330 likes | 344 Views
Satellite precipitation validation methodology with ground truth observation over Turkey. İbrahim SÖNMEZ 1 , Ali Ümran Kömüşçü 1 Ahmet ÖZTOPAL 2 , 1 Turkish State Meteorological Service, Turkey 2 Istanbul Technical University, Turkey. Content. Ground observation Network
E N D
Satellite precipitation validation methodology with ground truth observation over Turkey İbrahim SÖNMEZ1, Ali Ümran Kömüşçü1 Ahmet ÖZTOPAL2, 1Turkish State Meteorological Service, Turkey 2Istanbul Technical University, Turkey
Content • Ground observation Network • Validation metedology • Scatter & pdf plots • Validation statistics • Results • Discussion
Ground Observation Netwok Sites : AWOS (Automated Weather ObservationStation)
Ground Observation Netwok Sites : AWOS (Automated Weather ObservationStation) • Observed at every 1 minute: • Precipitation • Temperature • Relative humidity • Wind speed & direction • Observed at every 10 minute: • Pressure • Soil temperature at different depths • Evaporation
Flag Value Status Brief Description 0 Good Datum has passed all QA Test 1 Suspect There is concern about accuracy of datum 2 Warning Datum is very questionable 3 Failure Datum is unstable RG Quality Control Procedures • Tests applied to the precipitation data: • Range Test (climatological comparison) • Step Test (sequential observation comparison) • Persistance Test (group of observation comparison)
Validation Methodology FOV for AMSU sensor AMSU(H02) product FOV centers 20090201 @ 00:28
Validation Methodology AMSU(H02) product FOV centers [20090201 @ 00:28] * : awos + : product
Validation Methodology AMSU(H02) product FOV centers [20090201 @ 00:28] * : awos + : FOV center
Either, Or, Validation Methodology • Observation Product * : awos + : FOV center point area How to compare these two?
Validation Methodology * : awos + : FOV center
? Validation Methodology * : awos + : FOV center
S2 S1 W2 W1 SE SN WN W3 S3 . . W. W. Observation value at an unobserved location • W(r) is assumed to be: • Function of distance • Closer sites indicating higher weight • Farther sites indicating less weight!
Weighting functions • Weighting functions: • Theorocially defined • does not depend on observation itself Cressman (1951) Barnes (1964) W
Weightingfunctionalternative: Variogram Definition: • Variogram (V)counts on the relative change between two observations for the spatial variability and defined by the function of distance as, • Semivariogram(SV)is proposed by Matheron (1963) to provide the spatial dependence. Clark(1979) defined the mathematical formulation as,
AlternativeWeightingfunction: • Cumulative Semivariogram(CSV)concept has been proposed to the earth sciences by Sen(1989) for depicting spatial correlation structure of regionalized variables.
CSV study by Şen(1997) Şen, Z., 1997: Objective Analysis by Cumulative Semivariogram Technique and Its Application in Turkey, J. Of Applied Meteorology, 36, 1712-1724. • Monthly rainfall amountsin Marmara region(north-west part of Turkey) are used • Sites with at least 30 years of data are selected
AlternativeWeightingfunction: Point cumulative semivariogram (PCSV)was proposed by Sen (1995) where it measures the spatial variability around a site to provide the regional effect of all other sites within the area on the site of concern. * Specific SWF for every station can be defined
W(r) determined by PCSV (Sen,1996): • The observations are used rather than theorotical functions, such as Cressman or Barnes. Why PCSV is needed? r r * : awos + : FOV center
Summary • Before validation study is performed, there are couple of things to consider: • Sampling feature of the sensor OR geographical representation of the product. • What to use in validation (ground truth, or some other product) • The strategy to perform the validation(point to area OR area to point) • Use of the descriptive statistics(e.g., correlation, std, FAR, POD, etc)
Summary & Conclusions • H01 :overestimation is dominant • H01 : Relatively lower RR are observed over cost than land • H02 :overestimation is dominant • H02 : RR over cost ~= RR over land • H02 :increasing RR amounts through summer is observed • RR amount of H02 < RR amount of H01 for a particular month
Summary & Conclusions • H03 :underestimation is dominant • H03 : no significant difference over land & cost • HO3 : relatively lower RR are observed respect to H01 & H02 product
Summary & Conclusions • JFM period : highest correlation @ H02 among three products • AMJ period :highest correlation @ H01 among three products • H02& H03: Highercorrelation occured in winter. • Decreasingstd trend through summer months is observed for H01&H02 • Increasingstd trend through summer months is observed for H03 • Higheststd in winter months is observed in H01 while H03 has the highest amount in summer time.