Assessment of data quality obtained in groundwater monitoring and the application of chemometric methods for determination of measuring sites

V doktorski nalogi smo preučevali kakovost podatkov meritev fizikalno kemijskih parametrov podzemne vode. S pomočjo kemometrijske karakterizacije smo določili reprezentativnost merilnih mest v programih spremljanja kakovosti podzemne vode. Podzemne vode so vode, ki se nahajajo pod površino tal in so v več kot 90% vir pitne vode v Sloveniji. Prav tako so podzemne vode vir vseh naravnih mineralnih voda ter izvirskih voda v Sloveniji. V nalogi smo uporabili rezultate državnega monitoring podzemne vode, ki ga izvaja Agencija Republike Slovenije za okolje, podatke spremljanja podzemne vode v Mestni občini Ljubljana, podatke notranjega nadzora pitne vode Mariborskega vodovoda, podatke notranjega nadzora pitne vode Vodovoda Murska Sobota, podatke državnega monitoringa pitne vode, podatke iz deklaracij naravnih mineralnih vod. Za proučevanje smo uporabili kemometrijske metode kot so analiza variance, korelacijska analiza, hierarhično razvrščanje, linearna diskriminantna analiza, metoda glavnih osi. Parametri, ki se spremljajo v posameznih vrstah vod, so različni. V podzemni vodi in mineralni vodi se spremljajo osnovni parametri, ki opredeljujejo mineraloško sestavo vode. V podzemni in pitni vodi se določajo tudi onesnaževala, kot so pesticidi, halogenirana organska topila, aromatska topila, skupni organski ogljik, težke kovine ipd. Spremljanje kakovosti voda mora zagotoviti verodostojne podatke, programi spremljanja kakovosti pa morajo biti tudi stroškovno sprejemljivi. Na eni strani se laboratoriji soočajo z vedno večjimi zahtevami naročnikov po nižanju meje določanja za posamezna onesnaževala, po drugi strani pa se soočajo tudi z ekonomskim pritiskom in konkurenčnostjo z zasebnimi in tujimi laboratoriji, hkrati pa mora biti zagotovljena visoka kakovost storitev. V študiji smo proučili, kakšen vpliv na klasifikacijo vzorcev je imelo zmanjševanje števila vzorcev in izpuščanje parametrov v programu spremljanja. Ocenili smo tudi ustreznost zamenjave določenih merilnih mest tako za pitno kot podzemno vodo. Sistemi za oskrbo s pitno vodo so lahko enostavni (en vodni vir, eno oskrbovalno območje) ali pa bolj kompleksni (več vodnih virov). V doktorski nalogi smo opisali metodologijo spremljanja kakovosti pitne vode in metodologijo izbire merilnih mest, ki bi najbolje odražale značilnosti vodnega vira in učinke morebitnega mešanja različnih vodnih virov. Poseben poudarek smo namenili tudi oceni kakovosti vzorčenja in izvedbe terenskih meritev. Ugotovili smo, da lahko s kemometrijskimi analizami hitro ugotovimo ali so meritve zanesljive, natančne in točne ali pa obstajajo morebitna odstopanja rezultatov meritev, ki jih pri rutinskem delu nismo mogli zaznati. V nalogi smo tudi dokazali uporabnost kemometrijskih metod za vode, ki so v postopku pridobivanja naziva “naravna mineralna voda”. Zakonodaja s tega področja je nejasna in omogoča različne interpretacije. Metodologija, kdaj se voda z nazivom “naravna mineralna voda” razlikuje od pitne vode, ni določena. Na vprašanje, ali gre v določenih primerih za isto vodo, lahko damo odgovor tudi s pomočjo kemometrijskih metod. Pri razvrščanju podzemnih voda smo uporabili različne kemometrijske metode. Ugotovili smo, da je zelo uporabna metoda linearna diskriminantna analiza, še zlasti na primerih kraških vodonosnikov, kjer z drugimi metodami nismo dobili pričakovane klasifikacije. Lahko potrdimo, da so kemometrijske metode odlično orodje pri načrtovanju in racionalizaciji programa spremljanja kakovosti podzemnih vod.The doctoral dissertation includes an assessment of the quality of data obtained by measuring physical and chemical parameters of groundwater. By means of chemometric characterisation we determined representativeness of monitoring points for groundwater monitoring programmes. In studying, we applied chemometric methods such as analysis of variance, correlation analysis, cluster analysis, linear discriminant analysis, and principal component analysis. Groundwater is the water under the ground surface, the source of more than 90% of drinking water in Slovenia. Groundwater is also the source of all natural mineral water and spring water in Slovenia. The dissertation relies on the results of national groundwater monitoring conducted by the Slovenian Environment Agency, the data about groundwater monitoring of the Ljubljana Municipality, the data on internal control of drinking water of the Maribor Waterworks and of the Murska Sobota Waterworks, the data from national drinking water monitoring, and the data from declarations of natural mineral water. The parameters monitored in individual types of water differ: In groundwater and mineral water we observe basic parameters that define the basic mineral composition of water. In groundwater and potable water, we also determine pollutants, such as pesticides, halogenated organic solvents, aromatic hydrocarbons, total organic carbon, heavy metals, etc. Water quality monitoring should provide reliable data and quality monitoring programmes have to be acceptable in terms of cost. On the one hand, laboratories are faced with ever greater requirements to lower the limit of quantitation for individual pollutants and assure quality, whereas on the other hand they are subject to economic pressure and competition from private and foreign laboratories. The study examines the impact on sample classification due to the reduced number of samples and omitting parameters in the monitoring programme. We also assessed the adequacy of replacement of certain monitoring points both for potable and groundwater. Potable water supply systems can be simple (one water source, one water supply zone) or more complex (several water sources). The doctoral dissertation describes the methodology of monitoring the quality of drinking water and selection of monitoring points that would best reflect the characteristics of a water source and the effects of potential mixing of various water sources. Special emphasis was put on assessing the quality of sampling and field measurements. We established that chemometric analyses can quickly tell us if measurements are reliable and correct or if there are any deviations in measurement results that we were unable to detect during routine work. The dissertation proves the usefulness of chemometric methods for waters that are in the process of acquiring the title of “natural mineral water”. The legislation in this field is not completely specific. The methodology as to when water called “natural mineral water” differs from drinking water is not specified. The question as to whether this is the same water in certain cases can also be answered by means of chemometric methods. In the classification of groundwater, we applied various chemometric methods. We established that linear discriminant analysis is especially useful, particularly for karst aquifers, where other methods have not resulted in expected classification. We can confirm that chemometric methods are an excellent tool in planning and streamlining the groundwater monitoring programme