We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
人工神经网络与相容性生物量模型 预测单木地上生物量的比较.
- Authors
梁瑞婷; 王轶夫; 邱思玉; 孙玉军; 谢运鸿
- Abstract
ForcsL biomass is an imporLanL index in fores L dcvclopmcnL planning and forcsL rcsoLircc moni Loring. In order Lo provide a more cflicicnL and low-biased mcLhod for csLimaLing individual Lrcc biomass, wc inLroduccd arLificial neural noLwork here. Wc used Lhc daLa of abovegmund biomass of 101 Larix olgensis Lrccs hai'vcsLcd from Lhc Dongzhclcnghc ForcsL Farm in Heilongjiang Province Lo develop four aggregaLion model svsLcms (AMS), based on diflerenL combinaLion of Lhc variables (diamcLcr aL breasL heighL, Lrcc heigh--crown widLh)・ The weighLed funcLions were used Lo climinaLc hcLcmsccdasLiciLy ・ Then, we Lrained ailificial neural neLwork (ANN) biomass model based on Lhc opLimal combinaLion ・ The models were Les Led by Lhc leave-one-on L cross-validation mcLhod Lo compare Lhc accLiracy of Lhc Lwo biomass esLimaLion meLhods・ llic rcsulLs showed LhaL biomass model based on only one variable, diamcLcr aL breasL heighL, could accuraLcly csLimaLc Lhc biomass of L・ olgaisis・ Adding Lwo indices, Lrcc heighL and crown widLh, could improve Lhc fiLLing pcTformanccj of models, wiLh AMS4 performing Lhc besL among Lhc four addicLivc model sysLems・ The biomass models developed by Lhc Lwo meLhods boLh could csLimaLc biomass aL lrcc level accLiraLcly, wiLh Lhc cocfllcicnL of dcLcrminaLion (R?) of each componcnL was higher Lhan ().87. Compared wiLh Lhc AMS4, R2 of leaf biomass model was abouL 0.05 higher, and LhaL of oLher organs were also aboiiL 0.01 higher in arLillcial neural neLwork model sysLem・ In addiLion, Lhc rooL mean square error (RMSE) and oLher indicaLors were also significanLly smaller. For example, Lhc RMSE of Lrcc sLem and aboveground biomass were smaller by 2.135 kg and 3.908 kg, rcspccLivcly. The modelJ s validaLion sLaLisLics mean rclaLivc error (MRE) performed better. In general, ANN was a flexible and reliable biomass esLimaLion mcLhod, which was worLhy considcraLion when prcdicLing Lrcc componcnL biomass or abovegmund biomass.
- Subjects
HEILONGJIANG Sheng (China); BIOMASS; LARCHES; PROVINCES; FARMS
- Publication
Chinese Journal of Applied Ecology / Yingyong Shengtai Xuebao, 2022, Vol 33, Issue 1, p9
- ISSN
1001-9332
- Publication type
Article
- DOI
10.13287/j.1001-9332.202201.001