Preview only show first 10 pages with watermark. For full document please download

Xrwis: The Use Of Geomatics To Predict Winter Road Surface Temperatures In Poland

XRWIS: the use of geomatics to predict winter road surface temperatures in Poland




  Meteorol. Appl. 12, 83–90 (2005)  doi:10.1017/S135048270500157X XRWIS: the use of geomatics to predict winterroad surface temperatures in Poland  John E. Thornes, Gina Cavan & Lee Chapman School of Geography, Earth and Environmental Sciences, University of Birmingham, UK  Email: [email protected]  A new method (XRWIS) to predict the minimum road surface temperature for the winter maintenanceof roads, using geomatics, has been tested in Poland. A geographical database was constructed for a200km test route from Krakow to the Slovakian border. A GPS survey to measure sky-view factor wascarried out as part of the COST 719 project. A computer energy balance model ‘IceMiser’ was runretrospectively to predict road surface temperature every 20 minutes for every 20 m along the road usinghourly weather data recorded at two adjacent climate stations: at low elevation in Krakow and at highelevation in Zakopane, in March 2003. A GIS was used to visualise the predicted road surfacetemperatures. The IceMiser model was verified by comparing the predicted road surface temperatureswith measured road surface temperatures at a number of road weather outstations along the route. Theresults for four road weather outstations are discussed. The best results are for Libertow (264m) and Myslenice (304m) close to Krakow, whereas the results for Skawa (518m) and Piatkowa Gora (649m)at high elevation are not as good, probably due to their distance from the Zakopane climatestation. 1. Introduction Traditionallyhighwayengineers aroundtheworldhaveusedRoadWeatherInformationSystems(RWIS)tohelpthem decide when to salt roads under their jurisdiction.The engineer has to interpret road weather forecasts forsite-specific locations in order to decide which saltingroutes need salting, at what time, and how much salt orbrine to use. Thermal mapping has been used to givethe engineer a spatial view of the likely minimum roadsurface temperature along the salting routes (Thornes1991). This RWIS approach has been used in Europefor almost 20 years and has recently become thestandard in North America as well. It is now time toconsider how this RWIS model can be improved totake on board more recent developments in geomatics,computer processing power, communications andespecially the use of the internet. A new paradigm –neXt generation Road Weather Information Systems(XRWIS) has been developed at the University of Birmingham. This new system uses geomatic surveysto help create a geographical database every 20 m to200 m (depending on topography) along each saltingroute (latitude, longitude, altitude, sky-view factor,thermal map residual, aspect, slope, cold air drainage,road construction, traffic, land use, etc.). A mesoscaleweather forecast is then used to provide a road weatherforecast (using the IceMiser energy balance model)for each individual salting route so that the engineeris presented with a route-based forecasting system(Thornes et al. 2005). This paper represents the resultsof a retrospective trial of XRWIS route-based forecast-ing in Poland. It is important to show that the metho-dology works in climates outside the UK where themethod was developed. Other XRWIS trials havebeen carried out in Japan on Highway 17, where themajor problem is snow rather than ice (Thornes et al.2004). 2. The Polish XRWIS trial Poland has a transitional climate with both maritimeand continental influences. The dominance of maritimeor continental air masses causes annual variations inthe seasons. Winters are variable, sometimes mild andsometimes severe, and similarly, summers can be cooland rainy or hot and dry. Sub-zero road temperaturesare usually recorded between November and March.Annual precipitation varies between the lowlands andmountains, with totals ranging between 500mm and600 mm. New technology has recently been investedinPoland’s roadnetwork, including theinstallationofanetwork of about 70 road weather information systems(RWIS). This paper considers the use of a geomaticsurvey and the new road weather prediction model‘IceMiser’ that has been tested on a study road thatlinks the city of Krakow to the mountainous borderwith Slovakia, as shown in Figure 1. This is a very busyroad, approximately 200 km in length, marked by a 83   John E. Thornes, Gina Cavan & Lee Chapman Figure 1.  Study area in Krakow, Poland. varietyoflanduses,differenttypesofroadconstructionand changes in topography. There are 10 RWIS alongthe road and four have been used to verify the model.Krakow has an average winter temperature of   − 3 ◦ Cand an average temperature in March of   − 1 ◦ C. Thismarginal winter weather in Poland creates a problemfor road maintenance, and there is little emphasis on icepredictiontooptimisesaltuse.Thismeansthatroadsareoften treated up to three or four times a day (Bartlett &Logiewa, 2002), at a great cost to the road maintenancebudget and to the environment.Therapidgrowthofcommercial‘offtheshelf’geomaticstechnology, including GIS and GPS, has enabled thedevelopment of a new ice prediction technique calledIceMiser(Chapman&Thornes2003).TheGISprogramArcViewisusedinthisstudytoruntheIceMisermodel. 3. IceMiser numerical model Numerical models provide a tool for investigatinginaccessible phenomena, such as environments in thefuture, which cause concern with regard to possibleimpacts of decisions made for future generations (Lane2003). There are two approaches to mathematicalmodelling: empirical and physically based. Lane (2003)notesthatitispossibletoconceiveofacontinuumfrommodelsthatarelargelybuiltaroundasetofobservations(empirical models) to models that are built around aset of laws (physically based models). The IceMisermodel was developed by Chapman et al. (2001b), andcomprises an empirical geographical database within aphysically based numerical model. IceMiser accountsfortheinfluenceofsite-specificgeographicalparameterson the micro-climatology of the road. The modelsimulates the energy transfer regime at a location bycalculating the unique equilibrium temperature whichbalances the energy flow across a road surface:(1 − α )( Q + q ) + σ  T  4 sky − σ  T  40  =  LE + H  + S where  α  is surface albedo,  Q  is direct beam solarradiation,  q  is diffuse radiation,  σ   is the Boltzmanconstant,  T  sky  is the radiation temperature of the skyhemisphere,  T  0  is the surface temperature,  LE  is latentheat flux,  H   is sensible heat flux and  S  is heatflux to the road construction. Thus, it uses a zero-dimensional energy balance approach (Chapman &Thornes 2001) and meteorological data is combinedwith a high resolution geographical parameter database(incorporating sky view factor, land use and elevationdata) in the forecast model, to predict the road surfaceconditionatthousandsofsitesaroundtheroadnetwork(at spatial and temporal resolutions of approximately20 metres and 20 minutes respectively). This enablesthe RST to be displayed for any site along the roadnetwork, at any particular time (Chapman et al. 2001b).Thismodelisavastimprovementonexistingtechniquesthat utilise forecast thermal maps.Recent research on geographical parameters affectingRSThasfocusedonthemeasurementofskyviewfactor(SVF) (e.g. Chapman et al. 2001a; 2001b; Bradley et al.2002; Chapman & Thornes 2002), which has beenfound to be the dominant control on RST togetherwith land use, at high levels of atmospheric stability(Chapman et al. 2001b). The SVF is the proportionof visible sky at a location, and varies between zero(when the sky is completely obscured) and 1 (inan open area where the whole sky is visible). TheSVF has an important role in the radiation budget,and influences RST by reducing the incoming solarradiationandcontrollingthelossoflong-waveradiation 84  XRWIS and the prediction of winter road surface temperatures Table 1.  Meteorological and geographical database inputs. Survey technique to obtainMeteorological data Geographical data geographical dataRST at noon Latitude GPSAir temperature ∗ Longitude GPSDew point ∗ Altitude GPSWind speed ∗ Sky-view factor GPSRainfall ∗ Cloud cover ∗∗ Cloud type ∗∗∗ Nine values at 12:00, 15:00, 18:00, 21:00, 00:00, 03:00, 06:00, 09:00, 12:00. ∗∗ Eight values averaged over the time periods 12:00–15:00, 15:00–18:00,18:00–21:00, 21:00–00:00, 00:00–03:00, 03:00–06:00, 06:00–09:00, and09:00–12:00. at night, through shading of road surfaces (Thornes1991). Until recently, this parameter was difficultand time consuming to include in climate studies.However,advancesintechnologyhaveallowedexistingtechniques using fish-eye imagery to be processedalgorithmically, providing an instant measurement inreal-time, as well as rapid and inexpensive methods of SVF collection entirely by using GPS (Chapman et al.2002).Chapman & Thornes (2002) note that due to climatedata being typically point-source in nature, one of the biggest challenges facing meteorology is theextrapolation of point climate data across a wide spatialdomain. This can be overcome by the extraction of climatedatausingdigitalterrainmodels(DTMs),whichenable a good estimate of an area’s climatology withoutthe need for extensive climate records and networksof weather stations. Unfortunately, a DTM was notavailableforthisstudy,andinstead,meteorologicaldatafromtwoclimatestationswerechosenasrepresentativeof two geographical regions. 4. Meteorological database The meteorological database is comprised of retrospec-tive data from two climate stations in Poland: Krakowand Zakopane. The meteorological data input into themodel are listed in Table 1. These parameters wererecorded hourly and were obtained for the month of March 2003. Retrospective data were chosen becausethis would identify the validity of the IceMiser model,rather than highlight errors in the forecasting of meteorological variables.Road surface temperature data was obtained from 10RWIS along the route, and was recorded at 10-minuteintervals. The locations of these RWIS are shown inFigure 2. Validation data for road surface temperaturewere also used from the RWIS. A simple cloudclassification procedure was used to assign cloud typestothreeclassesassuppliedbythePolishMeteorologicalOffice. 5. Geographical database The geographical database is comprised of variablescollected using a Geomatic Survey carried out in April2003, during a short-term scientific mission (STSM) forCOST 719. The geographical variables collected andthe method of data collection are listed in Table 1.Landuse,roadconstructiontype,aspect,slope,drainageand topography were not collected for this trial. Theseparameters have been set to default values in the model.This does affect the accuracy of the results, as resultswill be most accurate with the maximum input intothe model. These parameters could be collected at alater stage and input into the model to improve thesepreliminary results.The geographical variables provided included eastings,northings, altitude, latitude and sky view factor, andenabledthecreationofashapefile.ThisenabledtheroadtobedisplayedspatiallyinArcViewasaseriesofpoints.EachpointhasafieldrepresentingthegeographicaldatawhichwereinputintothedatabasecreatedinExcel.Thisallowed the creation of two separate shapefiles for eachof the geographical regions of Krakow and Zakopane,based on altitude. Experiments with the number of classes (using the natural breaks technique) indicated aclear divide in the road at an altitude of 400 m. Krakowwaschosentoberepresentativeofthelowertopography(0–400 m). Zakopane is situated on the mountainousPolishborder,andwaschosentorepresentmountainoustopography and climate (400+ m). 6. The accuracy of the IceMiser model forecasts The IceMiser model was run for two RWIS from eachsection of the road (four RWIS in total) and the resultswere compared with actual RST data from the RWIS. A24-hr forecast was made for each day in March for eachof the meteorological stations (Krakow and Zakopane),and this was displayed in ArcView as a series of colour-coded maps. The nearest coordinate to each of the fourselected RWIS along the road was obtained, and actualand forecast data were compared for each day. 85   John E. Thornes, Gina Cavan & Lee Chapman Figure 2.  Location of the study area and RWIS. 7. Results from the IceMiser model IceMiserforecastsweredisplayedinArcViewashourlythermal projections for each night. Figure 3 shows thatthe RST falls below zero first at the higher altitudesites around 22:00h (at the settlement of Klikuszowa)and remains below freezing for about 12 hours. Theseare the coldest areas of the route throughout thenight. Krakow city centre remains above freezingthroughout the night, and the RST along the route doesnot fall below freezing until it reaches the settlementof Lubien (see Figure 2 for location). It is clear thatthe route in the higher altitude area of Zakopane iscolder than Krakow throughout the night. At 09:00h 86  XRWIS and the prediction of winter road surface temperatures Figure 3.  Hourly time-slices of the numerically predicted RST along the study route. 87