Wide Band Power and Harmonic Amplitude of Precipitation

Alex RuaneJohn RoadsScripps Institution of Oceanography / UCSDRamat Gan, Israel: July, 2006This work supported by NOAA and NASA

OverviewIntroduction and Background

Datasets and Methodology

General Wide Band Behavior

Discussion

Motivation for Diurnal ExaminationCurrently, atmospheric parameterizations in global reanalyses tend to focus on longer-term means at the expense of extreme eventsToo many small storms, not enough floods or droughtsAGCMs are not producing proper statistical distributions of sub-seasonal events

Goal is to simulate proper evolution, exchanges, and statistics of water and energy cycles throughout the day over diverse regions at multiple spatial and temporal scalesWhat are the frequency characteristics of reanalysis output? of satellite products?Are the limitations in parameterization or dynamics?How are biases passed throughout the water and energy cycles?

Data SetsGSM-based Global Reanalyses T62L28, 3-hourly output, initialized 4x dailyCovering 200107 200412Weekly variation in SSTSatellite Precipitation ProductsNCEP / DOE Reanalysis-2ECPC SFM Reanalysis CMORPH60 N/S, 3-hourly valuescovering 200301 - 200512 PERSIANN50 N/S, 6-hourly valuescovering 200107 - 200412TRMM 3B4250 N/S, 3-hourly values covering 200107 - 200412

MethodologyHarmonic Reconstruction AnalysisFit diurnal and semidiurnal harmonics to 3.5 years of precipitation at 3-hr resolutionNo filling requiredFourier-based Wide BandsFilled in missing points with a cubic spline interpolationPerformed Fast Fourier Transform on 3-hr time seriesAveraged frequency bands into several wide bandsLow-Frequency: Periods longer than 30 daysSynoptic: Periods between 2 days and 30 daysHigh-Frequency: Periods shorter than 2 daysSummertime Precipitation Example

Downward SW Radiation Flux from SFMLow-FreqSynopticHigh-FreqHarmonic

Prior ExpectationsGeneral theory for precipitation would predict:Low-frequency power in ITCZ and at high-latitudes Synoptic power over mid-latitude storm tracks High-frequency power in ITCZ and over landEspecially arid regions

Strong diurnal and semidiurnal harmonics over:CoastlinesMountainsITCZ

Diurnal harmonic often described as explaining high percentage of variance over land areas

TRMM BehaviorLow-FreqSynopticHigh-FreqHarmonic

CMORPH BehaviorLow-FreqSynopticHigh-FreqHarmonic

PERSIANN BehaviorLow-FreqSynopticHigh-FreqHarmonic

SFM BehaviorLow-FreqSynopticHigh-FreqHarmonic

RII BehaviorLow-FreqSynopticHigh-FreqHarmonic

Zonal Mean Land/Sea Variance Explained

Discussion (1)General Frequency CharacteristicsLow-frequency power is very weak for precipitationLikely due to unchanging baseline value of non-precipitating eventsHigh-frequency variability dominates most locationsLikely due to short-lived convective eventsLand has more power in high-frequency than oceanParticularly over arid regionsOceans have more synoptic power than landParticularly over sinking portion of Hadley Cell

General Harmonic CharacteristicsExtremely low variance describedLikely due to sporadic nature of precipitationMost variance explained over mountainous regions and regions where there were no strong events

Discussion (2)Both approaches are enlighteningFourier approachFull frequency characteristicsRelies on interpolation of missing valuesStrongly affected by temporal averagingHarmonic approachCan be used on irregularly observed dataPerforms very well when applied to strongly averaged seriesMay be restricted to physically understood variationsDifferencesFourier approach brings out much more of the expected high-frequency variationCharacteristic precipitation behaviors may be more readily identified using Fourier approach

Discussion (3)Large biases evident in frequency characteristicsAmong satellite productsTRMM appears to over-emphasize high-frequency variationsCMORPH appears to behave the most like our expectations, but is overwhelmed by particularly strong eventsPERSIANN has the whitest spectrumAmong GSM reanalysesSFM has spuriously large low-frequency and synoptic power over tropical oceansRII has a spectrum that looks much more like we expected

Future Work

Apply these approaches to each component of the water and energy cycle

Run long-term experiments to test the sensitivity of these components frequency characteristics to: Land-surface schemesCloud schemesConvective schemes