Published on February 25, 2008
Slide1: MW-IR satellite rainfall measurements over Northern Italy for the monitoring of heavy rains V. Levizzani1, F. Torricella1, R. Amorati1,2, F. J. Turk3, F. Meneguzzo4, and P. P. Alberoni2 1CNR, Institute of Atmospheric and Climate Sciences, Bologna, Italy 2ARPA, Regional Meteorological Service, Bologna, Italy 3Naval Research Laboratory, Marine Meteorology Div., Monterey, CA, USA 4CNR, Institute of Biometeorology, Firenze, Italy The Mesoscale Alpine Programme (MAP) is an international research initiative devoted to the study of atmospheric and hydrological processes over mountainous terrain. A large-scale field phase in the Alpine region took place from 7 September to 15 November 1999. During this Special Observing Period (SOP) a number of Intensive Observation Periods (IOP) were conducted and a large database of observations was created, including surface, upper air, aircraft and remote sensing data. Primary scientific objectives were: 1a. Improve the understanding of orographically influenced precipitation events and related flooding episodes involving deep convection, frontal precipitation and runoff. 1b. Improve the numerical prediction of moist processes over and in the vicinity of complex topography, including interactions with land-surface processes. 2a. Improve the understanding and forecasting of the life-cycle of Foehn-related phenomena, including their three-dimensional structure and associated boundary layer processes. 2b. Improve the understanding of three-dimensional gravity wave breaking and associated wave drag in order to improve the parameterization of gravity wave drag effects in numerical weather prediction and climate models. 3. Provide data sets for the validation and improvement of high-resolution numerical weather prediction, hydrological and coupled models in mountainous terrain. IOP-2B: Frontal system and heavy rain over the Lago Maggiore region 18-21 September 1999 From the Operations Science Director Report: “The trough that was far to the west during IOP2A moved rapidly over the region of northern Italy during the period of 18-21 September 1999. When the trough affected northern Italy on 20 September it extended far south, into northern Africa, and the southwesterly jet ahead of the trough lay directly over the the Lago Maggiore region. A frontal cloud system with apparent embedded convective elements swept across northern Italy in association with the trough.” The METEOSAT 7 image shows the convective (cyan) elements. The through reached the MAP area around 1200 UTC on September 20. Rainfall was heavy throughout the region. The accumulation of rainfall in the mountains on the western side of the Lago Maggiore region was particularly large. This case had strong similarities with previous flooding storms, especially the famous Piemonte flood (November 1994). IOP-5: Frontal passage and cyclogenesis over Northern Italy 2-5 October 1999 From the Operations Science Director Report: “2 October: The Swiss model indicated that an upper trough would be over western France by 1200 UTC, with the predicted southwesterly flow over the Maritime Alps and Ligurian Appennines producing mountain waves and lee-side subsidence. 3 October: The front moved very slowly during the day. Precipitation appeared in a SW-NE line of moderate convective cells along the foothills of the Lago Maggiore region by 0835 UTC. The line remained in this position for many hours. By 1235 UTC the northeastern portion of the line had widened into a contiguous rain area over the mountains north and northeast of the Lago Maggiore. The precipitation was greatest in intensity and most widespread at about 1635 UTC. By 2035 UTC the precipitation still showed the same pattern but was somewhat weaker. A total of about 100 mm of rain fell at the Magadino airport and 280 mm at Tamaro. The dual-Doppler winds from the RONSARD and Monte Lema radars showed how the flow at the 2.5 km level approached the region. On the eastern side of the region, southwesterly flow turned to a more southerly direction as it approached the mountains at the northern end of the Lago Maggiore region. The flow was evidently responding to the mountain barrier on the eastern side of the area. Over the center of the region, directly over the southwest-northeast oriented radar reflectivity maximum, the flow on the eastern side converged with the flow on the western side of the region. 4 October: Precipitation moved out of the Lago Maggiore area early this morning and re-intensified over the eastern Po Valley. 5 October: The forecast for the night of the 4th and the day of the 5th was for northerly downslope flow and drying over the Lago Maggiore region. The northerly flow aloft past the western end of the Alps formed a cyclone. In conjunction with this cyclonic circulation, intense convective cells formed in the easterly flow over the mountains to the northeast of Milan. These cells formed a squall line which moved over Milan during the early morning hours, and the convection moved on across northern Italy.” Satellite rainfall measurements The satellite rainfall estimation algorithm of Ferraro (1997) was applied to the overpasses of the Special Sensor Microwave/Imager (SSM/I) during the two IOPs. The overpasses of the microwave (MW) instrument are, however, very limited in number and cannot capture the evolution of the precipitation systems. An extension to the infrared (IR) sensors on board the geostationary satellite METEOSAT is necessary to ensure the adequate space/time coverage (e.g. Levizzani et al., 2001). The histogram matching technique of Turk et al. (1999) was used for calibrating the METEOSAT IR brightness temperature imagery at 30 min frequency by means of the SSM/I derived rainfall maps. This approach poses a pressing information transfer problem, that is: “How much and for how long is microphysical information from past MW overpasses maintained? What are the best techniques to forward-propagate past information (MW observations, multispectral IR observations)?” The present work on MAP IOPs aims at defining performances in critical situations of heavy orographic influence on the precipitation systems. Such tests are necessary for identifying the weaknesses of the method, which are related to the rapidly changing microphysical and dynamical conditions of the rainfall. Slide2: References Ferraro, R. R., 1997: Special sensor microwave imager derived global rainfall estimates for climatological applications. J. Geophys. Res., 102 (D14), 16715-16735. Levizzani, V., J. Schmetz, H. J. Lutz, J. Kerkmann, P. P. Alberoni, and M. Cervino, 2001: Precipitation estimations from geostationary orbit and prospects for METEOSAT Second Generation. Meteor. Appl., 8, 23-41. Turk, J. F., G. Rohaly, J. Hawkins, E. A. Smith, F. S. Marzano, A. Mugnai, and V. Levizzani, 2000: Meteorological applications of precipitation estimation from combined SSM/I, TRMM and geostationary satellite data. In: Microwave Radiometry and Remote Sensing of the Earth’s Surface and Atmosphere, P. Pampaloni and S. Paloscia Eds., VSP Int. Sci. Publisher, Utrecht (The Netherlands), 353-363. Acknowledgements Work funded by the project EURAINSAT, a shared-cost project (contract EVG1-2000-00030) co-funded by the Research DG of the European Commission within the RTD activities of a generic nature of the Environment and Sustainable Development sub-programme (5th Framework Programme). Partial support is acknowledged from CARPE DIEM, a research project (contract EVG1-CT-2001-0045)supported by the European Commission under the 5th Framework Programme and contributing to the implementation of the proposal "Research and Technology Development Activities of a Generic Nature, Fight against Major Natural Hazards" within the Energy, Environmental and Sustainable Development. IOP-2b IOP-5 MW rainfall estimations. The rainfall fields as estimated using the MW technique of Ferraro (1997) match quite closely those of the Alpine radar composite. Note that not all the MW/radar pairs are absolutely coincident. The ability of the MW method to estimate rainfall is evidently more pronounced in the case of the rainfall maxima, while the light rain background as seen by the radar is generally not estimated. This is a well known bias of the MW estimation methods whose performances in light/inconspicuous rain are generally not satisfactory. When talking about monitoring applications, MW techniques are thus more applicable to heavy rains and storm showers, although a general progress is to be registered towards measuring lower rainrates. In the present case it is quite clear the difficulty in delimiting the frontal rain field, especially during IOP-2 west of the Alps in Eastern France. The field is missed altogether even though a substantial amount of rain is seen by the radar. For example at 0830 UTC during IOP-2 the MW method captures the convective activity over the Po Valley in Northern Italy, while it misses the frontal rain over France and Switzerland where only an isolated cell is delimited. The problem of this shallow, light, frontal rain is present also in IOP-5. Rapid update MW-IR rainfall estimates IOP-2b The image sequence of the MW-IR rainfall estimations show how this kind of technique works in an extreme case and with a limited number of SSM/I MW passages. The MW passages happened around 0613, 0836 and 1533 UTC, and are indicated by the red arrows. The estimates prior to the 0613 UTC passage show a low rainrate system positioned between Sardinia and the Lago Maggiore area in NW Italy. Right after the SSM/I passage the estimation system calibrates itself with the MW estimate and IR brightness temperatures are matched with MW rainrates. The result is a better distinction of the extensive cells located between Central Italy and the Po Valley. The cells are then followed in their ageing every half hour and the evolution is described by the changes in cloud top brightness temperature. The next MW passage at 0836 UTC depresses somehow the rainrates while maintaining the shape and location of the system, which afterwards dissipates more or less constantly for the rest of the day. In other words, given the MW calibration of 0836 UTC, the system constantly decreases its raining potential. However, this is not completely true since the next calibration at 1533 UTC kicks the rainrates once more up over the whole domain. A dramatic difference is detected in fact between the 1530 and the 1600 UTC rapid update estimates. This is the major risk in retaining the calibration for such a long time in cases when the rain systems evolve substantially. Finally, we must note that the method failed to measure precipitation over the Lago Maggiore area, while raingauges detected it. This is clearly due to the lack of SSM/I passages over the area at the right time of the rainfall event causing a loss of the proper calibration of the IR brightness temperature. Ongoing work The above described results suggest that more frequent MW passages would be needed for a more effective calibration strategy. However, this is not possible under the current conditions and we will have to wait for the Global Precipitation Measurement (GPM) mission to have a 3-hourly global MW coverage. Other possibilities of improving the rapid update algorithm performances are: Use multispectral IR and near IR (NIR) observations such as those that will shortly become available from METEOSAT Second Generation’s SEVIRI sensor. These observations allow for describing the cloud top microphysics of the raining clouds thus permitting a better rainfall assignment even in absence of MW passages. - Use dynamical information from mesoscale model forecasts. This has the advantage of following the evolution of the rain system and also adjusting for the topographic effects such as in the case of the Alps. - Use neural networks (NN) methods. NN can be effectively used to simulate the evolution of precipitating systems and the learning process takes care of part of the problems arising from the loss of information content between a MW passage and the next one.