基于哑变量的云南松蓄积生长模型
|
摘要
为了解云南松林分蓄积生长规律,为更好的经营云南松林分提供理论依据,笔者基于云南省云南松一类清查数据,以Schumacher和Richards模型为基础模型,选取不同的林分密度指标,来拟合云南松林分蓄积生长模型,并在最优基础模型上引入哑变量,将不同林分类型合并,建立同时适用于间伐林分和未间伐林分的蓄积生长模型。结果表明:4个传统生长模型的模拟效果都较好,R~2和预估精度最高分别为0.971 6、96.79%,不同的林分密度指标对模型的预测效果有较大影响,且以选用林分断面积作为林分密度指标的Richards模型的预估效果最好,其预估精度为96.79%。在最优传统生长模型中引入哑变量后,模型的R~2和预估精度都比传统的生长模型稍高,分别达到0.973 0、96.84%。通过对模型的适应性检验,所建的哑变量模型的预估精度超过98%,且对间伐林分的拟合效果更佳,可以用来描述不同措施下云南省云南松林分的蓄积生长规律,也解决了不同类型林分合并建模不相容的问题。
In order to understand the laws of volume growth of Pinus yunnanensis Franch and provide a basis for a better management,by using the inventory data of P. yunnanensis in Yunnan Province, based on the Richards and the Schumacher models,through choosing different density indexes,the volume growth models were simulated. On the basis of the best conventional model,the dummy variables were introduced to combine the different type of stands trying to build a volume growth model which is applicable for both thinned stands and un-thinned stands. The results showed that all the four conventional models were fitting well,the maximum R~2 and forecasting precision reached 0.971 6,96.79% respectively. Besides,different density index has a significant impact on the forecast effect of the model,and the one based on the Richards as well as taking the stand basal area as the index proved to be the best with its forecasting precision as high as 96.79%. The R~2 and forecasting accuracy were slightly improved after introducing dummy variables into the best traditional model,the value of R~2 was 0.973 0 and the forecasting accuracy was 96.84%.Independent test of the models indicated that the forecasting precisions of the dummy variables model was higher than 98% and the fitting effect of thinned stands was better. The dummy variable model can be used to describe the volume growth rule of P. yunnanensis with different management measures in Yunnan Province,and can be compatible with unified modeling for different types stands.
In order to understand the laws of volume growth of Pinus yunnanensis Franch and provide a basis for a better management,by using the inventory data of P. yunnanensis in Yunnan Province, based on the Richards and the Schumacher models,through choosing different density indexes,the volume growth models were simulated. On the basis of the best conventional model,the dummy variables were introduced to combine the different type of stands trying to build a volume growth model which is applicable for both thinned stands and un-thinned stands. The results showed that all the four conventional models were fitting well,the maximum R~2 and forecasting precision reached 0.971 6,96.79% respectively. Besides,different density index has a significant impact on the forecast effect of the model,and the one based on the Richards as well as taking the stand basal area as the index proved to be the best with its forecasting precision as high as 96.79%. The R~2 and forecasting accuracy were slightly improved after introducing dummy variables into the best traditional model,the value of R~2 was 0.973 0 and the forecasting accuracy was 96.84%.Independent test of the models indicated that the forecasting precisions of the dummy variables model was higher than 98% and the fitting effect of thinned stands was better. The dummy variable model can be used to describe the volume growth rule of P. yunnanensis with different management measures in Yunnan Province,and can be compatible with unified modeling for different types stands.
引文
[1]《中国森林》编辑委员会.中国森林:第2卷[M].北京:中国林业出版社,1999:985.
[2]戴开结,何方,沈有信,等.云南松研究综述[J].中南林学院学报,2006,26(2):138-142.
[3]郑元,黄新会,王大玮,等.云南松研究进展及展望[J].现代农业科技,2013(13):169-171.
[4]蔡年辉,李根前,朱存福,等.云南松人工林与天然林群落结构的比较研究[J].西北林学院学报,2007,22(2):1-4,163.
[5]王健敏,刘娟,陈晓鸣,等.云南松天然林及人工林群落结构和物种多样性比较[J].林业科学研究,2010,23(4):515-522.
[6]孙宝刚,陈飞,王健敏,等.不同径级云南松各部位生物量及其分布规律[J].林业科学研究,2012,25(1):71-76.
[7]杨阳,冉飞,王根绪,等.西藏高原云南松生物量模型及碳储量[J].生态学杂志,2013,32(7):1 674-1 682.
[8]冉啟香,邓华锋,黄国胜,等.云南松地上生物量模型研究[J].浙江农林大学学报,2016,33(4):605-611.
[9]BALDWIN V C J R,PETERSON K D,CLARK A I I I.The effect of spacing and thinning on stand and tree characteristic of38-year-old loblolly pine[J].Forest Ecology and Management,2000,137(3):91-102.
[10]HYTNEN J,ISSAKAINEN J.Effect of repeated harvesting on biomass production and sprouting of Betula pubescens[J].Biomass and Bioenergy,2001,20(4):237-245.
[11]胥富强,王齐,吕寻,等.小陇山林区抚育间伐对华山松人工林生长的影响[J].西北林学院学报,2014,29(6):176-179.
[12]鲍斌,丁贵杰.抚育间伐对马尾松林分生长与植物多样性的影响[J].中南林业科技大学学报,2013,33(3):30-33,46.
[13]孟宪宇.测树学[M].3版.北京:中国林业出版社,2006:115-116.
[14]郎荣,许建初,TENNIGKEIT T,等.基于样方数据的云南松林分生长模型研究:以云南省保山市杨柳白族彝族乡为例[J].植物分类与资源学报,2011,33(3):357-363.
[15]王少杰,邓华锋,黄国胜,等.基于哑变量的油松人工林和天然林生长模型[J].森林与环境学报,2016,36(3):325-331.
[16]李忠国,孙晓梅,陈东升,等.基于哑变量的日本落叶松生长模型研究[J].西北农林科技大学学报(自然科学版),2011,39(8):69-74.
[17]李春明.利用非线性混合模型进行杉木林分断面积生长模拟研究[J].北京林业大学学报,2009,31(1):44-49.
[18]高东启,邓华锋,程志楚,等.蒙古栎间伐林分和未间伐林分生长模型研究[J].中南林业科技大学学报,2014,34(2):50-54.
[19]冉啟香,邓华锋,吕常笑,等.油松林分断面积与蓄积量生长模型研究[J].西北林学院学报,2016,31(5):217-223.
[2]戴开结,何方,沈有信,等.云南松研究综述[J].中南林学院学报,2006,26(2):138-142.
[3]郑元,黄新会,王大玮,等.云南松研究进展及展望[J].现代农业科技,2013(13):169-171.
[4]蔡年辉,李根前,朱存福,等.云南松人工林与天然林群落结构的比较研究[J].西北林学院学报,2007,22(2):1-4,163.
[5]王健敏,刘娟,陈晓鸣,等.云南松天然林及人工林群落结构和物种多样性比较[J].林业科学研究,2010,23(4):515-522.
[6]孙宝刚,陈飞,王健敏,等.不同径级云南松各部位生物量及其分布规律[J].林业科学研究,2012,25(1):71-76.
[7]杨阳,冉飞,王根绪,等.西藏高原云南松生物量模型及碳储量[J].生态学杂志,2013,32(7):1 674-1 682.
[8]冉啟香,邓华锋,黄国胜,等.云南松地上生物量模型研究[J].浙江农林大学学报,2016,33(4):605-611.
[9]BALDWIN V C J R,PETERSON K D,CLARK A I I I.The effect of spacing and thinning on stand and tree characteristic of38-year-old loblolly pine[J].Forest Ecology and Management,2000,137(3):91-102.
[10]HYTNEN J,ISSAKAINEN J.Effect of repeated harvesting on biomass production and sprouting of Betula pubescens[J].Biomass and Bioenergy,2001,20(4):237-245.
[11]胥富强,王齐,吕寻,等.小陇山林区抚育间伐对华山松人工林生长的影响[J].西北林学院学报,2014,29(6):176-179.
[12]鲍斌,丁贵杰.抚育间伐对马尾松林分生长与植物多样性的影响[J].中南林业科技大学学报,2013,33(3):30-33,46.
[13]孟宪宇.测树学[M].3版.北京:中国林业出版社,2006:115-116.
[14]郎荣,许建初,TENNIGKEIT T,等.基于样方数据的云南松林分生长模型研究:以云南省保山市杨柳白族彝族乡为例[J].植物分类与资源学报,2011,33(3):357-363.
[15]王少杰,邓华锋,黄国胜,等.基于哑变量的油松人工林和天然林生长模型[J].森林与环境学报,2016,36(3):325-331.
[16]李忠国,孙晓梅,陈东升,等.基于哑变量的日本落叶松生长模型研究[J].西北农林科技大学学报(自然科学版),2011,39(8):69-74.
[17]李春明.利用非线性混合模型进行杉木林分断面积生长模拟研究[J].北京林业大学学报,2009,31(1):44-49.
[18]高东启,邓华锋,程志楚,等.蒙古栎间伐林分和未间伐林分生长模型研究[J].中南林业科技大学学报,2014,34(2):50-54.
[19]冉啟香,邓华锋,吕常笑,等.油松林分断面积与蓄积量生长模型研究[J].西北林学院学报,2016,31(5):217-223.