1.Heilongjiang Provincial Institute of Wood Science,Harbin 150081,China
2.Harbin Far East lnstitute of Technology,Harbin 150025
3.Heilongjiang Institute of Ecology,Harbin 150081,China
4.Heilongjiang Institute of Meteorological Science,Harbin 150030,China
Show less
文章历史+
Received
Published
2025-01-13
2025-09-15
Issue Date
2025-10-30
PDF (1670K)
摘要
森林碳储量是全球碳循环研究的重要内容,应对气候变化具有重要意义。以黑龙江省张广才岭北坡部分区域为研究对象,结合地面观测数据与Landsat TM(thematic mapper)/OLI(operational land imager)传感器数据,采用多种机器学习模型,结合引导聚集Bagging(bootstrap aggregating)集成学习算法,进行森林碳储量模拟。结果表明,1990—2022年研究区森林碳储量呈现显著增加趋势,年平均碳储量达到(80.77±0.27)Mg C/hm2,且空间分布表现出明显的异质性特征,高碳储量区域集中于平坦或半山坡地带。此外,生长季平均气温与森林碳储量呈极显著的正相关关系(P<0.01),表明气温是影响碳储量变化的主要气候因子。研究结果可为森林碳储量的精准模拟和碳汇管理提供新思路。
Abstract
Forest carbon storage is a critical component of global carbon cycle research and plays a significant role in addressing climate change. This study focused on the northern slope of the Zhangguangcai Mountains in Heilongjiang Province. By combining ground observation data with Landsat TM (thematic mapper)/OLI (operational land imager) data, multiple machine learning models were applied, along with the bootstrap aggregating ensemble learning algorithm, to simulate forest carbon storage. The results showed that from 1990 to 2022, the forest carbon storage in the study area exhibited a significant increasing trend, with an annual average carbon storage of (80.77±0.27) Mg C/hm2. The spatial distribution demonstrated notable heterogeneity, with high carbon storage areas concentrated in flat and semi-mountainous regions. Additionally, the mean growing-season temperature was found to have a highly significant positive correlation with forest carbon storage (P<0.01), indicating that temperature was the primary climatic factor influencing carbon storage changes. This study provides a novel approach for forest carbon storage accurate simulation carbon sink management.
Improved-Neural-Net采用4层Improved-Neural-Net网络结构,简化模型复杂度并满足数据建模要求;隐藏层使用ReLU激活函数进行非线性映射,输出层采用线性激活函数以适应回归任务;模型训练采用自适应矩估计优化器(adaptive moment estimation,Adam)(学习率0.001),通过暂退法(Dropout)和早停机制防止过拟合,同时使用小批量(32)和100轮次迭代确保模型高效收敛与稳定。
与传统统计模型,如日尺度碳循环模型(daily century model,DAYCENT)、洛桑土壤碳模型(rothamsted carbon model,ROTHC),以及单一机器学习模型相比,本研究创新性地结合了Improved-Neural-Net、LGBM和XGBoost 3种机器学习模型,并通过Bagging集成学习算法显著提高了模拟精度。例如与中国森林碳汇模型(forest ecosystem carbon budget model for China,FORCCHN)相比[6],深度学习方法在数据驱动的基础上减少了对过程参数依赖,表现出更强的适用性与精确度。与全球尺度模型(如MOD17A产品)的粗略估算不同,本研究在中等分辨率(30 m)遥感数据的支持下,结合机器学习方法,对小尺度森林碳储量的空间异质性进行精确模拟,补充了全球尺度数据在局部区域的精度不足问题[3,10]。
LIUW R, LIX T, LIT,et al.Spatiotemporal variations of forest NPP and related driving factors based on MODIS and CASA models in Yichun[J].Chinese Journal Of Ecology,2022,41(1):150-158.
LYUF C, MAJ Y, CAOY,et al.Carbon fluxes simulation of China′s typical forest ecosystem based on FORCCHN model[J].Acta Ecologica Sinica,2022,42(7):2810-2821.
YUG R, GAOY, WANGQ F,et al.Discussion on the key processes of carbon-nitrogen-water coupling cycles and biological regulation mechanisms in terrestrial ecosystem[J].Chinese Journal of Eco-Agriculture,2013,21(1):1-13.
PARTONW J, SCHIMELD S, COLEC V,et al.Analysis of factors controlling soil organic matter levels in great plains grasslands[J].Soil Science Society of America Journal,1987,51(5):1173-1179.
[14]
PARTONW J, SCURLOCKJ M O, OJIMAD S,et al.Observations and modeling of biomass and soil organic matter dynamics for the grassland biome worldwide[J].Global Biogeochemical Cycles,1993,7(4):785-809.
[15]
COLEMANK, JENKINSOND S.RothC-26.3-A model for the turnover of carbon in soil:model description and user's guide[M]//POWLSON D S,SMITH P,SMITH J U.Evaluation of Soil Organic Matter Models.NATO ASI Series.Berlin:Springer,1995.
[16]
LIY, CHEND, ZHANGY,et al.Comparison of three modeling approaches for simulating denitrification and nitrous oxide emissions from loam-textured arable soils[J].Global Biogeochemical Cycles,2005,19(3):GB3002.
[17]
LIC, FROLKINGS, FROLKINGT A.A model of nitrous oxide evolution from soil driven by rainfall events:1.Model structure and sensitivity[J].Journal of Geophysical Research:Atmospheres,1992,97(D9):9759-9776.
TANGH, FUP, SHERMANC S,et al.A deep learning-accelerated data assimilation and forecasting workflow for commercial-scale geologic carbon storage[J].International Journal of Greenhouse Gas Control,2021,112:103488.
[20]
TANGZ, LIM, WANGX.Mapping tea plantations from vhr images using OBIA and convolutional neural networks[J].Remote Sensing,2020,12(18):2935.
[21]
TUBIELLOF N, SALVATOREM, FERRARAA F,et al.The contribution of agriculture,forestry and other land use activities to global warming,1990—2012[J].Global Change Biology,2015,21(7):2655-2660.
[22]
董利虎.东北林区主要树种及林分类型生物量模型研究[D].哈尔滨,东北林业大学.2015.
[23]
DONGL H.Study on biomass models of dominant tree species and forest stand types in northeast China's forest region[D].Harbin,Northeast Forestry University,2015.
YUY, FANW Y, LIM Z.Forest carbon rates at different scales in northeast China forest area[J].Chinese Journal of Applied Ecology,2012,23(2):341-346.
[26]
MAOJ, SHIX, THORNTONP E,et al.Causes of spring vegetation growth trends in the northern mid-high latitudes from 1982 to 2004[J].Environmental Research Letters,2012,7(1):014010.
[27]
XIANGK, GUOQ, ZHANGB,et al.Impact of preseason climate factors on vegetation photosynthetic phenology in mid–high latitudes of the northern hemisphere[J].Plants,2024,13(9):1254.
[28]
KATULG G, ORENR, MANZONIS,et al.Evapotranspiration:A process driving mass transport and energy exchange in the soil-plant-atmosphere-climate system[J].Reviews of Geophysics,2012,50(3):RG3002.
[29]
ZHANGZ Z, ZHAOP, MCCARTHYH R,et al.Influence of the decoupling degree on the estimation of canopy stomatal conductance for two broadleaf tree species[J].Agricultural and Forest Meteorology,2016,221:230-241.
[30]
HURTEAUM D.Chapter 27-The role of forests in the carbon cycle and in climate change[M]//LETCHER T M.Climate Change(Third Edition).Elsevier,2021:561-579.
[31]
BONANG B.Forests and climate change:Forcings,feedbacks,and the climate benefits of forests[J].Science,2008,320(5882):1444-1449.
[32]
KILPELÄINENA, PELTOLAH.Carbon sequestration and storage in european forests[M]//HETEMäKI L,KANGAS J,PELTOLA H.Forest Bioeconomy and Climate Change.Cham; Springer International Publishing.2022:113-128.
[33]
LOGUERCIOG A, SIMONA, WINTERA N,et al.Carbon density and sequestration in the temperate forests of northern Patagonia,Argentina[J].Frontiers in Forests and Global Change,2024,7:1373187.
[34]
COMASL H, BECKERS R, CRUZV M,et al.Root traits contributing to plant productivity under drought[J].Frontiersin Plant Science,2013,4:442.
[35]
O'CONNORJ, DEKKERS, STAALA,et al.Forests buffer against variations in precipitation[J].Global Change Biology,2021,27(19):4686-4696.