Kõrguse-diameetri mudelite analüüs Järvselja lehisepuistute andmeil
Laen...
Kuupäev
2020
Kättesaadav alates
19.11.2020
Autorid
Ajakirja pealkiri
Ajakirja ISSN
Köite pealkiri
Kirjastaja
Eesti Maaülikool
Abstrakt
Lehis on atraktiivne okaspuu, mille kasvatamisele ja kasutusele on hakatud järjest enam
tähelepanu pöörama. Kuid suhteliselt tagasihoidlikult on tegeletud lehiste kasvu
modelleerimisega, kuna algandmestik ei ole piisav. Käesoleva lõputöö eesmärgiks on testida
ja võrrelda erinevaid kõrguskõveraid lehiste andmeil.
Antud tööks koguti lehiste mõõtmisandmed 2019 sügisel ja 2020 talvel Järvselja Õppe- ja
Katsemetsakonna 17. lehisepuistust. Välitööde käigus kordusmõõdeti üheksa ringproovitükki
ja kaheksa vana katseala. Andmete töötlemisel kasutati vabavara R, kõrguskõverate testimiseks
paketti lmfor. Töös testiti 17. kõrguskõverat, milles kaheksa olid 2-parameetrilised (Näslund,
Curtis, Schumacher, Meyer, Power, Michaelis-Menten, Wykoff) ja üheksa 3-parameetrilist
(Prodan, Logistic, Chapman-Richards, Weibull, Gomperz, Sibbesen, Korf, Ratkowsky,
Hossfeld IV ja Padari). okku mõõdeti ringproovitükkidelt ja vanadelt katsealadelt 1393 elusa
puu diameetrid, mõõdetud puudest 741 olid lehised ja 182 lehiselt mõõdeti ka kõrgused
(mudelpuud).
Töö tulemusel selgus, et kõige madalama prognoosiveaga oli 2-parameetriliste kõrguskõverate
mudelite seast Meyeri funktsioon (AIC 807,0855 ja Loglik -400,5427) ja 3-parameetrilistest
mudelitest Prodani funktsioon (AIC 706,5593 ja Loglik -348,2796). Meyeri funktsioon aga ei
arvesta, et puu ei kasva kõrgusesse lõputult, samas on suurenenud puu diameeter. Prodani
mudel ei sobi kasutamiseks, sest kui mõõdetud mudelpuid on vähe, prognoosib see kõrgust
kõvasti madalamaks. Padari kõrguskõverate mudelid hindavad lehisepuistute kõrgusi samuti
tegelikkusest madalamaks, seega on see antud juhul liiga ebatäpne. Seega täpsemaks kõrguse
prognoosimiseks tuleks siiski rohkem lehiste mudelpuid mõõta nii erinevates kasvukohtades
kui ka vanuses.
Larch is an attractive conifer, and more attention has been paid to the cultivation of it. Relatively little has been used larch for modelling purpose, as the initial dataset is modest. The aim of the thesis is to test and compare different height curves on the larch data. For this thesis the measurements of 17 larch tree plots were collected in the autumn of 2019 and winter 2020 from Järvselja Study and Experimental forests. During the fieldwork 9 circular sample plots and 8 old sample sites were re-measured. The data analysis was done with freeware R, for testing height-diameter models was used package lmfor. A total of 17 height-diameter models were tested, of which 8 equations contain two parameters (Näslund, Curtis, Schumacher, Meyer, Power, Michaelis-Menten, Wykoff) and 9 had three parameters (Prodan, Logistic, Chapman-Richards, Weibull, Gomperz, Sibbesen, Korf, Ratkowsky, Hossfeld IV and Padari). The results of the analysis revealed that the lowest predictive error had Meyer (AIC 807,0855 and Loglik -400,5427) function among the two parameters models and the Prodan (AIC 706,5593 and Loglik -348,2796) function among the three parameters models. However, the Meyer function does not consider that the tree does not grow in height indefinitely, while the diameter of the tree has increased. Prodan model is not suitable for use when the measured model tree amount on the sample area is low, then it predicts it a lot lower. Padari height curve models estimate the heights of larch stands lower than they actually are, so in this case it is too imprecise. Therefore, for a more precise height prediction, more larch sample trees should be measured included for further studies.
Larch is an attractive conifer, and more attention has been paid to the cultivation of it. Relatively little has been used larch for modelling purpose, as the initial dataset is modest. The aim of the thesis is to test and compare different height curves on the larch data. For this thesis the measurements of 17 larch tree plots were collected in the autumn of 2019 and winter 2020 from Järvselja Study and Experimental forests. During the fieldwork 9 circular sample plots and 8 old sample sites were re-measured. The data analysis was done with freeware R, for testing height-diameter models was used package lmfor. A total of 17 height-diameter models were tested, of which 8 equations contain two parameters (Näslund, Curtis, Schumacher, Meyer, Power, Michaelis-Menten, Wykoff) and 9 had three parameters (Prodan, Logistic, Chapman-Richards, Weibull, Gomperz, Sibbesen, Korf, Ratkowsky, Hossfeld IV and Padari). The results of the analysis revealed that the lowest predictive error had Meyer (AIC 807,0855 and Loglik -400,5427) function among the two parameters models and the Prodan (AIC 706,5593 and Loglik -348,2796) function among the three parameters models. However, the Meyer function does not consider that the tree does not grow in height indefinitely, while the diameter of the tree has increased. Prodan model is not suitable for use when the measured model tree amount on the sample area is low, then it predicts it a lot lower. Padari height curve models estimate the heights of larch stands lower than they actually are, so in this case it is too imprecise. Therefore, for a more precise height prediction, more larch sample trees should be measured included for further studies.
Kirjeldus
Bakalaureusetöö
Loodusvarade kasutamise ja kaitse õppekaval
Märksõnad
bakalaureusetööd, lehis, kõrguskõverad, modelleerimine, mudelpuud, proovitükid