Ndiyabulela ngokundwendwela i-Nature.com.Uguqulelo lwesikhangeli oyisebenzisayo lunenkxaso elinganiselweyo ye-CSS.Ukufumana amava angcono kakhulu, sincoma ukuba usebenzise i-browser ehlaziyiweyo (okanye ucime imodi yokuhambelana kwi-Internet Explorer). Okwangoku, ukuqinisekisa inkxaso eqhubekayo, siya kubonisa isayithi ngaphandle kwezitayela kunye neJavaScript.
Ungcoliseko lomhlaba yingxaki enkulu ebangelwa yimisebenzi yabantu.Ukwabiwa kwendawo yezinto ezinokuba yityhefu (PTEs) kuyahluka kwiindawo ezininzi ezisezidolophini nakwimimandla ekufutshane nedolophu.Ngoko ke, kunzima ukuqikelela ngokwesithuba umxholo we-PTEs kwimihlaba enjalo.Iisampulu ezili-115 zizonke zafunyanwa kwi-Frydek Mistek kwiRiphabhliki yase-Czech), i-magnesium ye-Nick) Ukugxininiswa kunqunywe ngokusebenzisa i-spectrometry emission inductively coupled plasma emission spectrometry.Impendulo eguquguqukayo yi-Ni kunye ne-predictors yi-Ca, Mg, kunye ne-K.I-matrix yokulungelelaniswa phakathi kwempendulo eguquguqukayo kunye ne-predictor variable ibonisa ukulungelelaniswa okwanelisayo phakathi kwezinto.Iziphumo zokubikezela zibonise ukuba i-Support Vector Machine Regression (SVMR) yenza kakuhle, nangona i-root. Impazamo epheleleyo (MAE) (166.946 mg/kg) ibiphezulu kunezinye iindlela ezisetyenzisiweyo.Imifuziselo exutyiweyo ye-Empirical Bayesian Kriging-Multiple Linear Regression (EBK-MLR) iqhuba kakubi, njengoko kungqinwa yi-coefficients yokumisela ngaphantsi kwe-0.1.Imodeli ye-Empirical Bayesian Kriging-Sups, imodeli ye-Empirical Bayesian Kriging-Sups nge-RMSE ephantsi (95.479 mg/kg) kunye ne-MAE (77.368 mg/kg) amaxabiso kunye ne-coefficient ephezulu yokuzimisela (R2 = 0.637) kunye nemihlaba ejikeleze idolophu. Iziphumo zibonisa ukuba ukudibanisa i-EBK kunye ne-SVMR bubuchule obusebenzayo bokuqikelela ukugxila kwe-Ni kwimihlaba yasezidolophini nakwimimandla ekufutshane nedolophu.
I-Nickel (Ni) ithathwa njenge-micronutrient kwizityalo ngenxa yokuba igalelo kwi-atmospheric nitrogen fixation (N) kunye ne-urea metabolism, zombini ezifunekayo ekuhlumeni kwembewu.Ukongezelela kwigalelo layo ekuhlumeni kwembewu, i-Ni inokusebenza njenge-fungal kunye ne-bacterial inhibitor kwaye ikhuthaze ukukhula kwezityalo. zidinga ukufakwa kwezichumiso ezisekelwe kwi-nickel ukuze kuphuculwe initrogen.2.Ukusetyenziswa okuqhubekayo kwezichumisi ezisekelwe kwi-nickel ukutyebisa umhlaba kunye nokwandisa amandla eediyitrojeni ekulungiseni i-nitrogen emhlabeni ngokuqhubekayo kwandisa ugxininiso lwe-nickel emhlabeni.Nangona i-nickel i-micronutrient kwizityalo, ukufakwa kwayo ngokugqithiseleyo emhlabeni kunokwenza i-pH ye-nick enetyhefu ngaphezu komhlaba. Ukuthathwa kwentsimbi njengesondlo esibalulekileyo ekukhuleni kwezityalo1.Ngokutsho kweLiu3, i-Ni ifunyenwe iyinto ye-17 ebalulekileyo efunekayo ekuphuhliseni nasekukhuleni kwezityalo.Ukongezelela kwindima ye-nickel ekuphuhlisweni kwezityalo kunye nokukhula, abantu bayayidinga kwiintlobo ezahlukeneyo ze-applications.I-Electroplating, ukuveliswa kwe-nickel-based alloys, kunye nokuveliswa kwezixhobo zokusetyenziswa kwe-autopark kwi-autopark. Ukongeza, i-alloys esekwe kwi-nickel kunye namanqaku e-electroplated asetyenziswa kakhulu kwi-kitchenware, izixhobo ze-ballroom, iimpahla zeshishini lokutya, umbane, intambo kunye nentambo, ii-jet turbines, implants zotyando, amalaphu, kunye nokwakhiwa kweenqanawa5. kune-anthropogenic4,6.Imithombo yendalo yenikeli ibandakanya ugqabhuko-dubulo lwentaba-mlilo, uhlaza, imililo yamahlathi, kunye neenkqubo zejoloji; nangona kunjalo, imithombo ye-anthropogenic ibandakanya i-nickel / cadmium iibhetri kwi-industry yensimbi, i-electroplating, i-arc welding, i-diesel kunye ne-oyile ye-fuel, kunye ne-atmospheric emissions evela ekutshisweni kwamalahle kunye nenkunkuma kunye nokutshisa i-sludge accumulation ye-Nickel7,8.Ngokutsho kwe-Freedman kunye ne-Hundchinson9 kunye ne-Manyiwa et. I-10, imithombo ephambili yongcoliseko lomhlaba kwindawo esondeleyo kunye nommandla osondeleyo ikakhulu i-nickel-copper-based smelters kunye nemigodi. Umhlaba waseNorway11.Ngokutsho kweAlms et al. I-12, isixa se-nickel ye-HNO3-extractable kumhlaba ophezulu olimekayo wommandla (imveliso ye-nickel eRashiya) yayisuka kwi-6.25 ukuya kwi-136.88 mg / kg, ehambelana nenani le-30.43 mg / kg kunye nesiseko soxinaniso lwe-25 mg / kg. Ngokutsho kwe-kabata ye-11, ukusetyenziswa komhlaba we-phosphorus-urban in phosphorus-urban-application of phosphorus-urban umhlaba. amaxesha ezityalo ezilandelelanayo zinokufaka okanye zingcolise umhlaba.Iziphumo ezinokuthi zenzeke zenickel ebantwini zingakhokelela kumhlaza nge-mutagenesis, umonakalo we-chromosomal, isizukulwana se-Z-DNA, ukulungiswa kwe-DNA excision evaliweyo, okanye iinkqubo ze-epigenetic13.Kwizilingo zezilwanyana, i-nickel ifunyenwe inamandla okubangela iintlobo ezahlukeneyo ze-tumor, kunye ne-carcinogenic nickel nickel complexes ingaba ne-epigenetic.
Uvavanyo lokungcoliseka komhlaba luye lwanda kumaxesha amva nje ngenxa yoluhlu olubanzi lwemiba enxulumene nempilo evela kubudlelwane bezityalo zomhlaba, umhlaba kunye nobudlelwane bebhayoloji yomhlaba, ukuthotywa kwe-ecological, kunye novavanyo lwempembelelo yendalo esingqongileyo.Ukuza kuthi ga ngoku, ukubikezelwa kwendawo yezinto ezinokuthi zibe yityhefu (PTEs) njengeNi emhlabeni iye yanzima kwaye idla ixesha kunye nokusetyenziswa kweendlela zedijithali ze-DS1M zangoku. iphuculwe kakhulu ipredictive soil mapping (PSM).Ngokutsho kukaMinasny kunye noMcBratney16, ipredictive soil mapping (DSM) ibonakalise ukuba luluhlu olubalaseleyo lwenzululwazi yomhlaba.Lagacherie and McBratney, 2006 define DSM as “the creative and filling of spatial soil information systems through use of the spatial-lab- observation- iinkqubo zokuthelekelela”.McBratney et al. I-17 ichaza ukuba i-DSM yangoku okanye i-PSM yeyona ndlela isebenzayo yokuqikelela okanye imephu yokusasazwa kwendawo ye-PTEs, iintlobo zomhlaba kunye neempawu zomhlaba.I-Geostatistics kunye ne-Machine Learning Algorithms (MLA) ziindlela ze-DSM zemodeli ezenza iimephu zedijithali ngoncedo lweekhompyutha usebenzisa idatha ebalulekileyo kunye nencinci.
I-Deuts18 kunye ne-Olea19 zichaza i-geostatistics “njengengqokelela yeendlela zobuchule zamanani ezijongana nokumelwa kweempawu zendawo, ikakhulu zisebenzisa iimodeli zestochastic, ezinje ngendlela uhlahlelo lwexesha olubonisa ngayo idatha yexeshana.” Ngokuyintloko, i-geostatistics ibandakanya uvavanyo lwee-variograms, ezivumela ukulinganisa kunye nokuchaza ukuxhomekeka kwamaxabiso esithuba ukusuka kwi-dataset nganye20.Gumiaux et al. I-20 ibonise ngakumbi ukuba uvavanyo lwe-variograms kwi-geostatistics lusekelwe kwimigaqo emithathu, kubandakanywa (a) ukubala isikali sokulungelelaniswa kwedatha, (b) ukuchonga kunye ne-computing anisotropy kwi-dataset disparity kunye (c) ukongeza kwi-Ukongeza ekuthatheleni ingqalelo impazamo engokwemvelo yokulinganisa idatha eyahlulwe kwimiphumo eqikelelwayo yendawo, i-international effects, i-compute ye-international. zisetyenziswa kwi-geostatistics, kuquka ikriging jikelele, i-co-kriging, i-kriging eqhelekileyo, i-empirical Bayesian kriging, indlela ye-kriging elula kunye nezinye iindlela ezaziwayo zokudibanisa imephu okanye ukuqikelela i-PTE, iimpawu zomhlaba, kunye neendidi zomhlaba.
Machine Learning Algorithms (MLA) bubuchule obutsha ngokwentelekiso osebenzisa iiklasi ezinkulu zedatha non-linear, ziphenjelelwa algorithms ngokuyintloko esetyenziselwa imigodi data, ukuchonga iipateni data, kwaye ngokuphindaphindiweyo isetyenziswe kuhlelo kwiinkalo zenzululwazi ezifana nenzululwazi yomhlaba kunye nokubuyisela imisebenzi.Amaphepha ophando amaninzi axhomekeke kwiimodeli ze-MLA ukuqikelela i-PTE kwimihlaba, efana ne-Tan et al. I-22 (amahlathi angaqhelekanga oqikelelo lwentsimbi enzima kwimihlaba yezolimo), Sakizadeh et al. I-23 (umzekelo usebenzisa oomatshini be-vector yenkxaso kunye nothungelwano lwe-neural eyenziweyo) ukungcoliswa komhlaba) .Ukongezelela, i-Vega et al. I-24 (I-CART yokwenza imodeli yokugcinwa kwesinyithi esinzima kunye ne-adsorption emhlabeni) I-Sun et al. I-25 (ukusetyenziswa kwe-cubist kukusasazwa kwe-Cd emhlabeni) kunye nezinye iindlela zokuziphatha ezifana nommelwane okufutshane no-k, ukuhlehla okwandisiweyo ngokubanzi, kunye nokunciphisa ukubuyisela Imithi nayo isebenzise i-MLA ukuqikelela i-PTE emhlabeni.
Ukusetyenziswa kwe-algorithms ye-DSM ekuqikeleleni okanye kwimephu ijongene nemingeni emininzi.Ababhali abaninzi bakholelwa ukuba i-MLA iphezulu kwi-geostatistics kunye ne-vice versa.Nangona enye ingcono kunomnye, ukudibanisa kwezi zibini ziphucula izinga lokuchaneka kwemephu okanye ukubikezela kwi-DSM15.Woodcock kunye neGopal26 Finke27; I-Pontius kunye ne-Cheuk28 kunye ne-Grunwald29 baphawula malunga neentsilelo kunye neempazamo ezithile kwimephu yomhlaba eqikelelweyo.Iinzululwazi zomhlaba ziye zazama iindlela ezahlukeneyo zokuphucula ukusebenza, ukuchaneka, kunye nokuqikelelwa kwe-DSM yemephu kunye noqikelelo.Ukudibanisa ukungaqiniseki kunye nokuqinisekiswa yenye yezinto ezininzi ezahlukeneyo ezidityanisiweyo kwi-DSM kunye nokunciphisa i-alfectye. 15 ichaza ukuba ukuziphatha kokuqinisekisa kunye nokungaqiniseki okuvezwe ngokuyilwa kweemephu kunye noqikelelo kufuneka kuqinisekiswe ngokuzimeleyo ukuze kuphuculwe umgangatho wemephu.Imida ye-DSM kungenxa yomgangatho womhlaba osasazwe ngokwejografi, obandakanya icandelo lokungaqiniseki; nangona kunjalo, ukungabikho kwengqiniseko kwi-DSM kunokuvela kwimithombo emininzi yephutha, okuyiphutha le-covariate, impazamo yomzekelo, impazamo yendawo, kunye nempazamo yohlalutyo 31. Ukungachaneki kweModelling eyenziwe kwi-MLA kunye neenkqubo ze-geostatistical zihambelana nokungabikho kokuqonda, ekugqibeleni kukhokelela ekugqithiseni kwenkqubo yokwenyani32. Ukuqikelelwa kwemodeli yeemathematika, okanye i-interpolation33.Kutshanje, kuye kwavela umkhwa omtsha we-DSM okhuthaza ukuhlanganiswa kwe-geostatistics kunye ne-MLA kwimephu kunye nokubikezela.Iinzululwazi ezininzi zomhlaba kunye nababhali, njengoSergeev et al. 34; I-Subbotina et al. 35; Tarasov et al. 36 kunye noTarasov et al. Abangama-37 basebenzise umgangatho ochanekileyo we-geostatistics kunye nokufunda koomatshini ukuvelisa iimodeli ezixubileyo eziphucula ukusebenza kakuhle koqikelelo kunye nemephu. umgangatho.Eminye yale mifuziselo ye-hybrid okanye edibeneyo ye-algorithm i-Artificial Neural Network Kriging (ANN-RK), i-Multilayer Perceptron Residual Kriging (MLP-RK), i-Generalized Regression Neural Network Residual Kriging (GR- NNRK)36, i-Artificial Neural Network Kriging-Multilayer Perceptron-Multilayer-Multilayer Perceptron (ANKsi-Multilayer Perceptron) kunye ne-ANKsi-3K Inkqubo ye-Multilayer-Multilayer kunye ne-ANKsi-3K ye-ANNRK Ukuhlehla38.
Ngokutsho kukaSergeev et al., Ukudibanisa iindlela ezahlukeneyo zokubumba kunamandla okuphelisa iziphene kunye nokwandisa ukusebenza kakuhle kwemodeli ye-hybride enesiphumo kunokuba iphuhlise imodeli yayo enye.Kulo mxholo, eli phepha litsha libonisa ukuba kuyimfuneko ukusebenzisa i-algorithm edibeneyo ye-geostatistics kunye ne-MLA ukudala imifuziselo efanelekileyo ye-hybrid ukuxela kwangaphambili ukutyetyiswa kwe-Ni kwimimandla yasezidolophini nakwi-periurban Bay. (EBK) njengomzekelo wesiseko kwaye uyixube kunye noMashini weVector yeNkxaso (SVM) kunye neMininzi yeMilayini yokuRegression (MLR) imifuziselo.I-Hybridization ye-EBK nayo nayiphi na i-MLA akwaziwa.Imifuziselo exubeneyo emininzi ebonwayo yindibaniselwano yesiqhelo, intsalela, i-regression kriging, kunye ne-MLA.EBK yinkqubo ye-geostatistical interpolation esebenzisa inkqubo ye-spatistic esetyenziswa njenge-spatic intsimi engahlaliyo/emileyo enemilinganiselo echaziweyo yendawo phezu kwentsimi, evumela ukwahluka kwendawo39.EBK isetyenziswe kwiintlobo ngeentlobo zezifundo, kuquka ukuhlalutya ukuhanjiswa kwekhabhoni ephilayo kwimihlaba yeefama40, ukuhlola ukungcoliseka komhlaba41 kunye neepropati zemephu yomhlaba42.
Ngakolunye uhlangothi, i-Self-Organising Graph (SeOM) yi-algorithm yokufunda esetyenziswe kumanqaku ahlukeneyo afana noLi et al. 43, Wang et al. 44, Hossain Bhuiyan et al. 45 kunye noKebonye et al.46 Ukumisela iimpawu zendawo kunye nokuhlelwa kwezinto.Wang et al. I-44 ichaza ukuba i-SeOM yindlela yokufunda enamandla eyaziwayo ngokukwazi ukuhlanganisa kunye nokucinga iingxaki ezingezona umgca.Ngokungafaniyo nezinye iindlela zokuqaphela iipatheni ezifana nokuhlalutya kwenqununu yecandelo, ukuhlanganisana okungaqondakaliyo, ukuhlanganiswa kwe-hierarchical, kunye nokwenziwa kwezigqibo ezininzi, i-SeOM ingcono ekuququzeleleni nasekuchongeni iipatheni ze-PTE. I-44, i-SeOM inokwenza iqela lendawo yokusabalalisa i-neurons ehambelanayo kwaye ibonelele ngokubonwa kwedatha ephezulu.
Eli phepha lijolise ekuveliseni imodeli yemephu eyomeleleyo ngokuchaneka okuphezulu kokuqikelelwa komxholo we-nickel kwimihlaba yasezidolophini nakwimimandla yedolophu. Siqikelela ukuba ukuthembeka kwemodeli edibeneyo kuxhomekeke ikakhulu kwimpembelelo yezinye iimodeli ezincanyathiselwe kwimodeli yesiseko. ngoko ke, siya kuzama ukuphendula imibuzo yophando enokuthi ivelise imizekelo exubeneyo.Nangona kunjalo, ichaneke kangakanani imodeli ekuqikeleleni into ekujoliswe kuyo?Kwakhona, lithini inqanaba lovavanyo olusebenzayo olusekwe kungqinisiso kunye novavanyo oluchanekileyo?Ngoko ke, iinjongo ezithile zolu phononongo yayikukuba (a) yenze imodeli yomxube odityanisiweyo we-SVMR okanye iMLR usebenzisa i-EBK (c) uthelekise imodeli yesiseko sepro Ugxininiso lwe-Ni kwimihlaba yasezidolophini okanye ekufutshane nedolophu , kunye (d) nokusetyenziswa kwe-SeOM ukuyila imephu enesisombululo esiphezulu sokwahluka kwesithuba senikeli.
Uphononongo lwenziwa kwiRiphabhliki yaseCzech, ngokukodwa kwisithili saseFrydek Mistek kwingingqi yaseMoravia-Silesian (jonga umfanekiso 1) .Ijografi yendawo yophononongo inzima kakhulu kwaye iyinxalenye yengingqi yaseMoravia-Silesian Beskidy, eyinxalenye yomngcelele ongaphandle weeNtaba zeCarpathian.Indawo yophononongo ibekwe phakathi kwe-N1′10 ° 4 kunye ne-2'19 ° ′ 49 ° ′ 49 0′ E, kwaye ukuphakama kuphakathi kwe-225 kunye ne-327 m; nangona kunjalo, inkqubo yokuhlelwa kweKoppen yemeko yemozulu yommandla inikwe umlinganiselo njengeCfb = imozulu epholileyo yolwandle, Kukho imvula eninzi nangeenyanga ezomileyo.Amaqondo obushushu ahluka kancinci unyaka wonke phakathi -5 °C kunye nama-24 °C, kunqabile ukuba abe ngaphantsi -14 °C okanye ngaphezulu kwe-avareji ye-30 °C kunye ne-5 °C yonyaka, ngelixa i-avareji ye-5 °C kunye ne-28 yemozulu. mm47.Ummandla wovavanyo oqikelelweyo wendawo yonke yi-1,208 yeekhilomitha zeekhilomitha, kunye ne-39.38% yomhlaba olinywayo kunye ne-49.36% ye-coverage yehlathi.Kwelinye icala, indawo esetyenziswe kolu phononongo malunga ne-889.8 yeekhilomitha zee-square.In kunye ne-Ostrava, imboni yentsimbi kunye ne-metal mills isetyenziselwa i-steel mills isetyenziswa kakhulu. (umzekelo ukuxhathisa umhlwa emoyeni) kunye neentsimbi ingxubevange (nickel kwandisa amandla ingxubevange ngelixa ukugcina ductility yayo elungileyo kunye nokuqina), kunye nezolimo olunzulu ezifana isicelo isichumiso iphosphate kunye nemveliso yemfuyo yimithombo yophando enokubakho nickel kummandla (umzekelo, ukongeza nickel ukuba amatakane ukwandisa amazinga okukhula kwimizi-mveliso yeegusha kunye neendawo zokusetyenziswa kwe-nickel ye-nickel ephantsi). i-electroplating, kubandakanywa ne-nickel ye-electroplating kunye ne-electroless nickel plating systems.Iipropati zomhlaba zibonakala ngokulula kumbala womhlaba, isakhiwo, kunye nesiqulatho se-carbonate.Ubume bomhlaba buphakathi ukuya kwi-fine, buvela kwizinto zomzali.Ziyi-colluvial, i-alluvial okanye i-aeolian ngokwendalo.Eminye imimandla yomhlaba ibonakala i-mottled kwi-surface kunye ne-subsoil, idla ngokuphindaphindiweyo kunye neentlobo ze-concrete kunye ne-stagnol. Ummandla we48.Ngokuphakama ukusuka kwi-455.1 ukuya kwi-493.5 m, ii-cambisols zilawula i-Czech Republic49.
Imephu yendawo yokufunda [Imephu yendawo yokufunda yenziwe kusetyenziswa i-ArcGIS Desktop (ESRI, Inc, inguqulelo 10.7, URL: https://desktop.arcgis.com).]
Iisampulu ze-115 zomhlaba eziphezulu zifunyenwe kwimihlaba yasezidolophini kunye ne-peri-urban kwisithili saseFrydek Mistek.Ipateni yesampula esetyenzisiweyo yayiyigridi eqhelekileyo kunye neesampuli zomhlaba ezihlukaniswe i-2 × 2 km ngaphandle, kunye nomhlaba ongaphezulu ulinganiswe kubunzulu be-0 ukuya kwi-20 cm usebenzisa isixhobo se-GPS esibanjwe ngesandla (i-Leica Zeno 5 ipakishwe ngokufanelekileyo kwi-Samples ye-GPS, i-Samples ifakwe kwi-GPS). ukuya kwibhubhoratri.Iisampulu zomiswe ngomoya ukuze zivelise iisampuli ezixutyiweyo, zixutywe yinkqubo yomatshini (i-Fritsch disc mill), kunye ne-sieved (ubungakanani be-sieve 2 mm) .Beka i-1 gram yeesampulu zomhlaba omisiweyo, i-homogenized kunye ne-sieved kwiibhotile ze-teflon ezibhalwe ngokucacileyo. I-HNO3 (usebenzisa i-dispenser ezenzekelayo - enye ye-asidi nganye), gubungela ngokukhawuleza kwaye uvumele iisampuli ukuba zime ubusuku bokusabela (iprogram ye-aqua regia) .Beka i-supernatant kwi-plate yensimbi eshushu (ubushushu: 100 W kunye ne-160 ° C) i-2 h ukuququzelela inkqubo yokugaya iisampulu, emva koko i-supernalutant 5 ml kunye ne-flavour ye-50 ipholile kwi-flat. ml ngamanzi ahlanjululweyo. Emva koko, hluza i-supernatant ehlanjululweyo kwi-tube ye-PVC ye-50 ml ngamanzi ahlanjululweyo. Ukongezelela, i-1 ml yesisombululo se-dilution ihlanjululwe nge-9 ml yamanzi adibeneyo kwaye ihlulwe kwi-tube ye-12 ml elungiselelwe i-PTE pseudo-concentration.I-concentrations ye-PTEs, i-Curn, i-Curn, i-Cred, i-Curn, i-Cred, i-Curn, i-Cred I-Mg, K) inqunywe yi-ICP-OES (i-Inductively Coupled Plasma Optical Emission Spectroscopy) (i-Thermo Fisher Scientific, eU.SA) ngokweendlela eziqhelekileyo kunye nesivumelwano.Ukuqinisekisa uQinisekiso loMgangatho kunye noLawulo (QA / QC) iinkqubo (SRM NIST 2711a Montana II Umhlaba) yayingu-0.0004.(wena) .Ukongezelela, ulawulo lomgangatho kunye nenkqubo yokuqinisekisa umgangatho wohlalutyo ngalunye luqinisekiswa ngokuhlalutya imigangatho yereferensi.Ukuqinisekisa ukuba iimpazamo zancitshiswa, uhlalutyo oluphindwe kabini lwenziwa.
I-Empirical Bayesian Kriging (EBK) yenye yeendlela ezininzi zokufakelwa kwe-geostatistical ezisetyenziswa kwimodeli kwiinkalo ezahlukeneyo ezifana nenzululwazi yomhlaba.Ngokungafaniyo nezinye iindlela zokudibanisa ikriging, i-EBK iyahluka kwiindlela zekriging zemveli ngokuqwalasela impazamo eqikelelwa yimodeli ye-semivariogram. I-semivariogram.Iindlela zokuguqulela zenza indlela yokungaqiniseki kunye neprogramu ehambelana noku kucwangciswa kwe-semivariogram eyenza inxalenye entsokothileyo yendlela eyaneleyo ye-kriging.Inkqubo yofakelo lwe-EBK ilandela iindlela ezintathu ezicetywayo ngu-Krivoruchko50, (a) imodeli iqikelela i-semivariogram esekelwe kwi-semivariogram kwi-dataset eqikelelweyo yendawo nganye kwidathasethi entsha eqikelelweyo isemivariogram kunye (c) nemodeli yokugqibela engu-A ibalwe kwidatha efanisiweyo.
Apho \(Ingxaki\ekhohlo(A\ekunene)\) imele okuphambili, \(Ingxaki\ekhohlo(B\ekunene)\) amathuba omda ahoywa kwiimeko ezininzi, \(Prob (B,A)\ ) .Ubalo lwesemivariogram lusekwe kumgaqo weBayes, obonisa ukuthambekela koqwalaselo loqwalaselo lwedatha enokuthi yenziwe kwisemiovari yedatha enokuthi yenziwe kwisemiovari yedatha. Umthetho we-Bayes, ochaza ukuba kunokwenzeka kangakanani ukwenza isethi yedatha yokuqatshelwa kwi-semivariogram.
Umatshini we-vector yenkxaso yi-algorithm yokufunda yomatshini eyenza i-hyperplane yokwahlula ngokufanelekileyo ukwahlula iiklasi ezifanayo kodwa ezingekho ngomgca ezizimeleyo.Vapnik51 idale i-algorithm yokuhlelwa kwenjongo, kodwa isetyenziswe kutshanje ukusombulula iingxaki ezijoliswe kwi-regression-oriented. IVector Machine Regression - SVMR) isetyenziswe kolu hlalutyo.I-Cherkassky kunye ne-Mulier53 iphayona i-SVMR njenge-kernel-based regression, i-comutation eyenziwa ngokusebenzisa imodeli yokunciphisa umgca kunye nemisebenzi yamazwe amaninzi. kuVohland et al. I-55, i-epsilon (ε) -SVMR isebenzisa idatha yedatha eqeqeshiweyo ukufumana imodeli yokumela njengomsebenzi we-epsilon-insensitive esetyenziswa ukwenza imephu yedatha ngokuzimeleyo ngeyona ndlela ingcono ye-epsilon bias ukusuka kuqeqesho kwidatha ehambelanayo.Impazamo yomgama osetyenzisiweyo ayihoywa kwixabiso langempela, kwaye ukuba impazamo inkulu kuno-ε(ε), ipropati yoqeqesho lomhlaba ihlawulela imodeli ebanzi yoqeqesho lwedatha ibuye ihlawule imodeli ebanzi yoqeqesho lwedatha. i-subset ye-vectors yenkxaso.I-equation ecetywayo yiVapnik51 iboniswe ngezantsi.
apho b imele i-scalar threshold, \(K\left({x}_{,}{ x}_{k}\right)\) imele umsebenzi wekernel, \(\alpha\) imele iLagrange multiplier, N Imele iseti yedatha yamanani, \({x}_{k}\) imele igalelo ledatha, kwaye \.Ornels isetyenzisiweyo sedatha, kwaye \(y\rnels) yidata esetyenzisiweyo ngumsebenzi wesiseko se-Gaussian radial (RBF) .I-RBF kernel isetyenziselwa ukumisela imodeli ye-SVMR efanelekileyo, ebaluleke kakhulu ekufumaneni isohlwayo esifihlakeleyo se-C kunye ne-kernel parameter gamma (γ) yedatha yoqeqesho lwe-PTE. Okokuqala, sivavanye isethi yoqeqesho kwaye emva koko sivavanya imodeli yokusebenza kwisethi yokuqinisekisa esetyenzisiweyo.
Imodeli yokubuyisela imigca emininzi (MLR) ngumzekelo wokubuyisela umva omele ubudlelwane phakathi kwenguquko yempendulo kunye nenani leenguqu ze-predictor ngokusebenzisa i-linear pooled parameters ezibalwe kusetyenziswa indlela ye-square encinci.Kwi-MLR, imodeli yesikwere esincinci ngumsebenzi oqikelelwayo weempawu zomhlaba emva kokukhethwa kwezinto ezichazayo.Kuyimfuneko ukusebenzisa impendulo ukuseka ubudlelwane obusetyenzisiweyo njengempendulo yomgca we-explanation ubudlelwane be-P.P. ngeenguqu ezicacisayo.Inxaki yeMLR yi
apho u-y yimpendulo eguquguqukayo, \(a\) yi-intercept, n linani lezinto eziqikelelwayo, \({b}_{1}\) lubuyiselo olungaphelelanga lwe-coefficients, \({x}_{i}\) imele ipredictor okanye ingcaciso eguquguqukayo, kwaye \({\varepsilon }_{i}\) imele impazamo eyaziwa ngokuba yimodeli.
Imifuziselo exutyiweyo yafunyanwa ngokuxutywa kwe-EBK kunye ne-SVMR kunye ne-MLR. Oku kwenziwa ngokukhupha amaxabiso aqikelelweyo ukusuka kwi-EBK interpolation. Amaxabiso aqikelelweyo afunyenwe kwi-interpolated Ca, K, kunye ne-Mg afunyanwa ngenkqubo yokudibanisa ukufumana izinto ezintsha eziguquguqukayo, ezifana neCaK, iCaMg, kunye ne-KMg. I-CaKMg. Ngokubanzi, iinguqu ezifunyenweyo ziyi-Ca, K, Mg, CaK, CaMg, KMg kunye ne-CaKMg. Ezi ziguquko zaba zizibikezelo zethu, zinceda ukuqikelela ukugxininiswa kwe-nickel kwimihlaba yasezidolophini kunye ne-peri-urban.I-algorithm ye-SVMR yenziwa kwii-predictors ukufumana imodeli edibeneyo Empirical Bayesian Kriging-Supports, I-EBmilar ye-Support ye-Support ye-SVMR . ngokusebenzisa i-algorithm yeMLR ukufumana imodeli exubeneyo ye-Epirical Bayesian Kriging-Multiple Linear Regression (EBK_MLR). Ngokwesiqhelo, izinto eziguquguqukayo Ca, K, Mg, CaK, CaMg, KMg, kunye neCaKMg zisetyenziswa njenge covariates njengezibikezelo ze Ni kumxholo wedolophu kunye nommandla wedolophu ofunyenweyo okanye i-EBKM ebonwayo imodeli efunyenweyo. usebenzisa igrafu ezilungelelanisayo.Ukuhamba komsebenzi kwesi sifundo kuboniswe kuMfanekiso 2.
Ukusebenzisa i-SeOM ibe sisixhobo esidumileyo sokulungelelanisa, ukuvavanya, kunye nokubikezela kwedatha kwicandelo lezemali, ukhathalelo lwempilo, ishishini, izibalo, isayensi yomhlaba, kunye nokunye.I-SeOM idalwe ngokusebenzisa i-neural networks eyenziweyo kunye neendlela zokufunda ezingalawulwayo zombutho, ukuvavanya, kunye nokubikezela.Kwesi sifundo, i-SeOM yayisetyenziselwa ukujonga i-Ni concentrations ye-Niu ye-predictioning ye-ingqikelelo ye-Niu yedatha kwi-model yedatha yedolophu. kuvavanyo lwe-SeOM zisetyenziswa njenge-n input-dimensional vector variables43,56.Melssen et al. I-57 ichaza uxhulumaniso lwe-input vector kwinethiwekhi ye-neural ngokusebenzisa i-input input layer kwi-vector ye-output kunye ne-vector eyodwa yobunzima.Imveliso eyenziwa yi-SeOM yimephu ye-dimensional-dimensional equkethe i-neurons eyahlukeneyo okanye i-nodes ezilukiweyo kwiimephu ze-hexagonal, i-circular, okanye i-square topological map ngokusondele. kunye ne-0.086 kunye ne-0.904, ngokulandelanayo, ikhethiweyo, eyiyunithi yemephu ye-55 (5 × 11) .Isakhiwo se-neuron sinqunywe ngokwenani leenodi kwi-empical equation.
Inani ledatha esetyenzisiweyo kolu phononongo ziisampuli ze-115. Indlela engahleliweyo isetyenziselwe ukwahlula idatha kwidatha yovavanyo (i-25% yokuqinisekiswa) kunye neeseti zedatha yoqeqesho (i-75% yokulinganisa) .Idatha yoqeqesho isetyenziselwa ukuvelisa imodeli yokubuyisela (ukulinganisa), kunye nedatha yovavanyo isetyenziselwa ukuqinisekiswa kwekhono lokudibanisa ngokubanzi58.Oku kwenziwa ukuvavanya umxholo we-nickel esetyenzisiweyo kwimodeli esetyenzisiweyo. inkqubo yokuqinisekiswa kwe-cross-fold-fold-convention, iphindwe kahlanu.Iinguqu eziveliswa yi-EBK interpolation zisetyenziselwa ukuqikelela okanye ukuguquguquka okuchazayo ukuqikelela ukuguquguquka okujoliswe kuyo (PTE) .Imodeli iphathwa kwi-RStudio usebenzisa ilayibrari yeephakheji (Kohonen), ilayibrari (caret), ilayibrari (modelr), ilayibrari ("e1071"), ilayibrari ("plyr"), ilayibrari ("plyr"), ilayibrari ("iprospect"). (“IiMetrikhi”).
Iiparameters ezahlukeneyo zokuqinisekisa zisetyenziselwe ukugqiba imodeli efanelekileyo yokuqikelela ukugxininiswa kwe-nickel emhlabeni kunye nokuvavanya ukuchaneka kwemodeli kunye nokuqinisekiswa kwayo.Iimodeli ze-Hybridization zavavanywa kusetyenziswa impazamo echanekileyo (MAE), i-root mean error square (RMSE), kunye ne-R-squared okanye i-coefficient determination (R2) . imodeli.RMSE kunye nobukhulu bentlukwano kumanyathelo azimeleyo achaza amandla okuqikelelwa komzekelo, ngelixa i-MAE inquma ixabiso langempela lobungakanani.Ixabiso le-R2 kufuneka libe phezulu ukuvavanya imodeli engcono kakhulu yomxube usebenzisa i-parameters yokuqinisekisa, ixabiso elisondeleyo kwi-1, liphezulu ukuchaneka.Ngokutsho kweLi et al. 59, ixabiso lekhrayitheriya ye-R2 ye-0.75 okanye ngaphezulu ithathwa njengento efanelekileyo yokuqikelela; ukusuka kwi-0.5 ukuya kwi-0.75 imodeli yokusebenza eyamkelekileyo, kwaye ngaphantsi kwe-0.5 ayivumelekanga ukusebenza kwemodeli.Xa ukhetha imodeli usebenzisa i-RMSE kunye ne-MAE yokuqinisekisa iindlela zokuvavanya iindlela zokuvavanya, amaxabiso aphantsi afunyenweyo anele kwaye athathwa njengeyona nto ingcono kakhulu.I-equation elandelayo ichaza indlela yokuqinisekisa.
apho i-n imele ubungakanani bexabiso elijongiweyo\({Y}_{i}\) imele impendulo elinganisiweyo, kwaye \({\widehat{Y}}_{i}\) ikwamele ixabiso lempendulo eqikelelweyo, ngoko ke, kuqwalaselo lokuqala i.
Iinkcazo zeenkcukacha-manani ze-predictor kunye neenguqu zempendulo zinikezelwe kwiThebhile 1, ebonisa intsingiselo, ukutenxa okusemgangathweni (SD), i-coefficient of variation (CV), ubuncinci, ubuninzi, i-kurtosis, kunye ne-skewness.Ubuncinane kunye nobuninzi bexabiso lezinto ziphantsi komyalelo weMg
Ukulungelelaniswa kwezinto eziguquguqukayo ze-predictor kunye nezinto zokuphendula zibonise ukulungelelaniswa okwanelisayo phakathi kwezinto (jonga uMfanekiso 3) Ulungelelwaniso lubonise ukuba i-CaK ibonise ukulungelelaniswa okuphakathi kunye nexabiso le-r = 0.53, njengoko kwenza i-CaNi. Nangona i-Ca kunye no-K bebonisa ubudlelwane obuthobekileyo kunye nomnye, abaphandi abafana no-Kingston et al. I-68 kunye ne-Santo69 zibonisa ukuba amanqanaba abo emhlabeni aphikisana ngokungafaniyo.Nangona kunjalo, i-Ca kunye ne-Mg iphikisana ne-K, kodwa i-CaK idibanisa kakuhle.Oku kusenokuba ngenxa yokusetyenziswa kwezichumisi ezifana ne-potassium carbonate, eyi-56% ephezulu kwi-potassium.I-Potassium yayihambelana ngokuphakathi kunye ne-magnesium (KM6 3, i-potassium kufuphi, i-industry ye-potassium . i-magnesium sulfate, i-potassium magnesium nitrate, kunye ne-potash zisetyenziswa kwimihlaba ukwandisa amanqanaba abo okunqongophala.I-Nickel ihambelana ngokuphakathi kunye ne-Ca, K kunye ne-Mg kunye nexabiso le-r = 0.52, 0.63 kunye ne-0.55, ngokulandelanayo.Ubudlelwane obubandakanya i-calcium, i-magnesium, kunye ne-PTEs ezifana ne-calcium nickel, i-calcium nickel, i-calcium nickel, i-calcium nickel kunciphisa iziphumo ze-magnesium engaphezulu, kwaye zombini i-magnesium kunye ne-calcium zinciphisa iziphumo ezinobuthi ze-nickel emhlabeni.
I-matrix yokulungelelanisa izinto ezibonisa ubudlelwane phakathi kwee-predictors kunye neempendulo (Qaphela: lo mfanekiso uquka i-scatterplot phakathi kwezinto, amanqanaba okubaluleka asekelwe kwi-p <0,001).
Umzobo 4 ubonisa ukusasazwa kwendawo yezinto.Ngokutsho kukaBurgos et al70, ukusetyenziswa konikezelo lwesithuba bubuchule obusetyenziselwa ukulinganisa nokuqaqambisa iindawo ezishushu kwiindawo ezingcolisekileyo.Amanqanaba okutyebisa eCa kumfanekiso wesi-4 angabonwa kumntla-ntshona wemephu yokusasazwa kwendawo.Umfanekiso ubonisa ukutyetyiswa kwekhalsiyam ephakathi ukuya kwintshona yeCap. kungenzeka ngenxa yokusetyenziswa kwe-quicklime (i-calcium oxide) ukunciphisa ubumuncu bomhlaba kunye nokusetyenziswa kwayo kwizixhobo zentsimbi njengeoksijini yealkaline kwinkqubo yokwenza intsimbi.Kwelinye icala, amanye amafama akhetha ukusebenzisa i-calcium hydroxide kwimihlaba ene-acidic ukuze ingathathi hlangothi i-pH, ekwanyusa isiqulatho se-calcium kumhlaba71.I-Potassium ikwabonisa iindawo ezishushu kumntla-ntshona kunye nomntla-ntshona woluntu lwe-potaziyamu oluphakathi kwimodethi yezolimo. inokuba ngenxa ye-NPK kunye nezicelo ze-potash.Oku kuhambelana nezinye izifundo, ezifana ne-Madaras kunye ne-Lipavský72, i-Madaras et al.73, i-Pulkrabová et al.74, i-Asare et al.75, eyaqaphela ukuba ukuzinza komhlaba kunye nonyango kunye ne-KCl kunye ne-NPK kubangele umxholo ophezulu we-K emhlabeni. Ukutyetyiswa kwe-Spatial Potassium kumntla-ntshona wemephu yokuhanjiswa kunokuba ngenxa yokusetyenziswa kwezichumisi ezisekelwe kwi-potassium ezifana ne-potassium chloride, i-potassium sulfate, i-nitrate ye-potassium, i-potash, kunye ne-potash ukwandisa umxholo we-potassium wemihlaba ehluphekileyo.Zádorová et al. 76 kunye noTlustoš et al. I-77 ichaze ukuba ukusetyenziswa kwezichumisi ezisekelwe kwi-K kwandisa umxholo we-K emhlabeni kwaye kuya kwandisa kakhulu isiqulatho sesondlo somhlaba ekuhambeni kwexesha, ngakumbi i-K kunye ne-Mg ebonisa indawo eshushu emhlabeni. I-chlorosis. Izichumisi ezisekelwe kwi-magnesium, ezifana ne-potassium sulfate, i-magnesium sulfate, kunye ne-Kieserite, zinyanga iintsilelo (izityalo zibonakala zimfusa, zibomvu, okanye zimdaka, zibonisa ukunqongophala kwe-magnesium) kwimihlaba enoluhlu oluqhelekileyo lwe-pH6. imveliso78.
Usasazo lwendawo yezinto [imephu yonikezelo lwesithuba yenziwe kusetyenziswa i-ArcGIS Desktop (ESRI, Inc, Version 10.7, URL: https://desktop.arcgis.com).]
Iziphumo zesalathisi sokusebenza komzekelo kwizinto ezisetyenzisiweyo kolu phononongo ziboniswe kwiThebhile 2. Ngakolunye uhlangothi, i-RMSE kunye ne-MAE ye-Ni zombini zisondele kwi-zero (0.86 RMSE, -0.08 MAE) . Ngakolunye uhlangothi, zombini i-RMSE kunye ne-MAE ixabiso le-K liyamkeleka. I-RMSE kunye ne-MAE iziphumo zazinkulu kwi-calcium kunye ne-magnesium esetyenzisiweyo kunye neziphumo zedatha ye-KSE kunye ne-K. I-RMSE kunye ne-MAE yolu phononongo usebenzisa i-EBK ukuqikelela i-Ni yafunyanwa ingcono kuneziphumo zikaJohn et al. 54 usebenzisa i-synergistic kriging ukuqikelela ugxininiso lwe-S emhlabeni usebenzisa idatha eqokelelweyo efanayo. Iziphumo ze-EBK esizifundileyo zihambelana nezo zikaFabijaczyk et al. 41, Yan et al. 79, Beguin et al. 80, uAdhikary et al. 81 kunye noYohane et al. 82, ingakumbi uK noNi.
Ukusebenza kweendlela zomntu ngamnye zokuqikelela umxholo we-nickel kwimihlaba yasezidolophini kunye ne-peri-urban yavavanywa kusetyenziswa ukusebenza kweemodeli (Itheyibhile 3).Ukuqinisekiswa kwemodeli kunye novavanyo oluchanekileyo luqinisekisile ukuba i-Ca_Mg_K ixela kwangaphambili idityaniswe nemodeli ye-EBK SVMR ivelise ukusebenza kakuhle.Umzekelo we-Calibration we-square we-square kunye nomzekelo we-square we-RM2 imodeli ye-square ye-square, i-SVM_KRM2 imodeli ye-square ye-RM2 Impazamo epheleleyo (MAE) yayiyi-0.637 (R2), 95.479 mg/kg (RMSE) kunye ne-77.368 mg/kg (MAE) Ca_Mg_K-SVMR yayingu-0.663 (R2), 235.974 mg/kg (RMSE) kunye ne-16kg (lessthewereE) ixabiso le-R2 elifunyenweyo. Ca_Mg_K-SVMR (0.663 mg / kg R2) kunye neCa_Mg-EBK_SVMR (0.643 = R2); iziphumo zabo ze-RMSE kunye ne-MAE zaziphezulu kunezo ze-Ca_Mg_K-EBK_SVMR (R2 0.637) (jonga iThebhile 3) .Ukongezelela, i-RMSE kunye ne-MAE ye-Ca_Mg-EBK_SVMR (RMSE = 1664.64 kunye ne-MAE = 1031.49) imodeli enkulu kunye ne-17.5 enkulu kunye ne-17. Ca_Mg_K-EBK_SVMR.Ngokunjalo, i-RMSE kunye ne-MAE ye-Ca_Mg-K SVMR (RMSE = 235.974 kunye ne-MAE = 166.946) imodeli ziyi-2.5 kunye ne-2.2 ezinkulu kunezo ze-Ca_Mg_K-EBK_SVMR zibonisa ukuba idatha ibekwe njani ngokwe-RMSE kunye ne-MAE. inomgca weyona nto ifanelekileyo.Ephakamileyo RSME kunye ne-MAE zabonwa.Ngokutsho kukaKebonye et al. 46 kunye noYohane et al. 54, okukhona i-RMSE kunye ne-MAE zisondela kuqanda, kokukhona iziphumo ziba ngcono.SVMR kunye ne-EBK_SVMR zinamaxabiso aphezulu e-RSME kunye ne-MAE.Kwaphawulwa ukuba uqikelelo lwe-RSME lwaluthe gqolo luphezulu kunamaxabiso e-MAE, ebonisa ubukho babangaphandle.Ngokutsho kweLegates, i-McCabso ye-McCabso egqithe kwi-8 imposiso ye-McCabe3 (i-MAE) iyanconywa njengesalathisi sobukho bezinto ezingaphandle.Oku kuthetha ukuba ubuninzi beesethi zedatha, ukuphakama kwe-MAE kunye ne-RMSE ixabiso.Ukuchaneka kovavanyo lokuqinisekiswa kwe-cross-validation ye-Ca_Mg_K-EBK_SVMR imodeli exubeneyo yokuqikelela umxholo we-Ni kwimihlaba yasezidolophini nakwi-suburban yayiyi-63.70 ye-alccord ukuya kwi-Etccord. I-59, eli nqanaba lokuchaneka liyimodeli yokusebenza eyamkelekileyo.Iziphumo ezikhoyo zifaniswa nesifundo sangaphambili sikaTarasov et al. I-36 imodeli yayo ye-hybride yakha i-MLPRK (i-Multilayer Perceptron Residual Kriging), ehambelana ne-EBK_SVMR yokuvavanya ukuchaneka kwesalathisi esichazwe kwisifundo sangoku, i-RMSE (210) kunye ne-MAE (167.5) yayiphezulu kuneziphumo zethu kwisifundo samanje (RMSE 95.479, MAE68 i-R2 yangoku) . (0.637) naleyo kaTarasov et al. I-36 (0.544), kucacile ukuba i-coefficient of determination (R2) iphezulu kule modeli edibeneyo. Umda wephutha (RMSE kunye ne-MAE) (EBK SVMR) imodeli edibeneyo iphantsi kabini. Ngokufanayo, uSergeev et al.34 urekhode i-0.28 (R2) ye-hybrid model ephuhlisiwe kwi-record imodeli ye-Recordron ye-Recording yangoku, i-Multila irekhodi ye-Recording ye-Recording yangoku 0.637 (R2) .Inqanaba lokuchaneka kokuchaneka kwalo mzekelo (EBK SVMR) yi-63.7%, ngelixa ukuchaneka kokuchaneka okufunyenwe nguSergeev et al. I-34 yi-28%.Imephu yokugqibela (umzobo 5) idalwe kusetyenziswa imodeli ye-EBK_SVMR kunye ne-Ca_Mg_K njenge-predictor ibonisa ukuqikelelwa kweendawo ezishushu kunye ne-moderate to nickel kuyo yonke indawo yokufunda.Oku kuthetha ukuba ukugxininiswa kwe-nickel kwindawo yophando ikakhulu kumodareyitha, kunye noxinzelelo oluphezulu kwezinye iindawo ezithile.
Imephu yoqikelelo lokugqibela imelwe kusetyenziswa imodeli engumxube EBK_SVMR kwaye kusetyenziswa iCa_Mg_K njengoqikelelo.[Imephu yonikezelo yendawo yenziwe kusetyenziswa iRStudio (uguqulelo 1.4.1717: https://www.rstudio.com/).]
Iboniswe kwi-Figure 6 yi-PTE yogxininiso njengendiza yokuqulunqa equka i-neurons nganye.Akukho nanye yeeplani zecandelo elibonisa ipateni yombala ofanayo njengoko kubonisiwe.Nangona kunjalo, inani elifanelekileyo le-neurons kwimephu ezotyiweyo yi-55.SeOM iveliswa kusetyenziswa imibala eyahlukahlukeneyo, kunye neepatheni zombala ezifanayo, ukuthelekisa ngakumbi iipropati zeesampuli, ngokwe-MCale, ngokwe-MCale element, ngokwe-MCa, umbala kunye ne-KCale, ngokuhambelana nombala wabo. Iipateni zombala ezifanayo kwi-neurons ephezulu enye kunye ne-neurons ephantsi kakhulu.Ngoko, i-CaK kunye ne-CaMg zabelana ngokufana kunye ne-neurons ephezulu kakhulu kunye neepatheni zemibala ephantsi ukuya kwimodareyitha.Zombini iimodeli ziqikelela ukuxinana kwe-Ni emhlabeni ngokubonisa imibala ephakathi ukuya phezulu yemibala enjengobomvu, i-orenji kunye nephuzi. ipateni yamacandelo emodeli ibonise ipateni ephezulu yombala ebonisa ukuxinwa kwe-nickel emhlabeni (jonga umfanekiso 4) .I-CakMg imodeli yecandelo lendiza ibonisa ipateni yombala ohlukeneyo ukusuka kwi-low ukuya phezulu ngokubhekiselele kwisikali sombala ochanekileyo.Ngaphezu koko, ukubikezelwa komzekelo we-nickel umxholo (i-CakMg) ifana nokusabalalisa kwendawo ye-nickel ephakathi kunye ne-prografu ephantsi ye-nickel ebonisiweyo. Ukugxininiswa kwenickel kumhlaba wasezidolophini nasemdeni wedolophu.Umfanekiso wesi-7 ubonisa indlela yecontour kumaqela e-k-means kwimephu, yahlulwe yangamaqela amathathu ngokusekelwe kwixabiso eliqikelelweyo kumzekelo ngamnye.Indlela yecontour imele elona nani liphezulu lamaqela.Kwiisampuli ezili-115 eziqokelelweyo, udidi loku-1 lufumene uninzi lweisampulu ezi-2 zomhlaba, i-c3luster efumene uninzi lwesampulu ezi-2, i-c3luster efunyenweyo I-3 ifumene iisampulu ze-8. I-7-component planar predictor indibaniselwano yenziwe lula ukuvumela ukutolikwa okuchanekileyo kweqela.Ngenxa yeenkqubo ezininzi ze-anthropogenic kunye nezendalo ezichaphazela ukubunjwa komhlaba, kunzima ukwahlula ngokufanelekileyo iipateni ze-cluster kwi-SeOM map78 esasazwayo.
Imveliso yenqwelomoya yecandelo ngumatshini ngamnye we-Empirical Bayesian Kriging Support Vector Machine (EBK_SVM_SeOM) oguquguqukayo.[Iimephu zeSeOM zenziwe kusetyenziswa iRStudio (uguqulelo 1.4.1717: https://www.rstudio.com/).]
Amacandelo ahlukeneyo okuhlelwa kweqela [iimephu zeSeOM zenziwe kusetyenziswa i-RStudio (uguqulelo 1.4.1717: https://www.rstudio.com/).]
Uphononongo lwangoku lubonisa ngokucacileyo ubuchule bobuchwephesha bokugxininiswa kwenickel kumhlaba wasezidolophini nasekupheleni kwedolophu.Uphononongo luvavanye iindlela ezahlukeneyo zokubonisa, ukudibanisa izinto ezinobuchwephesha bomzekelo, ukufumana eyona ndlela ingcono yokuqikelela ugxininiso lwe-nickel emhlabeni.Iimpawu ze-SeOM zeplani yeplani yomhlaba yendlela yokwenza imodeli ibonise ipateni yombala ophezulu ukusuka ezantsi ukuya phezulu kwisikali sombala we-Nickel esichanekileyo iqinisekisa ukuhanjiswa kwendawo okucwangcisiweyo kwamacandelo aboniswe yi-EBK_SVMR (jonga uMfanekiso 5) .Iziphumo zibonisa ukuba imodeli yokubuyisela umatshini wevector yenkxaso (Ca Mg K-SVMR) iqikelela ukuxinana kwe-Ni emhlabeni njengomzekelo omnye, kodwa ukuqinisekiswa kunye nokuchaneka kweeparameters kubonisa iimpazamo eziphezulu kakhulu malunga ne-RMSE yendlela yokulinganisa kunye ne-MAE yesandla kunye ne-maE. Imodeli ye-EBK_MLR nayo iphosakele ngenxa yexabiso eliphantsi le-coefficient of determination (R2) .Iziphumo ezilungileyo zifunyenwe ngokusebenzisa i-EBK SVMR kunye nezinto ezidibeneyo (CaKMg) kunye ne-RMSE ephantsi kunye neempazamo ze-MAE kunye nokuchaneka kwe-63.7%.Kuvela ukuba ukudibanisa i-algorithm ye-EBK kunye ne-algorithm ye-hybrid yoxinaniso inokuvelisa i-algorithm ye-algorithm yokucoca umatshini inokuvelisa i-algorithm yokufunda ngomatshini. usebenzisa i-Ca Mg K njengoqikelelo lokuqikelela ugxininiso lwe-Ni kummandla wophononongo lunokuphucula ukuqikelelwa kwe-Ni kwimihlaba.Oku kuthetha ukuba ukusetyenziswa okuqhubekayo kwezichumisi ezisekelwe kwi-nickel kunye nongcoliseko lwemizi-mveliso yomhlaba ngoshishino lwentsimbi lunotyekelo lokunyusa ukuxinana kwe-nickel emhlabeni.Olu phononongo lubonise ukuba imodeli ye-EBK inokunciphisa inqanaba lempazamo kwaye iphucule ukuchaneka komgangatho womhlaba wedolophu-kwimodeli yomhlaba. ngokubanzi, sicebisa ukusebenzisa imodeli ye-EBK-SVMR ukuvavanya nokuqikelela i-PTE emhlabeni; ukongeza, sicebisa ukusebenzisa i-EBK ukudibanisa ngeendlela ezahlukeneyo zokufunda koomatshini. nangona kunjalo, ukusebenzisa i-covariates ezininzi kuya kuphucula kakhulu ukusebenza komzekelo, onokuthi kuthathelwe ingqalelo umda womsebenzi wangoku.Omnye umda wolu phononongo kukuba inani leedatha liyi-115.Ngoko ke, ukuba idatha eninzi inikezelwe, ukusebenza kwendlela ecetywayo ephuculweyo ye-hybridization inokuphuculwa.
I-PlantProbs.net.Nickel kwiZityalo nakuMhlaba https://plantprobs.net/plant/nutrientImbalances/sodium.html (Ifikeleleke ngowama-28 kuTshazimpuzi 2021).
I-Kasprzak, i-Nickel ye-KS iqhubela phambili kwi-toxicology yendalo yanamhlanje.surroundings.toxicology.11, 145-183 (1987).
I-Cempel, M. & Nikel, G. Nickel: Ukuhlaziywa kwemithombo yayo kunye ne-toxicology yendalo.I-Polish J. Environment.Stud.15, 375-382 (2006).
Freedman, B. & Hutchinson, TC Igalelo elingcolisayo elivela kwiatmosfera kunye nokuqokelelana emhlabeni kunye nezityalo kufutshane nesinyibilikisi senickel-copper eSudbury, Ontario, Canada.can.J. Bot.58(1), 108-132.https://doi.org/10.1139/b80-014 (1980).
Manyiwa, T. et al. Isinyithi esinzima emhlabeni, izityalo kunye nemingcipheko eyayanyaniswa nezilwanyana ezetyisayo kufutshane nomgodi wenickel weSelebi-Phikwe eBotswana.surroundings.Geochemistry.Health https://doi.org/10.1007/s10653-021-00918-x (2021).
Cabata-Pendias.Kabata-Pendias A. 2011. Landela izakhi emhlabeni kunye… – Isifundiswa sikaGoogle https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=Kabata-Pendias+A.+2011.+Trace+ Izinto+kwimihlaba+kunye+nezityalo+4th+ed.+New+York+%28NY%29%3A+CRC+Press&btnG= (Ifikeleleke nge-24 Nov 2020).
Almås, A., Singh, B., Agriculture, TS-NJ of & 1995, undefined.Iziphumo zoshishino lwenickel yaseRashiya kugxininiso lwentsimbi enzima kumhlaba wezolimo kunye nengca eSoer-Varanger, Norway.agris.fao.org.
U-Nielsen, GD et al.Nickel ukufunxa kunye nokugcinwa kwamanzi okusela kuhambelana nokutya kunye ne-nickel sensitivity.toxicology.application.Pharmacodynamics.154, 67-75 (1999).
I-Costa, M. & Klein, i-CB Nickel carcinogenesis, ukuguqulwa, i-epigenetics okanye i-electro.surroundings.I-Health Perspective.107, 2 (1999).
iAjman, iPC; iAjado, SK; Borůvka, L.; IBhini, i-JKM; Sarkody, VYO; Cobonye, NM; Uhlalutyo lwendlela yezinto ezinokuthi zibe yityhefu: uphononongo lwebhayibhiliometri.I-Environmental Geochemistry and Health.Springer Science & Business Media BV 2020.https://doi.org/10.1007/s10653-020-00742-9.
Minasny, B. & McBratney, AB Digital Soil Mapping: Imbali emfutshane kunye neZifundo ezithile.Geoderma 264, 301–311.https://doi.org/10.1016/j.geoderma.2015.07.017 (2016).
McBratney, AB, Mendonça Santos, ML & Minasny, B. Kwimephu yomhlaba wedijithali.Geoderma 117(1-2), 3-52.https://doi.org/10.1016/S0016-7061 (03) 00223-4 (2003).
Deutsch.CV Geostatistical Reservoir Modeling,… – Google Scholar 28 kuTshazimpuzi 2021).
Ixesha lokuposa: Jul-22-2022
