Zikomo poyendera Nature.com. Mtundu wa msakatuli womwe mukugwiritsa ntchito uli ndi chithandizo chochepa cha CSS. Kuti mupeze chidziwitso chabwino, tikukulimbikitsani kuti mugwiritse ntchito msakatuli wosinthidwa (kapena zimitsani mawonekedwe ogwirizana mu Internet Explorer). Pakadali pano, kuti titsimikizire kuti chithandizo chikupitilizabe, tidzawonetsa tsamba lopanda masitayelo ndi JavaScript.
Kuipitsidwa kwa nthaka ndi vuto lalikulu lomwe limayambitsidwa ndi zochita za anthu. Kufalikira kwa zinthu zomwe zingakhale zoopsa (PTEs) kumasiyana m'madera ambiri akumatauni ndi m'mphepete mwa mizinda. Chifukwa chake, n'kovuta kuneneratu kuchuluka kwa PTEs m'nthaka zotere. Zitsanzo 115 zonse zinapezeka kuchokera ku Frydek Mistek ku Czech Republic. Kuchuluka kwa Calcium (Ca), magnesium (Mg), potassium (K) ndi nickel (Ni) kunapezeka pogwiritsa ntchito inductively combined plasma emission spectrometry. Response variable ndi Ni ndipo zoneneratu ndi Ca, Mg, ndi K. Matrix yolumikizana pakati pa response variable ndi predictor variable ikuwonetsa mgwirizano wokwanira pakati pa zinthuzo. Zotsatira za kulosera zikuwonetsa kuti Support Vector Machine Regression (SVMR) idachita bwino, ngakhale kuti estimated root mean square error (RMSE) (235.974 mg/kg) ndi mean absolute error (MAE) (166.946 mg/kg) zinali zapamwamba kuposa njira zina zomwe zidagwiritsidwa ntchito. Mitundu yosakanikirana ya Empirical Bayesian Kriging-Multiple Linear Regression (EBK-MLR) imagwira ntchito. Zoyipa, monga momwe zasonyezedwera ndi ma coefficients of determination osakwana 0.1. Chitsanzo cha Empirical Bayesian Kriging-Support Vector Machine Regression (EBK-SVMR) chinali chitsanzo chabwino kwambiri, chokhala ndi ma RMSE otsika (95.479 mg/kg) ndi MAE (77.368 mg/kg) ​​ndi ma coefficient apamwamba otsimikiza (R2 = 0.637). Zotsatira za njira yopangira ma modeling ya EBK-SVMR zikuwonetsedwa pogwiritsa ntchito mapu odzikonzera okha. Ma neuron ozungulira mu ndege ya gawo la hybrid model CakMg-EBK-SVMR akuwonetsa mitundu yambiri yomwe imaneneratu kuchuluka kwa Ni m'nthaka za m'mizinda ndi m'madera a m'mizinda. Zotsatira zake zikuwonetsa kuti kuphatikiza EBK ndi SVMR ndi njira yothandiza yoneneratu kuchuluka kwa Ni m'nthaka za m'mizinda ndi m'madera a m'mizinda.
Nickel (Ni) imaonedwa kuti ndi michere ya zomera chifukwa imathandizira kukhazikika kwa nayitrogeni mumlengalenga (N) ndi kagayidwe ka urea, zonse ziwiri zomwe zimafunika kuti mbewu zimere. Kuphatikiza pa kuthandiza kwake kumera kwa mbewu, Ni ikhoza kugwira ntchito ngati choletsa bowa ndi mabakiteriya ndikulimbikitsa kukula kwa zomera. Kusowa kwa nickel m'nthaka kumalola chomera kuyamwa, zomwe zimapangitsa kuti masamba azikula. Mwachitsanzo, nyemba zobiriwira ndi nyemba zobiriwira zimafuna kugwiritsa ntchito feteleza zochokera ku nickel kuti ziwonjezere kukhazikika kwa nayitrogeni 2. Kupitiliza kugwiritsa ntchito feteleza zochokera ku nickel kuti ziwonjezere nthaka ndikuwonjezera mphamvu ya nyemba kuti zikhazikitse nayitrogeni m'nthaka nthawi zonse kumawonjezera kuchuluka kwa nickel m'nthaka. Ngakhale nickel ndi michere ya zomera, kudya kwake mopitirira muyeso m'nthaka kungapangitse kuvulaza kwambiri kuposa zabwino. Kuopsa kwa nickel m'nthaka kumachepetsa pH ya nthaka ndikulepheretsa kutenga chitsulo ngati michere yofunika kwambiri pakukula kwa zomera 1. Malinga ndi Liu3, Ni yapezeka kuti ndi chinthu cha 17 chofunikira pakukula ndi kukula kwa zomera. Kuphatikiza pa ntchito ya nickel pakukula ndi kukula kwa zomera, anthu amafunikira kuti agwiritse ntchito zosiyanasiyana. Kutulutsa magetsi, kupanga nickel Ma alloy, komanso kupanga zida zoyatsira moto ndi ma spark plugs mumakampani opanga magalimoto zonse zimafuna kugwiritsa ntchito nickel m'magawo osiyanasiyana amafakitale. Kuphatikiza apo, ma alloy okhala ndi nickel ndi zinthu zopangidwa ndi ma electroplated akhala akugwiritsidwa ntchito kwambiri mu ziwiya za kukhitchini, zowonjezera za ballroom, zinthu zamafakitale azakudya, zamagetsi, waya ndi zingwe, ma jet turbines, ma implants opangira opaleshoni, nsalu, ndi zomangamanga za sitima5. Kuchuluka kwa Ni m'nthaka (monga, nthaka ya pamwamba) kwanenedwa kuti kumachitika chifukwa cha zinthu zachilengedwe komanso zachilengedwe, koma makamaka, Ni ndi gwero lachilengedwe osati la anthu4,6. Magwero achilengedwe a nickel akuphatikizapo kuphulika kwa mapiri, zomera, moto wa m'nkhalango, ndi njira za geological; komabe, magwero a anthu amaphatikizapo mabatire a nickel/cadmium mumakampani opanga zitsulo, electroplating, arc welding, dizilo ndi mafuta, ndi mpweya wochokera mumlengalenga wochokera ku kuyaka kwa malasha ndi zinyalala ndi matope. Kuchuluka kwa Nickel7,8. Malinga ndi Freedman ndi Hutchinson9 ndi Maniwa et al. 10, magwero akuluakulu a kuipitsa nthaka pamwamba pa nthaka m'malo oyandikira ndi omwe ali pafupi ndi malo osungunula ndi migodi yopangidwa ndi nickel-copper. Dothi lapamwamba lozungulira fakitale ya Sudbury nickel-copper ku Canada linali ndi kuipitsidwa kwakukulu kwa nickel pa 26,000 mg/kg11. Mosiyana ndi zimenezi, kuipitsidwa komwe kumachitika chifukwa cha kupanga nickel ku Russia kwapangitsa kuti nthaka ya ku Norway ikhale ndi nickel yambiri.11 Malinga ndi Alms et al. 12, kuchuluka kwa nickel yochotsedwa ya HNO3 m'malo abwino kwambiri olima m'chigawochi (kupanga nickel ku Russia) kunayambira pa 6.25 mpaka 136.88 mg/kg, zomwe zikugwirizana ndi avareji ya 30.43 mg/kg ndi kuchuluka koyambira kwa 25 mg/kg. Malinga ndi kabata 11, kugwiritsa ntchito feteleza wa phosphorous m'nthaka zaulimi m'matauni kapena m'mphepete mwa mizinda nthawi yanyengo yokolola motsatizana kungapangitse nthaka kuipitsidwa. Zotsatira zomwe nickel ingayambitse mwa anthu zitha kubweretsa khansa kudzera mu mutagenesis, kuwonongeka kwa chromosome, kupanga Z-DNA, kukonza DNA excision, kapena njira za epigenetic13. Mu zoyeserera za nyama, nickel yapezeka kuti ingayambitse mitundu yosiyanasiyana ya zotupa, ndipo ma nickel complexes a khansa amatha kukulitsa zotupa zotere.
Kuwunika kuipitsidwa kwa nthaka kwakula posachedwapa chifukwa cha mavuto osiyanasiyana okhudzana ndi thanzi omwe amabwera chifukwa cha ubale wa nthaka ndi zomera, ubale wa nthaka ndi zamoyo, kuwonongeka kwa chilengedwe, komanso kuwunika momwe chilengedwe chimakhudzira. Mpaka pano, kulosera malo a zinthu zomwe zingakhale zoopsa (PTEs) monga Ni mu nthaka kwakhala kovuta komanso kotenga nthawi pogwiritsa ntchito njira zachikhalidwe. Kubwera kwa mapu a nthaka a digito (DSM) ndi kupambana kwake kwamakono15 kwasintha kwambiri mapu a nthaka oneneratu (PSM). Malinga ndi Minasny ndi McBratney16, mapu a nthaka oneneratu (DSM) atsimikizira kukhala gawo lodziwika bwino la sayansi ya nthaka. Lagacherie ndi McBratney, 2006 amafotokoza DSM ngati "kupanga ndi kudzaza machitidwe odziwitsa nthaka m'malo pogwiritsa ntchito njira zowonera m'malo ndi m'ma laboratories komanso machitidwe owunikira nthaka m'malo ndi m'malo".McBratney et al. 17 ikufotokoza kuti DSM kapena PSM yamakono ndiyo njira yothandiza kwambiri yodziwira kapena kulongosola momwe PTEs, mitundu ya nthaka ndi zinthu za nthaka zimagawidwira. Ma Geostatistics ndi Machine Learning Algorithms (MLA) ndi njira za DSM zomwe zimapanga mamapu okhala ndi digito pogwiritsa ntchito makompyuta pogwiritsa ntchito deta yofunika komanso yochepa.
Deutsch18 ndi Olea19 amafotokoza za geostatistics ngati "kusonkhanitsa njira zamanambala zomwe zimagwira ntchito yowonetsera makhalidwe a malo, makamaka pogwiritsa ntchito mitundu ya stochastic, monga momwe kusanthula kwa nthawi kumafotokozera deta ya nthawi." Choyamba, geostatistics imaphatikizapo kuwunika kwa variograms, zomwe zimalola Kuyeza ndikufotokozera kudalira kwa mitengo ya malo kuchokera ku deta iliyonse20.Gumiaux et al. 20 akuwonetsanso kuti kuwunika kwa variograms mu geostatistics kumadalira mfundo zitatu, kuphatikiza (a) kuwerengera kukula kwa kulumikizana kwa deta, (b) kuzindikira ndi kuwerengera anisotropy mu kusiyana kwa deta ndi (c) kuwonjezera pa Kuphatikiza pa kuganizira zolakwika zamkati za deta yoyezera yolekanitsidwa ndi zotsatira zakomweko, zotsatira za dera zimayesedwanso. Pomanga pamalingaliro awa, njira zambiri zolumikizira zimagwiritsidwa ntchito mu geostatistics, kuphatikiza kriging wamba, co-kriging, kriging wamba, empirical Bayesian kriging, njira yosavuta ya kriging ndi njira zina zodziwika bwino zolumikizira kuti ziwonetse kapena kulosera PTE, mawonekedwe a nthaka, ndi mitundu ya nthaka.
Ma Algorithms Ophunzirira Makina (MLA) ndi njira yatsopano yomwe imagwiritsa ntchito magulu akuluakulu a deta osalunjika, olimbikitsidwa ndi ma algorithms omwe amagwiritsidwa ntchito makamaka pofufuza deta, kuzindikira mawonekedwe mu deta, komanso kugwiritsidwa ntchito mobwerezabwereza pakugawa m'magawo asayansi monga sayansi ya nthaka ndi ntchito zobweza. Mapepala ambiri ofufuza amadalira mitundu ya MLA kuti alosere PTE m'nthaka, monga Tan et al. 22 (nkhalango zosawerengeka zowerengera zitsulo zolemera m'nthaka zaulimi), Sakizadeh et al. 23 (kujambula pogwiritsa ntchito makina othandizira ma vector ndi ma network opangira mitsempha) kuipitsa nthaka). Kuphatikiza apo, Vega et al. 24 (CART yowonetsera kusunga ndi kulowetsedwa kwa zitsulo zolemera m'nthaka) Sun et al. 25 (kugwiritsa ntchito cubist ndiko kugawa kwa Cd m'nthaka) ndi ma algorithms ena monga k-pafupi ndi mnansi, generalized boosted regression, ndi boosted regression Mitengo idagwiritsanso ntchito MLA kuti ilosere PTE m'nthaka.
Kugwiritsa ntchito ma algorithm a DSM poneneratu kapena kupanga mapu kumakumana ndi zovuta zingapo. Olemba ambiri amakhulupirira kuti MLA ndi yabwino kuposa geostatistics ndipo mosemphanitsa. Ngakhale chimodzi chili bwino kuposa china, kuphatikiza kwa ziwirizi kumawongolera mulingo wolondola wa kupanga mapu kapena kulosera mu DSM15. Woodcock ndi Gopal26 Finke27; Pontius ndi Cheuk28 ndi Grunwald29 akunena za zofooka ndi zolakwika zina pakujambula mapu a nthaka yoloseredwa. Asayansi a nthaka ayesa njira zosiyanasiyana kuti akonze bwino, kulondola, komanso kulosera kwa mapu a DSM ndi kulosera. Kuphatikiza kusatsimikizika ndi kutsimikizira ndi chimodzi mwazinthu zosiyanasiyana zomwe zimaphatikizidwa mu DSM kuti ziwongolere kugwira ntchito bwino ndikuchepetsa zolakwika. Komabe, Agyeman et al. 15 akufotokoza kuti khalidwe lotsimikizira ndi kusatsimikizika komwe kumayambitsidwa ndi kupanga mapu ndi kulosera kuyenera kutsimikiziridwa paokha kuti kuwongolere khalidwe la mapu. Zofooka za DSM zimachitika chifukwa cha mtundu wa nthaka wofalikira m'malo osiyanasiyana, womwe umaphatikizapo gawo la kusatsimikizika; Komabe, kusowa kwa chitsimikizo mu DSM kungachokere ku magwero angapo a cholakwika, monga cholakwika cha covariate, cholakwika cha chitsanzo, cholakwika cha malo, ndi Cholakwika cha analytical 31. Kuyerekeza zolakwika zomwe zimachitika mu MLA ndi njira za geostatistical zimagwirizanitsidwa ndi kusamvetsetsa, zomwe zimapangitsa kuti njira yeniyeni ikhale yosavuta kwambiri32. Mosasamala kanthu za mtundu wa chitsanzo, zolakwika zitha kuyambitsidwa ndi magawo a chitsanzo, maulosi a chitsanzo cha masamu, kapena interpolation33. Posachedwapa, njira yatsopano ya DSM yawonekera yomwe imalimbikitsa kuphatikiza kwa geostatistics ndi MLA mu mapu ndi kulosera. Asayansi ndi olemba angapo a nthaka, monga Sergeev et al. 34; Subbotina et al. 35; Tarasov et al. 36 ndi Tarasov et al. 37 agwiritsa ntchito bwino mtundu wa geostatistics ndi kuphunzira kwa makina kuti apange mitundu yosakanikirana yomwe imapangitsa kuti ntchito yolosera ndi mapu ikhale yogwira mtima. khalidwe. Zina mwa ma algorithm awa osakanikirana kapena ophatikizana ndi Artificial Neural Network Kriging (ANN-RK), Multilayer Perceptron Residual Kriging (MLP-RK), Generalized Regression Neural Network Residual Kriging (GR-NNRK)36, Artificial Neural Network Kriging-Multilayer Perceptron (ANN-K-MLP)37 ndi Co-Kriging ndi Gaussian Process Regression38.
Malinga ndi Sergeev et al., kuphatikiza njira zosiyanasiyana zochitira chitsanzo kuli ndi kuthekera kochotsa zolakwika ndikuwonjezera magwiridwe antchito a mtundu wosakanizidwa womwe umabwera m'malo mopanga mtundu wake umodzi. Pachifukwa ichi, pepala latsopanoli likunena kuti ndikofunikira kugwiritsa ntchito njira yophatikizana ya geostatistics ndi MLA kuti apange mitundu yabwino kwambiri yosakanikirana kuti ilosere kuchuluka kwa Ni m'mizinda ndi m'madera ozungulira mizinda. Kafukufukuyu adzadalira Empirical Bayesian Kriging (EBK) ngati chitsanzo choyambira ndikusakaniza ndi Support Vector Machine (SVM) ndi Multiple Linear Regression (MLR) mitundu. Kusakaniza kwa EBK ndi MLA iliyonse sikudziwika. Mitundu yosiyanasiyana yosakanikirana yomwe ikuwoneka ndi kuphatikiza kwa wamba, wotsalira, regression kriging, ndipo MLA.EBK ndi njira yolumikizira geostatistical yomwe imagwiritsa ntchito njira yokhazikika yomwe imayikidwa ngati munda wosakhazikika/wokhazikika wokhala ndi magawo odziwika bwino a malo pamunda, kulola kusintha kwa malo39.EBK yagwiritsidwa ntchito m'maphunziro osiyanasiyana, kuphatikizapo kusanthula kufalikira kwa kaboni wachilengedwe m'nthaka ya famu40, kuwunika kuipitsidwa kwa nthaka41 ndi kupanga mapu a nthaka. katundu42.
Kumbali inayi, Self-Organizing Graph (SeOM) ndi njira yophunzirira yomwe yagwiritsidwa ntchito m'nkhani zosiyanasiyana monga Li et al. 43, Wang et al. 44, Hossain Bhuiyan et al. 45 ndi Kebonye et al. 46 Dziwani makhalidwe a malo ndi magulu a zinthu. Wang et al. 44 akufotokoza kuti SeOM ndi njira yamphamvu yophunzirira yomwe imadziwika ndi kuthekera kwake kophatikiza magulu ndi kulingalira mavuto osakhala a mzere. Mosiyana ndi njira zina zozindikira mapangidwe monga kusanthula kwakukulu kwa zigawo, kusonkhanitsa zinthu mopanda nzeru, kusonkhanitsa zinthu motsatira malamulo, komanso kupanga zisankho zambiri, SeOM ndi yabwino kwambiri pakukonza ndi kuzindikira mapangidwe a PTE. Malinga ndi Wang et al. 44, SeOM imatha kugawa magawo a ma neuron okhudzana ndi malo ndikupereka mawonekedwe apamwamba a deta. SeOM idzawonetsa deta yolosera ya Ni kuti ipeze chitsanzo chabwino kwambiri chofotokozera zotsatira kuti imasuliridwe mwachindunji.
Pepala ili likufuna kupanga chitsanzo cholimba cha mapu cholondola kwambiri cholosera kuchuluka kwa nickel m'matauni ndi m'madera akutali. Tikuganiza kuti kudalirika kwa chitsanzo chosakanikirana kumadalira kwambiri mphamvu ya mitundu ina yolumikizidwa ku chitsanzo choyambira. Tikuvomereza mavuto omwe DSM ikukumana nawo, ndipo ngakhale mavutowa akuyankhidwa m'mbali zosiyanasiyana, kuphatikiza kwa kupita patsogolo kwa geostatistics ndi mitundu ya MLA kukuwoneka kuti kukuchulukirachulukira; chifukwa chake, tiyesa kuyankha mafunso ofufuza omwe angapereke mitundu yosakanikirana. Komabe, kodi chitsanzochi ndi cholondola bwanji poneneratu chinthu chomwe chikufunidwa? Komanso, kodi kuwunika kogwira ntchito bwino kutengera kutsimikizika ndi kuwunika kolondola ndi kotani? Chifukwa chake, zolinga zenizeni za kafukufukuyu zinali (a) kupanga chitsanzo chosakanikirana cha SVMR kapena MLR pogwiritsa ntchito EBK ngati chitsanzo choyambira, (b) kuyerekeza mitundu yomwe yatsatira (c) kupereka chitsanzo chabwino kwambiri chosakanikirana cholosera kuchuluka kwa Ni m'matauni kapena m'madera akutali, ndi (d) kugwiritsa ntchito SeOM kuti apange mapu apamwamba a kusiyana kwa malo a nickel.
Kafukufukuyu akuchitika ku Czech Republic, makamaka m'chigawo cha Frydek Mistek m'chigawo cha Moravia-Silesian (onani Chithunzi 1). Malo a malo ophunzirirawa ndi olimba kwambiri ndipo nthawi zambiri ndi gawo la dera la Moravia-Silesian Beskidy, lomwe ndi gawo la m'mphepete mwa mapiri a Carpathian. Malo ophunzirirawa ali pakati pa 49° 41′ 0′ N ndi 18° 20′ 0′ E, ndipo kutalika kwake kuli pakati pa 225 ndi 327 m; Komabe, dongosolo la Koppen logawa nyengo m'derali limawerengedwa kuti Cfb = nyengo yofunda ya m'nyanja, Pali mvula yambiri ngakhale m'miyezi youma. Kutentha kumasiyana pang'ono chaka chonse pakati pa −5 °C ndi 24 °C, nthawi zambiri kumatsika pansi pa −14 °C kapena kupitirira 30 °C, pomwe mvula yapakati pachaka imakhala pakati pa 685 ndi 752 mm47. Malo oyerekeza ofufuza a dera lonselo ndi 1,208 masikweya kilomita, ndi 39.38% ya malo olimidwa ndi 49.36% ya nkhalango. Kumbali ina, dera lomwe lagwiritsidwa ntchito mu kafukufukuyu ndi pafupifupi 889.8 masikweya kilomita. Mu Ostrava ndi mozungulira, makampani opanga zitsulo ndi ntchito zachitsulo ndi otanganidwa kwambiri. Zitsulo, makampani opanga zitsulo komwe nickel imagwiritsidwa ntchito mu zitsulo zosapanga dzimbiri (monga kukana dzimbiri mumlengalenga) ndi zitsulo za alloy (nickel imawonjezera mphamvu ya alloy pamene ikusunga ductility yake yabwino komanso kulimba), ndi ulimi wovuta monga kugwiritsa ntchito feteleza wa phosphate ndi kupanga ziweto ndi magwero ofufuza omwe angayambitse nickel mu dera (monga kuwonjezera nickel ku ana a nkhosa kuti awonjezere kuchuluka kwa kukula kwa ana a nkhosa ndi ng'ombe zosadya bwino). Ntchito zina za nickel m'mafakitale m'magawo ofufuza zimaphatikizapo kugwiritsa ntchito kwake popangira ma electroplating, kuphatikizapo nickel yopangira ma electroplating ndi njira zopangira ma nickel opanda ma electroless. Makhalidwe a nthaka amatha kusiyanasiyana mosavuta ndi mtundu wa nthaka, kapangidwe kake, ndi kuchuluka kwa carbonate. Kapangidwe ka nthaka ndi kapakati mpaka kosalala, kochokera ku zinthu zomwe zimapangidwa ndi makolo ake. Ndi zachilengedwe za colluvial, alluvial kapena aeolian. Madera ena a nthaka amawoneka ngati otuwa pamwamba ndi pansi pa nthaka, nthawi zambiri okhala ndi konkriti ndi bleaching. Komabe, cambisols ndi stagnosols ndi mitundu yofala kwambiri ya nthaka m'derali48. Ndi kutalika kuyambira 455.1 mpaka 493.5 m, cambisols amalamulira Czech Republic49.
Mapu a malo ophunzirira [Mapu a malo ophunzirira adapangidwa pogwiritsa ntchito ArcGIS Desktop (ESRI, Inc, mtundu 10.7, URL: https://desktop.arcgis.com).]
Zitsanzo zonse 115 za nthaka yapamwamba zinapezeka kuchokera ku dothi la m'mizinda ndi m'madera ozungulira tawuni m'chigawo cha Frydek Mistek. Chitsanzo chomwe chinagwiritsidwa ntchito chinali gridi yokhazikika yokhala ndi zitsanzo za nthaka zomwe zinali kutali ndi 2 × 2 km, ndipo nthaka yapamwamba inayesedwa pa kuya kwa 0 mpaka 20 cm pogwiritsa ntchito chipangizo cha GPS chogwiridwa ndi manja (Leica Zeno 5 GPS). Zitsanzo zimapakidwa m'matumba a Ziploc, zolembedwa bwino, ndikutumizidwa ku labotale. Zitsanzozo zinaumitsidwa ndi mpweya kuti zipange zitsanzo zopukutidwa, zopukutidwa ndi makina (Fritsch disc mill), ndikusefedwa (kukula kwa sieve 2 mm). Ikani 1 gramu ya zitsanzo za nthaka zouma, zosakanikirana komanso zosefedwa m'mabotolo a teflon olembedwa bwino. Mu chotengera chilichonse cha Teflon, perekani 7 ml ya 35% HCl ndi 3 ml ya 65% HNO3 (pogwiritsa ntchito chopatsira chokha - chimodzi pa asidi iliyonse), phimbani pang'ono ndikulola zitsanzozo kuti ziyime usiku wonse kuti zichitike (pulogalamu ya aqua regia). Ikani supernatant pa mbale yotentha yachitsulo (kutentha: 100 W ndi 160 °C) kwa maola awiri kuti muchepetse kugaya kwa zitsanzo, kenako muziziziritse. Samutsani supernatant ku botolo la volumetric la 50 ml ndikuchepetsa mpaka 50 ml ndi madzi oyeretsedwa. Pambuyo pake, sefa supernatant woyeretsedwayo mu chubu cha PVC cha 50 ml ndi madzi oyeretsedwa. Kuphatikiza apo, 1 ml ya yankho loyeretsedwayo idachepetsedwa ndi 9 ml ya madzi oyeretsedwa ndikusefedwa mu chubu cha 12 ml chokonzedwa kuti chikhale ndi PTE pseudo-concentration. Kuchuluka kwa PTEs (As, Cd, Cr, Cu, Mn, Ni, Pb, Zn, Ca, Mg, K) kudatsimikiziridwa ndi ICP-OES (Inductively Coupled Plasma Optical Emission Spectroscopy) (Thermo Fisher Scientific, USA) malinga ndi njira zokhazikika ndi mgwirizano. Onetsetsani kuti njira zotsimikizira ndi kulamulira (QA/QC) (SRM NIST 2711a Montana II Soil). PTEs zokhala ndi malire ozindikira pansi pa theka sizinaphatikizidwe mu kafukufukuyu. Malire ozindikira a PTE Zomwe zinagwiritsidwa ntchito mu kafukufukuyu zinali 0.0004.(inu). Kuphatikiza apo, njira yowongolera khalidwe ndi kutsimikizira khalidwe pa kusanthula kulikonse zimatsimikiziridwa pofufuza miyezo yowunikira. Kuti zitsimikizire kuti zolakwika zachepetsedwa, kusanthula kawiri kunachitika.
Empirical Bayesian Kriging (EBK) ndi imodzi mwa njira zambiri zolumikizirana za geostatistical zomwe zimagwiritsidwa ntchito popanga zitsanzo m'magawo osiyanasiyana monga sayansi ya nthaka. Mosiyana ndi njira zina zolumikizirana za kriging, EBK imasiyana ndi njira zachikhalidwe za kriging poganizira cholakwika chomwe chimayesedwa ndi semivariogram model. Mu interpolation ya EBK, mitundu ingapo ya semivariogram imawerengedwa panthawi yolumikizirana, m'malo mwa semivariogram imodzi. Njira zolumikizirana zimapangitsa kuti pakhale kusatsimikizika ndi mapulogalamu okhudzana ndi kufotokozera uku kwa semivariogram komwe kumakhala gawo lovuta kwambiri la njira yokwanira ya kriging. Njira yolumikizirana ya EBK ikutsatira njira zitatu zomwe zaperekedwa ndi Krivoruchko50, (a) chitsanzocho chimayesa semivariogram kuchokera ku dataset yolowera (b) mtengo watsopano wonenedweratu wa malo aliwonse a dataset yolowera kutengera semivariogram yopangidwa ndi (c) chitsanzo chomaliza cha A chimawerengedwa kuchokera ku dataset yoyeserera. Lamulo la equation la Bayesian limaperekedwa ngati posterior
Pamene \(Prob\left(A\right)\) ikuyimira mwayi wapambuyo, \(Prob\left(B\right)\) wa marginal probability umanyalanyazidwa nthawi zambiri, \(Prob (B,A)\ ) .Kuwerengera kwa semivariogram kumachokera ku lamulo la Bayes, lomwe limasonyeza kukonda kwa deta yowonera yomwe ingapangidwe kuchokera ku semivariograms. Mtengo wa semivariogram umatsimikiziridwa pogwiritsa ntchito lamulo la Bayes, lomwe limafotokoza momwe zingathere kupanga deta yowonera kuchokera ku semivariogram.
Makina othandizira ma vector ndi njira yophunzirira makina yomwe imapanga hyperplane yolekanitsa bwino kuti isiyanitse magulu ofanana koma osadziyimira pawokha. Vapnik51 adapanga njira yogawa cholinga, koma posachedwapa yagwiritsidwa ntchito kuthetsa mavuto ozungulira. Malinga ndi Li et al.52, SVM ndi imodzi mwa njira zabwino kwambiri zogawa ndipo yagwiritsidwa ntchito m'magawo osiyanasiyana. Gawo lowongolera la SVM (Support Vector Machine Regression - SVMR) linagwiritsidwa ntchito mu kusanthula uku. Cherkassky ndi Mulier53 adayambitsa SVMR ngati njira yowongolera yochokera ku kernel, yomwe kuwerengera kwake kunachitika pogwiritsa ntchito njira yowongolera yolunjika yokhala ndi ntchito za malo ambiri. John et al54 akunena kuti SVMR modeling imagwiritsa ntchito hyperplane linear regression, yomwe imapanga ubale wosagwirizana ndi mzere ndipo imalola ntchito za malo. Malinga ndi Vohland et al. 55, epsilon (ε)-SVMR imagwiritsa ntchito deta yophunzitsidwa kuti ipeze chitsanzo choyimira ngati ntchito yosakhudzidwa ndi epsilon yomwe imagwiritsidwa ntchito pojambula detayo padera ndi epsilon bias yabwino kwambiri kuchokera ku maphunziro pa deta yolumikizidwa. Cholakwika cha mtunda wokonzedweratu chimanyalanyazidwa kuchokera ku mtengo weniweni, ndipo ngati cholakwikacho chili chachikulu kuposa ε(ε), katundu wa nthaka amachilipiritsa. Chitsanzochi chimachepetsanso zovuta za deta yophunzitsira kukhala gulu lalikulu la ma vector othandizira. Equation yomwe yaperekedwa ndi Vapnik51 ikuwonetsedwa pansipa.
pomwe b ikuyimira malire a scalar, \(K\left({x}_{,}{ x}_{k}\right)\) ikuyimira ntchito ya kernel, \(\alpha\) ikuyimira Lagrange multiplier, N ikuyimira seti ya data ya manambala, \({x}_{k}\) ikuyimira kulowetsa deta, ndipo \(y\) ndi kutulutsa deta. Chimodzi mwa ma key kernels omwe amagwiritsidwa ntchito ndi ntchito ya SVMR, yomwe ndi ntchito ya Gaussian radial basis (RBF). RBF kernel imagwiritsidwa ntchito kuti idziwe mtundu woyenera wa SVMR, womwe ndi wofunikira kwambiri kuti mupeze penalty set factor C ndi kernel parameter gamma (γ) ya data yophunzitsira ya PTE. Choyamba, tidayesa seti yophunzitsira kenako tidayesa magwiridwe antchito a chitsanzo pa seti yotsimikizira. Gawo lowongolera lomwe likugwiritsidwa ntchito ndi sigma ndipo mtengo wa njira ndi svmRadial.
Chitsanzo cha regression cha mzere wambiri (MLR) ndi chitsanzo cha regression chomwe chikuyimira ubale pakati pa response variable ndi ma variable angapo odziwira pogwiritsa ntchito ma parameter olumikizidwa omwe amawerengedwa pogwiritsa ntchito njira ya least squares. Mu MLR, chitsanzo cha least squares ndi ntchito yolosera ya katundu wa nthaka pambuyo posankha ma variable ofotokozera. Ndikofunikira kugwiritsa ntchito yankho kukhazikitsa ubale wa mzere pogwiritsa ntchito ma variable ofotokozera. PTE idagwiritsidwa ntchito ngati response variable kukhazikitsa ubale wa mzere ndi ma variable ofotokozera. Equation ya MLR ndi
kumene y ndi yankho losinthika, \(a\) ndi intercept, n ndi chiwerengero cha zoneneratu, \({b}_{1}\) ndi gawo la regression la ma coefficient, \({x}_{ i}\) ikuyimira choneneratu kapena chofotokozera chosinthika, ndipo \({\varepsilon }_{i}\) ikuyimira cholakwika mu chitsanzo, chomwe chimadziwikanso kuti chotsalira.
Mitundu yosakanikirana idapezeka pophatikiza EBK ndi SVMR ndi MLR. Izi zimachitika pochotsa ma values ​​omwe adanenedweratu kuchokera ku interpolation ya EBK. Ma values ​​omwe adanenedweratu omwe adapezeka kuchokera ku interpolation ya Ca, K, ndi Mg omwe adalowetsedwa amapezeka kudzera mu njira yophatikizana kuti apeze ma variable atsopano, monga CaK, CaMg, ndi KMg. Zinthu Ca, K ndi Mg zimaphatikizidwa kuti zipeze variable yachinayi, CaKMg. Ponseponse, ma variable omwe adapezeka ndi Ca, K, Mg, CaK, CaMg, KMg ndi CaKMg. Ma variable awa adakhala ma predictors athu, kuthandiza kulosera kuchuluka kwa nickel m'matauni ndi m'mphepete mwa tawuni. Algorithm ya SVMR idachitidwa pa ma predictors kuti apeze mixed model Empirical Bayesian Kriging-Support Vector Machine (EBK_SVM). Mofananamo, ma variable amayendetsedwanso kudzera mu MLR algorithm kuti apeze mixed model Empirical Bayesian Kriging-Multiple Linear Regression (EBK_MLR). Kawirikawiri, ma variable Ca, K, Mg, CaK, CaMg, KMg, ndi CaKMg zimagwiritsidwa ntchito ngati zinthu zomwe zimasonyeza kuchuluka kwa Ni m'nthaka za m'mizinda ndi m'madera a m'mphepete mwa mizinda. Chitsanzo chovomerezeka kwambiri chomwe chapezeka (EBK_SVM kapena EBK_MLR) chidzawonetsedwa pogwiritsa ntchito graph yodzikonzera yokha. Kayendetsedwe ka ntchito ka kafukufukuyu kakuwonetsedwa pa Chithunzi 2.
Kugwiritsa ntchito SeOM kwakhala chida chodziwika bwino chokonzekera, kuwunika, ndi kulosera deta mu gawo lazachuma, chisamaliro chaumoyo, mafakitale, ziwerengero, sayansi ya nthaka, ndi zina zambiri. SeOM imapangidwa pogwiritsa ntchito maukonde opanga mitsempha ndi njira zophunzirira zosayang'aniridwa pakukonza, kuwunika, ndi kulosera. Mu kafukufukuyu, SeOM idagwiritsidwa ntchito kuwona kuchuluka kwa Ni kutengera chitsanzo chabwino kwambiri cholosera Ni m'nthaka za m'mizinda ndi m'mphepete mwa mizinda. Deta yomwe imagwiritsidwa ntchito mu kuyesa kwa SeOM imagwiritsidwa ntchito ngati n input-dimensional vector variables43,56.Melssen et al. 57 ikufotokoza za kulumikizana kwa vekitala yolowera mu netiweki ya mitsempha kudzera mu gawo limodzi lolowera ku vekitala yotuluka yokhala ndi vekitala yolemera imodzi. Chotuluka chomwe chimapangidwa ndi SeOM ndi mapu okhala ndi magawo awiri okhala ndi ma neuron kapena ma node osiyanasiyana opangidwa mu mamapu a hexagonal, circular, kapena square topological malinga ndi kuyandikira kwawo. Poyerekeza kukula kwa mamapu kutengera metric, quantization error (QE) ndi topographic error (TE), chitsanzo cha SeOM chokhala ndi 0.086 ndi 0.904, motsatana, chimasankhidwa, chomwe ndi gawo la mamapu 55 (5 × 11). Kapangidwe ka neuron kamatsimikiziridwa malinga ndi kuchuluka kwa ma node mu equation yoyeserera.
Chiwerengero cha deta yomwe yagwiritsidwa ntchito mu kafukufukuyu ndi zitsanzo 115. Njira yosankhana mitundu idagwiritsidwa ntchito kugawa detayo m'ma data oyesera (25% yotsimikizira) ndi ma data ophunzitsira (75% yowunikira). Data yophunzitsira imagwiritsidwa ntchito popanga regression model (calibration), ndipo data yoyesera imagwiritsidwa ntchito kutsimikizira kuthekera kwa generalization58. Izi zidachitika kuti ziwunikire kuyenerera kwa mitundu yosiyanasiyana yolosera kuchuluka kwa nickel m'nthaka. Mitundu yonse yomwe idagwiritsidwa ntchito idadutsa njira yotsimikizira ya cross-fold, yobwerezedwa kasanu. Mitundu yopangidwa ndi EBK interpolation imagwiritsidwa ntchito ngati zolosera kapena zosintha zofotokozera kuti zidziwike zomwe zili mu target variable (PTE). Modeling imayendetsedwa mu RStudio pogwiritsa ntchito ma packages library (Kohonen), library (caret), library (modelr), library (“e1071″), library (“plyr”), library (“caTools”), library (“prospectr”) ndi library (“Metrics”).
Ma parameter osiyanasiyana otsimikizira adagwiritsidwa ntchito kuti adziwe chitsanzo chabwino kwambiri choyenera kulosera kuchuluka kwa nickel m'nthaka komanso kuwunika kulondola kwa chitsanzocho ndi kutsimikizira kwake. Ma model a Hybridization adayesedwa pogwiritsa ntchito mean absolute error (MAE), root mean square error (RMSE), ndi R-squared kapena coefficient determination (R2). R2 imafotokoza kusiyana kwa magawo mu yankho, loyimiridwa ndi regression model. RMSE ndi kukula kwa kusiyana mu miyeso yodziyimira payokha zimafotokoza mphamvu yolosera ya chitsanzocho, pomwe MAE imatsimikiza mtengo weniweni wa kuchuluka. Mtengo wa R2 uyenera kukhala wokwera kuti uwonetse chitsanzo chabwino kwambiri chosakaniza pogwiritsa ntchito ma parameter otsimikizira, mtengo ukakhala pafupi ndi 1, kulondola kumakhala kwakukulu. Malinga ndi Li et al. 59, mtengo wa R2 wa 0.75 kapena kupitirira umaonedwa ngati wolosera wabwino; kuyambira 0.5 mpaka 0.75 ndiye magwiridwe antchito ovomerezeka a chitsanzo, ndipo pansi pa 0.5 ndiye magwiridwe antchito osavomerezeka a chitsanzo. Posankha chitsanzo pogwiritsa ntchito njira zowunikira za RMSE ndi MAE, mitengo yotsika yomwe idapezeka inali yokwanira ndipo idaonedwa ngati chisankho chabwino kwambiri. Equation yotsatirayi ikufotokoza njira yotsimikizira.
pomwe n ikuyimira kukula kwa mtengo wowonedwa\({Y}_{i}\) ikuyimira yankho loyesedwa, ndipo \({\widehat{Y}}_{i}\) ikuyimiranso mtengo woyankhidwa wonenedweratu, chifukwa chake, pakuwona koyamba kwa i.
Mafotokozedwe a ziwerengero za zinthu zodziwira ndi zoyankha aperekedwa mu Table 1, kusonyeza avereji, kupotoka kokhazikika (SD), coefficient of variation (CV), osachepera, apamwamba, kurtosis, ndi kusweka. Ma mtengo ocheperako ndi apamwamba a zinthu ali mu dongosolo lotsika la Mg < Ca < K < Ni ndi Ca < Mg < K < Ni, motsatana. Kuchuluka kwa yankho losinthika (Ni) lomwe linatengedwa kuchokera kudera lophunzirira kunayambira pa 4.86 mpaka 42.39 mg/kg. Kuyerekeza kwa Ni ndi avareji yapadziko lonse (29 mg/kg) ndi avareji ya ku Europe (37 mg/kg) kunawonetsa kuti avereji yonse yowerengedwa ya geometric ya dera lophunzirira inali mkati mwa malire olekerera. Komabe, monga momwe Kabata-Pendias11 yasonyezera, kuyerekeza kwa avareji ya nickel (Ni) mu kafukufuku wapano ndi nthaka yaulimi ku Sweden kukuwonetsa kuti avareji ya nickel yomwe ilipo ndi yayikulu. Momwemonso, kuchuluka kwa Frydek Mistek m'nthaka zamatauni ndi za m'mphepete mwa mizinda mu kafukufuku wapano (Ni 16.15 mg/kg) kunali kokwera kuposa kololedwa. malire a 60 (10.2 mg/kg) a Ni m'nthaka za m'mizinda ya ku Poland omwe adanenedwa ndi Różański et al. Kuphatikiza apo, Bretzel ndi Calderisi61 adalemba kuchuluka kochepa kwambiri kwa Ni (1.78 mg/kg) m'nthaka za m'mizinda ku Tuscany poyerekeza ndi kafukufuku wapano. Jim62 adapezanso kuchuluka kochepa kwa nickel (12.34 mg/kg) m'nthaka za m'mizinda ku Hong Kong, komwe kuli kotsika kuposa kuchuluka kwa nickel komwe kulipo mu kafukufukuyu. Birke et al63 adanenanso kuti kuchuluka kwa Ni kwapakati pa 17.6 mg/kg m'dera lakale la migodi ndi mafakitale akumatauni ku Saxony-Anhalt, Germany, komwe kunali kokwera ndi 1.45 mg/kg kuposa kuchuluka kwa Ni m'derali (16.15 mg/kg). Kafukufuku wapano. Kuchuluka kwa nickel m'nthaka m'madera ena akumatauni ndi akumidzi m'dera lophunzirira kungakhale chifukwa cha makampani achitsulo ndi zitsulo komanso mafakitale achitsulo. Izi zikugwirizana ndi kafukufuku wa Khodadoust et al. 64 kuti makampani opanga zitsulo ndi ntchito zachitsulo ndiye magwero akuluakulu a kuipitsidwa kwa nickel m'nthaka. Komabe, zomwe zinanenedweratu zinali kuyambira 538.70 mg/kg mpaka 69,161.80 mg/kg ya Ca, 497.51 mg/kg mpaka 3535.68 mg/kg ya K, ndi 685.68 mg/kg mpaka 5970.05 mg/kg ya Mg.Jakovljevic et al. 65 adafufuza kuchuluka kwa Mg ndi K m'nthaka yapakati pa Serbia. Adapeza kuti kuchuluka konse (410 mg/kg ndi 400 mg/kg, motsatana) kunali kotsika kuposa kuchuluka kwa Mg ndi K m'kafukufuku wapano. Mosadabwitsa, kum'mawa kwa Poland, Orzechowski ndi Smolczynski66 adayesa kuchuluka konse kwa Ca, Mg ndi K ndipo adawonetsa kuchuluka kwa Ca (1100 mg/kg), Mg (590 mg/kg) ndi K (810 mg/kg). Kuchuluka kwa nthaka pamwamba ndi kotsika kuposa chinthu chimodzi mu kafukufukuyu. Kafukufuku waposachedwa wa Pongrac et al. 67 adawonetsa kuti kuchuluka konse kwa Ca komwe kudafufuzidwa m'nthaka zitatu zosiyanasiyana ku Scotland, UK (nthaka ya Mylnefield, nthaka ya Balruddery ndi nthaka ya Hartwood) kudawonetsa kuchuluka kwa Ca mu kafukufukuyu.
Chifukwa cha kuchuluka kosiyanasiyana kwa zinthu zomwe zayesedwa, kugawa kwa deta ya zinthuzo kumawonetsa kupendekeka kosiyana. Kupendekeka ndi kurtosis ya zinthuzo zinali kuyambira 1.53 mpaka 7.24 ndi 2.49 mpaka 54.16, motsatana. Zinthu zonse zowerengedwa zili ndi kupendekeka ndi kurtosis pamwamba pa +1, motero zikusonyeza kuti kugawa deta sikuli kofanana, kopendekeka mbali yoyenera komanso kufika pachimake. Ma CV oyerekezeredwa a zinthuzo akuwonetsanso kuti K, Mg, ndi Ni amawonetsa kupendekeka pang'ono, pomwe Ca ili ndi kupendekeka kwakukulu kwambiri. Ma CV a K, Ni ndi Mg amafotokoza kugawa kwawo kofanana. Kuphatikiza apo, kugawa kwa Ca sikofanana ndipo magwero akunja angakhudze kuchuluka kwake kowonjezera.
Kugwirizana kwa zinthu zolosera ndi zinthu zoyankhira kunasonyeza kugwirizana kokwanira pakati pa zinthuzo (onani Chithunzi 3). Kugwirizanako kunasonyeza kuti CaK inasonyeza kugwirizana kwapakati ndi r value = 0.53, monga momwe CaNi inachitira. Ngakhale kuti Ca ndi K zimasonyeza kugwirizana pang'ono, ofufuza monga Kingston et al. 68 ndi Santo69 zikusonyeza kuti milingo yawo m'nthaka ndi yosiyana. Komabe, Ca ndi Mg zimatsutsana ndi K, koma CaK imagwirizana bwino. Izi zitha kukhala chifukwa cha kugwiritsa ntchito feteleza monga potassium carbonate, yomwe ndi yokwera ndi 56% mu potaziyamu. Potaziyamu inali yogwirizana pang'ono ndi magnesium (KM r = 0.63). Mumakampani opanga feteleza, zinthu ziwirizi zimagwirizana kwambiri chifukwa potaziyamu magnesium sulfate, potaziyamu magnesium nitrate, ndi potashi zimagwiritsidwa ntchito m'nthaka kuti ziwonjezere kusowa kwawo. Nickel imagwirizana pang'ono ndi Ca, K ndi Mg ndi r values ​​​​= 0.52, 0.63 ndi 0.55, motsatana. Ubale womwe ulipo pakati pa calcium, magnesium, ndi PTE monga nickel ndi wovuta, komabe, magnesium imaletsa kuyamwa kwa calcium, calcium imachepetsa zotsatira za magnesium wochulukirapo, ndipo magnesium ndi calcium zonse zimachepetsa zotsatira za poizoni za nickel m'nthaka.
Matrix yogwirizana ya zinthu zomwe zikuwonetsa ubale pakati pa zoneneratu ndi mayankho (Dziwani: chithunzichi chikuphatikiza scatterplot pakati pa zinthu, milingo yofunikira imachokera pa p < 0,001).
Chithunzi 4 chikuwonetsa kugawa kwa zinthu m'malo osiyanasiyana. Malinga ndi Burgos et al70, kugwiritsa ntchito kugawa kwa malo ndi njira yomwe imagwiritsidwa ntchito poyesa ndikuwonetsa malo otentha m'malo oipitsidwa. Kuchuluka kwa Ca mu Chithunzi 4 kumawoneka kumpoto chakumadzulo kwa mapu ogawa malo. Chithunzichi chikuwonetsa malo otentha a Ca pakati mpaka okwera. Kuchuluka kwa calcium kumpoto chakumadzulo kwa mapu mwina chifukwa cha kugwiritsa ntchito quicklime (calcium oxide) kuchepetsa acidity ya nthaka ndi kugwiritsidwa ntchito kwake m'mafakitale achitsulo ngati mpweya wa alkaline popanga zitsulo. Kumbali ina, alimi ena amakonda kugwiritsa ntchito calcium hydroxide m'nthaka ya acidic kuti achepetse pH, zomwe zimawonjezeranso kuchuluka kwa calcium m'nthaka71. Potaziyamu imasonyezanso malo otentha kumpoto chakumadzulo ndi kum'mawa kwa mapu. Kumpoto chakumadzulo ndi dera lalikulu laulimi, ndipo mawonekedwe apakati mpaka apamwamba a potaziyamu akhoza kukhala chifukwa cha NPK ndi potashi. Izi zikugwirizana ndi maphunziro ena, monga Madaras ndi Lipavský72, Madaras et al.73, Pulkrabová et al.74, Asare et al.75, omwe adawona kuti kukhazikika kwa nthaka ndi chithandizo ndi KCl ndi NPK kunapangitsa kuti pakhale kuchuluka kwa K m'nthaka. Kuchuluka kwa Potassium m'malo akumpoto chakumadzulo kwa mapu ogawa kungakhale chifukwa chogwiritsa ntchito feteleza wokhala ndi potaziyamu monga potassium chloride, potassium sulfate, potassium nitrate, potaziyamu, ndi potaziyamu kuti awonjezere kuchuluka kwa potaziyamu m'nthaka yofooka. Zádorová et al. 76 ndi Tlustoš et al. 77 inafotokoza kuti kugwiritsa ntchito feteleza wochokera ku K kunawonjezera kuchuluka kwa K m'nthaka ndipo kungawonjezere kwambiri kuchuluka kwa michere m'nthaka pakapita nthawi, makamaka K ndi Mg zomwe zikuwonetsa malo otentha m'nthaka. Malo otentha pang'ono kumpoto chakumadzulo kwa mapu ndi kum'mwera chakum'mawa kwa mapu. Kukhazikika kwa colloidal m'nthaka kumachepetsa kuchuluka kwa magnesium m'nthaka. Kusowa kwake m'nthaka kumapangitsa kuti zomera ziwonetse chlorosis yachikasu ya intervein. Feteleza wochokera ku magnesium, monga potassium magnesium sulfate, magnesium sulfate, ndi Kieserite, zimathandiza kusowa kwa michere (zomera zimawoneka zofiirira, zofiira, kapena zofiirira, zomwe zikusonyeza kusowa kwa magnesium) m'nthaka yomwe ili ndi pH yokhazikika6. Kuchuluka kwa nickel pamalo a m'mizinda ndi m'mphepete mwa mizinda kungakhale chifukwa cha zochita za anthu monga ulimi ndi kufunika kwa nickel popanga zitsulo zosapanga dzimbiri78.
Kugawa kwa malo kwa zinthu [mapu ogawa malo adapangidwa pogwiritsa ntchito ArcGIS Desktop (ESRI, Inc, Version 10.7, URL: https://desktop.arcgis.com).]
Zotsatira za chitsanzo cha magwiridwe antchito a zinthu zomwe zagwiritsidwa ntchito mu kafukufukuyu zawonetsedwa mu Gome 2. Kumbali inayi, RMSE ndi MAE za Ni zonse zili pafupi ndi zero (0.86 RMSE, -0.08 MAE). Kumbali inayi, ma RMSE ndi MAE a K onse ndi ovomerezeka. Zotsatira za RMSE ndi MAE zinali zazikulu pa calcium ndi magnesium. Zotsatira za Ca ndi K MAE ndi RMSE ndi zazikulu chifukwa cha ma datasets osiyanasiyana. RMSE ndi MAE za kafukufukuyu pogwiritsa ntchito EBK kulosera Ni zidapezeka kuti ndizabwino kuposa zotsatira za John et al. 54 pogwiritsa ntchito synergistic kriging kulosera kuchuluka kwa S m'nthaka pogwiritsa ntchito deta yomweyi yosonkhanitsidwa. Zotsatira za EBK zomwe tidaphunzira zikugwirizana ndi za Fabijaczyk et al. 41, Yan et al. 79, Beguin et al. 80, Adhikary et al. 81 ndi John et al. 82, makamaka K ndi Ni.
Kugwira ntchito kwa njira zosiyanasiyana zodziwira kuchuluka kwa nickel m'matauni ndi m'mphepete mwa tawuni kunayesedwa pogwiritsa ntchito magwiridwe antchito a zitsanzozo (Table 3). Kutsimikizira kwa chitsanzo ndi kuwunika kolondola kunatsimikizira kuti cholosera cha Ca_Mg_K pamodzi ndi chitsanzo cha EBK SVMR chinapereka magwiridwe antchito abwino kwambiri. Chitsanzo cha Calibration Ca_Mg_K-EBK_SVMR chitsanzo R2, cholakwika cha root mean square (RMSE) ndi cholakwika cha mean absolute (MAE) chinali 0.637 (R2), 95.479 mg/kg (RMSE) ndi 77.368 mg/kg (MAE) Ca_Mg_K-SVMR inali 0.663 (R2), 235.974 mg/kg (RMSE) ndi 166.946 mg/kg (MAE). Komabe, ma R2 values ​​​​abwino adapezeka a Ca_Mg_K-SVMR (0.663 mg/kg R2) ndi Ca_Mg-EBK_SVMR (0.643 = R2); zotsatira zawo za RMSE ndi MAE zinali zapamwamba kuposa za Ca_Mg_K-EBK_SVMR (R2 0.637) (onani Gome 3). Kuphatikiza apo, RMSE ndi MAE za Ca_Mg-EBK_SVMR (RMSE = 1664.64 ndi MAE = 1031.49) ndi 17.5 ndi 13.4, motsatana, zomwe ndi zazikulu kuposa za Ca_Mg_K-EBK_SVMR. Momwemonso, RMSE ndi MAE za Ca_Mg-K SVMR (RMSE = 235.974 ndi MAE = 166.946) ndi zazikulu 2.5 ndi 2.2 kuposa za Ca_Mg_K-EBK_SVMR RMSE ndi MAE, motsatana. Zotsatira za RMSE zowerengedwa zikuwonetsa momwe deta ilili yolumikizidwa ndi mzere woyenera bwino. RSME ndi MAE zapamwamba zidawonedwa. Malinga ndi Kebonye et al. 46 ndi john et al. 54, pamene RMSE ndi MAE zili pafupi kufika pa zero, zotsatira zake zimakhala zabwino kwambiri. SVMR ndi EBK_SVMR zili ndi ma RSME ndi MAE oyezedwa kwambiri. Zinaonedwa kuti ziwerengero za RSME zinali zapamwamba nthawi zonse kuposa ma MAE, zomwe zikusonyeza kukhalapo kwa zinthu zosafunikira. Malinga ndi Legates ndi McCabe83, momwe RMSE imapitilira cholakwika chapakati (MAE) chikulimbikitsidwa ngati chizindikiro cha kukhalapo kwa zinthu zosafunikira. Izi zikutanthauza kuti deta ikasiyana kwambiri, ma MAE ndi RMSE amakhala apamwamba kwambiri. Kulondola kwa kuwunika kotsimikizika kwa chitsanzo chosakanikirana cha Ca_Mg_K-EBK_SVMR cholosera kuchuluka kwa Ni m'matauni ndi m'midzi kunali 63.70%. Malinga ndi Li et al. 59, mulingo uwu wa kulondola ndi mulingo wovomerezeka wa magwiridwe antchito a chitsanzo. Zotsatira zomwe zilipo zikuyerekezeredwa ndi kafukufuku wakale wa Tarasov et al. 36 omwe chitsanzo chawo chosakanikirana chinapanga MLPRK (Multilayer Perceptron Residual Kriging), chogwirizana ndi index yowunikira molondola ya EBK_SVMR yomwe inanenedwa mu kafukufuku wapano, RMSE (210) ndi The MAE (167.5) zinali zapamwamba kuposa zotsatira zathu mu kafukufuku wapano (RMSE 95.479, MAE 77.368). Komabe, poyerekeza R2 ya kafukufuku wapano (0.637) ndi ya Tarasov et al. 36 (0.544), n'zoonekeratu kuti coefficient of determination (R2) ndi yokwera mu chitsanzo chosakanikiranachi. Malire a cholakwika (RMSE ndi MAE) (EBK SVMR) cha chitsanzo chosakanikiranachi ndi otsika kawiri. Momwemonso, Sergeev et al.34 adalemba 0.28 (R2) cha chitsanzo chosakanikirana chopangidwa (Multilayer Perceptron Residual Kriging), pomwe Ni mu kafukufuku wapano adalemba 0.637 (R2). Mulingo wolondola wa chitsanzo ichi (EBK SVMR) ndi 63.7%, pomwe kulondola kwa kulosera komwe Sergeev et al. 34 adapeza ndi 28%. Mapu omaliza (Chithunzi 5) omwe adapangidwa pogwiritsa ntchito chitsanzo cha EBK_SVMR ndi Ca_Mg_K ngati cholosera akuwonetsa kulosera kwa malo otentha komanso pakati pa nickel pagawo lonse lophunzirira. Izi zikutanthauza kuti kuchuluka kwa nickel m'dera lophunzirira kuli pakati, ndi kuchuluka kwakukulu m'malo enaake.
Mapu omaliza oneneratu akuimiridwa pogwiritsa ntchito chitsanzo cha hybrid EBK_SVMR ndikugwiritsa ntchito Ca_Mg_K ngati choneneratu. [Mapu ogawa malo adapangidwa pogwiritsa ntchito RStudio (mtundu 1.4.1717: https://www.rstudio.com/).]
Zomwe zafotokozedwa mu Chithunzi 6 ndi kuchuluka kwa PTE monga kapangidwe kake kokhala ndi ma neuron payekhapayekha. Palibe gawo lililonse la magawo lomwe linawonetsa mtundu womwewo monga momwe zasonyezedwera. Komabe, chiwerengero choyenera cha ma neuron pamapu ojambulidwa ndi 55. SeOM imapangidwa pogwiritsa ntchito mitundu yosiyanasiyana, ndipo mawonekedwe amitundu akafanana kwambiri, mawonekedwe a zitsanzo amafanana kwambiri. Malinga ndi muyeso wawo wolondola wamitundu, zinthu zamtundu uliwonse (Ca, K, ndi Mg) zikuwonetsa mitundu yofanana ndi ma neuron apamwamba amodzi ndi ma neuron ambiri otsika. Chifukwa chake, CaK ndi CaMg zimafanana ndi ma neuron apamwamba kwambiri komanso mitundu yotsika mpaka yocheperako. Mitundu yonse iwiri imaneneratu kuchuluka kwa Ni m'nthaka powonetsa mitundu yapakati mpaka yayitali monga yofiira, lalanje ndi yachikasu. Mtundu wa KMg umawonetsa mitundu yambiri yapamwamba kutengera kuchuluka kolondola ndi mitundu yotsika mpaka yapakati. Pa muyeso wolondola wamitundu kuyambira yotsika mpaka yokwera, mawonekedwe ogawa a zigawo za mtunduwo adawonetsa mtundu wapamwamba womwe ukuwonetsa kuchuluka kwa nickel m'nthaka (onani Chithunzi 4). Mtundu wa CakMg ukuwonetsa mitundu yosiyanasiyana kuyambira yotsika mpaka yokwera malinga ndi mtundu wolondola. sikelo. Kuphatikiza apo, kulosera kwa chitsanzo cha kuchuluka kwa nickel (CakMg) kuli kofanana ndi kugawa kwa nickel komwe kwawonetsedwa pa Chithunzi 5. Ma graph onsewa akuwonetsa kuchuluka kwa nickel m'nthaka za m'mizinda ndi m'mphepete mwa mizinda. Chithunzi 7 chikuwonetsa njira ya contour mu gulu la k-means pamapu, logawidwa m'magulu atatu kutengera mtengo womwe wanenedweratu mu chitsanzo chilichonse. Njira ya contour ikuyimira chiwerengero chabwino cha magulu. Pa zitsanzo 115 za nthaka zomwe zasonkhanitsidwa, gulu 1 linapeza zitsanzo zambiri za nthaka, 74. Gulu 2 linalandira zitsanzo 33, pomwe gulu 3 linalandira zitsanzo 8. Kuphatikiza kwa magawo asanu ndi awiri okonzekera kunasinthidwa kuti kulola kutanthauzira kolondola kwa magulu. Chifukwa cha njira zambiri zobadwa nazo komanso zachilengedwe zomwe zimakhudza mapangidwe a nthaka, n'kovuta kukhala ndi mapangidwe osiyana bwino a magulu mu mapu ogawidwa a SeOM78.
Kutulutsa kwa gawo la gawo ndi mtundu uliwonse wa Empirical Bayesian Kriging Support Vector Machine (EBK_SVM_SeOM). [Mapu a SeOM adapangidwa pogwiritsa ntchito RStudio (mtundu 1.4.1717: https://www.rstudio.com/).]
Zigawo zosiyanasiyana za magulu [Mapu a SeOM adapangidwa pogwiritsa ntchito RStudio (mtundu 1.4.1717: https://www.rstudio.com/).]
Kafukufuku wapano akuwonetsa bwino njira zowonetsera kuchuluka kwa nickel m'matauni ndi m'madera ozungulira mizinda. Kafukufukuyu adayesa njira zosiyanasiyana zowonetsera, kuphatikiza zinthu ndi njira zowonetsera, kuti apeze njira yabwino yodziwira kuchuluka kwa nickel m'nthaka. Mawonekedwe a malo ozungulira a SeOM omwe ali mu njira yowonetsera adawonetsa mtundu wapamwamba kuyambira wotsika mpaka wapamwamba pa sikelo yolondola yamitundu, kusonyeza kuchuluka kwa Ni m'nthaka. Komabe, mapu ogawa malo amatsimikizira kugawa kwa malo ozungulira kwa zigawo zomwe zawonetsedwa ndi EBK_SVMR (onani Chithunzi 5). Zotsatira zake zikuwonetsa kuti chitsanzo chothandizira makina othandizira (Ca Mg K-SVMR) chimaneneratu kuchuluka kwa Ni m'nthaka ngati chitsanzo chimodzi, koma magawo owunikira otsimikizira ndi kulondola akuwonetsa zolakwika zazikulu kwambiri malinga ndi RMSE ndi MAE. Kumbali inayi, njira yowonetsera yomwe imagwiritsidwa ntchito ndi chitsanzo cha EBK_MLR ilinso ndi zolakwika chifukwa cha mtengo wotsika wa coefficient of determination (R2). Zotsatira zabwino zidapezeka pogwiritsa ntchito EBK SVMR ndi zinthu zophatikizana (CaKMg) ndi zolakwika zochepa za RMSE ndi MAE ndi kulondola kwa 63.7%. Zikuoneka kuti kuphatikiza EBK Njira yogwiritsira ntchito makina ophunzirira ikhoza kupanga njira yosakanikirana yomwe ingalosere kuchuluka kwa ma PTE m'nthaka. Zotsatira zake zikusonyeza kuti kugwiritsa ntchito Ca Mg K ngati njira zodziwira kuchuluka kwa Ni m'nthaka kungathandize kuneneratu kuchuluka kwa Ni m'nthaka. Izi zikutanthauza kuti kugwiritsa ntchito feteleza wopangidwa ndi nickel nthawi zonse komanso kuipitsa nthaka m'mafakitale achitsulo kumapangitsa kuti nthaka ikhale yochuluka kwambiri. Kafukufukuyu adawonetsa kuti njira ya EBK ikhoza kuchepetsa kuchuluka kwa zolakwika ndikuwonjezera kulondola kwa njira yogawa malo m'nthaka m'matauni kapena m'mphepete mwa tawuni. Kawirikawiri, tikupereka lingaliro logwiritsa ntchito njira ya EBK-SVMR kuti tiwunikire ndikulosera PTE m'nthaka; kuphatikiza apo, tikupereka lingaliro logwiritsa ntchito EBK kuti tisakanize ndi ma algorithms osiyanasiyana ophunzirira makina. Kuchuluka kwa Ni kunanenedweratu pogwiritsa ntchito zinthu ngati covariates; komabe, kugwiritsa ntchito ma covariates ambiri kungawongolere kwambiri magwiridwe antchito a chitsanzocho, chomwe chingaganizidwe ngati malire a ntchito yomwe ilipo. Cholepheretsa china cha kafukufukuyu ndichakuti chiwerengero cha ma dataset ndi 115. Chifukwa chake, ngati deta yochulukirapo yaperekedwa, magwiridwe antchito a njira yosakanikirana yomwe ikuperekedwayo akhoza kuwongoleredwa.
PlantProbs.net.Nickel mu Zomera ndi Nthaka https://plantprobs.net/plant/nutrientImbalances/sodium.html (Yapezeka pa 28 Epulo 2021).
Kasprzak, KS Nickel ikupita patsogolo mu toxicology yamakono yokhudza chilengedwe. surroundings.toxicology.11, 145–183 (1987).
Cempel, M. & Nikel, G. Nickel: Kuwunikanso magwero ake ndi poizoni wa chilengedwe. Polish J. Environment.Stud.15, 375–382 (2006).
Freedman, B. & Hutchinson, TC Kuipitsidwa kwa zinthu kuchokera mumlengalenga ndi kusonkhana m'nthaka ndi zomera pafupi ndi chosungunula cha nickel-copper ku Sudbury, Ontario, Canada.can.J. Bot.58(1), 108-132.https://doi.org/10.1139/b80-014 (1980).
Manyiwa, T. et al. Zitsulo zolemera m'nthaka, zomera ndi zoopsa zokhudzana ndi kudyetsa ziweto pafupi ndi mgodi wa mkuwa ndi nickel ku Selebi-Phikwe ku Botswana. surroundings.Geochemistry.Health https://doi.org/10.1007/s10653-021-00918-x (2021).
Cabata-Pendias.Kabata-Pendias A. 2011. Zinthu zotsatizana mu nthaka ndi… – Google Scholar https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=Kabata-Pendias+A.+2011.+Trace+ Elements+in+soils+and+plants.+4th+ed.+New+York+%28NY%29%3A+CRC+Press&btnG= (Yapezeka pa 24 Nov 2020).
Almås, A., Singh, B., Agriculture, TS-NJ ya & 1995, yosadziwika. Zotsatira za makampani aku Russia a nickel pa kuchuluka kwa zitsulo zolemera m'nthaka ndi udzu waulimi ku Soer-Varanger, Norway.agris.fao.org.
Nielsen, GD et al. Kuyamwa ndi kusunga nickel m'madzi akumwa kumagwirizana ndi kudya komanso kukhudzidwa ndi nickel.toxicology.application.Pharmacodynamics.154, 67–75 (1999).
Costa, M. & Klein, CB Nickel carcinogenesis, kusintha kwa majini, epigenetics kapena kusankha. malo ozungulira. Malingaliro Aumoyo.107, 2 (1999).
Ajman, PC; Ajado, SK; Borůvka, L.; Bini, JKM; Sarkody, VYO; Cobonye, ​​​​NM; Kusanthula kwa Trend kwa zinthu zomwe zingakhale zoopsa: ndemanga ya bibliometric. Environmental Geochemistry and Health. Springer Science & Business Media BV 2020. https://doi.org/10.1007/s10653-020-00742-9.
Minasny, B. & McBratney, AB Kujambula Mapu a Dothi la Digito: Mbiri Yachidule ndi Maphunziro Ena. Geoderma 264, 301–311.https://doi.org/10.1016/j.geoderma.2015.07.017 (2016).
McBratney, AB, Mendonça Santos, ML & Minasny, B. Pa mapu a nthaka ya digito.Geoderma 117(1-2), 3-52.https://doi.org/10.1016/S0016-7061(03)00223-4 (2003).
Kujambula kwa Deutsch.CV Geostatistical Reservoir Modeling,… – Google Scholar https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=CV+Deutsch%2C+2002%2C+Geostatistical+Reservoir+Modeling%2C +Oxford+University+Press%2C+376+pages.+&btnG= (Yapezeka pa 28 Epulo 2021).