ࡱ> ikh5@=Lbjbj22 !PXX~ ~ ~ ~ ~ ~ ~  :%:%:%8r%l%L <6&L&b&b&b&='='=';;;;;;;$=R?;Q~ )='='));~ ~ b&b&O<,,,)~ b&~ b&;,);,0,-V4@~ ~ L5b&*& :%*p4 \8|e<0<4xsA +fsAL5 ~ ~ ~ ~ sA~ L5='h'J,'<+(Y='='=';;  r,d Geostatistical mapping of the air temperature and temperature related climate elements for the period 1961-1990 in Croatia Melita Per ec Tadi Meteorological and hydrological service of Croatia +385 1 45 65638 +385 1 45 65630  HYPERLINK "mailto:melita.percec@cirus.dhz.hr" melita@cirus.dhz.hr In climatology, one of the most important information about the climate of a place or a region is the information about the CLImate NOrmals (CLINO), the average values of the meteorological elements for the 30-year period. For the CLINO period 1931-1960, the handmade maps were published for selected parameters in Climate Atlas of Croatia. During last year, the preparation of the Croatian climate atlas for the 1961-1990 period started, in which the 24 maps concerning temperature, precipitation, humidity, solar irradiation, sunshine duration, cloudiness and biometeorological phenomena are going to be presented. Here, the data exploration, methodology and final maps of temperature and related parameters are presented on the examples of January and July monthly mean temperature (T_Jan, T_Jul), mean annual temperature (T_ann) and mean annual number of cold (tmin<0C), warm (tmax25C) and days with warm nights (tmin20C). According to climatological stations density (152 meteorological stations) and cartographic rules, the finest possible map resolution of 1000 m is selected for the mapping area of approximately 56,000 km2. Frequency distribution of weather stations altitudes and altitudes of the digital elevation model (DEM) is comparable, but still some places at higher elevation or less populated areas are underrepresented. For approximately 1/3 of the stations, coordinates that are stated in the meta-database in deg and min had to be re-evaluated, mostly because of the two reasons: when overlaid to 100 m resolution DEM some of the stations fell into the sea, and some had the difference in height stated in the meta-database and DEM larger then 200 m. Distributions of monthly and annual averages are in general bimodal, with almost clear differentiation between continental and maritime part of the country. In some cases data is also highly skewed or come from non-normal distributions. For the mapping of climatological fields the regression kriging (RK) has been applied. Regression-kriging is a spatial interpolation technique that combines a regression of the dependent variable on predictor variables with simple kriging of the regression residuals. The predictors in the regression model have been selected according to the stepwise procedure among DEM, weighted distance to the sea, longitude and latitude or factor components (FC) derived from those. Besides the elevation, weighted distance to the sea, (with weights proportional to altitudes that introduce a resistance to distance calculation) proved to be important predictor in Croatia where large mountains are situated in the vicinity of the sea, preventing the maritime influence to spread to the continent. The shape of the country is not very favourable considering the spatial distribution of the stations but this negative effect can be reduced by considering the data from the neighbouring countries (85 additional stations). Model using only Croatian data is compared with the one using also the neighbours. The improvements are visible near the borders, but are not significant according to performance statistics. Different statistical measures have been applied for assessing the quality of the maps: adjusted coefficient of determination for multiple regression and cross-validation statistics and prediction error variance maps for regression kriging. Finally, the important part of the evaluation is visual inspection of the maps by experienced climatologist. Table 1: Summary table with descriptive statistics of climate variables and cross-validation statistics. Statistics of the climate variables used for spatial modelling and mapping: number of stations N, average Avg, minimum Min, maximum Max, variance Var, standard deviation StD, skewness Skew and kurtosis Kurt. R a2 of the regression models and cross-validation statistics of the applied regression kriging models: mean prediction error ME, root mean square prediction error RMSE, normalized RMSE RMSEr. DESCRIPTIVE STATISTICSMODEL VALIDATIONNMeanMinMaxVarStDSkewKurtRa2MERMSERMSErT_Jan (C)1521.8-4.79.715.13.90.5-1.30.98-0.010.740.21T_Jul (C)15221.012.225.48.22.9-0.70.30.93-0.020.640.24T_an$&Zbd|    ) * , w x y ׽׫׏thYhWhHWzCJaJmH sH hXCJaJmH sH hWh^GCJaJmH sH h^GCJaJmH sH hWhp>CJaJmH sH hd h 0J,mH sH #jhd h UmH sH jhd h UmH sH hd hq>mH sH hd h mH sH hd h7mH sH hDmH sH hd h9mH sH  + , \,$$Ifa$gd l-.`gd#o'gdldegdW_"gd.G"gdCgd gd gd gdq>)gd gd gdH#LCJaJmH sH hWh`CJaJmH sH  y    # , 1 ĵ񵦵ėymy^yRy^yFh8jQCJaJmH sH hDCJaJmH sH hWh+o4CJaJmH sH hCJaJmH sH hWh eCJaJmH sH hWh9CJaJmH sH hWhXfCJaJmH sH hWhRCJaJmH sH hWh hCJaJmH sH hWh XCJaJmH sH hWhhCJaJmH sH hWh# _CJaJmH sH hWh2CJaJmH sH 1 < K L R Y e m  @DFLNVXZ\ 辮辮辮ٌ}n}bhCJaJmH sH hWhrCJaJmH sH hWh]CJaJmH sH hWh^CJaJmH sH $h.Gh.GCJOJQJaJmH sH h.Gh.GCJH*aJmH sH h.Gh.GCJaJmH sH hDCJaJmH sH hWh eCJaJmH sH h.GCJaJmH sH hnZCJaJmH sH ! $6HJRxz'05F_bcÓxxl]Q]lhcCJaJmH sH hWhxlCJaJmH sH hZCJaJmH sH hjLCJaJmH sH hWh%CJaJmH sH "hWh]6CJH*aJmH sH hWh]CJaJmH sH hWhrCJaJmH sH hWh]6CJaJmH sH hWh CJaJmH sH hWh]CJaJmH sH hWhCJaJmH sH &*.12<>?HIǵǵ⩝ufWHuhWhLCJaJmH sH hWh~CJaJmH sH hWhnDCJaJmH sH hE@yh]6CJaJmH sH hE@yCJaJmH sH h MCJaJmH sH hVhCJaJmH sH hVoeCJaJmH sH "hWh 6CJ]aJmH sH hWh CJaJmH sH hSCJaJmH sH hWh]CJaJmH sH hWhxlCJaJmH sH &'GJstyӵөӵӚӋ|m^O^hWh_CJaJmH sH hWh&*CJaJmH sH hWh MCJaJmH sH hWhGQCJaJmH sH hWhCJaJmH sH hWhiCJaJmH sH h MCJaJmH sH hWhmCJaJmH sH hWh YCJaJmH sH hWh9"CJaJmH sH hWhCpCJaJmH sH hWh CJaJmH sH 5?_*5;[_rwx03<BqĸĩĎĎsdXXLXLhiCJaJmH sH ha^)CJaJmH sH hWh9SCJaJmH sH hWhGMCJaJmH sH hCJaJmH sH hWh%CJaJmH sH hCCJaJmH sH hChCCJaJmH sH h!CJaJmH sH hWhCpCJaJmH sH hWh9"CJaJmH sH hWh&*CJaJmH sH hWhk3CJaJmH sH qrAHIKĵyjj[L[j=hWhA"CJaJmH sH hWhd CJaJmH sH hWh)CJaJmH sH hWhW=CJaJmH sH hWh&WCJaJmH sH hWhCJaJmH sH hWh-CJaJmH sH hWh9"CJaJmH sH hWhF8CJaJmH sH hWh}-CJaJmH sH hWhlCJaJmH sH hWhCpCJaJmH sH hWh4CJaJmH sH 78TYep┅yj[j[jODhW_hW_mH sH h1CJaJmH sH hWh CJaJmH sH hWhk*CJaJmH sH hdECJaJmH sH hWh5CCJaJmH sH hWhe CJaJmH sH hWhwCJaJmH sH hWh8*CJaJmH sH "hWhCp5CJ\aJmH sH hWhoCJaJmH sH hWhCpCJaJmH sH hWhCJaJmH sH "OVej>BSUVno=>?`abcdrºu[Au2hDh 5B*CJH*OJQJ\^JaJph2hDh 5B*CJH*OJQJ\^JaJph/hDh 5B*CJOJQJ\^JaJphh mH sH /hDh 5B*CJOJQJ\^JaJphhh CJmH sH h^mH sH hPc hldemH sH hxuhldeH*mH sH hxuhldeH*mH sH hldemH sH hxuhldemH sH hmH sH ,=>?A*$$If]^a$gd l-$Ifgd l-kd$$IflF-^ t06    44 la$$Ifa$gd l-.`AFJNRV[`dglrs~$$Ifa$gd l-\$Ifgd l-\FfK$$If]^a$gd l-rs~@ @@V@^@@@@@@AAA$AraYBB,hDh B*CJH*OJQJ^JaJphh rnmH sH  hhh rnCJOJQJ^JaJh rnCJOJQJ^JaJ hh rnCJOJQJ^JaJ)hDh rnB*CJOJQJ^JaJphU#h rnB*CJOJQJ^JaJph/hDh 5B*CJOJQJ\^JaJph hDh CJOJQJ^JaJ)hDh B*CJOJQJ^JaJphh mH sH $$Ifa$gd l-$Ifgd l-Ff$$If^a$gd l-\$$Ifa$gd l-\@@ @(@2@:@B@L@V@`@l@v@$$Ifa$gdaQl-\$$Ifa$gdaQl-$Ifgd l-FfY $$Ifa$gd l-$$If^a$gd l-n (C)15211.53.516.38.72.9-0.2-0.50.92-0.020.620.23tmin<0C (d)13967.91.0159.01799.842.4-0.1-1.10.90-0.3411.080.26tmax25C (d)13976.40.2119.9766.527.7-1.10.90.880.257.710.31tmin20C (d)13911.00.072.0286.916v@@@@@@@@@@@@@AAAAFfm$$If^a$gd l-$$Ifa$gd l-$Ifgd l-Ff $$Ifa$gdaQl-\$A&A8AAAAAAABL!L"L#L$L&L'L)L*L,L-L/L0L6L7L8L9L;L*B*ph@ ^ Table Grid7:V-0-dh5$7$8$9DH$44 D Table Grid 2:V.0jjj#j .dh5$7$8$9DH$,5\5\5\5\P{+,m \   $(-16:>CHNSXYdhmrw{ $*/0>BGKQW\aejotyz00)00000p"0"0"0"0"0"0"0"00'0008 08 08 0< 08 08 08 08 08 08 08 08 08 08 08 08 08 0< 08 08 08 08 08 08 08 08 08 08 08 08 08 0< 08 08 08 08 08 08 08 08 08 08 08 08 08 0< 08 08 08 08 08 08 08 08 08 08 08 08 08 0< 08 08 08 08 08 08 08 08 08 08 08 08 08 0< 08 08 08 08 08 08 08 08 08 08 08 08 08 0< 08 08 08 08 08 08 08 08 08 08 08 08 08 0< 0@0@0@0@0@0@00` $(-16:>CNXYdhmrw{ $/>BGKQW\aeoyOy00wOy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00@0 @0 @0 @0 Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00@0 @0 Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy030Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy0x1Oy0x1 Oy0 0Oy0 0Oy0 0Oy0 0Oy0 0Oy0 0Oy0 0Oy0 0Oy0 0Oy0 0Oy0 0Oy0U0Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy0 0Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00 Oy00 Oy00 Oy00 0w  1 qr$A=L",Av@AA:L=L!#$'CEJfkY^*#3,,$$,1<Kem$.;>\ \ 9C*Melita Per ec TadiTD42$~t+69h ^`hH.h ^`hH.h pLp^p`LhH.h @ @ ^@ `hH.h ^`hH.h L^`LhH.h ^`hH.h ^`hH.h PLP^P`LhH.h ^`hH.h ^`hH.h pLp^p`LhH.h @ @ ^@ `hH.h ^`hH.h L^`LhH.h ^`hH.h ^`hH.h PLP^P`LhH.h ^`hH.h ^`hH.h pLp^p`LhH.h @ @ ^@ `hH.h ^`hH.h L^`LhH.h ^`hH.h ^`hH.h PLP^P`LhH.T2$+VA&&mCPc z/ ` e 5C'5^9oA"hJ eDHWm! 9"iO#a^)q,*8*G+N?+ ,}-b.u+3k3+o4/5350n6W89<fg=p>q>l?DnDdEF^GB KGM M%N' OGQaQ8jQ9S XnZo[p$\:m\2]=^# _ _``H bcb[acldeVoepeVfXf h}yh|dilxl rn#oCp/#tjtUuxuVvZ7wWkxE@y+MyoyHWzK}C[k*?XI70W=Gg\Vh^ub4v:R] Gh&*G3_ Z@RK.G<`UkjLD^4%-i,1S 2zg ,+ Y?)ZF8iK8):JL^!:pwd  &WDSH2~rcW_67OW/-]m   $(-16:>CHNSXYdhmrw{ $*/0>BGKQW\aejotyzi0@** **LPP@PP @P@P@UnknownGz Times New Roman5Symbol3& z Arial5& zaTahoma"1HfC{ņ+l " "!xx4d 2QHX ?pe;C:\DOCUME~1\Melita\LOCALS~1\Temp\Springer_word_template.dot(Author template for normal English books2Copyright Springer-Verlag Heidelberg Berlin 1997 Melita Percec TadicMelita Per ec Tadi   Oh+'0,P lx    )Author template for normal English books WouthMelita Percec TadiceliSpringer Heidelberg 2005 ESpringer_word_template.dotEMelita Perec Tadi43iMicrosoft Word 10.0@2@J<{@5d@.g՜.+,D՜.+,d  px  oJ" A )Author template for normal English books Title 8@ _PID_HLINKSA|tH"mailto:melita.percec@cirus.dhz.hr  !"#$%&'(*+,-./012345689:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWYZ[\]^_abcdefgjRoot Entry Fp?lData )1Table7AWordDocument!PSummaryInformation(XDocumentSummaryInformation8`CompObjj  FMicrosoft Word Document MSWordDocWord.Document.89q