Data preprocessing
The Terrestrial Ecosystem Service Value Distribution Database used in this study was constructed by Xie using attitudinal methods (http://www.resdc.cn) (Table S1).
According to China’s green development policy (Ji et al. 2017), the factors related to ecological conditions and sustainability were divided based on six aspects: resource utilization, environmental governance, environmental quality, ecological protection, growth quality, and green life. Due to the spatial scale of this study, natural conditions such as temperature, perception, and humidity are relatively similar across the city, so they were not used as model input for differentiation. Taking the correlation with ESV and anthropogenic controllability into account, 23 indexes from 2015 were chosen and altered from three perspectives (Table S1) as input for the following ESV deep learning model. The data resources and spatial revolution of raster data are shown in Table S1.
Multidata fusion on the same scale (Openshaw 1984; Perkins 2017; Hodgkinson and Andresen 2019) is necessary to make labeled samples meet the common format and quantity requirements for deep learning model training. Considering the area of Nanjing and the ecological significance of ESV, we used a 2*2 km grid for data processing by grid transformation. A total of 2191 grid units were obtained as samples. The socioeconomic data were allocated by key weighting factors (Table S2) using Eq. 1. Spatial socioeconomic data with a resolution of 2*2 km were obtained.
\(\text{Af}_{i}=\text{As}_{i}\times\frac{\text{Wf}_{i}}{\text{Ws}_{i}}\)(Eq. 1)