Abstract:

It is of great significance for regional ecological construction to grasp the dynamic change characteristics of ecosystem service value (ESV) and clarify the evolution trend of its leading factors. Based on the remote sensing interpretation data of 2005, 2010, 2015 and 2019, the characteristics of ESV dynamic changes in Hubei Province were analyzed using the equivalent factor method and spatial statistical tools. Furthermore, the random forest model was used to clarify the dominant factors of ESV, and the spatio-temporal weighted regression model was further used to explore the dominant factors. The results show that (1) the amount of ESV in Hubei Province from 2005 to 2019 decreased by 124.31×10⁸ yuan, but the ESV’s decreasing rate at 2005-2010 decreased from 1.14% to 0.57% during 2015-2019. Over the years, ESVs presented a distribution feature of “higher in the surrounding and lower in the middle”. During the study period, ESV in 83.4% regions fluctuated slightly and presented at a stable spatial distribution. (2) The results of random forest sequencing showed that social and economic factors were more important to ESV, followed by natural factors. The most important factors were human activity (60.16%), net primary productivity (50.66%), annual rainfall (46.85%), gross regional product density (34.00%), population density (29.53%) and construction land proportion (29.36%), respectively, and the topographic and temperature ranked behind of them. The results of the importance of ESV leading factors changes showed that factors increased in varying degrees, except the human activities and precipitation. (3) The strength and fluctuation trends of the dominant factors for ESV in different regions varied. Human activities had the strongest driving effect on ESV, showing a global negative correlation, which was most obvious in western Hubei. The positive impact area of net primary productivity was mainly located in the east of Enshi County, the west of Yichang City and the west of Shiyan City. The negative impact of annual rainfall was gradually stabilized in the Jianghan Plain. The fluctuation trends of population density and GDP density in the Three Gorges and Danjiangkou reservoir areas diverged since 2010. The fluctuation trend of the proportion of construction land was positive at west and negative at east. The research results can provide a theoretical basis for the formulation of ecological protection policies and sustainable development in Hubei Province.

Key words: ecosystem service value, temporal-spatial evolution characteristics, random forest model, influencing factor, GTWR model

摘要：

把握生态系统服务价值（ESV）动态变化特征，厘清其主导因素的演变趋势，对于区域生态建设意义重大。基于2005、2010、2015和2019年的遥感解译数据，结合当量因子法和空间统计工具，分析了湖北省ESV的动态变化特征，利用随机森林模型识别ESV的主导因素，并运用时空地理加权回归（GTWR）模型对其主导因素的驱动作用进行探究。结果显示，（1）2005—2019年湖北省ESV整体减少124.31×10⁸元，但是下降率由2005—2010年的1.14%缩减至2015—2019年的0.57%。历年ESV呈现“四周高中间低”的分布特征，研究期间有83.4%区域的ESV处于上下轻微波动的状态，空间分布较为稳定。（2）随机森林排序结果显示，社会经济因素对ESV的重要程度更高，自然因素次之。重要性排名前六位的分别是人类活动（60.16%）、净初级生产力（50.66%）、年降雨量（46.85%）、地区生产总值密度（34.00%）、人口密度（29.53%）和建设用地占比（29.36%），地形因素及年均温度排名相对较低。ESV主导因素重要性变化结果表明，研究时段内，除人类活动和降水等因素的重要性呈现下降趋势外，其他因素的重要性均有不同程度的上升。（3）不同区域ESV主导因素的作用强度和波动方向各不相同。其中，人类活动对ESV的驱动作用最强，呈现全局负相关，在鄂西最为明显；净初级生产力的正向影响区域主要集中在恩施州东部、宜昌市西部以及十堰市西部；年降雨量的负向影响在江汉平原逐渐稳定；人口密度和GDP密度在三峡和丹江口库区的波动方向于2010年后呈相背离趋势；建设用地占比的波动方向呈现明显的西正东负。研究结果可为湖北省制定针对性的生态保护政策提供参考，同时为生态与经济协调可持续发展提供理论依据。

关键词: 生态系统服务价值, 时空演变, 随机森林模型, 影响因素, 时空地理加权回归模型