以盐城自然保护区核心区的ETM+遥感影像和同期野外实测生物量为数据源,建立BP神经网络模型反演研究区地上生物量。结果表明:运用BP神经网络模型反演生物量湿重、干重精度分别达到了70%和74%;研究区生物量总量湿重为4.996×10⁸kg,干重为9.370×10⁷kg;在空间上呈现出海陆分异明显,海岸方向变化缓慢;米草、芦苇、碱蓬这3种植物的单位面积生物量呈现由高到低的特征,碱蓬的生物量干重集中在0~1.5kg/m²,湿重集中在0~6kg/m²;米草的生物量干重集中在1~2kg/m²,湿重集中在8kg/m²以上;芦苇的生物量干重集中在0~2kg/m²,湿重集中在2~6kg/m²,生物量与株高、盖度呈正相关,干重与二者相关性更强;生物量与生态位、土壤环境要素呈正相关,尤其与土壤养分相关性最强。

Taking the ETM + remote sensing image of the core area of Yancheng Nature Reserve and the field aboveground biomass measured in the same period as the data source,we built BP artificial neural network model and simulated biomass distribution of the study area.The conclusions of the study are as follows:The BP neural network model was used to retrieve the biomass of wet weight and dry weight,their accuracy arrived at 70% and 74% respectively.The total biomass of dry weight is 9.370×10⁷kg and the total biomass of wet weight is 4.996×10⁸kg.In the spatial variation,it mainly presents obvious difference from the land to the sea and slow change along the coast.The biomass of Spartina,reed and Suaeda range from high to low.The dry weight of Suaeda salsa’s biomass mainly concentrated in 1.5kg/m²,while its biomass of wet weight is concentrated in 0~6kg/m²; The dry weight of Spartina alterniflora’s biomass mainly concentrated in 1~2kg/m²,while its biomass of wet weight is over 8kg/m²; The dry weight of Reed’s biomass mainly concentrated in 0~2kg/m²,while its biomass of wet weight is concentrated in 2~6kg/m².The biomass is positively correlated with plant height and coverage,and the correlation between dry weight and the two was stronger.The biomass was positively correlated with ecological niche and soil environmental factors,especially with soil nutrients.