1 Validation for CRR (PGE05) NWC SAF PAR Workshop 17-19 October 2005 Madrid, Spain A. Rodríguez.

Slides:



Advertisements
Similar presentations
Validation of Satellite Rainfall Estimates over the Mid-latitudes Chris Kidd University of Birmingham, UK.
Advertisements

Quantification of Spatially Distributed Errors of Precipitation Rates and Types from the TRMM Precipitation Radar 2A25 (the latest successive V6 and V7)
CDOP NWC SAF Workshop on Physical Retrieval Gabriela Cuevas Tascón, INM (Spain) November, 28th CDOP NWCSAF Workshop on Physical Retrieval.
1 Workshop on Physical Retrieval of Clear Air Parameters from SEVIRI SAF NWC Requirements November 2007 P. Fernández.
Validation of Satellite Precipitation Estimates for Weather and Hydrological Applications Beth Ebert BMRC, Melbourne, Australia 3 rd IPWG Workshop / 3.
1 GOES-R Precipitation Products July 27, 2011 Presented By: Bob Kuligowski NOAA/NESDIS/STAR Thanks to: Richard Barnhill, Yaping Li, and Zhihua Zhang.
A Combined IR and Lightning Rainfall Algorithm for Application to GOES-R Robert Adler, Weixin Xu and Nai-Yu Wang University of Maryland Goal: Develop and.
Convective Initiation Studies at UW-CIMSS K. Bedka (SSAI/NASA LaRC), W. Feltz (UW-CIMSS), J. Sieglaff (UW-CIMSS), L. Cronce (UW-CIMSS) Objectives Develop.
Combining GLM and ABI Data for Enhanced GOES-R Rainfall Estimates Robert Adler, Weixin Xu and Nai-Yu Wang CICS/University of Maryland A combination of.
Monitoring the Quality of Operational and Semi-Operational Satellite Precipitation Estimates – The IPWG Validation / Intercomparison Study Beth Ebert Bureau.
Satellite Rainfall Estimation Robert J. Kuligowski NOAA/NESDIS/STAR 3 October 2011 Workshop on Regional Flash Flood Guidance System—South America Santiago.
Paul Fajman NOAA/NWS/MDL September 7,  NDFD ugly string  NDFD Forecasts and encoding  Observations  Assumptions  Output, Scores and Display.
Univ of AZ WRF Model Verification. Method NCEP Stage IV data used for precipitation verification – Stage IV is composite of rain fall observations and.
PERFORMANCE OF THE H-E ALGORITHM DURING THE CENTRAL AMERICAN RAINY SEASON OF 2001.
Evaluation of NWC SAF Precipitation Products for the Adriatic Region Izidor Pelajić, P. Mikuš Jurković, I. Smiljanić, N. Strelec Mahović Meteorological.
How low can you go? Retrieval of light precipitation in mid-latitudes Chris Kidd School of Geography, Earth and Environmental Science The University of.
1 GOES-R AWG Hydrology Algorithm Team: Rainfall Probability June 14, 2011 Presented By: Bob Kuligowski NOAA/NESDIS/STAR.
The Evaluation of a Passive Microwave-Based Satellite Rainfall Estimation Algorithm with an IR-Based Algorithm at Short time Scales Robert Joyce RS Information.
1 On the use of radar data to verify mesoscale model precipitation forecasts Martin Goeber and Sean Milton Model Diagnostics and Validation group Numerical.
Testing of V1. GPM algorithm of rainfall retrieval from microwave brightness temperatures - preliminary results using TRMM observations Chuntao Liu Department.
LMD/IPSL 1 Ahmedabad Megha-Tropique Meeting October 2005 Combination of MSG and TRMM for precipitation estimation over Africa (AMMA project experience)
Anthony DeAngelis. Abstract Estimation of precipitation provides useful climatological data for researchers; as well as invaluable guidance for forecasters.
1 GOES-R AWG Hydrology Algorithm Team: Rainfall Potential June 14, 2011 Presented By: Bob Kuligowski NOAA/NESDIS/STAR.
Towards an object-oriented assessment of high resolution precipitation forecasts Janice L. Bytheway CIRA Council and Fellows Meeting May 6, 2015.
Fine-scale comparisons of satellite estimates Chris Kidd School of Geography, Earth and Environmental Sciences University of Birmingham.
International Centre for Integrated Mountain Development Validation of satellite rainfall estimation in the summer monsoon dominated area of the Hindu.
SEASONAL COMMON PLOT SCORES A DRIANO R ASPANTI P ERFORMANCE DIAGRAM BY M.S T ESINI Sibiu - Cosmo General Meeting 2-5 September 2013.
We carried out the QPF verification of the three model versions (COSMO-I7, COSMO-7, COSMO-EU) with the following specifications: From January 2006 till.
SOME EXPERIENCES ON SATELLITE RAINFALL ESTIMATION OVER SOUTH AMERICA. Daniel Vila 1, Inés Velasco 2 1 Sistema de Alerta Hidrológico - Instituto Nacional.
1 The GOES-R Rainfall Rate / QPE Algorithm Status May 1, 2011 Presented By: Bob Kuligowski NOAA/NESDIS/STAR.
Julie Haggerty National Center for Atmospheric Research Friends and Partners of Aviation Weather October July 2014.
National Lab for Remote Sensing and Nowcasting Dual Polarization Radar and Rainfall Nowcasting by Mark Alliksaar.
IPWG, M. Desbois, AMMA, Melbourne, October satellite rainfall estimation and validation in the frame of AMMA (the African Monsoon Multidisciplinary.
Latest results in verification over Poland Katarzyna Starosta, Joanna Linkowska Institute of Meteorology and Water Management, Warsaw 9th COSMO General.
1 GOES-R AWG Aviation Team: Convective Initiation June 14, 2011 Presented By: John R. Walker University of Alabama in Huntsville In Close Collaboration.
The Rapid Developing Thunderstorm (RDT) product CDOP to CDOP2
VALIDATION AND IMPROVEMENT OF THE GOES-R RAINFALL RATE ALGORITHM Background Robert J. Kuligowski, Center for Satellite Applications and Research, NOAA/NESDIS,
Evaluation of Passive Microwave Rainfall Estimates Using TRMM PR and Ground Measurements as References Xin Lin and Arthur Y. Hou NASA Goddard Space Flight.
Evaluation of gridded multi-satellite precipitation (TRMM -TMPA) estimates for performance in the Upper Indus Basin (UIB) Asim J Khan Advisor: Prof. Dr.
GII to RII to CII in South Africa Estelle de Coning South African Weather Service Senior Scientist.
TOULOUSE (FRANCE), 5-9 September 2005 OBJECTIVE VERIFICATION OF A RADAR-BASED OPERATIONAL TOOL FOR IDENTIFICATION OF HAILSTORMS I. San Ambrosio, F. Elizaga.
Fine tuning of Radar Rainfall Estimates based on Bias and Standard Deviations Adjustments Angel Luque, Alberto Martín, Romualdo Romero and Sergio Alonso.
Short Range Ensemble Prediction System Verification over Greece Petroula Louka, Flora Gofa Hellenic National Meteorological Service.
1 PGE04-MSG Precipitating Clouds Product Presented during the NWCSAF Product Assessment Review Workshop October 2005 Prepared by : Anke Thoss, Anna.
Page 1© Crown copyright 2004 The use of an intensity-scale technique for assessing operational mesoscale precipitation forecasts Marion Mittermaier and.
VERIFICATION Highligths by WG5. 2 Outlook Some focus on Temperature with common plots and Conditional Verification Some Fuzzy verification Long trends.
1 GOES-R AWG Product Validation Tool Development Hydrology Application Team Bob Kuligowski (STAR)
Convective Rainfall Rate improvements 2 nd Convection Workshop Landshut, Germany, 8-10 October 2009 Cecilia Marcos Antonio Rodríguez.
Data Analysis of GPM Constellation Satellites-IMERG and ERA-Interim Precipitation Products over West of Iran Ehsan Sharifi 1, Reinhold Steinacker 1, and.
RUC Convective Probability Forecasts using Ensembles and Hourly Assimilation Steve Weygandt Stan Benjamin Forecast Systems Laboratory NOAA.
VALIDATION OF HIGH RESOLUTION SATELLITE-DERIVED RAINFALL ESTIMATES AND OPERATIONAL MESOSCALE MODELS FORECASTS OF PRECIPITATION OVER SOUTHERN EUROPE 1st.
Evaluation of Precipitation from Weather Prediction Models, Satellites and Radars Charles Lin Department of Atmospheric and Oceanic Sciences McGill University,
Application of Probability Density Function - Optimal Interpolation in Hourly Gauge-Satellite Merged Precipitation Analysis over China Yan Shen, Yang Pan,
Dec 12, 2008F. Iturbide-Sanchez Review of MiRS Rainfall Rate Performances F. Iturbide-Sanchez, K. Garrett, S.-A. Boukabara, and W. Chen.
Validation of Satellite Rainfall Estimates over the Mid-latitudes Chris Kidd University of Birmingham, UK.
EVALUATION OF SATELLITE-DERIVED HIGH RESOLUTION RAINFALL ESTIMATES OVER EASTERN SÃO PAULO AND PARANÁ, BRAZIL Augusto J. Pereira Filho 1 Phillip Arkin 2.
The Convective Rainfall Rate in the NWCSAF
Comparing a multi-channel geostationary satellite precipitation estimator with the single channel Hydroestimator over South Africa Estelle de Coning South.
PGE05 CRR Convective Rainfall Rate
Precipitation Classification and Analysis from AMSU
Dipartimento della Protezione Civile Italiana
LPW & SAI Layer Precipitable Water and Lifted Index
Verifying Precipitation Events Using Composite Statistics
Combining GOES-R and GPM to improve GOES-R rainrate product
15 October 2004 IPWG-2, Monterey Anke Thoss
PGE06 TPW Total Precipitable Water
EUMETSAT Precipitation Week
Analysis of NASA GPM Early 30-minute Run in Comparison to Walnut Gulch Experimental Watershed Rain Data Adolfo Herrera April Arizona Space Grant.
Validation for TPW (PGE06)
Presentation transcript:

1 Validation for CRR (PGE05) NWC SAF PAR Workshop October 2005 Madrid, Spain A. Rodríguez

2 Contents of presentation General validation objectives for CRR product Summary description of the validation process Example of visual validation. General case study Example of visual validation. Usefulness on non-radar coverage areas Example of visual validation. 3D vs. 2D calibration behaviour Visual validation. Summary Results of accuracy statistics: ME, MAE, RMSE Results of categorical statistics: FAR, POD, CSI, PC Objective validation. Summary Planned activities

3 General Validation Objectives for CRR product The goal of this presentation is to compare visual and objectively the CRR values obtained from SEVIRI data with the radar information considered as the “truth data”. For this purpose convective episodes from the 01/06/05 to the 07/09/05 have been selected. The datasets used for this validation are the following:  CRR values from MSG SEVIRI SPAIN region (SAFNWC software package version v1.2)  Composite Radar imagery from the Spanish National Radar Centre: Echotop and Rainfall Rate (RFR) from PPI (at about 2Km of resolution).  IR10.8 SEVIRI imagery (at full resolution). The geographical area to match the Spanish radar area is about 1500x1500 Km centred in 40N 3W.  I.N.M Lightning data base (LDB) (Only visual validation) List of days (up to 26) used in this validation, in Julian date: 152,161,162,163,169,170,171,172,173,174,178,179,185,208,209,212,213,214,220,222,223,228,229,230, 232,250

4 General Validation Objectives for CRR product All the corrections with the default values have been applied. The fields for the moisture and orographic corrections have been extracted from ECMWF at 0.5 x 0.5 degree spatial resolution, every 6h. McIDAS software has been used in order to obtain the collocated datasets and to generate the objective validation process. The visual validation has been performed by displaying, analyzing and comparing the products in an interactive manner by using also the McIDAS environment.

5 Summary description of the objective validation process Image projection: Radar images have been re-projected to the satellite projection. Ground echoes detection: A rain image has been obtained from the IR10.8 data using the basic AUTOESTIMATOR algorithm (Vicente, G.A. et al, 1998). A pixel with significant radar echo is considered to be a ground echo and set to zero if no significant value is found in a 15x15 centred box in the AUTOESTIMATOR image. Potential convective pixels detection: When in the ECHOTOP image the ratio between the number of echoes greater than 6 Km and the ones greater than 0 Km is lower than 1% the Radar images are rejected. Convective image: A filtered Radar image has been obtained to choose the area of validation. The pixels in the RFR image are set to zero if all the nearest pixels in a 11x11 grid centred on the pixel do not reach a top of 6 Km in the ECHOTOP image and a rainfall rate of 3 mm/hr in the RFR image. Then the Radar rainrate data contained in the convective image have been matched pixel by pixel with the CRR data and accuracy and categorical statistics have been calculated for all the selected case studies. The CRR>0 classes have been assigned to the rainfall rate corresponding to the center of the class to compare the radar information (mm/hr) with the CRR values. CRR=0 has been assigned to 0 mm/hr.

6 Example of visual validation (01/08/2005 at 17:00 UTC). General case study. Comparison of CRR image (top left), Convective radar image (top right), INM lightning data (bottom left) and RFR radar image (bottom right) on 1 August 2005 at 17:00 UTC

7 Example of visual validation ( 20/08/2005 at 13:00 UTC ). Usefulness on non-radar coverage areas Comparison of CRR image (top left), Convective radar image (top right), INM lightning data (bottom left) and RFR radar image (bottom right) on 20 August 2005 at 13:00 UTC

8 Example of visual validation. (17/08/2005 at 18:00 UTC). 3D vs. 2D calibration behaviour Comparison of CRR image (top left), Convective radar image (top right), INM lightning data (bottom left) and RFR radar image (bottom right) on 17 August 2005 at 18:00 UTC

9 Example of visual validation. (17/08/2005 at 18:30 UTC). 3D vs. 2D calibration behaviour Comparison of CRR image (top left), Convective radar image (top right), INM lightning data (bottom left) and RFR radar image (bottom right) on 17 August 2005 at 18:30 UTC

10 Visual validation. Summary The convective area chosen for the objective validation is well supported by the lightning data. The CRR data are in general also well supported by the lightning data. In some cases CRR=0 when the radar shows precipitation and there is also presence of lightning data. Behaviour associated with warmer tops, mainly using 2-D calibration. In some cases CRR gives rain where radar doesn't. The CRR generally underestimates the rainfall rate and sometimes overestimates the area of precipitation. High values of radar rainfall rate are not reached by CRR. The CRR behaviour is better on well developed convection associated with the colder tops. The area of CRR precipitation corresponds mainly to the higher echoes in the ECHOTOP image. The product is useful on non-radar coverage areas. Better behaviour when using solar channel. Some areas of precipitation detected with the solar channel are lost when using 2-D cal.

11 Accuracy statistics. 3D vs. 2D Cal.NMEANMEMAERMSE 3D ,85-0,161,133,28 2D ,89-0,341,113,01 Statistics affected for a high number of pixels with no rain in both sources in the validation area :409009(3D), (2D) MEAN and MAE quite similar. ME: general underestimation. Better results using solar channel (3D). RMSE slightly greater using 3-D cal.

12 Categorical Statistics. Contingency Tables Estimated (CRR) yesno Observed (RADAR) yesH (hits)M (misses) no FA (false alarms) CN (correct negatives) Five contingency tables have been obtained using five rainfall rate thresholds: 0, 1, 3, 5 and 7 mm/h. Yes event means the rain rate (estimated or observed) is greater than the threshold. No event means the rain rate (estimated or observed) is not greater than the threshold. Categorical statistics : False Alarm Ratio FAR = [FA/(H+FA)] Probability of Detection POD = [H/(H+M)] Critical Success Index CSI = [H/(H+M+FA)] Percentage of Corrects PC = [(H+CN)/(H+M+FA+CN)]

13 Categorical Statistics. 3D Number of pixels processed: ThresholdContingencytable FARPODCSIPC 0mm ,350,420,350, mm ,690,490,240, mm ,820,200,10, mm ,840,100,060, mm ,850,030,020,

14 Categorical Statistics. 2D Number of pixels processed: ThresholdContingency tableFARPODCSIPC 0mm ,330,290,250, mm ,680,330,190, mm ,840,140,080, mm ,890,03 0, mm ,910,01 0,

15 Categorical Statistics. 3D vs. 2D: FAR, POD FAR is quite similar in both calibrations. Slightly higher when using 2D cal. for rates greater than 3 mm/hr. POD is always better when using 3D cal. Slightly higher detecting rates greater than 1mm/hr.

16 Categorical Statistics. 3D vs. 2D: CSI, PC CSI is always better when using 3D cal. PC slightly higher when using 3D cal.

17 Objective validation. Summary Better results in general using the solar channel (3D). General underestimation of the algorithm. No CRR estimated rainfall rate greater than 20 mm/hr on this study. Many cases on Radar rates. The probability of detection if the observed rates overcomes a threshold is better for the low ones. The false alarm ratio and the critical success index have also better values in these cases. The value of these scores are very poor because the pixels with precipitation in both (CRR and Radar) images doesn't matched very well geographically. CRR based on cloud top measurements Vs. the radar first elevation data. The percentage of corrects have better values because includes the correct negatives which are greater for high thresholds. Better results are expected if using an instantaneous validation by grouping pixels in lat/lon boxes or working with accumulations.

18 Planned activities To use other methods of validation: Validate instantaneous rainrates using lat/lon boxes Validate rainfall accumulations Study the impact of applying or not the corrections on the validation result. Calibration based on mm/hr instead of classes in order not to lose information on basic and corrected CRR values. Study the Impact of using of other channels