"user manual"
Contents
1. eb amp 2 z O 5 4 O a Be So k O 2 gt LLI O 2 oS c O CMCC Research Papers Bias Correction Layers yi ws_orvieto_prec_day19212012 Type r l cosmo_italy8km_era40_tprec_day19712000_ lw Macro Areas Area n ITALY_CENTRO_CLM Add Layers Control Time Projection Time NOTE layers must be selected from catalog first Start Date Time Start Date Time Observation Grid ws_orvieto_prec_day19212012 1971 01 01 00 00 00 z 1971 01 01 00 00 00 z Model Grid Control cosmo_italy8km_era40_tprec_day1 9712000_Iw End Date Time End Date Time booo 12 30 00 00 00 F booo 12 30 00 00 00 zi Model Grid Projected cosmo_italykm_era40_tprec_day 9712000_Iw Output Table Name biascorected_table Linear Scaling Quantile Mapping O O c eb z amp 2 z i O 5 O eb c a gt ks ko h c3 O 2 gt LLI O J c O N points per square side 5 must be odd Sel point Reference Station Model Grid Nearest Nb LON 12 140639 12 100000 12 131549 LAT 42 719600 42 716999 42 684989 z lt Set Nearest gt gt Select process Multiplicati nt idati Multiplicative I cross validation SEs In order to run the following processes Fi software is needed Download it from here Figure 23 Bias Correction form with Linear Scaling process menu Vow each related to a single month of the year2. Edtor mPa lat iT Ho LS De ob gt Bl AES Import Export Data Eg Raster List diff winter 1113 16330296 Export Import Choose Format IMOD Figure 30 Import Export form CONCLUSIONS This report describes all functionalies currently implemented into Clime software developed at CMCC REMHI division such features have been conceived and realised with focusing on the actual needs of any end user expected to perform climate analysis without a specific ex pertise in this field Climate data considered in Clime are either observed and simulated per mitting a large use of this software for different purposes Furhermore this tool can be im proved with new features following needs and feedbacks provided by all communities adopt ing it as a possible standard for climate analy S S O O z c eb z amp 2 z O 5 O c La ts Sn cb oO e3 O pa gt LLI O 3 c c O O O eb z amp 2 z i O 5 O eb c ks ko h e3 O ho gt LLI O B c O 6 CMCC Research Papers Bibliography 1 V Conrad and Pollack C Methods in Cli matology Harvard University Press page 459 1950 2 C D andreis C Pag P Braconnot L Barring E Bucchignani W de Cerff R Hutjes S Joussaume C Mares S Planton and M Plieger Towards a dedicated impact portal to bridge the gap
3. Resolution ALL x Field Temperature x Show list LS Time Aggregation cosmo_italy8km_om45_t2m_day20012070 Reference period Field name Figure 8 Catalog form displaying all layers available for Clime processes Centro Euro Mediterraneo sui Cambiamenti Climatici gle layer selected in as Reference Layer lrer with all the other ones ln cal culating a point by point difference lrer Thus every plot contains the spa tial mean of such difference bias repre sented at the selected time aggregation Trend option available tion it works faster with already monthly aggregated data PDF the Probability Density Function is obtained with counting the occurrences within every bin interval and then nor malized by the total number of values discrete approach Bin resolution is se lectable as well of total range of values Scatter Plot this process compares two but the latter could also be automatically detected overall minimum and maximum are chosen in this case Reject val ues allows to reject all values below se lected threshold before building the PDF By checking Spatial Mean Values option input values are actually averaged over the domain area for every time step as described before otherwise all single point values are processed m Bias this process compares the sin datasets by placing their value domains onthe X Y axes and drawing a point as re sult for every time step o
4. collected over the entire time period In this way every value of input grid V cp is sub ject to a correction depending on its position and the month it belongs to in order to obtain a table of corrected values Vogg Output is gen erated on a square grid of the same dimension and position of the one defined as the reference station neighbourhood V orr Vzom Von Viicm l l Vi Voorr Vrem oe 2 VROM 6 2 QUANTILE MAPPING Differently from previous case this process compares datasets by focusing on their statisti cal characterization Given a modeled variable Vm and an observed one V their relationship can be expressed through the transformation operator h in the following way Vo Va 3 Given that the distribution of modeled variable is known the latter equation could be displayed as follows Vo HTa aD 4 Where F is the CDF related to Vm and F7 is the inverse CDF also defined as quantile func tion of V 4 Since there are several ways to approximate quantile function user is allowed to choose among a wide range of algorithms 3 More specifically this process Figure 25 com pares a selected station with a spatial mean of the surrounding square grid so only single point grids are produced as output Currently the whole process is focused on one manually chosen station point so only single point grids are produced as output Some algorithms be longing to this cla
5. Days Year with prec gt 10mm Number of Days Year with prec gt 20mm Largest number of consecutive days with prec lt 1mm Largest number of consecutive days with prec gt 1mm Annual Precipitation during Wet Days prec gt 1mm aries percentile evaluation is confined within the chosen interval with all other values being completely ignored from Statistic count Station check is an ad ditional control to activate in case input layers do not present values at every time step missing data which frequently hap pens for station data Since such option comes at a higher computational cost it is recommended to enable it only if nec essary If desired it is also possible to enable the option to evaluate percentile differences from reference layer Seasonal Differences Extreme Indices Percentiles Trend Tests Differences Errors BEN Select percentiles T Lower boundary Upper boundary e Station check recommended for station layers alculate differences Ae Add layer s fic m Aefererice Layer Figure 18 Percentiles menu Multiple thresholds could within a single process Trend tests in this section Figure 19 it is possible to run trend tests on ev ery point of the input layers and show their results on the map Currently the only test available is Mann Kendall see Centro Euro Mediterraneo sui Cambiamenti Climatici Centro Euro Mediterraneo sui Cambiamenti C
6. chosen by the user Type the variable considered e g Tem perature m Name the short name of the variable e g T_2M m Unit measure the desired Unit measure e g K for Kelvin degrees Transformation if data in the original file are in a different Unit measure from the desired one it is necessary to specify the Transformation factor e g K gt C if orig inal data are in Kelvin degrees but are requested to be in C degrees Partitioned option should be selected if a large amount of data must be imported e g daily data over a 30 year period During the process temporary CSV files are created in clot_daily1 91 983 Start Date Time Periods fia79 01 01 00 00 00 MONTH _ 5 A Server Dump Path Di dumpdbais Import i Figure 2 Import section Clime Data Manager a dump folder whose path is chosen by the user and has two different names the one directly related to destination device Server Dump Path and the other as seen from user machine Local Dump Path generally includes an IP address these paths are equal if data is imported into the same device localhost Finally the import process can be started by clicking on Import button and the related layer is copied into the database selected in Server address box 3 2 AGGREGATE Since data contain values characterized by a regular time step e g hour day month and so on it is also possible to create objects start
7. extend the application field for high resolution climate models The main focus of research in this last field is to improve downscaling processes in order to have them grant high standards of technical performances and reliability Plus the identified dowsncaling method is expected be implemented through a fast algorithm without high hardware require ments once it is finally selected it is neces sary to perform an extensive validation of its results produced by comparing them with time series collected from weather stations radars and satellite data Comparison of an usually large number of permutations along with the processes for data homologation requires au tomated and generalized procedures that must also be equipped with interfaces to link them into the operating pipeline All these needs have brought to the development of the CMCC Clime software which provides several meth ods for post processing and validation function alities featuring the above described interoper ability Assuming that the base structure of a GIS is characterized by a set of layers in vec tor or raster format collimating square cells where any climate model and dataset could be easily stored Clime has been implemented as an extension for ESRI ArcGIS Desktop and is launched from a plugin user interface bar anchored to the main toolbar allowing users to take full advantage of the high level primi tives e g block functions for interpolation al
8. gebra on raster reference systems transforma tions and many other features provided by the base software as a result this combined ar ray of processes is expected to cover all the steps concerning the phases of validation and data processing Finally analysis results are displayable in a variety of formats and stan dards with any assignment of classifications histograms and legends 2 SOFTWARE ARCHITECTURE Clime is classified by its nature of extension as a special purpose GIS software integrated in the consolidated and evolved ESRI ArcGIS Desktop 10 X thus providing in this mode a dynamic linkage library DLL that is compat ible with Microsoft Windows operating systems all versions NT compliant and functions pro vided by ArcGIS Desktop As shown in Figure 5 Clime tightly integrates its graphical user inter face plugin mode with the host system through an anchored bar with function buttons each one related to a distinctive feature of the soft ware Besides it is designed to act mostly as a stand alone utility in order to meet easy porta bility requirements the implementation of his algorithms has been coded separately from the GIS portions for which routine calls to the na tive environment are used On a closer look Clime operates on an internal database pow ered by Microsoft Access RDBMS which can be accessed only through a SQL 9 2 declara tive language linked to a catalog dedicated on tracking any tas
9. high res olution regional climate models The CMCC REMHI division also collaborates with local in stitutions interested in climate change impacts on the soil such as river basin authorities in the Campania region ARPA Emilia Romagna and ARPA Calabria Hence Clime a Geographic Information System G I S developed add in tool is the result of such close collaboration with impact communities with the main goal to grant the use of climate data also to users with little experience in this field It features a reli able interface allowing to easily manage climate data and evaluate their reliability over any ge ographical entity of interest by accepting mul tiple sources of different formats like observa tions and or numeric model outputs and using them as inputs for traditional models hydro logic slope stability etc The latter feature is of particular interest for different end users be Cause Spatial resolution of modern regional cli mate models e g COSMO CLM MM5 WRF Model is currently of about 10 km which is too poor for impact studies or other activities which may involve civil protection cultural heritages historical studies of impact in limited areas that need input data at a resolution of about 100 me ters For this reason climate data are usually processed with any of the downscaling meth ods provided by literature It is clear that down scaling approach represents a crucial research activity in order to
10. reference station point with Add Layers both model and observation grids ap pear on the screen then the Get a Point func tion from Clime toolbar enables user to choose a given point by mouse click Figure 24 whose position is registered and used to evaluate the nearest point relative to every grid under exam lt lt Set Nearest gt gt Since the process takes into account other points surrounding the reference ones within a square neighbourhood the user must determine the size of such area by de fault it is a square with a 5 points side Finally it is possible to choose the algorithm to use for the bias correction Quantile Mapping and Linear Scaling each one with its own set tings 6 1 LINEAR SCALING This method consists in correcting the daily se ries starting from monthly values For all the 12 months the ratio between simulated and ob served values is evaluated and then applied to the input series as a correction factor 14 It is possible to run a process in cross validation mode in order to have a correction mask ap plied to the same model dataset used to create it Anyway this stage is just for performance evaluation purposes and not strictly required As shown on Figure 23 two distinct algorithms are available to evaluate and apply a correction mask Additive and Multiplicative 2 The mask is a gridded layer evaluated from the means of observations V g and model data O O Cc
11. time specific data field s three dimensional CSV Header Clime climate data processing in GIS environment Clime Data Manager ersion 1 60 lel es In Data Raster 1 name type_lyr resolution type_field name_field dimension type_periods aj ee cosmo_persmed14km_era40_t2m_day19712000 COSMO 14 00 Temperature T_2M G DAY cosma_persmed1 4km_era40_tmaxtmin_day1 9712000 COSMO 14 00 Temperature TMAX_2M i DAY ___ cosmo_persmed1 4km_era40_tmaxtmin_day1 9712000 COSMO 14 00 Temperature TMIN_2M f DAY cosmo_persmed 4km_erainterim_t2m_day19792011 COSMO 14 00 Temperature T_2M i DAY Lee cosmo_persmed25km_eraint_day19792011 COSMO 25 00 Temperature iE MAx_2M f DAY oa cosmo_persmed25km_eraint_day19792011 COSMO 25 00 Temperature T_2M i DAY Nel cosmo_persmed25km_eraint_day19792011 COSMO 25 00 Temperature TMIN_2M i DAY 4 Filter by Server address Enter keyword Grid Type JaLL 7 Time Aggregation DAY gt C Localhost Resolution falL gt Field Temperature gt Remote IP 156 14 147 2 Refresh Edit Delete Database cemce isc 7 Filter AddtoMap J Start from dump Files Theo Bh Correction Import Aggregate Export TasksLog Clear m Add Type Layer Nominal Res Km Start Date Time Periods cosmo v 28 X fi979 01 01 00 00 00 zj DAY Select file D c T_2M_daily19791983 ne Layer Name Fal Suggest Iv Partitioned No for tiny data Lon Lat Loca
12. CMCC Research Papers Homogenise oix name type_lyr resolution type_field dimension type_periods date_from ws_bolzano_day19812012 WSTATIONS 1 00 Temperature i DAY 01 01 1981 ws_bolzano_day19812012 WSTATIONS 1 00 Temperature E 0101 1981 ws_bolzana_day19812012 WSTATIONS tec_sum nm day JAY 01 01 1981 ws_bolzano_prec_mntl9212012 WSTATIONS 1 00 Days wetdays days month MONTH 01 01 1921 ws_bolzano_prec_mnt19212012 WSTATIONS 1 00 Rainfall prec_sum mm month MONTH or 011921 ws_bolzana_temp_mnt19612012 WSTATIONS 4 00 Temperature tmin c MONTH 0120171961 me halanna feran ront1 9019017 UWICTATIONG 1An Temnerstiiro tra ay o gt KANKITU ni sn 41981 Filter by IV Enter keyword bolzano Grid Type fit ts Time Aggregation fat Add layer Refresh Resolution aLL gt Field JALL Selected point Candidate Station Name Jws_boleano_day 13812012 Max distance from candidate km Si a e lar Field prec_sum Dimension mm day N reference stations tice Mana Period DAY N min reference stations Minimum correlation Start 0170171981 00 00 00 End 31 12 2012 00 00 00 Multi Station Outliers percentile 95 ID fasto Report folder C Users Public Dacuments ClimeSHomogenise ty Start fi981 01 01 00 00 00 Outlets seach lt lt Breakpoints analysis End bo12 12 31 00 00 00 Start Save input RData file Start Daily precipitation data Fig
13. Cmcc Centro Euro Mediterraneo sui Cambiamenti Climatici Research Papers Issue RP0257 June 2015 Regional Models and geo Hydrological Impacts Division REHMI By Luigi Cattaneo Regional Models and geo Hydrological Impacts Division CMCC Via Maiorise s n c l 81043 Capua cattaneo cira it Valeria Rillo Regional Models and geo Hydrological Impacts Division CMCC Via Maiorise s n c l 81043 Capua v rillo cira it Maria Paola Manzi Regional Models and geo Hydrological Impacts Division CMCC Via Maiorise s n c l 81043 Capua m manzi cira it Veronica Villani Regional Models and geo Hydrological Impacts Division CMCC Via Maiorise s n c l 81043 Capua v Villani cira it and Paola Mercogliano Italian Aerospace Research Centre CIRA Regional Models and geo Hydrological Impacts Division CMCC p mercogliano cira it The work here presented has been carried out in close cooperation with dr Francesco Cotroneo Clime climate data processing in GIS environment SUMMARY Clime is an extension software for ArcMap 10 environment featuring multiple tools for observed and simulated climate data analysis Since a large number of functionalities is featured in Clime this report has been intended as an introductive guide for any user which could be interested on its practical purposes Due to its nature a background knowledge of ArcGIS software is required The paper is structured as follows se
14. O r O eb z amp 2 z i O 5 O eb c a gt ks ko h e3 O ho gt LLI O c O CMCC Research Papers through the Server address box and searching the requested data by filtering results selecting m Grid type numerical climate models as COSMO CLM 11 gridded observational dataset as CRU 5 9 Resolution spatial resolution in km Time aggregation DAY MONTH YEAR SEASON Field e g Temperature Rainfall Wind speed Multiple choices are allowed for example COSMO 8 km Temperature It is also pos sible to filter results by name selecting a part or the whole name of the desired layer Once that the search criteria have been defined it takes to click on Refresh to visualize the re quested layers Then he desired element is selected by clicking on the gray square in the leftmost column in this way the entire row will be highlighted The lower half part is dedicated to three different processes Import Aggre gate and Export It is also possible to fil ter results by selecting a keyword included in the name of desired table for instance the search can be focused on every object contain ing daily aggregated temperature data with italy in its name with COSMO grid 8 Km resolution Once that all search criteria have been defined it takes to click on Refresh to visualize the re quested layers Then the desired element is selected by clic
15. Temperature C IV Show Legend Ey x Move v 1972 1974 1976 1978 RI year Export to Excel Export Image Figure 11 Sample Plot form with Time Series The window is named after plot domain area Temperature PDF 0 14 E 088 E ccccm MB Raan 0 12 0 1 0 08 PDF o 5 10 15 20 25 30 Temperature C Figure 12 PDF sample Scatter Plot 24 22 20 oO 6 20 25 15 EOBS Temperature data C Figure 13 Figure 14 Scatter Plot sample Taylor Diagram sample Centro Euro Mediterraneo sui Cambiamenti Climatici O pre O c eb z amp 2 z i O 5 O eb c ks ko h e3 O ho gt LLI O B c O CMCC Research Papers 5 2 ELABORATIONS This macro section form shown in Figure 13 includes all processes which produce graphi cal outputs point feature or raster which are all available in the Table Of Contents win dow while their source file is saved at the path chosen during Start options All maps are geo referenced with the projected system WGS 1984 Web Mercator Auxiliary Sphere The output map can be obtained from a single layer or can be the result of a difference between map and maprer Map test map is related to any object from the input layer list while maprer reference map is the one selected from the Reference Layer box A difference output always originates raster objects Laye
16. W TO START CLIME LOGIN Once ArcMap 10 is started Clime toolbar can be made visible by checking Clime CMCC on toolbar list mouse right click on screen dis playing multiple buttons Figure 5 each one related to a form characterized by its distinct Clime climate data processing in GIS environment Import Aggregate Export Table Name cosmo_alps2km_eraint_run2_day1980 t_2m tmin_2m tmax_2m Select All V tot_prec Remove All Figure 4 Export section Clime Data Manager set of processes except for Get Point which enables an interactive mode with ArcMap en vironment and is assumed to co operate with other functions At the beginning the only one active is Get Started opening the login form Figure 6 user can edit the database list and select the ones to connect to before logging in for any ongoing process Moreover it is possi ble to choose the folder path where to save out put raster objects In order to properly run all further operation Climate should be selected as Primary Mode After setting these pref erences Clime session can be started through CMCC Clime button Overhaul and Compare Get Point Get jail Started nterpo ation techniques Import and Export Figure 5 Clime toolbar in ArcGIS Desktop 10 environment Each button calls a different form 5 OVERHAUL amp COMPARE In Overhaul amp Compare form Figure 7 it is possible to select the
17. alues within the base pe riod Count number of time steps within the base period where chosen condition oc curs Clime climate data processing in GIS environment Table 2 Complete list of statistic operators evaluated by Errors with their implementation Mean p Sere ke Vari E _ 2 ariance oy o n 1 Xn UX 2 Covariance o y len SVG 57 Standard deviation ox o 2 ORV Correlation pxy OY Bias BIAS 45A Xn Yn Mean Absolute Error MAE 5AL Xn Yn Root Mean Square Error RMSE Centred Root Mean Square Error CRMS If requested the process can be limited only to a particular period or season of the year Month selection allows determining the months to be observed Input data is treated accord ing to Monthly base settings which produces monthly values Then if the user chooses to aggregate by year lower left option box data are ready to be processed by Yearly base op erator A spatial mean is performed in order to have output represented on time plot In this way it is possible to analyse critical events within a chosen period by evaluating their ex General Errors Indices Month selection Select all DJF MAM Mjn Wap Miu Moc M feb M may M aug M nov Deselect al JJA SON ao go ae die Monthly base i m Yearly base value constraints value constraints mean mean C min sum l Coan Sn gs C max C count 42 7 C max count l
18. an of every element then performs index eval uation which may involve comparison be tween two layers At the end of process all indices are shown on a table Draw Taylor Diagram Taylor Diagram is a quick way of comparing the behaviour of multiple datasets with respect to a ref erence one 13 All datasets including the reference one are represented as points inside a circle being their radial distance proportional to the standard de viation The reference point is located on the abscissa axis The distance of Centro Euro Mediterraneo sui Cambiamenti Climatici Centro Euro Mediterraneo sui Cambiamenti Climatici CMCC Research Papers Table 1 Complete list of plots and their related available options Min Max St Dev Least Sq Time Series Yes Yes Yes Seasonal Cycles Yes Yes No PDE No No No Bias No No Yes Scatter Plot No No Yes Correlation Yes Yes Yes Verif Meas No No No Plus each point from the reference one mea sures the centred root mean square error CRMS correlation depends on the an gle and varies as cosine The diagram is in normalised form and all distances are divided by the reference standard de viation so the reference point always is always located in 1 0 the following options are available Indices are evalu ated by default starting from spatial means of each dataset temporal dia gram choosing spatial diagram it is possible to perform time mean instead The la
19. between the impact and climate commu nities Lessons from use cases Climatic Change 125 3 4 333 347 2014 L Gudmundsson Package qmap Sta tistical transformations for post processing climate model output CRAN January 2014 L Gudmundsson J B Bremnes J E Haugen and T Engen Skaugen Techni cal Note Downscaling RCM precipitation to the station scale Hydrology and Earth System Sciences pages 3383 3390 2012 I Harris P D Jones T J Osborn and D H Lister fdjrsjs International Journal of Cli matology 34 3 623 642 2014 ISPRA Elaborazione delle serie tempo rali per la stima delle tendenze climatiche Stato dell Ambiente 32 2012 July 2012 ISPRA Linee guida per lanalisi e l elaborazione statistica di base delle serie storiche di dati idrologici Manuali e Linee Guida 84 2013 2013 K Manoj and K Senthamarai Kannan Comparison of methods for detecting out liers International Journal of Scientific amp En gineering Research 4 09 14 2013 T Mitchell and P Jones An improved method of constructing a database of monthly climate observations and associ ated high resolution grids Int J Climate 25 6 693 712 2005 3 4 lL 5 La 6 e 7 iN 8 N 9 L 10 11 12 13 14 15 16 17 M Montesarchio A L Zollo E Bucchig nani Mercogliano and S P Castellari Performance evaluation of high res
20. ction 1 Introduction briefly explains the reasons who brought to software development along to its general purposes section 2 is an overall description of software internal architecture section 3 deals about all data import and managing processes to run before analysis in section 4 database connection settings are described section 5 shows all processes involving output image rendering like plots and maps section 6 explains Bias Correction tools section 7 is about homogenization of station data finally section 8 describes all remaining processes dealing primarily on graphic interpolation and format conversion Keywords Climate Data Analysis Geographic Information Systems O poe O eb z amp 2 z i O 5 O eb c ks ko h c3 O ho gt LLI O B c O CMCC Research Papers 1 INTRODUCTION REMHI Capua division had several collabora tion experiences with impact communities in cluding European Projects such as IS ENES VIL FP Infrastructure 2008 2 SafeLand VII FP Environment 2008 about the study of landslide risk in Europe ORIENTGATE South East Europe Transnational Cooperation Pro gramme 2012 and finally INTACT VII FP In frastructure 2013 These partnerships brought to the execution of different reseach activities concerning the quantitative analysis of the vari ous impacts of climate change which are mostly based on the use of high and very
21. e as well as the forms in detail any part not directly interact ing with ArcGIS in all major hardware platforms and software Unix like MacOS X86 SPARC In this way the real porting issues are expressly limited to the GIS modules geo processing map algebra spatial interpolation reference systems but all these components are well documented and their reimplementation is not strictly necessary since there is the possibility to use business forms and many available Open Source codes The described operations could either keep the original Clime layout intended as a mere extension e g with multiplatform as QGIS or GRASS or simply provide it with a stand alone execution mode 3 EDITING PHASE CLIME DATA MANAGER Clime is conceived to handle data with specific features so it is necessary to build and arrange data in a suitable way Such pre processing phase is carried out by Clime Data Manager Fig ure 1 a database interface which allows to im port new data and to edit existing ones This software is executed separately and does not need any environment application Data are stored as layers Afterwards itis possible to run any desired process with the main software In the upper part it is possible to manage all pre viously stored layers by selecting the database O p E O z eb amp 2 z O 5 O c La ts Sn cb oO e3 O pa gt LLI O 3 c c O
22. e name cat egory grid resolution time aggregation etc Once all preferences are chosen in the filter ing box it is possible to click on the Refresh button and all search results are shown After selecting a single layer the entire row will be highlighted it takes to click on the Add button in order to add it to the process list There is no restriction on the number of layers but it is im portant to notice that selection is limited to data sharing a common period and the same time aggregation day month etc Looking into the catalog form each data unit is characterised by a unique set of features and is representable as a grid of geo referenced points either regular or not evolving on a discrete period any number of time steps They can be easily viewed on ArcGIS as layers Back in Overhaul amp Compare form it is possible to choose a space domain or point from a list of reference areas mostly countries and continents with a more specific array for Italy imported from GIS shape files and a time period with season filter if desired Then a tab arranged menu explains which op eration could be run usually the output is ei Clime climate data processing in GIS environment one of the following processes listed in Table 1 Login and Settings Time Series this process displays the M localhost E 156 14 147 2 New Database Remove Database Log in A m Setting
23. e based on the comparison between the series under study candidate and a number of reference series These last ones are a representative series of the climate of the region in which the candidate gauge station is located and at the same time without non homogeneities The process can be invoked by selecting Homogenise function from Clime tool bar Figure b the panel shown in Figure 26 will show up 7 1 DATA INPUT amp OUTLIER RESEARCH This functionality is used on station data in order to check the presence of abnormal val ues among observations outliers whose high number may affect predictions and alter their Statistic distributions thus leading to a faulty estimation In the upper part of the panel it is possible to search the requested layer by using searching criteria based on Grid Type Reso lution Time Aggregation Field and clicking on Refresh button Then the selected layer is visu alised as an ArcGIS layer through the Add layer button The test is carried out on a single sta tion point selected with Get a Point button and elected as candidate The exact position which is needed for the process execution is obtained by clicking on the Set nearest button The ref erence stations are determined according with Centro Euro Mediterraneo sui Cambiamenti Climatici O e O c z amp 2 z O O 2 bm Oo O Be gt LLI O 2 ca O
24. erence maps evaluated within one or more distinct sea sons Each input dataset is filtered by seasons before performing a time mean over the selected period on every point of the grid If Enable layer correction op tion is enabled an additional layer SIN GLE MAP only is taken into account as layer correction map mapzic and there fore added to the test map in order to have map map mapzc Itis useful for example in order to perform a temper ature elevation correction related to the orography Extreme indices this section Figure 16 provides some basic tools to calculate ex treme indices by aggregating input data lt is either possible to select indices from a default set taken from ETCCDIllist as dis played in Table B or to define a custom version Custom index By selecting this option another form appears Figure 17 As for Indices from Plot menu the user can choose aggregation operator for every time base month year total or the single months to analyse through Month selection Index name will be the same of the field in output layer Clime climate data processing in GIS environment Table 3 List of ETCCDI extreme indices provided by Clime webpage http etccdi pacificclimate org list_27_indices shtml For periods longer than one year output map displays annual mean of index value Index name Frost Days FD Ice Days ID Summer Days SU Tropical Nights TR Hot Waves HW Si
25. et of functions in Plot Er rors see par 5 1 2 except for they are evaluated on temporal means and thus layered on point grids Since these opera tions require a point to point comparison only objects with similar grids are allowed to this process If a perfect match is not reached it is possible to set a tolerance level lon lat round though it is always recommended to have grids of the same resolution Each output layer includes all indices related to a single input dataset map compared with maprer if desired layer correction is applicable Seasonal Differences Extreme Indices Percentiles Trend Tests Differences Errors Grid compare perfect match C lon lat round to E subdegree digits J Enable layer correction Add layer s Figure 21 Errors menu for point to point layer comparing Figure 22 shows typical output maps rendered through ArcMap interface and saved as image files 6 BIAS CORRECTION Since modeled data may present unacceptable bias values for impact studies it is required to carry out further controls and improve the re liability of predicted values For this purpose a Bias Correction process generally involves a comparison between model output and an ob servational dataset in order to evaluate the bias rate and estimate correction parameters to be applied on the whole modelled stream More specifically Clime allows the user to run such process on any test layer from
26. f the interval e g values xo and yo related to the same step are mapped as point P zo yo Pearson coefficient correlation can be evaluated it is equal to 1 if the two datasets are the same as well as covariance factor It is possible to plot a least squares line indicating the overall relation between the two input datasets Usual inputs are cou ples of arrays like a modelled series and an observational dataset Correlation same as Time Series but used for plotting heterogeneous data with different measure units e g temperature and rainfall For a better comparison Pearson coefficient evaluation is available in this case Verification Measures only for daily rain fall this process is conceived to compare modeled data with the relative observed dataset Proceeding with a dichotomous yes no prediction where each value is compared in order to verify if it is equal or greater than a determined threshold respectively 1 2 5 10 mm day the mod eled values and the corresponding obser vations are represented in a contingency table with the following responses hit the event is both observed and predicted by the model false positive the event is pre dicted by the model but not observed missed the event is observed but not pre dicted by the model correct negatives the event is not observed and not predicted by the model Then key quality measures in this system are defined as PC Proportion correc
27. h provides an idea about the overall behavior of the data con sidered Plus choosing the Running Mean option the time series will be rep resented on annual scale each value av eraged on a selectable window of adja cent values only odd integer ranges al lowed For example choosing a window with range 3 will produce for the year yo the mean over the period yo 1 yo 1 3 steps Finally in order to assess the presence of a real trend in a dataset within test period a Mann Kendall signifi cance test could be performed The alpha threshold parameter is set to 0 05 by de fault meaning that the normalised trend rate must reside in the 5 tail of standard cumulative distribution function CDF in order to reject null hypothesis and have H 1 trend presence It is important to notice that such test has relevance as time step of every processed dataset is kept constant Seasonal Cycles this function produces a 12 steps plot synthesizing values re lated to every distinct month More Clearly the first value is the mean col lected over all the ones belonging to Jan uarys and so on Also here it is possible to evaluate extremes and standard devia O O z c eb z amp 2 z O 5 O c La ts Sn cb oO e3 O pa gt LLI O 3 c c O CMCC Research Papers Catalogs xi dimension type_perio lt name 2 type_lyr resolution type_field na
28. igure can be changed through the Change Pars button Figure 28 The current nominal level of confidence p lev must be chosen among the following values 0 75 0 80 0 90 0 95 0 99 0 9999 In the case of daily precipitation data it is also possible to set the lower precip itation threshold to be considered in the pro cess pthr Finally Transform Data converts daily data series in RClimDex standard format to monthly mean series in RHtestsV4 standard format 8 OTHER FUNCTIONS Despite most features have been described in previous sections there are also other function alities which interact with ArcGIS objects layers rasters and can be directly executed through Clime toolbar buttons Standard Interpola tion function Figure 29 basically reproduces Interpolation toolboxes provided by ArcGIS with Please enter the Missing Value Code Please enter the nominal conf level p ley value Please enter integer Iadj 0 to 10000 inclusive Please enter integer Mq of points for evaluating PDF Please enter integer Ny4a gt 5 or 0 for choosing the whole segment Please enter the lower precipitation threshold pthr gt 0 Figure 28 Change Pars menu pthr value is available only for daily precipitation data out running them from Catalog window More precisely available processes are IDW Natu ral Neighbour Spline Trend and Kriging Since the early development stage of this section there is still a limited
29. ing from existing ones by rearranging its con tent into a longer period through a set of ag gregation functions max min mean standard deviation sum Figure 8 For instance a ta ble of monthly means could be obtained from a dataset of daily values Season aggregation consists of four parts of year each one com posed of three months DJF MAM JJA and SON if True season option is enabled De cember data is taken from the year preceding January and February of the same block DJF If input data contain monthly means cumula tive values can be evaluated MonMean gt MonSum It is worth to point that Clime is capable of aggregation during processes but dealing with tinier tables helps users to save a significant amount of time Import Aggregate Export Aggregate To Periods MONTH M True season BJF shitt Layer Name test_month Select aggregation s M Mean T Min Max F StdDev l Sum Aggregate MonMean gt MonSum Figure 3 Aggregate section Clime Data Manager 3 3 EXPORT This functionality allows user to export a table from any database to a local device as a CSV file Figure 4 Once an object Table name is selected it is possible to choose the fields to include into the output file it is possible to select all fields by clicking on the button Select all and then to click on the button Export Data is always arranged by date and position even if these fields are not exported 4 HO
30. k to be carried out and its pro cessed data in order to historicize the associa tions between methods and validations as well as suitably mark the import data as a function of the source specific data The development lan guage used is a subset of C NET compatible with the MONO framework whereas the parts concerning the primitive GIS are native libraries that are accessed through the ArcGIS API Ar cObjects and ArcToolbox The chosen approach has noticeable advantages the C language independently manages the dynamic allocation of memory and calculations with heavier com putational load are executed by native mod ules which grants faster performances The execution speed is a key requirement for this type of product the validation phase may im ply the production of a large amount of data small scale forecast models with dense tem Clime climate data processing in GIS environment poral sampling time series of weather stations radars or satellites expected to be processed with multiple methods of downscaling then the production phase requires to provide forecasts in nearly real time to fit cases like emergency management At the present moment Clime re quires the support of ESRI ArcGIS Desktop 10 X an ArcView license Is sufficient with Spatial An alyst extension but its structure makes it open to further solutions In fact another advan tage of the choice of C language MONO is the complete portability of the cod
31. king on the gray square in the leftmost column in this way the entire row will be highlighted 3 1 IMPORT In order to process data in Clime it is neces sary to import the requested data into dedicate database clusters Figure 2 Hence the very first step consists on identifying the original for mat of the input data and properly converting it into a standard one currently managed data are in Network Common Data Form NetCDF or Comma Separated Values CSV format and are represented as discrete functions of space and time There are no particular constraints on the shape of the physical domains moreover val ues could either be distributed on a regular and time invariant grid usually for model data or be spread on an erratic cluster of points station data with setting nominal resolution as the av erage distance in kilometres between adjacent points 00 for irregular grids concerning the temporal evolution it only takes to determine a start date and a nominal step e g hour day month between adjacent time units Files con taining data on a single time step single maps and one point datasets are also allowed In or der to properly run this process overall struc ture of input files must be arranged as follows NetCDF Files are required to have the following fields longitude and latitude one or two dimensional 180 to 180 time one dimensional vector with in teger values must be named
32. l Dump Path Note 4 4156 14 147 2 repository dumpdbais dati repository dumpdbgis in is Import O O Cc eb amp 2 z O 5 4 O i gt a So k O 2 gt LLI O 2 c c O Figure 1 Clime Data Manager form All functionalities are directly called through this menu Delimiter id_stazione lon lat lt any number fields gt time_ shape idx idy itime lon lat lt any number fields gt time_ shape shape SRID 3857 POINT lt Mercator Sphere coordinates separated by space gt ex SRID 3857 POINT 1647529 7733594 Examples Types id_stazione lon lat hsurf time_ shape id_stazione character string max 20 idx idy itime integer lon lat real idx idy itime lon lat tmin tmax time _ shape data field real NaN allowed time_ yyyy MM dd hh mmi ssZ leave Z at the end ex 1970 01 01 00 00 00Z The file NetCDF or CSV containing the data to be imported can be selected from the file sys CMCC Research Papers Import Aggregate Export Add a Type Layer Nominal Res Km Cosmo r 2 0 Select file D c CLOT_daily1 9791983 ne Layer Name E Suggest M Partitioned No for tiny data Local Dump Fath Note DO dumpdbais tem by clicking on the button Select file Then it is necessary to specify the following attributes Centro Euro Mediterraneo sui Cambiamenti Climatici Cc Layer name
33. layers to be analyzed by clicking the button In this way the main O O z eb amp 2 z O 5 O c La ts Sn oO e3 O pa gt LLI O 3 c c O O r O c z amp 2 z O 5 O ks ko h Oo c3 O 2 gt LLI O B oJ c O CMCC Research Papers Overhaul amp Compare Climate lv eobs_day19702012 WV cosmo_italy8km_era40_t2m_day19712000_lw iV cosmo_italy8km_cm_t2m_day19712000_lw Filters Domain On Point Type Continents v fi 971 01 01 00 00 00 C Africa Asia Australian Area C Europe C North America C South America Add Layers Add domain shape s Time domain Start Date Time End Date Time booo 12 30 00 00 00 z Season filter DJF EJJA C MAM SGN Plot Settings Reference Layer eobs_day1 9702012 Plot Elaborations General Errors Indices Time Series Plot M Min Max Aggregate by year N Bl f Standard dev Least Squares Correlation coefficient Pearson Covariance I Mann Kendall Test alpha 0 05 Running Mean Figure 7 Login and Settings form User can choose databases and output folders here window form Figure 8 will appear Through this window every piece of data stored into se lected databases is visible as layer in alist each column showing a different featur
34. limatici CMCC Research Papers Plots since it is meant to analyse annual trends all dataset are averaged by year before being processed Selecting Run Test data undergoes also spatial mean and results are displayed as synthetic in dices on a window at the end of process With Add result grid test is performed on every point of input layer in order to re turn a map of responses p values and hypothesis Seasonal Differences Extreme Indices Percentiles Trend Tests Differences Errors Select test Mann Kendall alpha 0 05 Run Test Add result grid Figure 19 Trend Tests menu Differences this process Figure 20 pro duces simple difference maps evaluated in a similar way as Seasonal Differences but in this case the user has to choose test layers from the list in this tab along with a new time period whereas the refer ence layer is unchanged Such process is often used to compare datasets focused on two different time periods e g future minus past Seasonal Differences Extreme Indices Percentiles Trend Tests Differences Errors EA Layers mj cosmo_italy8km_cem45_t2m_day19712100 W cosmo_italy8km_cm85_t2m_day200621 00 C Seasons Start Date Time E aeon Jens aia as I San EES bo71 01 01 00 00 00 7 d End Date Time gt gt Diff Rasters 100 12 30 00 00 00 z Figure 20 Trend Tests menu Errors this section Figure 21 consists on the same s
35. me_field cosmo_italy8km_cm_t2m_day19712000_lw COSMO 8 00 Temperature T_2M He cosmo_italy8km_cm_tminmax_day1 9712000 COSMO 8 00 Temperature TMIN_2M casmo_italy8km_cm_tminmax_day1 9712000 COSMO 8 00 Temperature TMAX 2M cosmo_italy8km_cm45_t2m_day19712100 COSMO 5 00 Temperature T_2M T 7c iG cosmo_italy8km_cm45_t2m_day20012070 cosmo_italy8km_cm45_t2mtprec_day20712100 COSMO Temperature Temperature T_2M cosmo_italy8km_cm45_tminmax_day19712100 COSMO Temperature TMIN_2M cosmo_italy8km_cm45_tminmax_day19712100 COSMO Temperature TMAx_2M cosmo_italy8km_cm45_tminmax_day2001 2100 COSMO Temperature TMAX_2M _cosmo_italy8km_cm45_tminmax_day2001 2100 COSMO Temperature TMIN_2M cosmo_italy8km_cm85_t2m_day20062070 COSMO Temperature T_2M cosmo_italy8km_cm85_t2m_day20062100 COSMO Temperature T_2M cosmo_italy8km_cm85_t2mtprec_day20712100 COSMO Temperature T_2M cosmo_italy8km_cm85_tminmax_day200621 00 COSMO Temperature TMAX_2M cosmo_italy8km_cm85_tminmax_day20062100 COSMO Temperature TMIN_2M cosmo_italy8km_cmalb_day20062100 COSMO Temperature TMIN_2M i TEPE Fy PRL EFP a TE a E RE TE C I GARE REE EAI GRO REE Cas CREI Filter by V Enter keyword fitaly Refresh Add Done Grid Type ALL x Time Aggregation Day
36. mple Daily Intensity Index SDID Number of Heavy Precipitation Days R10 Number of Very Heavy Precipitation Days R20 Consecutive Dry Days CDD Consecutive Wet Days CWD Annual Total Wet Day Precipitation Prep Tot Seasonal Differences Extreme Indices Percentiles Trend Tests Differences Errors Temp indices FD CiD Prec indices Spl C AiG C p2 Dp OWD Prep Tat Add index layers Custom index Figure 16 Extreme Indices menu Available indices vary depending on selected input data temperature or precipitation Index name custom _index Month selection Select all DJF Mam M jan M apr IV feb M may Deselect all JJA SON mole ee Monthly base Total period Add index layer s Figure 17 Custom Index menu Percentiles this section Figure allows the calculation of different per centiles which can be selected by choos ing one or more threshold values Percentiles over the selected period are evaluated in every point in order to build percentile maps A single input object produces one layer for each chosen per centile By setting upper lower bound Definition Number of Days Year with Tmin lt 0 C Number of Days Year with Tmax lt 0 C Number of Days Year with Tmax gt 25 C Number of Days Year with Tmin gt 20 C Number of Days Year with Tmax gt 35 C Daily Precipitation Mean during Wet Days prec gt 1mm Number of
37. nt values are m Class B data measured with medium ac considered as a suspicious behaviour If data curacy instrumentation 3 5 e g me are sufficient the quality control is performed chanical recorder rain gauge ji His Ih ee UL Se IE Data class A 1 B 3 4 C 1 2 D 0 Table 5 Correspondence between quality rate and iQuaSl Quality iQuaSI HIGH 09 Quasi 1 GOOD 0 7 lt iQuaSI lt 0 9 SUFFICIENT C3 Qual 0T POOR O lt Quasi lt 03 BAD UNUSABLE O lt 7Ouasl lt 0 3 m Class C data measured with low accu racy instrumentation gt 5 or estimated through indirect variables e g simple rain gauge meteo radar for precipitation flow rate estimated through discharge scale m Class D missing data or reconstructed by mathematical modeling If metadata are not available an average value between quality coefficients bz and cz is at tributed by default JQuaSi index ranges be tween 0 and 1 and provides information about the series quality according to the five intervals shown in table 6 Only datasets with overall quality SUFFICIENT or above are considered valid reference stations Such control is carried out starting from the closest reference point until it collects a suf ficient number defined by the user of stations fulfilling this criteria if a minimum number is not reached the process is aborted Once all reference stations are gathered the candidate Clime climate data processing in GIS environmen
38. olution regional climate simulations in the Alpine space and analysis of extreme events J Geophys Res Atmos 119 3222 3237 2014 B Rockel A Wil and A Hense The regional climate model COSMO CLM CCLM Meteorol Z 17 4 347 348 2008 M Sajad R Majid G Ali E Abazar E Hasan M Maryam and M Yadollah Determination of A Some Simple Meth ods for Outlier Detection in Maximum Daily Rainfall Case Study Baliglichay Water shed Basin Ardebil Province lran Bull Env Pharmacol Life Sci 3 110 117 February 2014 K E Taylor Summarizing multiple aspects of model performance in a single diagram Journal of Geophysical Research pages 1 18 2000 C Teutschbein and J Seibert Bias cor rection of regional climate model simula tions for hydrological climate change im pact studies Review and evaluation of different methods Journal of Hydrology 456 457 12 29 2012 X L Wang Accounting for autocorrelation in detecting mean shifts in climate data se ries using the penalized maximal t or F test J Appl Meteor Climatol 47 2423 2444 2008 X L Wang Penalized maximal F test for detecting undocumented meanshifts with out trend change J Atmos Oceanic Tech 25 3 368 384 2008 X L Wang Q H Wen and Y Wu Penal ized maximal t test for detecting undocu mented mean change in climate data se ries J Appl Meteor Climatol 46 6 916 931 2007 Clime clima
39. operability after user de fines input data e g 12m tot_prec Clime picks the first point feature of Table Of Contents hav ing this field and interpolates it with selected algorithm As a future improvement it would be possible to select more grids in order to run multiple interpolations at once Import Export IDW Natural Neighbor RBF Spline Trend Polynomial Kriging Power Cell Size meters Insert input data Run Figure 29 Standard Interpolation form tool Figure 30 is conceived to handle output raster maps both converting them into IMOD RHtests 4 Clime climate data processing in GIS environment Transform Data PMT and t tests FindU wRef PMF and F tests FindU To adjust daily Gaussian data RHtests V4 Change Pars FindUD wRef StepSize wRef FindUD StepSize QMadj wRef Current Missing Yalue Code Current nominal level of confidence p lev Segment to which to adjust the series Iadj Current Mq of points for evaluating PDF Current Ny4a max of years of data For estimating PDF Current input Base series filename Current input Reference series filename Current data directory Current output directory Figure 27 RH Test interface with all parameters listed below format which is frequently used for impact stud ies Export or creating new objects from SAT Matlab files xls Import Ce EN RR RR CE JOM A A
40. process generates the following files E stationList txt list of reference stations position and table E yyyy MM dd_hh mm_Outliers txt O O z c eb z amp 2 z O 5 O c La ts Sn cb oO e3 O pa gt LLI O 3 c c eb O Centro Euro Mediterraneo sui Cambiamenti Climatici CMCC Research Papers list of outliers obtained with all the three methods E yyyy MM dd_hh mm_Table txt completeness results E yyyy MM dd_hh mm_BaseS csv candidate data list E yyyy MM dd_hh mm_RefS csv correlation weighted average of refer ence stations if the number of reference Stations is lesser than three it will not be created Important in order to correctly get the station coordinates only the selected grid must be vis ible on ArcGIS interface so its related layer in Table Of Contents should be left checked 7 2 BREAKPOINT TEST Afterwards it is possible to run the changepoint test through the RH Test V4 software package interface it automatically appears at the end of the process which enables the user to edit the various parameters used in the analysis and apply a corrective algorithm to the data series 17 if input data is a daily rainfall a slightly different version is executed RHtests_dlyPrep For a more detailed doc umentation on this software package a com plete guide is available 18 If needed the parameters displayed in the main form F
41. r differences can be evaluated in two different ways specified in Difference Representa tion box simple difference map maprrr and percentage map maprer MmaprReF 100 Options Create Raster by selecting this option every grid point is turned into a raster through an interpolation process Natural Neighbour provided by ArcMap system toolboxes Algebra operations can be performed only between rasters so this step always takes place before any sub traction to execute in case of differences Since the raster basic unit is a square its size is defined by input parameter Cell Size m Contour this option allows executing Contour With Barriers toolbox creating a feature layer with contour lines following values of output map The user may choose contour interval The complete list of functions is showed below Plots Elaborations Seasonal Differences Extreme Indices Percentiles Trend Tests Differences Errors fae Seasons lV DJF T MAM L WA T SON T All m Data Interpolator Natural Neighbour C Spline EEIE gt gt Seasonal diff Rasters NOTE differences will be evaluated by subtracting Reference Layer to all other layers I Enable layer correction other la M Difference representation e Simple difference Percentage Figure 15 Elaborations section with Seasonal Differences menu Seasonal Differences this process Fig ure 15 produces simple diff
42. s Primary Mode C Meteo Climate Select Project c temp contour bd Add Grid from Cosmo LM NetCDF file with interval 25 Wean Batch Jobs CMCC Clime Figure 6 Login and Settings form User can choose databases and output folders here ther a plot chart or a map represented as a layer raster object on the GIS All the options are briefly explained in the following sections 5 1 PLOTS In this tab lower part of the form show in Fig ure 7 it is possible to run any data analysis displayable as line plots mostly on a temporal scale Normally input data are averaged over the selected domain before undergoing further operations The graphic output is fully cus tomizable by choosing colour line width pen style marker shape and label name for the leg end through the Plot Settings window Once data are plotted the user can customize it by setting axes labels and legend and save it as an image or Excel file 5 1 1 GENERAL This section display a set of functions mostly used to analyse temporal trend of selected data Acombo box menu allows user to choose data evolution through the reference pe riod with a time scale determined by the user e g days months years so that every plotted point is evaluated as tempo ral mean over the given time unit Option ally extreme values max min and stan dard deviations are displayable along with the trend line obtained with least square method whic
43. ss may require a parameter to set the step of quantile probability vector Q Step SO user can choose it to have a default inverse of V length or custom value between 0 and 1 Clime climate data processing in GIS environment N points per square side 5 must be odd Reference Station Model Grid Nearest Nb lt lt Set Nearest gt gt Figure 24 Coordinates selection using Get Point operator Linear Sealing Quartile Mapping Select process Hl Step DIST bernE xp default Start C custom Figure 25 Quantile Mapping process menu 7 HOMOGENISE The study of climate variability and the eval uation of climate tendencies require the avail ability of long homogeneous series of climatic data A time series is homogenous if varia tions can be attributed only to climate factors 1 Real data series are usually affected by perturbations or non homogeneities due to external non climatic factors The time step in which a series starts to exhibit a perturba tion is usually defined as breakpoint or change point The information availability metadata that supports the history of gauge stations sim plifies the study of non homogeneities in a time series so in order to identify not documented non homogeneities and to correct their effect on the series several methodologies have been developed mainly statistical homogenisation methods Most of the widely used statistical methodologies ar
44. t Table 4 Table of coefficients for quality control Series lenght 3 4 1 2 1 4 0 lt 5 1 2 0 1 4 0 0 0 0 0 is scanned by means of three different algo rithms multi station quartile and mean sd in order to identify any possible outlier The multi station process is the only one using reference stations whereas others just focus on candi date It is carried out only in case that three or more reference stations are found otherwise it is skipped Each value from the candidate time series is compared to all the related values be longing to the reference series and is marked as outlier if itis too from them In this case a standardised value is evaluated and displayed in the output text file Such analysis is car ried out separately for all the four seasons but results are listed together 7 In the quartile method the 25 and 75 percentiles respec tively 1 and 3 quartiles are evaluated for every season in order to define outliers which can be either moderate or extreme depending on their exceeding amount from such values 8 The Mean sd algorithm is so called because it elects as outliers any value exceeding seasonal mean by more than 3 times its standard devia tion sd 12 Finally all results from the previ ous processes are then compared and outliers found by all the methods common outliers are written into the output file each represented as a day by day list in case of daily data The
45. t js Fr by C month year Plot Figure 10 All statistic operators evaluated by Errors with their implementation N eG Ge D Ao fine Oy fF treme values and occurrence rate Example starting from a dataset of daily rainfall data the requested output is the total number of days per year with a pre cipitation amount exceeding 1 mm day From Monthly base box select count and set value constraints to gt l From Yearly base box select sum Aggregate by year and plot 5 1 4 PLOT FORM At the end of process Clime displays a form for every selected area each one displaying re sults through a chart as shown in Figure 11 lt is possible to regulate the scale interval and the tick size of both the axes The label format can be properly customized with Label Settings menu legend could be moved or even hid den Finally plot image can be exported as Centro Euro Mediterraneo sui Cambiamenti Climatici CMCC Research Papers image file bmp or excel table file x sx Some examples of output images are shown below in Figures Centro Euro Mediterraneo sui Cambiamenti Climatici C COSMO Temperature data Clime climate data processing in GIS environment GI Time Series AdB Campania xi A i Label Setti Time series __tsbel Settings Axes Settings Axis min Arial x max l I Bold Size 75 Y Axis min fio arial 7 max fiz I Bold Size 75
46. t measures the model accuracy by consider ing the simple matching coefficient based on the proportion of total correct hits and rejections BIAS percentage of events mod eled to those observed and should be unity unbiased for a perfect sys tem In practice it generally dif fers from unity due to the presence of systematic biases errors in the model or observing system From a climatological point of view bias is defined as the systematic difference Clime climate data processing in GIS environment between the observed data and sim ulated results Such score is com puted in the Elaboration Form POD Probability of Detection per centage of observed events cor rectly modeled nl FAR False Alarm Ratio percent age of events predicted by the model and that do not verify CSI Critical Success Index per centage of observed and or mod eled values that were correctly pre dicted For each index a chart is plotted with values corresponding to distinct thresholds 5 1 2 ERRORS In this section Figure 9 Clime analyses the differences between a set of objects and a Ref erence Layer and displays the results as error indices listed in Table 2 By checking Process all seasons it is possible to perform such pro cess for every single season total 4 runs There are two main modes to run processes m Calculate Indices selecting this option software first evaluates spatial me
47. te data processing in GIS environment 18 Xiaolan L Wang and Yang Feng RHt ogy Directorate Science and Technology estsV4 User Manual Climate Research Di Branch Environment Canada Toronto vision Atmospheric Science and Technol Ontario July 2013 O E O z c amp 2 z O 5 O ao ho Bes one xe O im gt LLI O ho tad Cc cb O Centro Euro Mediterraneo sui Cambiamenti Climatici 2015 Visit www cmcc it for information on our activities and publications The Euro Mediteranean Centre on Climate Change is a Ltd Company with its registered office and administration in Lecce and local units in Bologna Venice Capua Sassari Viterbo Benevento and Milan The society doesn t pursue profitable ends and aims to realize and manage the Centre its promotion and research coordination and different scientific and applied activities in the field of climate change study cmcc Centro Euro Mediterraneo sui Cambiamenti Climatici
48. the database and create a bias corrected new one itis saved into the database where station data is stored The process can be started by clicking on the days year HE 150 170 Clime climate data processing in GIS environment a 3 zug lt Ti 6 y o o bee angat 26 60 Sf a ES S ees amp amp ae wv 2 amp 288s s88q eo t 3 Om bane Qa cI lt S ee ae eT S LN Oo eB Oo g Es NQ b S E a b amp E E E E C mm E o E G m 5 Ae bs bw A ih Figure 22 Examples of indices maps a number of weak precipitation days d yr provided by EURO4M APGD data bias of weak precipitation days of bo COSMO CLM0 0715 and c COSMO CLM 0 125 versus EURO4M APGD data d number of intense precipitation days d yr provided by EURO4M APGD data bias of intense precipitation days of e COSMO CLM 0 0715 and f COSMO CLM 0 125 versus EURO4M APGD data 10 button Bias Correction tool in the multiple but tons bar shown in Figure B The panel shown in Figure 23 will appear The general process consists in comparing Model Grid Control and Observation Grid within Control Time period over the selected domain in order to create a correc tion mask which is applied to Projected Grid within the Projection Time interval and eval uate a corrected grid whose values are saved into an Output Table First it is necessary to select the
49. tter option requires all layers to have the same grid Characterisation Enable layer correction a correction layer SINGLE MAP only is added to all elements except for reference layer be fore statistical processing 5 1 3 INDICES The following section deals with the index evalu ation Figure 10 Rather than displaying a sim Correlation No No No No Yes Yes No Covariance Run Mean Mann Kendall No Yes Yes No No No No No No No No No Yes No No No Yes No No No No General Errors indices This application evaluates the following indices mean bias mean absolute error MAE foot mean square error RMSE centered root mean square CAMS correlation and covariance All factors are evaluated by comparing selected layers with Reference Layer Taylor Diagram radius is normalised by reference standard deviation Calculate Indices Draw Taylor Diagram Process all seasons Taylor Diagram Circle radius fi 5 IV Draw legend Characterisation p Circle shape temporal diagram First quadrant 90 T Enable layer correction Upper half 180 C spatial diagram Figure 9 Errors menu ple trend it allows to aggregate data on monthly or an nual scale using a set of operators and assuming conditions for every temporal tier All aggregate functions are listed below Mean Min Max aver age minimum maximum value within the base period Sum sum of v
50. ure 26 Homogenise form including both catalog view and process options 4 Set Nearest gt gt the following parameters The result is expressed through iQuaS coeffi cient evaluated with the following formula Maximum distance from the candidate L L L station iQuaSI ar x by x a Cn amp L b L m Maximum and minimum number of refer d X Ce C x ee ence stations to be determined 5 m Minimum correlation value with respect to where coefficient values az br cr dz are de the candidate station fined by data class and series length L as listed in Errore Lorigine riferimento non stata trovata and ratios L L are relative to the per centage of metadata belonging to a single qual ity class i over total period L n of years Each quality class is defined as a function of the avail able metadata m Outlier percentile for multi station method 95 or above recommended especially for daily data Each reference point series must first pass a completeness test in order to ensure it con tains a satisfying percentage of valid data at least 75 with results shown in the file Ta ble txt 1 indicates a positive response 0 a negative response In the case of rainfall data m Class A data measured with high accu racy instrumentation lt 3 e g elec tronic recorder rain gauge in perfect effi ciency negatives are corrected to null and 4 days or longer streaks of non zero consta
Download Pdf Manuals
Related Search