Merken

Weather radar enhanced flash flood forecasting

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
and next presentation is from couple dominant and lake would go the other and is
where there rather than on the last lab by casting at it the coming at the at a lot of money that the but I mean
Iran very jealous of all the data we have available united States the what data as soon as the do that after 2 or 3 hours it's gone the your opportunities to do analysis in a very nice actually in my opinion and and ladies and so on I'm going to try to address the class for testing problems it's a it's a difficult problem because was of why the very nature of the in a very small should read at time and the due to varying intended that of intense rainfall event I am used being quite a bit about I think you're in you're I wrote recently developed to brainstorm that we want but 1 moment I my approach was used to use like after the he's not a hybrid can today which is a module built into the work included in the graph distribution and has been distributed with the graph synthesis and very early on it was 1 of the modules that was built for the Army Corps of Engineers are generally this the representation 1 of the probe data processing is necessary you acquire the radar reflector of do some magic cure to turn that into rainfall was I and distributed rainfall fields whether which is what the hydrologic model requires I based on the various parameters that are profit for the the watershed a watershed in you extract to some of the rainfall Due to infiltrations you're left with essentially the basin accumulation and you get run it went off compared to other parameters and and at that point you can issue a warning watch that's appropriate the radar imagery the numerator our data
is required using this kind of the called volumes scanned in the in the greater jargon each can each rater volume rather can and area of the atmosphere that earlier isolated and I only in my research I've used the highest quality and power or or density dataset which is called the digital hybrid that reflected that can be used largest reflected in in the lower volumes it ignores the the Irish of this is a typical radar
field overall my area of the United States and you can see there's a lot of small RNAs of future rainfall events occurring in the area but there are relatively forests and the relatively light and this is part of a the part of might but go on on this summer waited all summarization significant rain and single few years ago this small watershed indicated in the black outlines there were 100 150 500 years store not events that occurred so i'm gonna particularly the year primary input to the a hydro model and I can't do that just to deal with of course the digital elevation model and various datasets are derived from that the watershed boundaries can be defined a aspect that the flow direction channel networks such such things are all required for model model also has the capability of using additional data made available such as soil moisture content and forest and your crop cover and of course developed areas that certainly influence the amount of runoff occurs over watershed is an image of but this is essentially the same area that I showed you earlier with the overarching alone and you can see the areas properly complexes Denver Colorado areas located right on the edge of the eastern slope of the Rocky Mountains both of them have the area that I need to be a constant is relatively flat while the other half of it is very rugged mountain terrain my study area within the Buffalo Creek region indicated on the the yeah the use of radar
data as it was pointing out is is wonderful because you have a field of very dense that you have a very dense field of observations effectively that you can see of the a Buffalo Creek area my particular watershed that our
model has quite a few different radar bins are associated with that in reality the watershed only had a single automated rain gage and 3 or 4 of Cooperative Observer engages in a they don't really they don't really operate as a real-time engaged a little closer look at my work and its watershed and I was able to I was fortunate enough to obtain very high resolution 10 meter DDN data for this particular watershed which now allowed me to extract channel networks and such in in a great deal of detail but you can see yeah the significant detail here is actually that there quite a few rhetoric and all my particular watershed like in that's the nature of precipitation during already fairly accurate in processing the DDM and 1 of the things that you need to is extracted from the various watershed I'd been demonstrated at least in eastern Colorado that because of the convective nature of most of our storms there relatively small in a unified so we tried and so I have found that in order to give the best estimate for an hour from the the models necessary to have well defined watershed are smaller than 30 square kilometers and this this image which shows watershed that are in the range of 2 km 220 kilometers 1 of the other things that will allow but to extract and allow us to to do is to extract the specific points for forecasting well an intersection between a watershed and an extreme channels of course obvious forecast points but there may be impact features such as that of homes near the new near the river roads or bridges across over the river there could be and a potential risks for people crossing over the years over flooded extremes on that that the area they region fairly complex because amount
this is an image of available soil there was quite lucky to obtain them and I'm quite lucky to obtain these data publicly available in the United States and people complain about how difficult it is to painted these are variables and the and and that the model
itself to operate on the square grid In allows the use of spatial and temporal changes over time rainfall of course you know that a slope and flow direction grid usually a well-defined stream network the life the the the soil texture classifications it uses a kinematic wave over and channel flow algorithm to run out of water across the ground and then I want to get to the rivers down the on the Unix it abstracts from the precipitation using an endocrine and infiltration method which is a fairly sophisticated in and it does so to some degree from soil moisture the original model but was they a were an analysis tool the collected all the data before it began and validated that had all of the data available and then them more in spatial and temporal steps by modification was to try to adapt are not cascading into running in real time so we didn't have all the the data available for began so I had to reorganize the way it operates it now collects all the spatial data and a priori but I want to begin to computing the check to see if the next time step of data that are made available that is it going continues process the data that is and it waits until that date because they are available and then it goes on and on iterates over but as as I said I didn't have any significant rainfall events over my watersheds so when you hear a a calibration hydrograph this is the output of the what that this is a this a two-dimensional display of the output of flow from the I know the watershed itself OK and there were 3 small low rainfall events but you can see you how to respond to this watershed intended it has flooded the early on in the past the danger in the the 3rd column at the top which is and in the end a display of what rainfall intensity that the peak of the output of the watershed occurred almost immediately after the rain stopped and the dotted lines in there is the same rainfall events on the watershed without the infiltration of stress of the results they can
extract from the from the model are either tabular created the which makes our visualization convenient you get to look at the figure your favorite kind of output display this is the output can be generated for any particular point or a watershed defined I mean and various other kinds of output are possible and available for dealing an analogy and
although life my results are a bit incomplete lines I would strongly encouraged by where you know the the software operating I believe it's possible to to you're regional analysis using the software you may need a small onyx clusters something like that to do regional area perhaps all of the northeastern part of the state of Colorado would require maybe 16 of 30 plus all node cluster but it's certainly possible that today computer I so you can be to run all of this that the background processes then I have the application now let you know when something significant is occurring in that's feature that needs to be built and and that but but anyway um just to isn't
an ideal hydrologic model that has a variety of problems it's a a particularly unstable on steep slopes and which is obviously a problem in my area and it it crashes when the water flows become more critical and but certainly not a good thing in real time for testing environment alternative models need should be evaluated I think um elevating such a large area is difficult because there's so much uncertainty in the calibration constant parameters that are chosen so some of stochastic method for evaluating uncertainty needs to be included I think in the in the model as I
mentioned before automated notification is certainly a a that necessity I think especially if you're going to run it in the background forecasters are often very busy with other things especially when it's raining hard and so they need to be told the something occurring at some particular class something else that should be added use the verification How do we know that people can start really there aren't very good and these sorts of activities need to be here and but like to say that I wouldn't courage market to using the plot the primary compiler only to get a better but a lot of difficulty in translating the software to an operational because it wasn't so bad and maintained in the early nineties and I believe that and in pretty yet crude based plus plus you're that doesn't negate getting the that allows you to this standard code much of the time be found if needed and there you your model will grammar on a relatively small what effect you in your presentation you mentioned also the possibility of the extension over the the the model to the regional of scale was a lot and I did all of this work on my own and I had a very low power computer with not very much memory and I can easily exceed the capabilities so I limited my study this particular watershed also a matter of time and effort calibrating watershed calibration data of runoff data and those sorts of things a difficult problem and I just had the opportunity to you sort of thing or of great a much larger area the results principle but there is it's a fair
question I tried to do is take a typical watershed of that in the but the watershed that I was operating on working with actually
had a very very across run across that a few years ago which made it more prone to plus flooding but I get a higher probability of the events on unintelligible and J. have to questions and from the problem the book money the toward that if the kind of the forecasting model but you did indeed I can video games summation how you forecast that or how you look at to whether of the year but what i presented what assimilation simply because they didn't have or can be made available but the hydrograph and that you get off of a a watershed for instance it runs out over a long period of time and the watershed it may take a long period of time for water to to to reach a height of potential flooding and somebody at some particular point which may be in the future from the current period of time the model generates those data as well and in its output so what you get is an indication of what the current now but you also get an indication of what's going to occur in the future some of future point all those all those outputs that output hydrograph can be compared to non critical states or that particular location and if you exceed the point of flooding or some something close to the point of flying
you have the forecaster conditional forecast to
say we expect flooding to occur at this statement this time rather probably get to this height and will last for such and such amount of time you can issue very specific work at for specific flight and that have talking about that hidden edited kind of and so the creation and he didn't tell them apart but if you know about it you have to know something about the watershed and before you can use forecast of course then and it certainly would have taken measurements state when the water gets to be it's going to cost of
conclusion that you have book modified at the come to the what we today I want you even you can use the point and for that I mean that that he died the modified the model evolution then when it designed to accept radar data all right what it not designed to do is seperate our data in real time and that's my contribution is that I modified approach to that and radar data become available the model you can take advantage of the currently the that the applications there how available to in Graz currently is a is the analysis tool
only and used it to in any other way so you have to have all the data you're going to operate on In order to be competitive and at your work and that then it would move if the that and just kind of models have a lot of parameters and that it would be timely there would be a family that I and who not have they use of parameters and the sophistication of parameters of the wire cannot and the same problem that was mentioned the and by then the speaker the other speaker from both of which some land some I'm going to use in the program which is that available do some poor capping and then it child so that they I think that that that the people the big clear with the people that said if you have been instrumental to let you you don't you don't heart right away before a character the there there are a lot of variations in the area of the variables in the application and as Peter pointed out there a lot just in the way the data itself and the
radar data I I I should point out is not very accurate if we have seen that variations from the true estimates of from 50 to 200 per cent error so there's substantial areas potentially available in the way the data itself but radar meteorology is not my area of expertise so I have to assume that someone who knows much more about that will correct those problems at some point in the future I would just like to provoke congratulate got to do good work on them from a pretty good literature
meteorologists and felt that of that having such a model of the capability to run such models in Europe is really most of these to be functional feeding
grounds for the beach we are a module for something that a change in the process of growing the moment of attention attracted to watching developments dulcimer intended the development in the predators and could if look at that cross if was running will for sure because now we have across so heading in and and can broke that involved that's tremendous but cool the but
Besprechung/Interview
Vorlesung/Konferenz
Kombinatorische Gruppentheorie
Softwaretest
Distributionstheorie
Parametersystem
Bit
Punkt
Spiegelung <Mathematik>
Momentenproblem
Graph
Logiksynthese
Natürliche Zahl
Selbstrepräsentation
Klasse <Mathematik>
Kardinalzahl
Modul
Ereignishorizont
Informationsmodellierung
Datenfeld
Datenverarbeitung
Vorlesung/Konferenz
Hybridrechner
Analysis
Flächeninhalt
Spezifisches Volumen
Hybridrechner
Leistung <Physik>
Dichte <Physik>
Beobachtungsstudie
Wald <Graphentheorie>
Datennetz
Einfache Genauigkeit
Ein-Ausgabe
Datenfluss
Komplex <Algebra>
Ereignishorizont
Dichte <Physik>
Richtung
Überlagerung <Mathematik>
Randwert
Informationsmodellierung
Datenfeld
Flächeninhalt
Mereologie
Luenberger-Beobachter
Vorlesung/Konferenz
Inhalt <Mathematik>
Umwandlungsenthalpie
Schätzwert
Subtraktion
Vektorpotenzial
Punkt
Datennetz
Natürliche Zahl
Einfache Genauigkeit
Prozessautomation
Bridge <Kommunikationstechnik>
Computeranimation
Variable
Spannweite <Stochastik>
Informationsmodellierung
Flächeninhalt
Binärdaten
Meter
Vorlesung/Konferenz
Ordnung <Mathematik>
Extreme programming
Bildgebendes Verfahren
Bildauflösung
Resultante
Punkt
Prozess <Physik>
Wasserdampftafel
Datensichtgerät
Wellenlehre
Mathematisierung
Kinematik
Richtung
Streaming <Kommunikationstechnik>
Informationsmodellierung
Lesezeichen <Internet>
Ganglinie
Algorithmus
Visualisierung
Vorlesung/Konferenz
Figurierte Zahl
Gerade
Analysis
Funktion <Mathematik>
Videospiel
Datennetz
Abstraktionsebene
Temporale Logik
Datenfluss
Ereignishorizont
Minimalgrad
Echtzeitsystem
Normalspannung
Softwaretest
Resultante
Parametersystem
Nichtlinearer Operator
Videospiel
Bit
Wasserdampftafel
Güte der Anpassung
Kartesische Koordinaten
Computer
Konstante
Informationsmodellierung
Knotenmenge
Echtzeitsystem
Flächeninhalt
Software
Mereologie
Äußere Algebra eines Moduls
Cluster <Rechnernetz>
Programmierumgebung
Gerade
Varietät <Mathematik>
Analysis
Aggregatzustand
Beobachtungsstudie
Resultante
Soundverarbeitung
Zentrische Streckung
Compiler
Klasse <Mathematik>
Besprechung/Interview
Formale Grammatik
Programmverifikation
Plot <Graphische Darstellung>
Computer
Kombinatorische Gruppentheorie
Code
Quick-Sort
Modallogik
Negative Zahl
Informationsmodellierung
Flächeninhalt
Software
Festspeicher
Vorlesung/Konferenz
Maßerweiterung
Leistung <Physik>
DoS-Attacke
Vektorpotenzial
Punkt
Gewichtete Summe
Wasserdampftafel
Strömungsrichtung
Frequenz
Ereignishorizont
Informationsmodellierung
Kritischer Punkt
Ganglinie
Computerspiel
Konditionszahl
Indexberechnung
Funktion <Mathematik>
Instantiierung
Umwandlungsenthalpie
Managementinformationssystem
Befehl <Informatik>
Wasserdampftafel
Kartesische Koordinaten
Teilmenge
Informationsmodellierung
Echtzeitsystem
Evolute
Vorlesung/Konferenz
Einflussgröße
Aggregatzustand
Analysis
TVD-Verfahren
Parametersystem
Punkt
Familie <Mathematik>
Besprechung/Interview
Kartesische Koordinaten
Variable
Informationsmodellierung
Flächeninhalt
Vorlesung/Konferenz
Optimierung
Ordnung <Mathematik>
Fehlermeldung
Informationsmodellierung
Prozess <Physik>
Momentenproblem
Mathematisierung
Besprechung/Interview
Vorlesung/Konferenz
Softwareentwickler
Modul

Metadaten

Formale Metadaten

Titel Weather radar enhanced flash flood forecasting
Serientitel Open source GIS - GRASS user conference 2002
Anzahl der Teile 45
Autor O'Donnell, Scott
Lizenz CC-Namensnennung - keine Bearbeitung 3.0 Deutschland:
Sie dürfen das Werk in unveränderter Form zu jedem legalen Zweck nutzen, vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen.
DOI 10.5446/21776
Herausgeber University of Trento
Erscheinungsjahr 2002
Sprache Englisch

Inhaltliche Metadaten

Fachgebiet Informatik

Ähnliche Filme

Loading...