A GIS application for weather analysis and forecasting

A GIS application for weather analysis and forecasting

SHARE

A GIS application for weather analysis and forecasting


Saseendran S. A., Harenduprakash L., Rathore L. S. and Singh S. V.

National Centre for Medium Range Weather Forecasting,
Department of Science and Technology, New Delhi -3
saseendransa@hotmail.com

Abstract
With technological progresses and associated need for more and more human comfort, the demands for accurate weather forecasts for different spatial and temporal scales are also increasing. In this context, application of emerging technologies for increasing accuracy and skill of the weather forecast calls for special attention. Use of Geographical Information Systems (GIS) software viz. Arc View to develop an application, for plotting, analysis, visualization, and interpretation of weather data, to serve as an aid in the prognostication of weather is attempted in this paper. The application developed can help the meteorologists in instantaneous plotting of synoptic weather data from different locations at various isobaric levels of the atmosphere. Analysis of this data, for visualization and interpretation of weather systems over wide geographic areas become possible with less effort and error. Facilities available include, provision for superimposition of synoptic weather maps of the past with the present for tracking of movement of weather systems, computation of their persistence, tendencies and trends. Weather maps at different levels, or different days (past, present and future) can be superimposed and removed with the click of the mouse for analysis and visualization of weather developments. Advancing the weather systems forward or backward geographically for visualization of past and future (as forecasted) movement of weather systems across geographical areas becomes easier. Climatological data can also be plotted, departures from normals, tendencies, etc. calculated and presented as charts. Satellite pictures, topographical information, etc. can also be plotted and superimposed with other weather parameters for assistance in weather forecasting.

Introduction
Atmosphere is the gaseous envelope of the earth in which all its flora and fauna survive. As weather is the statement of its physical conditions at an instant, its forecasting is of concern to one and all living over the earth. As such, since time immemorial weather forecasting was a subject of grave concern for the physical scientists. But, due to extremely complex nature of various physical processes of the atmosphere, which lead to weather, these endeavors have always been met with limited success.

Various methods were developed and used by meteorologists for weather forecasting. The most important methods in vogue currently are the conventional Synoptic, and Numerical Weather Prediction (NWP) methods. The former method is human subjective, and the latter is objective and deterministic. Skill of these forecasts can be enhanced through use of GIS by relating different features of the atmosphere and their proper visualization.

Conventional synoptic method
In this subjective method, conventional forecasting tools like, trend, persistence, Climatology, and analogue of weather systems, are popularly employed. Each of these methods makes use of some basic assumptions for extrapolating the weather into the future. The forecaster blends these extrapolations with his own experience and the location specific weather quirks like topography, land sea distributions etc.

None of these methods seems perfect, as the weather sometimes manifest differently, deviating considerably from the basic concepts on which these methods are founded. The inadequate human understanding of the various complex atmospheric processes leading to the weather development itself is one of the major problems associated with this method.

NWP method
To forecast weather, the NWP method makes use of numerical solutions (high speed super computers are generally required for this task) of complex system of mathematical prognostic equations/models representing both the physical and dynamical processes occurring in the atmosphere. These models are commonly known as Global Circulation Models (GCMs). In order to integrate the GCM forward in time, the model equations need initialization with precise knowledge of the current state or initial conditions of the atmosphere. To achieve this task, global observations of various atmospheric parameters e.g. temperature, wind speed and direction and humidity, made routinely at standard synoptic hours are usually assimilated into the model using a process known as Variation Analysis. The model integrations into the future automatically produce charts of important parameters such as surface pressure, wind circulations, etc. The forecaster interprets these charts for weather forecasting at the locations of his interest.

Medium Range Weather Forecasting in India
The National Centre for Medium Range Weather Forecasting (NCMRWF) was established in India under the Department of Science and
Technology for issuing weather forecasts in the medium range i.e. 3 to 10 days in advance. The GCM used in the Medium-range Analysis Forecast System (MAFS) of the NCMRWF is an adapted version from NCEP (NMC, 1988). It is a Global Spectral Model having T80 horizontal resolution (about 150 km) and 18 layers in the vertical. The model uses climatological boundary conditions for the sea surface temperature, albedo, ice, snow, soil moisture, soil temperature, and roughness length and plant resistance. The MAFS operational at NCMRWF consists of (1) data processing and quality control, (ii) utilization of non-conventional data, (iii) data assimilation, (iv) model integration, (v) post processing and diagnostic studies, and (vi) preparation of location specific forecasts.

 

 

 

A GIS application for weather analysis and forecasting

Limitations of NWP weather forecasts
As of today, there are many limitations in the GCM based method of weather forecasting, firstly, the GCM produced charts represent the atmosphere in general only at levels well above the land surface, due to inadequate representation of land surface processes and topography in the model. In addition, errors in the model forecasts are known to be due to 1) inadequate representations of the initial conditions of the atmosphere, 2) inadequate finite resolution of the model – presents difficulty in the representation of orography, giving rise to differences in station levels, 3) Inadequate parameterizations of the boundary layer and other physical processes, and 4) incorrect representation of ground conditions. These lead to deficiencies in the forecast fields, with the surface forecast being particularly erroneous. Owing to the inherent shortcomings of the NWP model forecasts, human interpretations of the meteorological data generated by the models become a necessity. There have been tremendous improvements in the dynamical and statistical models used in NWP in the last few decades, but still, it will be awhile before totally automated forecasts based on unmodified model output, will be of skill comparable with those produced by human forecasters who modify such output based on their broad meteorological knowledge and forecasting confidence (Doswell et al., 1995). Sousounis et al. (1999) also opines that despite improvements in GCMs, Statistical models, forecast decision making trees, and forecast rules of thumb, etc. in automated weather forecasting, human synoptic interpretation of meteorological information for a particular situation can yield superior results. It was also, reported that the value added by humans over model forecast quality can be significant at times, especially when the forecast involves convective situations or shallow cold air outbreaks, which operational models still do not handle well (Cortinas and Stensrud, 1995). As such, a man machine mix approach is currently advocated for preparation of weather forecasts from GCM runs, especially in the medium range.

Keeping the above limitations of NWP based medium range weather forecasts in view, this paper focuses at improving the human component of the man-machine mix philosophy for improving forecast skill by making use of new technological tools like Geographical Information System (GIS) software for plotting, analyses and visualization of observed meteorological parameters, superior to the conventional techniques otherwise followed for the purpose. The GIS software – ArcView has been made use of for developing the application tool for the purpose, and demonstrated how this tool can help the human forecaster in his efforts at producing a better forecast from the model output products.

Materials and Methods for the Study
The GIS used for the application development is the ArcView GIS of ESRI with extensions of Spatial analyst, and Image analyst.

The weather data utilized is the T80 model analysis (initial conditions) for the weather parameters viz. Wind Speed, Wind Direction, temperature and Geopotential height at vertical levels of the atmosphere at 850 hPa and 500 hPa at the T80 model grid points (approximately 150 km apart) over the globe. The 5-day weather forecasts for the same parameters based of the above initial conditions (analysis) were made use of as weather forecasts.

Discussion
The synoptic weather chart is the main tool of the forecaster. A forecaster need analyze, and interpret numerous weather charts of the past period, before he gets a grip of the current weather situation, in order to evolve likely changes in the weather systems as time advances. Weather analysis is the process of drawing isobars, isohyets, isotachs, etc. and locating pressure systems, fronts, etc. on a base map of an area on which the weather observations from a wide area are plotted, following meteorological conventions. The forecast in general is generated from the observations. Fig. 1. provide the locations of the synoptic weather observation stations over the globe.

Figure 1: Locations of global synoptic weather data observing stations (source: ECMWF, U.K.)

At NCMRWF, data are received from thousands of weather observing stations over the globe for weather analysis and forecasting. After plotting the observations following a synoptic model, the analyst checks the chart for erroneous and inconsistent values of weather parameters, frequently, it becomes necessary for the analyst the suspected reading with neighboring stations and or previous observations and analyses or observations at other levels in the vertical. The Arc View plotted maps can be utilized for comparing the values of a particular parameter with the neighboring stations, at different levels and between observations and analyses for past days, for properly assessing the accuracy of the observation for inclusion in the analysis or exclusion.

 

 

 

A GIS application for weather analysis and forecasting

The observations coming from the observing stations, through several processing steps are transferred into weather forecasts. First step in this process is to plot the observations over a base map of the region, global or regional according to the area of interest, after removing possible observational and communication errors.


Figure 2: Wind analysis for 500 hPa on 17th Dec. 2000 at 5.30 A. M. over India and neighborhood areas.


Figure 3: Geoptoential height analysis at 850 hPa level on 17th Dec. 2000 at 5.30 A. M. over India and neighborhood areas.

The plotted maps are analyzed to bring out different weather systems in action in the atmosphere at that particular instant of observation. Fig. 2. shows, such an analyzed chart generated using ArcView GIS at NCMRWF for 5.30 A.M. observation time on 17th December 2000. In the figure, the red arrows represent the wind at 500-hPa level of the atmosphere. The hollow head of the arrow represents the direction towards which the wind is blowing. Fig. 3. presents the geopotential height analysis at 850 hPa on the same day and time.

The forecaster draws contours or isobars of pressure and marks the fronts, lows highs etc. on the chart. The contouring can be done by Arc View plotted map using the necessary analysis tools. In this application, Troughs, ridges, highs and lows can be drawn on the chart in appropriate color and style, and saved as a part of the chart or a separate view. The successive movement of these systems also can be drawn on the same chart by super imposing the successive charts. The trends and rate of movement of the systems can be studied either directly from the charts or using the theme attribute table.


Fig. 4. Wind flow pattern over India and neighborhood at 500 hPa on 17th Dec. 2000 at 5.30 A.M (IST). The green line represents the low pressure troughs. The same trough as manifested at 850 hPa is also marked in the chart.

 

 

A GIS application for weather analysis and forecasting

Through combining analyses at various levels with past analyses, the weather forecaster endeavors to visualize in his mind the weather processes at work, like the large-scale vertical motion, convection, radiative or advective cooling, etc. Fig. 4. is an ArcView generated map of the wind flow pattern at 500 hPa with the low pressure trough marked in green. The same trough as manifested at 850 hPa also is marked in the same chart by superimposing the 850 hPa wind flow pattern over the 500 hPa wind flow pattern. The locations of the trough at various levels of the atmosphere helps the forecaster in understanding the tilt, if any of the trough with elevation, which has high bearing on the expected weather of the trough.

A forecaster studying the sequence of evolution of a weather system in a current chart requires analyzed chart for the previous hour of observation, and indicate the movement of weather systems like centers of low and high pressure, trough and ridges, etc. This is commonly done by marking positions of the centers of the system at the 6-hourly intervals and joining the successive points by a broken line, resulting tracks give ideas on movement of the system. In the Arc View plotted analyzed maps, the time sequence of maps can be super imposed and locations marked. The distance between successive locations is automatically obtained form the tools available. Conventionally, this work is done by the synoptitian using an illuminated tracing or ‘light’ table. The successive charts are placed one above the other and successive positions of the systems are marked.

In figure 5, the movement of the trough along the 500 hPa level during the period 11th to 19th December 2000 is shown. Superimposing the 500 hPa wind flow pattern successively, for these days in the ArcView GIS created the map. Such maps help the forecaster study the speed and intensity of the trough in the past days for forecasting its behavior in the coming days. In the GIS distance between successive locations of the trough is readily available through the click of the mouse at the locations of interest, for calculation of its speed of progression.


Fig. 5. Locations of a low pressure trough in the westerlies at 500 hPa at 5.30 A. M. IST during 11th to 19th Dec. 2000.

Once the forecaster is able to explain the recent weather evolutions taking place, he will be in a position to make prognosis of their future behavior from the knowledge of the physical and dynamical processes taking place in the atmosphere. It is of paramount importance that the forecaster is able to explain the processes realizing the current situation perfectly well before an attempt on forecasting its future behavior.

In the processes of generating satisfactory explanation for weather systems, the forecaster need examine the weather charts of different levels of past few days back and forth. There is a necessity of superimposing the analyzed maps of different weather parameters one above the other for studying the physical processes at work in the atmosphere leading to the manifested weather. . In the figure Fig. 6. the specific humidity analysis is superimposed on the wind flow pattern using GIS, to study the moisture distribution and convergence in the atmosphere, in order to identify potential areas for fog, cloud, and rain formation. Conventionally a forecaster does this manually, and connecting between the weather systems available in the charts is done mentally. Success of this – procedure
depends on the mental alertness of the forecaster, his experience and knowledge in the subject. In the case of the weather charts generated through GIS, these works can be carried out any number of times with the click of the mouse. The GIS platform provides opportunity for connecting the weather systems physically across the charts of different days and levels.


Fig. 6. Super imposition of the wind flow pattern and specific humidity (gm/kg) at 5.30 A.M. IST on 17th Dec. 2000. The red arrows show the wind flow, and the white contour lines with blue colored numbers across show the specific humidity distribution. The multicolored background is the surface created out of the specific humidity values.

 

 

 

A GIS application for weather analysis and forecasting

 


Fig. 7. Super imposition of the wind flow pattern and specific humidity (gm/kg) at 850 hPa level at 5.30 A.M. IST on 17th Dec. 2000, over the surface topography. The red arrows show the wind flow, and the green contour lines with blue colored numbers across show the specific humidity distribution. The multicolored background is the surface created out of the ten minute interval orography of the area.

A forecaster need examine the occurrence of all systems of cloud, precipitation, fog, stratus, drizzle, etc. in relation to orographic influences, coastal influences, etc, and also give special attention to diurnal cycles in weather behavior. Only by long experience and close examination of successive weather charts and particularly of current ones, which can be related by the forecaster to the existing weather in his own area, can ability be acquired in applying to actual weather forecasting the physical principles and other meteorological information underlying the weather process as such. The capability of GIS in superimposing the surface topographic features over the atmospheric analysis for interpretation of the dynamics and physical processed in the atmosphere is given in Fig. 7.


Fig. 8. Super imposition of the wind flow pattern and specific humidity (gm/kg) at 850 hPa level at 5.30 A.M. IST on 17th Dec. 2000, over the surface topography. The red arrows show the wind flow, and the green contour lines with blue colored numbers across show the specific humidity distribution. The purple colored contours with black numbers across represent the ten minute interval orography of the area.

In fig. 8, the same information provided in Fig. 7 is provided with contour values of topography instead of the orographic surface. This figure shows an example of various ways of presentation of the weather parameters in order to study and understand atmosphere with various angles and perspectives. It is also possible to superimpose the various satellite pictures of cloud, moisture etc. over the weather maps for better interpretation and forecasting of weather. Use of GIS in weather forecasting will certainly improve the capability of the meteorologist in forecasting weather with better skill and accuracy in the days to come.

Summary
A GIS application has been developed using ArcView of ESRI to serve as an aid in the man – machine – mix approach for preparation of medium range weather forecasts. Weather and climate are integral parts of the geography of a place. In order to prepare weather forecasts, the forecaster plots information of the atmosphere around the globe on a weather chart. The plotted data is analyzed with the help of contours and surfaces to bring out the weather systems in the atmosphere at various levels. The analyzed weather becomes initial condition for the conventional synoptic weather forecasts and GCMs for prognostication of weather in the coming days. The capability of GIS software in handling the spatial data in an easier way coupled with the different analysis tools available make it a viable tool for its adoption in weather forecasting.

References

  • Sousounis, et al., 1999. Forecasting during the Lake-ICE/SNOWBANDS Field Experiments. Weather and Forecasting, 14, 955-975.
  • Doswell et al., 1995. III, H, E. Brooks, and E. N. Rasmussen, 1995: Forecasting issues and implications from VORTEX-94 Project. Reprints, 14th Conf. on Weather Analysis and Forecasting, Dallas, TX, Amer. Meteor. Soc., 23-28.
  • National Meteorological Centre, 1988: Research Version of the Medium Range Weather Forecast Model, vol. 1.