Abstract:

In recent years, intensified climate change and human activity have posed severe threats to terrestrial ecosystems stability. Guangdong, one of the most important provinces in China’s economy, is rich in natural resources and has an extensive vegetation cover. However, rapid urbanization and industrialization have led to complex vegetation changes in the Guangdong Province. In the soil-plant-atmosphere continuum, vegetation plays a crucial role in balancing socioeconomic development and environmental protection. To better monitor and understand the characteristics of vegetation change in Guangdong, this study introduced a new vegetation index: the Kernel Normalized Difference Vegetation Index (kNDVI). Compared with traditional NDVI, kNDVI addresses the challenges in monitoring vegetation dynamics owing to vegetation saturation, bias, and complex phenological dynamics. It also excels in noise resistance and data stability, making it ideal for evaluating vegetation change across diverse biomes and climate zones. To verify the applicability of kNDVI in Guangdong, we first compared the changes in monthly NDVI and kNDVI between 2000 and 2022, particlarly from June to September 2021, during which the vegetation cover reached its peak. Based on this comparison, we used the NDVI saturation threshold of 0.80 as a reference, and found that the saturation effect of NDVI may prevent accurate representation of vegetation change in approximately 30% of Guangdong. In contrast, kNDVI reflects the spatiotemporal distribution of vegetation in Guangdong better than NDVI, particularly in areas with high vegetation cover. Using kNDVI as the main vegetation indicator, this study investigated the spatiotemporal characteristics of vegetation dynamics in Guangdong from 2000 to 2022 through Theil-Sen Median trend analysis and Mann-Kendall tests. The Shapley Additive Explanation (SHAP) method was adopted to explore the primary driving factors of kNDVI change. Finally, we analyzed future trends in vegetation change based on the Theil-Sen Median trend analysis of the Hurst index and verified the results with predictions from the Long Short-Term Memory (LSTM) model. Our results indicate the following. 1) Spatial distribution of kNDVI: The annual average kNDVI values in Guangdong ranged from 0.005 to 0.718, showing obvious spatial heterogeneity due to the combined influence of natural conditions and economic development. Generally, the kNDVI gradually increases outward from the Pearl River Delta region. Northern Guangdong, an ecological protection area, recorded the highest annual average kNDVI (0.542), indicating a high vegetation coverage. Western and eastern Guangdong had annual average kNDVI values of 0.507 and 0.446, respectively. In contrast, the Pearl River Delta had the lowest annual average kNDVI (0.373), particularly around Foshan, Dongguan, and Zhongshan. 2) Temporal change in kNDVI: During the tudy period, the proportions of areas with vegetation that experienced extremely significant degradation, significant degradation, non-significant degradation, stability, non-significant improvement, significant improvement, and extremely significant improvement in Guangdong were 3.62%, 1.75%, 10.0%, 0.04%, 24.8%, 12.9%, and 46.9%, respectively. Overall, 84.6% of Guangdong’s area has experienced vegetation improvement, driven mainly by the implementation of ecological protection policies, such as returning farmland to forests and tree planting, as well as the positive effects of green space construction during urbanization. Nonetheless, 15.4% of Guangdong still experiences vegetation degradation. Although the total proportion of significantly and extremely significantly degraded areas was limited, they mostly occurred in the Pearl River Delta and eastern Guangdong due to rapid economic development and the conversion of considerable farmland and natural land into urban residential and commercial areas. 3) Factors influencing kNDVI change: In Guangdong Province, human activity factors, such as GDP growth and population increase, surpassed natural factors, such as precipitation and air temperature, in determining vegetation dynamics. For the entire province, the contribution order of the different factors to vegetation dynamics was GDP>air temperature>population>precipitation. In different sub-regions of Guangdong, while GDP is always the primary factor influencing vegetation dynamics, the contribution order of the other factors varies. For example, in the Pearl River Delta, the contribution ratios of GDP, population, air temperature, and precipitation to vegetation dynamics were 36%, 22.1%, 22%, and 20%, respectively. 4) Future trends of kNDVI: Based on the analysis of the Sen-Hurst index, future vegetation change trends were categorized into four groups: continuous degradation (6.03%), continuous improvement (27.4%), reversal from continuous degradation (9.35%), and reversal from continuous improvement (57.2%). The regions expected to experience an increasing trend of vegetation cover accounted for 36.7% of Guangdong, whereas the remaining 63.3% of the regions were expected to experience a degradation of vegetation cover. Notably, future vegetation trends in different regions of Guangdong may revert from continuous improvements to degradation. This is consistent with the kNDVI predictions of the LSTM model over the next 50 years (2000‒2072). In addition, according to the significance test of future kNDVI changes simulated by the LSTM model, vegetation in Guangdong may degrade over the next 13 years, even without any changes in current natural and human factors. In conclusion, this study revealed spatiotemporal variations in vegetation in Guangdong from 2000 to 2022 by analyzing the kNDVI index using multiple methods. We also identified the main factors driving vegetation change and predicted future vegetation trends. Our results emphasize the remarkable impact of economic development and human activities on vegetation change, and reveal the potential risk of vegetation degradation in the future. To effectively protect and restore the ecological environment in Guangdong and ensure a balance between socioeconomic development and environmental protection, the government and relevant departments must strengthen the implementation of ecological protection policies, make better land use plans, and execute green and sustainable development.

Key words: kNDVI, spatiotemporal change of vegetation, human activity, future trend, Guangdong Province

摘要：

广东省经济发展迅速，植被资源丰富，研究其植被动态变化对于实现社会经济发展与生态环境保护之间的平衡具有重要意义。基于2000－2022年的核标准化植被指数（kNDVI），探讨了广东省植被的时空变化特征，通过SHAP方法探究了影响kNDVI变化的主导因子，并利用LSTM模型与Hurst指数预测了未来植被变化趋势。结果表明，1）广东省kNDVI整体呈现“中间低、四周高”的格局，粤北地区年平均kNDVI最高，反映了区域较高的植被覆盖率；珠三角地区年平均kNDVI最低，这可能是由于快速的城市化进程导致了自然植被减少。2）研究时段内，广东省仅0.04%的区域植被状况保持稳定；约84.6%的区域植被状况改善，这可能得益于一系列生态保护政策与城市化积极效应的共同作用；15.4%的区域出现植被退化，主要集中在珠三角及粤东地区，这与当地频繁的土地利用变化和城市扩张紧密相关。相较于自然因素（如气温和降水），人为因素（如GDP增长和人口增加）对广东省植被变化的影响更显著。3）未来，广东省植被呈现反持续性像元占比为66.6%，持续性像元占比为33.4%，表明大部分地区植被覆盖的变化趋势可能发生逆转。

关键词: kNDVI, 植被时空变化, 人类活动, 未来趋势, 广东省