Combining Meteosat data and weather radar products to improve the meteorological surveillance and nowcasting

Transcription

1 Combining Meteosat data and weather radar products to improve the meteorological surveillance and nowcasting R. Hernandez (1)(2), S. Gaztelumendi (1)(2), K. Otxoa de Alda (1)(2), I. R. Gelpi (1)(2), J. Egaña (1)(2) (1) Basque Meteorology Agency (EUSKALMET) Parque Tecnológico de Álava. Avda. Einstein 44 Ed.6 Of Miñano, Alava, Spain. (2) European Virtual Engineering Technological Centre (EUVE), Meteorology Division. Avda Huetos 79, Edificio Azucarera, Vitoria-Gasteiz, Alava, Spain. Abstract This paper is a first attempt to combine meteorological satellite data and weather radar products in the Basque Meteorology Agency (Euskalmet). An intensive use of different Meteosat-9 data, including some available products developed by EUMETSAT, and products from the C-band Doppler weather radar of Euskalmet is done. As daily routine, the combinations are inspected to get an expertise of errors. They also facilitate the surveillance and nowcasting tasks. A simple loop of the radar Plan Position Indicator (PPI) and infrared channel composition for itself is a valuable tool to get a general survey of a precipitation event. The combination with other radar products offers to the forecaster additional information for a better interpretation. The synergies between both sources of information are also shown useful in order to find spurious echoes in weather radar data. The new derived products and diagnosis images are part of the Surveillance Panel installed at the Euskalmet s office, a Mitsubishi s Display Wall Cube for monitoring hidrometeorological information in real time. INTRODUCTION From the standpoint of atmospheric monitoring, the key element is the observational data. Remote sensing data -satellite, radar and lightning- stands out because of their spatio-temporal resolution. It is very important to integrate all these sources of meteorological information for a better interpretation of meteorological events. New products can be generated by means of this merging. For example, they are numerous projects related to the calculation of multi-sensor rainfall fields, under the assumption that final multi-sensor precipitation estimation is better than any single sensor. The synergy between satellite and radar offers the opportunity to achieve a better exploitation in the radar operation. This work shows two case examples of errors detection: radomo attenuation and anomalous propagation. We would like to highlight the usefulness of the products developed by the Satellite Application Facility on Nowcasting and Very Short Range Forecasting (NWC SAF). In particular, the Precipitating Clouds product (PC1) can be applied as a mask of spurious radar echo. By other hand, the use of GIS is increasingly common in the visualization and spatial analysis of data from different disciplines, including meteorology. We have therefore made an effor in which the products are in standard exchange formats, such as geotiff. DATA The following table shows the different data used to carry out this work.

2 Data Description Temp. resol. Format Meteosat-9 EUMETCast system installed in Euskalmet. 15 min XPIF VCS Support to Nowcasting and Very Short Range Forecasting. NWC SAF 15 min MSG Precipitation & Convection Products Image from SAFNWC Desk METEOR 1500 Doppler Radar with Dual polarization capabilities from Gematronik. It ASCII Kapildui operates in C-band frequency. 10 min Rainbow from Radar Two volumetric scans are used with ranges of Gematronik 300 and 100 km respectively. Table 1: Data description. TOOLS We have implemented functions in IDL (Interactive Data Language, from ITTVIS) to process and display data. The routines for multi-sensor data reading, processing and product composition were developed in Euskalmet. We also used SPT software (SEVIRI Pre-processing Toolbox) (Y. Govaerts et al., 2007) for Meteosat-9 navigation purposes: to convert each line and column number into geographical latitude and longitude and vice versa. WARPING METEOSAT IMAGES The radar data was assumed to have the correct Earth-relative navigation compared to other datasets. For this reason, satellite images are remapped to the radar projection. Specifically, we used IDL MAP_PROJ_* functions to warp Meteosat images from Satellite to 'Azimuthal Equidistant' projection. The MAP_PROJ_IMAGE function warps an image from one map projection to another. The IMAGE_STRUCTURE keyword requires the map structure for the study image -the image we are warping-, while the MAP_STRUCTURE keyword requires the map structure of the overflight image (the map projection we are warping the image to). Such structures are defined with MAP_PROJ_INIT. This function initializes a mapping projection, that is, establishes the coordinate conversion mechanism for mapping points on a globe's surface to points on a plane. Figure 1: Meteosat-9 IR /03/ :15 UTC. Geostationary Proyection (top) and Azimuthal Equidistant (below). BLENDING

3 We used successfully the available alpha blending channel in object-graphic images, to combine the colours of the two images we were trying to overlay. Transparency is applied to radar image background. Figure 2: Example of image blending. QUALITY CONTROL OF WEATHER RADAR DATA Meteosat-9 imagery and derived products provide valuable assistance in identifying errors in radar data. Here we present two case examples. Radomo attenuation The analysis of Z and ZDR radar products confirmed that data was affected by radomo attenuation on 17th September 2007, due to a strong precipitation over Kapildui area (Maruri M. et al., 2006). In real time, overlapping SEVIRI channels, RDT, CRR, etc could reveal the discontinuity of the data, especially when the time series are analysed. In this sense, the vector field characteristics can be used as an input to the anomaly detection process, showing anomalous echoes having more or less chaotic behaviour from the movement detection point of view. Figure 3: Meteosat-9 IR 10.8 and Kapildui PCAPPI 100km from 17/09/2007 between 09:52 and UTC, with convective cell tracks and centroids. Central panel shows the radomo attenuation. Also a parallax effect can be observed. Anomalous propagation NWCSAF Precipitating Clouds product (PC1) is particularly interesting in relation to filtering spurious radar echo. It can act as a mask that is superposed on the radar data, identifying which echoes are not coming from a precipitating cloud. Thus, the basic principle to validate the radar observations is very simple: a radar echo must come from a region where a precipitating cloud is present, otherwise this echo is suspicious of being originated by effects of anomalous propagation (Magaldi A, 2006). A similar methodology has been carried out to calibrate the PGE05 product (CRR) of the SAFNWC/MSG software package. The process of removing false echoes is made through a rain image that has been obtained from the IR10.8 data using the basic AUTOESTIMATOR algorithm. A pixel with significant radar echo is considered to be a ground echo and set to zero if no significant

4 value is found in a 15x15 centered box in the AUTOESTIMATOR image. A more sophisticated alternative was pointed out by Bøvith et al. (2006) who performed a supervised classification of the radar echoes into clutter and precipitation classes. Moreover, clutter masks could be elaborated using this principle over clean air days. Figure 4: Kapildui radar PPI 300 Km 0.5º (left) and excerp of SAFNWC PC1 image (right) - 20/07/ :30 UTC. Figure 5: Identification of false echoes in PPI 300km 0.5º. Only reflectivity into red circle matchs with rain. This event happened in a period of mechanical breakdown of Kapildui radar. SHARING INFORMATION Some export utilities have been included that provide the images in geotiff (TIFF with geospatial information tags), a suitable format for sharing nowcasting products with other operational meteorologists for analysis and feedback. It can also be imported into most GIS software, including Google Earth pro. Nowcasting tasks can benefit from the capacities of visualization and spatial analysis of GIS. By other hand, Google Earth can be extremely useful in supporting civil protection activities. For instance, polarimetric Doppler C-Band radars are able to discriminate the phase of precipitation - solid, liquid, and mixed-, so we can locate areas of heavy snowfall and roads affected.

5 Figure 6: Meteosat-9 IR10.8 and Kapildui PCAPPI 300 km from 17/09/ :30 UTC viewed as fields in Google Earth. General view (left). Radar detail over Vitoria-Gasteiz (right) with road network. OUTLOOK The synergies between both sources of information are shown useful in order to find spurious echoes. These echoes could be removed of the radar image because they don t fulfil the basic principle to come from a precipitating cloud region. The new compositions also facilitate the surveillance and nowcasting tasks. We have adopted geotiff format and Google Earth platform as a way to share new weather products with other researchers and operational meteorologists for evaluation and feedback. The new derived products will be part of the Surveillance Panel installed at the Euskalmet s office. FUTURE WORK This is a preliminary study, so there are still many tasks to be developed. The first, apply parallax correction to SEVIRI channels and solve the different temporal resolution between Meteosat-9 and Kapildui radar images. We have also found PC1 product particularly useful in relation to filtering spurious radar echo, but an evaluation must be made. ACKNOWLEDGEMENTS We thank Meteorology and Climatology Directorate staff (Basque Government) for providing radar and Meteosat data, and EUMETSAT for the availability of NWC SAF products. We also would like to thank Interior Department of the Basque Government for operational service financial support, and all our colleagues from EUSKALMET for their daily effort in promoting valuable services for the Basque community. REFERENCES Bøvith T., Gill R. S., Overgaard S., Hansen L.K. Nielsen A.A. (2006) Detecting weather radar clutter using satellite-based nowcasting products. ERAD 2006, Barcelona, Spain. Celano M., Siviero F., Poli V., Alberoni P.P., Di Giuseppe F. (2008): Using Google Earth Visualization Platform to Support the Analysis of Severe Weather Case Studies. ERAD Gaztelumendi S., Egaña J., Gelpi I.R., Otxoa de Alda K., Maruri M., Hernández R. (2006) The new Radar of Basque Meteorology Agency: configuration and some considerations for its operative use. ERAD 2006, Barcelona, Spain.

6 Gaztelumendi S., Otxoa de Alda K., Hernández R., Egaña J., Gelpi I.R. (2010) Meteosat products for surveillance: the euskalmet case. EUMETSAT 2010, Córdoba, Spain. Govaerts Y., Wagner S., Clerici M. (2007) SEVIRI Native Format Pre-Processing Toolbox User s Guide Version 3.0. EUMETSAT, report EUM/OPS-MSG/TEN/03/0011, 88 pp. Magaldi A., Bech J., Delgado G., Lorente J. (2006) Filtering weather radar AP echoes with MSG observations and NWP data. ERAD Barcelona, Spain. Maruri M., Gaztelumendi S., Egaña J., Otxoa de Alda K., Hernández R., Gelpi I.R. (2008) Product quality monitoring of Kapildui weather radar during critical meteorological events. ERAD 2008, Helsinki, Finland.