Elsevier

Computers & Geosciences

Volume 33, Issue 8, August 2007, Pages 989-1007
Computers & Geosciences

A GIS tool for the evaluation of the precipitation forecasts of a numerical weather prediction model using satellite data

https://doi.org/10.1016/j.cageo.2006.12.001Get rights and content

Abstract

In this study, the possibility of implementing Geographic Information Systems (GIS) for developing an integrated and automatic operational system for the real-time evaluation of the precipitation forecasts of the numerical weather prediction model BOLAM (BOlogna Limited Area Model) in Greece, is examined. In fact, the precipitation estimates derived by an infrared satellite technique are used for real-time qualitative and quantitative verification of the precipitation forecasts of the model BOLAM through the use of a GIS tool named as precipitation forecasts evaluator (PFE). The application of the developed tool in a case associated with intense precipitation in Greece, suggested that PFE could be a very important support tool for nowcasting and very short-range forecasting of such events.

Introduction

Climatological and meteorological phenomena are naturally spatially variable and hence Geographic Information Systems (GIS) represent a useful solution to the management of vast spatial climate datasets for a wide number of applications. GIS has become a key management component in weather-processing systems allowing instantaneous plotting, interpolation and animation of weather data across any isobaric level of the atmosphere. The synoptic situation across different levels is then estimated by a forecaster, from which the GIS is used to rapidly calculate the speed of progression of weather systems. An extreme example of this is the relational positioning and monitoring of tornado's and tropical cyclones, where GIS is used to issue warnings to precise locations using remote sensing signatures (Kumar et al., 1998). An alternative use of GIS is the combination of different layers of weather information in expert classification systems. For example, specific humidity is often compared with wind flow to identify areas of fog, cloud and precipitation in relation to orographic and coastal influences. Moreover, interpolated climate datasets are used to set the boundary conditions for numerical weather prediction such as mesoscale forecast models and general circulation models. For example, Cheng and Shang (1998) used a GIS to manipulate topographic and roughness data to model wind fields using a numerical kinematic flow model. GIS has significantly facilitated the incorporation of numerical weather model output into weather processing systems, onto which satellite imagery and topography can be superimposed; an approach that greatly aids the skill of the weather forecaster. Overall, GIS partially automates forecasting by facilitating speed and processing of weather data in real time as well as providing support for traditional weather processing tasks such as contouring and superposition.

Numerical Weather Prediction (NWP) models have become more reliable over the last decades and constitute an indispensable and powerful tool in routine weather forecasting. However, in spite of a generally very high quality, NWP models occasionally fail to accurately predict intense precipitation, particularly on the smaller scales. Remote-sensing techniques, which generally support forecasting tools, have proven highly valuable in the diagnosis and nowcasting of convective systems. Infrared (IR) data from the present international constellation of operational geostationary meteorological satellites offer high-resolution (∼4–7 km) views from space and at frequent intervals (15–30-min) over the entire globe (∼60°N–60°S). These data are used extensively in weather forecasting both directly by forecasters and indirectly through their ingestion into model forecast assimilation systems.

The use of ground-based data for the monitoring of rainfall is restricted by its high spatial and temporal variability. The wide spatial coverage and high temporal resolution of geostationary satellite imagery has constituted an important challenge for the development of rainfall retrieval algorithms. When IR satellite data with high spatial and temporal resolution became available, precipitation was correlated to the cloud-top temperature (CTT) through empirical relationships. Numerous new precipitation estimation algorithms have been developed that use IR data as the only data input i.e. Arkin technique (Arkin, 1979), area time integral (ATI) technique (Doneaud et al., 1984; Lopez et al., 1989) and the GOES Precipitation Index (GPI; Arkin and Meisner, 1987). The satellite rainfall technique known as the Griffith–Woodley technique (GWT) was developed by Griffith and Woodley (Griffith et al., 1976, Griffith et al., 1978, Griffith et al., 1981) and it is based on the analysis of the temporal and spatial evolution of clouds, defined as individual bodies delimited by the isotherm of 253 K. Negri et al. (1984) have proposed a simplified version of this technique, known as Negri–Adler–Woodley technique (NAWT), which does not use the temporal evolution of the cloud, and is optimized for fast computer processing. Other methods, such as the convective stratiform technique (CST) (Adler and Negri, 1988), have demonstrated the ability to eliminate the cirrus contamination problem and are pixel-based algorithms and thus provide estimates at higher spatial resolution than the GPI. A common problem of the retrieval algorithm, when using these data, is that the coupling between observed quantities (IR brightness temperatures of the cloud tops) and the near surface precipitation is not direct. But, IR techniques have proven to be relatively successful, if convective rain events are dominating and the vertical cloud extend is related to the near-surface precipitation rate by the convective cloud dynamics (Adler et al., 1993). Due to the deficient physical basis of the IR satellite techniques, distinction of precipitation into convective and stratiform components is difficult and unreliable. The CST is the only IR-based method which potentially estimates both convective and stratiform precipitation and has been tested in northern South America and globally (Negri et al., 2002, Negri et al., 2003).

The development of an automatic integrated operational system for real-time validation of NWP precipitation forecasts is very important for nowcasting and very short-range forecasting of weather events associated with heavy rain. The real-time verification of the precipitation forecasts may significantly contribute to the evaluation of the validity of the model forecasts for the next hours after the first forecast fields have become available. An ideal case might be the real-time cross-comparison of precipitation forecasts with rain gauge records. Such data, however, are rarely available before a period of 12 h and ever more in real time. In addition, remote and uninhabited areas are not covered by conventional observation networks. Similar problems are associated with the use of weather radars. Ground-based radar systems provide fairly continuous coverage in space and time, but the quantitative range of radar measurements is generally limited to 150 km or less. Most importantly, both rain gauges and radars provide incomplete coverage on a global scale, particularly over the oceans, where such instruments are virtually non-existent. A particular case is the Mediterranean basin where topography and sea impose serious limitations in setting up a network of rain gauge stations and ground-based weather radars capable of providing a continuous spatial coverage. These inadequacies may be encountered by the high temporal resolution (15–30-min) of geostationary satellites such as Meteosat-7 and Meteosat-8 imagery, which in combination with its direct availability, allows the near real-time extraction of precipitation estimates for the model verification. Moreover, the visualization and the real-time qualitative and quantitative comparison of the precipitation forecasts with the precipitation estimates derived by a satellite technique, requires automated procedures that are favorably supported by a geographical information system. Up to now, there has been no attempt of developing such a system capable of operating, in real time, in the case of heavy rainfall in middle latitudes and especially in the eastern Mediterranean.

In this context, the aim of the present work is the development of a fully automatic GIS tool for the real-time evaluation and qualitative and quantitative verification of the precipitation forecasts provided by the NWP model BOLAM (BOlogna Limited Area Model), using as a comparative parameter the rainfall estimates derived by a satellite IR technique. This GIS tool was named precipitation forecasts evaluator (PFE) and intends to be used as a supporting tool for nowcasting and very short-range forecasting of heavy rainfall in areas such as the eastern Mediterranean Sea where there is a lack of ground-based radars and conventional observation networks over sea and mountainous zones. PFE system is applied in a case associated with intense precipitation in Greece, which occurred during the period of 7–8 November 1999, and the results are assessed with respect to the ability of the PFE tool to operate in operational terms. For this period, the rainfall estimates provided by a satellite IR technique are qualitatively and quantitatively intercompared with the model forecasted maximum precipitation totals.

Section snippets

Data and methodology

The PFE system uses as data input IR data from Meteosat-7 satellite, precipitation forecasts provided by BOLAM model and rain gauge observations.

The precipitation forecasts evaluator tool

PFE is a GIS tool developed for the real-time evaluation and qualitative and quantitative verification of the precipitation forecasts provided by a NWP, using IR Meteosat imagery and ground based data from meteorological stations. The application geographical domain covers the area of Greece spanning approximately from 35° to 42°N and 18° to 30°E (Fig. 1). In the preliminary stage, the system preprocesses IR Meteosat-7 image files, and estimates the 3-hourly accumulated precipitation in the

Application of the precipitation forecasts evaluator tool

The developed GIS tool described in the previous section was applied in the case of the period 7–8 November 1999, which was characterized by high precipitation rates in several areas over Greece.

On 7 November 1999, at 18:00 UTC, an occluded depression, with a center of 997 hPa, was crossing central Mediterranean in a northwest to southeast direction (Fig. 7a). At the same time, the associated cold front was traversing western Greece while a low-level convergence zone was situated over the eastern

Conclusions

The PFE system is a flexible and self-contained GIS tool capable of operating for the real-time evaluation and qualitative and quantitative verification of the precipitation forecasts provided by the NWP model BOLAM using as a comparative parameter the rainfall estimates derived by a satellite IR technique. The system is fully automatic and allows the visualization of the spatial distribution of all the relevant parameters along with the presentation of several tables with the statistical

References (36)

  • E. Cheng et al.

    Kinematic flow model based extreme wind simulation

    Journal of Wind Engineering and Industrial Aerodynamics

    (1998)
  • E.E. Ebert et al.

    Verification of precipitation in weather systems: determination of systematic errors

    Journal of Hydrology

    (2000)
  • R.F. Adler et al.

    A satellite infrared technique to estimate tropical convective and stratiform rainfall

    Journal of Applied Meteorology

    (1988)
  • R.F. Adler et al.

    Estimation of monthly rainfall over Japan and surroundings waters from a combination of low-orbit microwave and geosynchronous IR data

    Journal of Applied Meteorology

    (1993)
  • R.A. Anthès

    Regional models of the atmosphere in middle latitudes

    Monthly Weather Review

    (1983)
  • P.A. Arkin

    The relationship between fractional coverage of high clouds and rainfall accumulation during GATE over B-scale array

    Monthly Weather Review

    (1979)
  • P.A. Arkin et al.

    The relationship between large-scale convective rainfall and cold cloud over the Western Hemisphere during 1982–1984

    Monthly Weather Review

    (1987)
  • F. Atger

    Verification of intense precipitation forecasts from single models and ensemble prediction systems

    Nonlinear Processes in Geophysics

    (2001)
  • Bacchi, B., Ranzi, R. (Eds.), 2000. RAPHAEL—Runoff and Atmospheric Processes for Flood Hazard Forecasting and Control....
  • Bougeault, P., 2003. The WGNE survey of verification methods for numerical prediction of weather elements and severe...
  • A. Buzzi et al.

    Low level jet simulation over the Antarctic Ocean

    Tellus

    (1997)
  • A. Buzzi et al.

    Numerical simulations of the 1994 piedmont flood. Role of orography and moist processes

    Monthly Weather Review

    (1998)
  • A. Buzzi et al.

    Mesoscale meteorological features associated with heavy precipitation in the southern Alpine region

    Meteorology and Atmospheric Physics

    (2000)
  • R. De Elia et al.

    Distribution-oriented verification of limited-area model forecasts in a perfect-model framework

    Monthly Weather Review

    (2003)
  • A.A. Doneaud et al.

    The area time integral as an indicator for convective rain volume

    Journal of Climate and Applied Meteorology

    (1984)
  • H. Feidas

    Validating three infrared-based rainfall retrieval algorithms for intense convective activity over Greece

    International Journal of Remote Sensing

    (2006)
  • R. Ferretti et al.

    Analyses of the precipitation pattern on the Alpine region using different cumulus convection parameterisations

    Journal of Applied Meteorology

    (2000)
  • M. Georgelin et al.

    The second COMPARE exercise: a model intercomparison using a case of a typical mesoscale orographic flow, the PYREX IOP3

    Quarterly Journal of Royal Meteorological Society

    (2000)
  • Cited by (12)

    • Acceptance of GIS within ERP System: Research Study in Higher Education

      2022, ISPRS International Journal of Geo-Information
    • Climate Downscaling and Prediction Using GIS-Based Machine Learning

      2022, 2022 2nd International Conference on Computer Science, Engineering and Applications, ICCSEA 2022
    View all citing articles on Scopus
    View full text