-DSS USER MANUAL
Contents
1. VERSION 1 0 Feb 13 2006 USER MANUAL HELAINE UNIVERSITY OF TECHNO LOA HA ae Chine npg e cra entr alar Ceuta anda Current Predicted ore m A oo 0 0 A B uh 0 17130 40 ug 00 0 5 000 2 og 1020 wg 0150 30 4010 001040 0 00 aan biui hak A bert ere ke ered cn A What is the CLIME DSS Objective The objective of the CLIME Decision Support System CLIME DSS 1s to summarise and illustrate the main results of the CLIME project in such a form that they are easily accessible and comprehensible to interest groups both within and outside of the research community The CLIME DSS presents information about projected changes in the climate of Europe and in lake characteristics driven by the changing climate Development of the CLIME DSS had the following goals 1 capability of illustrating predicted changes in major climate characteristics across Europe 2 capability of illustrating main effects of the projected climate change on key variables characterising lakes across Europe 3 intuitive usability also for persons that are not experts in meteorological and limnological sciences 4 usability over the Internet 5 extensibility 1 e capability to readily incorporate a new decision support issue Brief description of the methodology behind the CLIME DSS The climatic projections used in the CLIME DSS are based on three regional climate models RCM More details are given in Samuelsson 2004 The2. New lake from map When New lake from map 1s active the user can pick any location on the map for analysis The Calibration lake drop down list Figure 5a allows the user to pick that lake whose data have been used in parameterising the Bayesian network a Scenario Lake of interest Topic Calibration lake Map type Model wO matron cron O R O estimate 10 psican O peoe O c Deviation in autumn runoff Control A2 D 113 167 0 00 0 00 113 167 60 113 0 10 0 13 60 113 7 60 0 20 0 30 7 60 46 7 0 63 0 40 46 7 99 46 0 07 0 17 99 46 un Pier e _ Show climatic similarity on map Climatic similarity LE Climate and Lake Impacts In 2071 2100 i Soar Lake of terest Topic ahh ston Lake Map tye Az CE Baaren ho CE Polopos CF Ctra E Pa E cao CP mm m LETEN r d Strong O00 LEEA bir a RELL METE Weak Figure 5 Changes in lakes panel Scenario Lake of Interest Topic Calibration lake Map type and Model drop down lists a Bayesian network figure b probability distribution of the selected variable c climatic similarity indicator d and Show climatic similarity on map check box e The CLIME DSS provides help in assessing how similar the climate is at the calibration site and at the site of interest The idea is that the uncertainty in the results increases when the diffe
3. derived using either simulation model applications or empirical equations Instead of using absolute values in the Bayesian networks of the CLIME DSS deviations from prescribed reference levels are adopted to describe the climate change impact on lake characteristics In other words variables in the Bayesian networks describe how annual seasonal values deviate from the reference level The reference level 1s the long term mean value of a variable for the control period from 1961 to 1990 Deviations are used as it is assumed that regionalisation of results based on deviations 1s more robust than regionalisation of results using absolute values Regionalisation means here application of the CLIME DSS to other European locations outside of those sites from which there were data available in CLIME The CLIME DSS however supports converting the projected deviations into future values The user may select the current level of an objective variable and the system adds the deviation distribution on top of the current level to produce the projected distribution of the future level Technical requirements CLIME DSS is a Java applet and therefore Java not same as Javascript has to be enabled in the Internet browser options The applet requires Java Runtime Environment 1 5 JRE 5 0 can be downloaded free of charge from java sun com CLIME DSS help system Whenever there is a question mark symbol in the CLIME DSS panel pointing to the question mark di
4. OC levels of a lake respond to the climate change To study the projected future DOC levels in Poulaphuca proceed exactly in the same manner as you did with Lough Feeagh in a and Poulaphuca in b You might want to do the analysis using more than one calibration lake in order to assess how sensitive the results are on the lake whose data have been used in quantifying the causal relationships in the Bayesian network References Pearl J 1988 Probabilistic Reasoning in Intelligent Systems Networks of Plausible Inference Morgan Kaufmann San Mateo CA Samuelsson P 2004 RCAO and HadRM3p simulation results with focus on CLIME lake sites SMHI Rossby Centre SE 602 36 Norrk ping Sweden CLIME deliverable nr 5 70 pp
5. RCM results include simulated daily time series of meteorological and hydrological variables for two time periods 1961 1990 control and 2071 2100 future The levels of greenhouse gas emissions are different in the control and future periods Within each period the emissions however remain at the same level The CLIME DSS visualises the projected difference in climate between these two periods In the CLIME DSS the impact of climate change on lake characteristics 1s described using probabilistic Bayesian networks A Bayesian network can be presented as a directed acyclic graph which consists of nodes representing variables and arcs representing dependencies between the variables e g Pearl 1988 All variables in a Bayesian network are described as probability distributions and dependences between the variables are described as conditional probabilities between a dependent node and its parent nodes A parent node refers to an explanatory variable that is directly connected to the dependent variable In the CLIME DSS variables in the Bayesian networks are discrete and the conditional probabilities between a set of variables are presented in the form of conditional probability tables CPT Structures of Bayesian networks 1 e selection of variables and the causal relationships amongst them are identified on the basis of expert knowledge and the structure of environmental simulation models used in CLIME The values for the CPTs are
6. an network For each topic the analysis is based on an application of a Bayesian network whose graphical representation is shown in the lower right corner of the panel Figure 5b Variables are presented as ovals and the causal relationships between them as arrows The user can select a variable by clicking the ovals Above the Bayesian network figure there is graph showing the probability distribution of the selected variable Figure 5c Each distribution is discretised into five classes Both the class ranges in parentheses and the probability value are given next to each class Each class has a different colour and these colours are also used in the Bayesian network figure to denote the mode class 1 e the class having the highest probability value If a node in the Bayesian network figure shows as white all five classes have equal probability values Similar to the Changes in climate panel the user can select the scenario map type and climate model from drop down lists Figure 5a Also there is a Lake of interest drop down list which includes all lakes that were studied in CLIME Figure 5a The user can select the lake of interest either from this list or interactively from the map All CLIME sites are shown as white squares on the map and the selected site is shown as a red square For selecting a location that is not included in the CLIME sites the user can select in the Lake of interest drop down list the first entry labelled as
7. den suggesting that lakes may have problems with DOC Now you need to select that calibration lake whose behaviour most resembles the lake of interest in southern Sweden As explained in b the CLIME DSS can assist you here just in terms of climatic similarity To have a look at the climatic similarity between the calibration lake and the lake of interest have the Show climatic similarity check box checked and keep an eye on the climatic similarity indicator above the Bayesian network figure Say you have selected RCAO E as the regional climate model from the Model drop down list When you now go through the four available calibration sites from the Calibration lake drop down list you can observe that none of the four available calibration lakes are climatically very far away from your lake of interest in southern Sweden This is because no matter which calibration lake you select your site of interest is not covered by the white mask When looking at the climatic similarity indicator Mustajoki seems to be climatically further away from southern Sweden than Lough Feeagh Lough Leane and Moor House Caution 1s warranted here for a couple of reasons Firstly different regional climate models give different results on climatic similarity For example the HadRM3p model suggests significantly smaller regions of similar climate Secondly and more importantly other factors like land use and soil type can be very important drivers in determining how D
8. dicator above the Bayesian network figure Figure 5d Moreover you can check the Show climatic similarity check box above the map which applies a mask that paints those regions white that are climatically far away from the calibration site Doing this when Lough Feeagh is selected as the calibration lake shows that the climate at Poulaphuca is not too different from that of Lough Feeagh the climatic similarity indicator shows a strong similarity and Poulaphuca is not covered by the white mask this is the case for the climate simulation results of any of the three available climate models This is not surprising as Lough Feeagh and Poulaphuca are geographically close to each other To study the projected future DOC levels in Poulaphuca proceed exactly in the same manner as you did with Lough Feeagh above in a c In the CLIME project there was no study site in southern Sweden In order to assess a site not included into the set of CLIME sites select from the Lake of interest drop down list the first entry labelled as New lake from map When New lake from map is active you can pick any location on the map for analysis by clicking at 1t After having selected New lake from map click in southern Sweden to pick your lake of interest You might want to select the Peatlands and lakes map from the Map type drop down list to guide you in locating the study site in southern Sweden The same map also shows that there are peat lands in southern Swe
9. e migration Selecting Climate migration radio button activates a mode where the user can click on a map and identify the place in Europe where the current climate most resembles the future climate at the clicked location Figure 3a In other words the climate migration arrow indicates from where the climate is going to migrate to the selected location When the user selects a location by clicking on the map an arrow with a circle appears Figure 3b The head of the arrow points at the selected location and the tail of the arrow indicates the location where the current climate most resembles the future climate in the selected location Strength of resemblance similarity is illustrated with a circle where the diameter of the red circle within the white circle increases with an increasing resemblance a Click on map to see the location s 2 Climate migration O Mean values of climate variables O Typical daily weather c Scenario Map type Model romeo O b GUME Climate and Lake Impacts in 2071 2100 Changes in labore EBM i Map type a2 Piersical rete sells ee Click vn prap bu see the hoes Climate migration 7 7 Mean values of climate variades Py D Typical day weather F Figure 3 Selecting Climate migration radio button a climate migration arrow on the map b and selecting scenario map type model c The user can select the scenario map type and climate model f
10. ll the five classes to have equal probabilities once again Here it was suggested that the current long term mean is between 5 and 10 mgC I so click the corresponding green class in the Current bar In the Predicted bar you can observe that a significant increase in winter DOC levels is projected for Lough Feeagh The winter DOC level is projected to be in 55 of years above 20 mgC I in 25 of years between 15 and 20 mgC I in 14 of years between 10 and 15 mgC I in 6 of years between 5 and 10 mgC I and in 1 of years below 5 mgC I A similar analysis for the remaining seasons reveals that DOC levels are projected to increase in all seasons b Select Poulaphuca as the lake of interest in the same manner as you selected Lough Feeagh in a As Poulaphuca is not one of the calibration lakes that one of the four available calibration lakes should be selected that 1s most similar in its behaviour to Poulaphuca The similarity in behaviour depends on several factors including soil type land use and vegetation Due to lack of data the CLIME DSS unfortunately does not help the user to assess if two sites are similar in terms of the above mentioned factors Also the analysis is assumed to more reliable for the region where the climate does not differ very much from that at the calibration site The CLIME DSS provides help in assessing how similar the climate is at the calibration site and at the site of interest There is a climatic similarity in
11. long term mean level is between 5 and 10 mgC I c in a lake in southern Sweden when the current long term mean level is assumed to be between 10 and 15 mgC I Answer Click first Changes in lakes button Figure 1 1f the Changes in lakes panel is not already active Select Water colour DOC from the Topic drop down list Figure 5a Selecting either A2 or B2 from the Scenario drop down list Figure 5a allows you to assess the projected precipitation change either under more A2 or less B2 severe atmospheric emission scenarios The Model drop down list Figure 5a provides the option to change the regional climate model which has been used for generating the results should you wish to assess the sensitivity of the results on the choice of the climate model Also you can select the background map type from the Map type drop down list Figure 5a You can choose between three different map types Physical relief Land cover and Peatlands and lakes a Select Lough Feeagh as the lake of interest There are two alternative ways for doing that You can either select Lough Feeagh from the Lake of interest drop down list Figure 5a or you can click on the map the white square at the location of Lough Feeagh When you point to any of the white squares on the map the name of the lake appears in a yellow message box You also need to select the calibration lake from the Calibration lake drop down list Figure 5a Calibration lake means tha
12. mples of use Changes in climate Q1 How is precipitation projected to change in the winter season in Europe from the control period 1960 to 1990 to the future period 2071 to 2100 Answer Click first Changes in climate button Figure 1 if the Changes in climate panel is not already active Also make sure that either Mean values of climate variables or Typical daily weather radio button is active Figure 2b Select Precipitation from the Variable drop down list and Winter from the Season drop down list If you want to see the projected change in millimetres select Absolute climate change and should you wish to assess the change in percentage select Relative climate change from Map type drop down list Figure 2a The map shows in red and blue colours the spatial distribution of the projected change in precipitation in Europe Red colour denotes increase and blue decrease The magnitude of the change is represented by the shade of the colour The legend is shown at the bottom on the right side of the map Selecting either A2 or B2 from the Scenario drop down list allows you to assess the projected precipitation change either under more A2 or less B2 severe atmospheric emission scenarios Figure 2a Please see e g Samuelsson 2004 for more information on the emission scenarios Changing the regional climate model on whose results the projections are based from the Model drop down list Figure 2a allows you
13. nge is computed as the difference between the 30 year mean values for control and future periods 2 Climate migration This tool displays where in Europe the current climate most resembles the future climate in the selected location 3 Changes in typical daily weather As opposed to single variables daily weather considers several variables at time Changes in mean values In order to study projected changes in mean values of climate variables the Mean values of climate variables radio button must be selected Figure 2b The system stores over 300 climate change maps which display results for different combinations of 1 Emission scenarios according to the Intergovernmental Panel of Climate Change o A2 and B2 o Scenario drop down list Figure 2a Variables o Precipitation runoff snow soil moisture air temperature wind speed o Variable drop down list Figure 2a Season o Autumn spring summer winter o Season drop down list Figure 2a Relative or absolute change o Some variables allow to study both relative and absolute changes o Map type drop down list Figure 2a Regional climate model o 3 different climate models labelled as HadRM3p RCAO E and RCAO H for more information see Samuelsson 2004 o Model drop down list Figure 2a The user selects from drop down lists Figure 2a which climate change map is displayed Pointing to the question marks next to the drop down lists displays information on the selected o
14. nges in climate button Figure 1 if the Changes in climate panel is not already active Also make sure that you have selected the Climate migration radio button Figure 3a Click location of Stockholm on the map and you see an arrow whose head 1s at Stockholm The tail of the arrow indicates where in Europe the current climate is most similar to the future climate in Stockholm Strength of resemblance similarity is illustrated with a circle where the diameter of the red circle within the white circle increases with an increasing resemblance From the Scenario drop down list Figure 3c you can change the emission scenario to study its effect on the location of the closest resemblance Moreover the regional climate model used for climate simulations can be used from the Model drop down list Figure 3c In order to assess how sensitive the climate migration result is on assumptions used in the analysis it is recommended to test different climate models and to click a couple of times in the neighbourhood of the study location in this case Stockholm Sometimes the tail of the arrow jumps a substantial distance as a result of changing the climate model or when the study location is changed only little Changes in lakes Q3 What is the projected future DOC concentration distribution a in Lough Feeagh in western Ireland when the current long term mean level is between 5 and 10 mgC 1 b in Poulaphuca in eastern Ireland when the current
15. ption a Scenario Variable Season Map type Model a2 y Precipitation autumn absolute climate change rcaoe b Click on map to see the location s O Climate migration B Mean values of climate variables O Typical daily weather CUME SOMA Variable Mayor Map hype Maget IRE Precpaanon zO Anua O Absohte climate change F ACAO E es c Absolute change in mean 430 430 215 a B 215 O g Units mmseason Variable Scenario 2 Precipitation minfseasar 298 4 311 1 275 9 117 4 E o 139 1 185 Air temperature C 5 Wind speed mis 4 Figure 2 Selecting the Scenario Variable Season Map type absolute or relative change and Model from drop down lists a selecting Mean values of climate variables radio button b legend for the map c and table of climate variables for the selected location d There is a legend in bottom right corner that explains what the colours in the climate change map mean Figure 2c In general blue means always decrease red increase and white that there is no change between the control and future periods In addition to looking at the spatial distribution of the projected change on the map the user can click to a location on the map orange dot for a more detailed table showing the seasonal values of climatic variables for the control period as well as for the future period under both A2 and B2 scenarios Figure 2d Climat
16. rence in climate between the calibration site and the site of interest increases There is a climatic similarity indicator above the Bayesian network figure Figure 5d Moreover by checking the Show climatic similarity check box above the map Figure 5e the user can apply a mask that paints those regions white that are climatically far away from the calibration site Assessing future levels While the DSS provides tools to assess the distribution of all variables in the Bayesian networks the most common way to use it 1s to select the current level of the target variable and then study the distribution of the projected future level At the bottom of each network there are nodes representing current levels of the target variables By selecting a Current level node and then clicking on any of the five classes the user can fix the current level to that class In addition to selecting a single class to represent the current level the user can drag the borders between the classes to specify a distribution for the current level When the mouse cursor is placed on the border between two classes the cursor changes into a double arrow to indicate that the class height can be adjusted Also to enable an easier assessment of future levels using different current level scenarios the same functionality of adjusting the current levels is available when a node representing the Predicted level is selected at the very bottom of the Bayesian network figure Exa
17. rom a drop down list Figure 3c Different map types include relief land cover and location of peat lands and lakes the size of lakes 1s exaggerated for better visibility Changes in typical daily weather Clicking the Typical daily weather radio button Figure 4a changes the table of climate variables to a table of typical daily weather Figure 4b In contrast to studying climate variables separately the typical daily weather considers several climate variables at one time Weather types have been determined using discrete classes of the following variables precipitation runoff snow water equivalent soil moisture air temperature and wind speed Discretisation intervals are shown when the user points to question marks in the last column of the table a Clink Olimale and Lake impact i 2071 3100 m De nm Click on map to SEE the location 5 O Climate migration CP 0 Mean values of climate variables P 8 Typical daily weather gt b Typical daily weather in spring a 16 5 11 4 B2 6 13 8 11 7 Control 3 17 6 17 Precipitation Figure 4 Selecting Typical daily weather radio button a and table of typical daily weathers for the selected location b In brief the only difference between Changes in mean values mode and Typical daily weather is that the table showing the table of climate variables is replaced with the table of typical daily weathers which shows the fo
18. splays information on the topic relevant to the context Two sides of the CLIME DSS When the user opens the CLIME DSS a load indicator appears while the CLIME DSS retrieves data from the server The CLIME DSS is divided into two main panels 1 Changes in climate and 2 Changes in lakes The user can switch between these two panels by clicking the two tabs that are labelled Changes in climate and Changes in lakes and are located at the top of the page Figure 1 Climate and Lake Impacts in 2071 2100 Changes in climate Changes in lakes Climate and Lake Impacts in 2071 2100 Ch 2 Changes intakes sii da Y Multa pason ap type fodel O ole IO e so IO 7 Click on map to see the location s O Climate migration Q ge dl 8 Mean values of climate variables O yk Tical daily weather 2 i if p aes Fe dete k l Mean values of climate variables in autumn k Variable Scenario A2 Control Precipitation mm season 298 4 311 1 275 9 56 117 0 106 51 1391 185 Air temperature C 10 Wind speed mis 4 4 ES 2 Absolute change in mean Fo 430 430 215 Changes in climate This panel provides tools for studying how the climate is projected to change in Europe from the control period 1960 to 1990 to the future period 2071 to 2100 The projected climate change is presented in terms of 1 Changes in mean values of a set of climate variables The cha
19. t lake or catchment of the lake as is the case in the DOC topic whose data have been used in quantifying the causal relationships in the Bayesian network for DOC For the topic of Water colour DOC there are four calibration lakes available Preferably you should choose as the calibration lake a lake that resembles the lake of interest In this case it is easy because Lough Feeagh 1s also one of the four available calibration lakes for this topic When you select Lough Feeagh as the calibration lake you see a blue edge to appear around the square showing the location of Lough Feeagh The blue edge denotes that a lake is the selected calibration lake for the current topic Now you are ready to proceed to assessing the projected future DOC levels At the bottom of the Bayesian network figure there are four nodes labelled as Predicted DOC winter Predicted DOC spring Predicted DOC summer and Predicted DOC autumn Select one of these nodes e g winter by clicking on it and you will see two distribution bars labelled as Current and Predicted on top of the Bayesian network figure The bar representing the current state is divided into five classes having an equal probability To set the current state click on any of the five classes When you do that the system shows the projected future distribution in the bar labelled as Predicted If you want to change the selected current level click again on the Current distribution which changes a
20. to study how consistent the three climate models used in CLIME project are in their projections for different parts of Europe For example it can be concluded that all model show an increase in precipitation for England Ireland Sweden Finland and Denmark But for example in case of Norwegian west coast RCAO E projects an increase in winter precipitation while RCAO H and HadRM3p suggest a decrease To obtain more detailed information on a specific location you can click on the map and look at the table showing either mean values of selected meteorological variables Mean values of climate variables radio button selected Figure 2b or typical daily weather Typical daily weather radio button selected Figure 2b See previous sections for more details on what is meant with the typical daily weather Note that the mean values for meteorological values are results of a simulation model which uses 50 x 50 km grid cells in the computation Due to local controls on climate it is quite possible that in the control period representing current climate you obtain results that seem to be different to your knowledge of a certain location within that grid cell It is recommended to pay more attention to the projected change in climate than to the absolute values of the climate variables in the control and future periods Q2 Where in Europe the current climate most resembles the future climate in Stockholm under scenario B2 Answer Click first Cha
21. ur most common daily weather types for the selected location and the selected season The percentage values in the first three rows give the frequency of the four daily weather types under control and future scenarios Changes in lakes General instructions This panel allows the user to analyse the effects of climate change on the key lake characteristics Those lake characteristics that are available for the analysis are listed in the Topic drop down list Figure 5 a which is located on top of the map of Europe In addition to the topic the user can select the Scenario the Map type and the climate Model from drop down lists Figure 5a similar to the Changes in climate panel Also there is a Lake of interest drop down list which includes all lakes that were studied in CLIME Figure 5a The user can select the lake of interest either from this list or by clicking a CLIME lake location on the map All CLIME lake locations are shown as white squares on the map and the selected site is shown as a red square For studying a location that is not included in the CLIME sites the user can select in the Lake of interest drop down list the first entry labelled as New lake from map When New lake from map is active the user can click any location on the map for analysis The Calibration lake drop down list Figure 5a allows the user to pick that lake or catchment of the lake whose data have been used in identifying values for the CPT s in the Bayesi
Download Pdf Manuals
Related Search