森林碳储量是研究森林生态系统功能的重要指标,同时也是人们经营森林的参考。珠三角为国内重要的发达地区,研究其森林碳储量分布格局,可以为珠三角森林城市群建设提供重要的理论依据。为精准估测大尺度森林碳储量空间分布格局,以2017年珠三角区域419块森林资源连续清查样地数据为研究对象,采用普通最小二乘法(OLS)模型和4种地理加权回归(GWR)模型对珠三角森林碳储量进行空间分布预测。结果表明:4种GWR模型均比OLS模型有更好的拟合效果和独立性样本检验结果,其中MGWR(Gaussian)的模型拟合效果最佳,GWR(Gaussian)的独立性样本检验结果最佳;运用GWR(Gaussian)模型估测的珠三角森林单位面积碳储量为6.15~81.52 t/hm²,平均值为33.46 t/hm²,且估测误差优于OLS模型,整体上呈现中心高、边缘低的空间分布格局。地理加权回归方法能较有效地估测森林碳储量,并且通过调整模型形式和空间核函数可进一步提高估测精度,为珠三角森林碳储量估测提供了新的方法和思路。

Based on the data of 419 consecutive forest resource inventory plots in the Pearl River Delta region in 2017,the spatial distribution of forest carbon storage in the Pearl River Delta region was predicted using the Ordinary Least Squares (OLS)model and four Geographically Weighted Regression (GWR)models.The results showed that all four GWR models had better model fitting performance and independent sample test results than OLS models,with MGWR (Gaussian)model having the best fitting performance and GWR (Gaussian)model having the best independent sample test results;Using the GWR (Gaussian)model to draw a spatial distribution map of forest carbon storage,the estimated carbon storage per unit area of the Pearl River Delta forest is 6.15-81.52 t/hm²,with an average value of 33.46 t/hm².The estimation error is better than the OLS model,and the overall spatial distribution pattern shows a high center and low edge.It also provides a new method and idea for the estimation of forest carbon stocks in the Pearl River Delta.