CGMS Inter-comparison of Satellite-tracked Winds

Transcription

1 CGMS Inter-comparison of Satellite-tracked Winds. Introduction CGMS has concentrated its efforts to enhance the utilization and improve the quality of satellite-tracked winds. Under the auspices of CGMS, Working Group on satellite-tracked winds organizes workshops on regular basis, at intervals of about 8 24 months that facilitate exchange of recent progress in the science and utilization of the products in this area. The CGMS Working Group on Cloud Motion Vectors has established a series of workshops that focus on the science and operational development and use of atmospheric motion vectors from geostationary satellites. 2. International Wind Workshops The CGMS Working Group on Cloud Motion Winds (WG-CMW) has been established to continue and emphasize the CGMS accomplishments and objectives in the area of operational extraction of Cloud Motion Winds (CMW) from satellite data. This includes coordination of complementary and compatible operational procedures, development of common verification and validation procedures, and encouragement of a robust program of scientific research in this technology. The objectives of the WG-CMW are: a) to devise and implement regular procedures for the exchange of data on inter-comparisons of operational CMW; b) to promote similar and, where feasible/practical, standard operational procedures for deriving CMW; c) to establish standards for verification and validation of CMW derived from satellite data, including - selection of data sources for validation, - standardization of statistical parameters to be used for verification and inter-comparison, and - standardization of verification criteria, i.e. standard windows in space and time for collocations and standard criteria for the acceptance (or consideration) of the validation data; d) to promote increased scientific activity in this field, and to establish routine means of exchanging scientific results and progress; e) to encourage regular scientific and operational production information exchange regarding - an agreed upon designation of data quality as a part of the delivery of the data (e.g. quality flags), - common data formats and codes, and - means for verifying the usefulness and quality of the data in numerical analyses and prediction.

2 The WG-CMW organizes workshops, co-sponsored by CGMS members. The Workshops promote the exchange of scientific and operational information between the producers of CMW, the research community, and the user community. 3. Utilization of Atmospheric Motion Vectors (AMVs) Global observations of atmospheric wind fields are potentially the most important data in the analysis for numerical weather prediction. Direct observations of wind fields are indispensable at low latitudes where winds cannot be inferred from the mass field. Wind observations from satellites also constitute the sole source of wind data over wide regions of the Southern Hemisphere. Very good has been made over the last fifteen years in the derivation of winds (or atmospheric motion vectors: AMVs) at satellite operating centers and, on the user side, at numerical weather prediction centers in their capability to assimilate the wind information into numerical forecast models. owadays AMVs are a well-established and important ingredient in the global observing system. CGMS has played the key role in fostering a continuous improvement in the winds product and its utilization. This progress has largely been achieved through the working groups that are part of the regular CGMS meetings. The establishment of the International Winds Workshops (IWW), with the first workshop held in 99 in Washington D.C., initiated a close cooperation between satellite operators, users at WP centers and the science community. Five IWWs have been held since then: the second in Tokyo, Japan in993, the third in Ascona, Switzerland in 996, the fourth in Saanenmöser, Switzerland in 998 and the fifth one in Lorne, Australia in The fruitful outcome of these workshops is published as workshop proceedings by EUMETSAT. An effort is also made to communicate the workshop results succinctly to the open science community through publication of a workshop report in a peer-reviewed journal (Schmetz et al., 997 and 999). 4. Wind Workshop Achievements and Issues The specific accomplishments of the most recent IWW5 were: (a) continued expansion of the winds user community resulting from enhanced Education and Training efforts as well as improved wind distribution with BUFR formats; (b) continued generation of global wind data sets from two GOES, two Meteosat, and one GMS satellite; (c) concurrence on the advantages of high density winds for various applications including hurricane trajectory forecasting; (d) convergence toward a combined set of automatic quality flags and progress toward improved utilization by WP centers; (e) expanded awareness of geometric approaches (stereo, shadows) for 2

3 validation of cloud height assignment with a combination of LEO-GEO sensors (f) agreement to distribute WP center satellite wind monitoring via the web; (g) recommendation that all satellite wind producers give priority to reprocessing winds from archived data to enable meaningful reanalysis of long-term data sets; and (h) anticipation of new technologies (wind lidars, interferometers, and scatterometers) for improved motion vector determinations. 5. International Comparison of Satellite Winds As one of the early joint activities of CGMS it was agreed to perform on a regular basis inter-comparisons of satellite tracked winds in order to assess the homogeneity and accuracy of this product. comparison had been proposed and accepted originally: Two forms of - direct inter-comparison between satellite winds in the areas of overlap between adjacent satellites, - inter-comparison with conventional data CGMS X agreed to continue the international comparison program with the following modifications: - biased Rawinsonde reports would be eliminated by analyzing Rawin stations to identify those with persistent errors and omitting those reports from the comparison program, - the "collocation box" would be refined. This could be done by using an elliptical collocation area whose major axis was oriented along the wind direction and whose length was proportional to the wind speed. Sharply reducing the length of the minor axis would minimize the comparison on winds that lay on opposite sides of shear lines and remove statistical differences, which were real-time space variations. CGMS then agreed on the collocation ellipse parameters in the following table. Table Collocation ellipse parameters agreed by CGMS Wind Speed Major Axis Minor Axis High & medium level winds less than 0 m/s 225 km 75 km from 0 to 25 m/s 250 km 60 km Greater than 25 m/s 300 km 00 km Low level winds any speed 225 km 75 km At CGMS XII, ESA pointed out that they were continuing to use latitude-longitude collocation boxes for the 3

4 comparison of Meteosat satellite winds and rawinsonde data. Several tests had been carried out using the elliptical collocation area as recommended at CGMS XI. However, the number of comparisons thus obtained was considered to be too few to have any statistical significance. At the CGMS XXIII the Working Group on Satellite Tracked Winds recommended that evaluation of operational wind production quality should be accomplished with a new standardized reporting method. They recommended three parts to the report. () Monthly means of speed bias and RMS vector difference between radiosondes and satellite winds for low- (> 700 hpa), medium- ( hpa), and high- (< 400 hpa) levels together with the radiosonde mean wind speed. This should be done for three latitude bands: north of 20, the tropical belt (20 to 20S), and south of 20 S. (2) Trends of the evaluation statistics for the monthly cloud motion vectors and water vapor motion vectors through the last 2 months. (3) Information on recent significant changes in the wind retrieval algorithm. This reporting has now been established as routine. The exact reporting format had been proposed at IWW3 in 996 and the relevant subsection from the Report of the Working Group on Verification is reproduced below: The WG started with a discussion of an appropriate reporting format for the comparison of Cloud Motion Vectors (CMV) with radiosonde data. The goal of the reporting is to assist in achieving international production of like quality motion vectors. It was noted that the working paper submitted by the US at CGMS 24 was a good starting point. The WG suggested reporting,,,, CMV, and C for low (>700 hpa), medium (700 to 400 hpa), and high (<400 hpa) levels for all winds as well as those segmented by latitude bands in the northern extratropics (north of 20), tropics (20 to 20S), and southern extratropics (south of 20S). Some definitions follow for clarification: The mean vector difference () is given by, ( ) = ( VD) i. i= where the vector difference (VD) i between an individual CMV report (i) and the collocated rawinsonde (r) report used for verification is, ( ) ( ) 2 2 ( VD) = U U + V V. i i r i r The root-mean-square error () traditionally reported is the square root of the sum of the squares of the mean vector difference and the standard deviation about the mean vector difference, ( ) ( ) 2 2 ( ) = + SD. where the standard deviation (SD) about the mean vector difference is, 2 (( VDi ) ( ) ) ( SD) =. i= 4

5 The speed bias () is given by ( ) i = ( Ui + Vi Ur + Vr ) i= The number of wind vectors produced is given by CMV and the number of collocations found with raobs is indicated by C. Collocation with radiosondes should be within 50 km. 5

6 Table 2 Example of a reporting template ALL REGIOS H EX-TROP TROP SH EX-TROP ALL LEVELS CMV C HI CMV C MID CMV C LOW CMV C This reporting should be done for Cloud Motion Vectors (CMV) derived from infrared window images, Water Vapor Motion Vectors (WVMV) derived from water vapor images (indicating whether only gradients in cloudy 6

7 regions were tracked or both cloudy and clear; separation of cloudy and clear statistics is desirable, if possible), motion vectors derived from visible images (VISMV), and the total combined wind field (TOTMV). Statistics should be reported for three-month segments (Dec to Feb, Mar to May, Jun to Aug, and Sep to ov) and should be submitted the month after the segment is finished. In addition, a plot of the monthly and of the full disk winds for each wind type category for the last twelve months also should be submitted to assist in indicating trends. A history of processing changes should also be appended. It is recognized that existing trend analysis produced locally at different operational wind production centers may be based on different collocation requirements or statistical parameters; maintaining these will require some additional effort at each site. The issue of the adequate collocation box for comparison of AMVs with radiosondes has been addressed the 4 th International Winds Workshop. A study was performed by JMA (Tokuno, 998) investigating the differences in collocation statistics between the elliptical collocation areas proposed earlier and a circular one. JMA noted that differences exist with regard to RMS vector difference but no difference was discerned for the bias. The relevant discussion at IWW4 should put at rest the continuing arguments for elliptical areas. It is suggested that satellite operators continue their well-established collocation statistics in order to maintain a consistent monitoring. It is also noted that the data monitoring performed at WP centers provides a unified database for quality monitoring of AMVs (c.f. statistics from WP SAF on web page). However, as note of caution it is added that WP monitoring also has caveats which potentially distort the statistics (e.g. positive feedback in terms data quality when data are used and little other data are available or to the contrary, too low a weight to AMVs in comparison to other wind data and the first guess, which might not be correct). Reference: CGMS Secretariat: Consolidated Report of CGMS Activities, 9 th edition, Version, 2 March 200 Tokuno, M., 998: Colocation area for comparison of satellite winds and radiosondes. Proceedings of the 4 th International Winds Workshop, Saanenmöser, Switzerland, October 998, EUM P 24, pp Remarks: This document fully depends on the Consolidated Report of CGMS Activities, 9 th edition, Version, 2 March 200, which is issued by CGMS Secretariat (EUMETSAT). 7