the typical and most obvious cases illustrating the poor QPF performance of the LM.
Task 1.4: Provide LM reference version. Task 1.5: Run test cases with LM reference version to confirm QPF problems. Task 1.6: Reduce list of test cases recommended for sensitivity studies to a maximum
of 3 cases for each LM implementation.
Motivation (4)What do we want to achieve at this workshop? Share the results we have obtained so far … … as well as the problems we encountered! Are we on track? A first step towards task 1.3. Outlook to tasks 1.4 – 1.6. Can we profit from the common verification
package?
…
First (online …) summary ‘continental’ results
overprediction of pp mean in winter overprediction of pp max in summer
‘mediterranean’ results modulation of results depending on upstream
conditions? (flow over sea or land)
First (online …) summary (2) pp related to frontal systems (cold & warm) pp related to orography (luv or lee) pp dependency on flow regime combination(s) of the above
pp dependency on upstream flow characteristics (e.g., over sea/land)
classify accordingly
Methodology synthesise! look at day 1 pp (& check synoptic situation) consider absolute as well as relative bias; emphasis on area
mean rather than single maxima (upscaling of obs!) look at relative contribution of stratiform and convective pp
in the model classify as stratiform or convectively dominated cases
for Italy (and Switzerland?): try to separate luv/lee effects from land/sea effects
Motivation (4)What do we want to achieve at this workshop? Share the results we have obtained so far … … as well as the problems we encountered! Do we need to re-formulate the task? A first step towards task 1.3. Outlook to tasks 1.4 – 1.6. Can we profit from the common verification
package?
A first step towards task 1.3 Agree on common verification measures. Agree on how the cases should be described
(deliverable for tasks 1.1 and 1.2). Agree on how the verification material should be
provided (deliverable for task 1.3)
A first step towards task 1.3 (2)Common verification measures (COSMO Standard): 6-hourly precipitation sums contingency tables for thresholds 0.1 mm, 2 mm,
and 10 mm Bias, POD, FAR (after Wilks 1996)
… plus individual scores.
…
Motivation (4)What do we want to achieve at this workshop? Share the results we have obtained so far … … as well as the problems we encountered! Do we need to re-formulate the task? A first step towards task 1.3. Outlook to tasks 1.4 – 1.6. Can we profit from the common verification
package?
Outlook to tasks 1.4 – 1.6 Task 1.1: Consolidate QPF related problem reports from all COSMO member states. Task 1.2: Consolidate QPF related verification results from all COSMO member states
and provide prototype cases reflecting the observed QPF problems. Task 1.3: Condense the lists provided by tasks 1.1 and 1.2 by selecting the typical and
most obvious cases illustrating the poor QPF performance of the LM.
Task 1.4: Provide LM reference version. Task 1.5: Run test cases with LM reference
version to confirm QPF problems. Task 1.6: Reduce list of test cases recommended
for sensitivity studies to a maximum of 3 cases for each LM implementation.
…
Motivation (4)What do we want to achieve at this workshop? Share the results we have obtained so far … … as well as the problems we encountered! Do we need to re-formulate the task? A first step towards task 1.3. Outlook to tasks 1.4 – 1.6. Can we profit from the common verification
package?
…