Abstract

Mapping and quantifying above ground carbon (AGC) Stocks reflect significant dynamics in the terrestrial carbon cycle and cascade climate change. Estimation of such key driver was performed for the dominant species (Bamboo, Eucalyptus and Teak) over Coimbatore and Nilgiris Biosphere (2006 − 2018 quadruple interval) of Tamilnadu, India with the developed global stepwise multiple linear regression (SMLR) and local geographically weighted regression (GWR) models using multi-dynamic variables. Evapotranspiration (ET) developed using Mapping Evapotranspiration at High Resolution with Internalized Calibration (METRIC) model for the region was analysed with the best-fitted AGC estimation model to understand AGC—ET synergistic pertinence dynamics. The study compared and validated the estimation by the models and indicated that AGC estimation using SMLR exhibiting a high degree of accuracy ($R^2=0.84,$ $\mathrm{RMSE}=5.67,$ $\mathrm{MAE}=4.61\text{ Mg }{\mathrm{ha}}^{-1}$) with nominal negative bias in the estimation ranging from $0.13-146.34\text{ Mg }{\mathrm{ha}}^{-1}$ with amplification of $1.87\pm 1.2\mathrm{\%}\mathrm{ }{\mathrm{year}}^{-4}.$ GWR prediction indicted positive bias with comparatively least mean accuracy ($R^2=0.59,$ $\mathrm{RMSE}=13.72,$ $\mathrm{MAE}=12.85\text{ Mg }{\mathrm{ha}}^{-1}$). The ET—AGC reciprocity for the dominant species resulted that, bamboo with lower AGC$\mathrm{ }\sim 25\text{ Mg }{\mathrm{ha}}^{-1}$ correlated with higher ET$\mathrm{ }\sim 3\text{ mm }{\mathrm{day}}^{-1}$ tailed by teak with higher AGC$\mathrm{ }\sim 45\text{ Mg }{\mathrm{ha}}^{-1}$ and ET$\mathrm{ }\sim 2.5\text{ mm }{\mathrm{day}}^{-1}$ and eucalyptus with relatively higher AGC and lower ET$\mathrm{ }\sim 50\text{ Mg }{\mathrm{ha}}^{-1}$ and $\sim 3\text{ mm }{\mathrm{day}}^{-1},$ respectively. The analysis resulted in minimal biasness in AGC mapping using SMLR, and both the model signifies that the region can potentially be considered a long-term carbon sink.

摘要

绘制和量化地上碳 (AGC) 储量反映了陆地碳循环和级联气候变化的重要动态。对印度泰米尔纳德邦 Coimbatore 和 Nilgiris 生物圈（2006 - 2018 年四重区间）的优势物种（竹子、桉树和柚木）进行了此类关键驱动因素的估计，并使用了发达的全球逐步多元线性回归（SMLR）和当地地理加权回归(GWR) 模型使用多动态变量。使用最适合的 AGC 估计模型分析该区域使用高分辨率蒸发蒸腾映射和内部校准 (METRIC) 模型开发的蒸散量 (ET)，以了解 AGC-ET 协同相关性动态。${\mathrm{电阻}}^2=0.84,$ $\mathrm{均方根误差}=5.67,$ $\mathrm{MAE}=4.61\text{ 镁 }{\mathrm{哈}}^{-1}$) 在估计范围内具有名义上的负偏差 $0.13-146.34\text{ 镁 }{\mathrm{哈}}^{-1}$ 随着放大 $1.87\pm 1.2\mathrm{\%}\mathrm{ }{\mathrm{年}}^{-4}.$ GWR 预测以相对最低的平均准确度（${\mathrm{电阻}}^2=0.59,$ $\mathrm{均方根误差}=13.72,$ $\mathrm{MAE}=12.85\text{ 镁 }{\mathrm{哈}}^{-1}$）。优势种的 ET-AGC 互易性导致，具有较低 AGC 的竹子$\mathrm{ }～25\text{ 镁 }{\mathrm{哈}}^{-1}$ 与更高的 ET 相关$\mathrm{ }～3\text{ 毫米 }{\mathrm{日}}^{-1}$ 尾随具有较高 AGC 的柚木$\mathrm{ }～45\text{ 镁 }{\mathrm{哈}}^{-1}$ 和 ET$\mathrm{ }～2.5\text{ 毫米 }{\mathrm{日}}^{-1}$ 和桉树具有相对较高的 AGC 和较低的 ET$\mathrm{ }～50\text{ 镁 }{\mathrm{哈}}^{-1}$ 和 $～3\text{ 毫米 }{\mathrm{日}}^{-1},$分别。该分析导致使用 SMLR 的 AGC 映射中的偏差最小，并且这两个模型都表明该区域可能被视为长期碳汇。