Quantitative the urban vegetation carbon storage by constructing seamless and dense time series remote sensing data

 Abstract Urban vegetation is an integral component of the terrestrial ecosystem, and accurate calculation of its carbon storage is essential. Net primary productivity (NPP), as a crucial element in the dynamic surface carbon cycle, plays a key role in characterizing the resilience of terrestrial ecosystems. It also serves as a fundamental indicator for measuring carbon storage. However, the insufficient coverage of effective image data within cities hampers the acquisition of high-quality Normalized Difference Vegetation Index (NDVI) data, posing a challenge for estimating carbon storage across different vegetation periods. Consequently, continuous monthly NPP estimation remains a challenge. In response to this issue, the primary objective of this study is to propose a novel approach for estimating intensive time series NPP by reconstructing NDVI data. We utilized the GF-SG algorithm to construct a densely sampled NDVI time series dataset by merging Landsat 8 and MODIS data, coupled with meteorological information. These inputs were then employed in an improved CASA model, enabling the generation of monthly NPP results at a 30m resolution. Using the Pearl River Delta (PRD) region as a case study, we investigated temporal and spatial variations in NPP resulting from changes in land cover and their impact on vegetation from 2017 to 2020. Our findings suggest that establishing a consistent time series of NDVI data contributes to reducing the uncertainty associated with NPP inversion outcomes. The annual cumulative NPP values varied significantly among different land cover types. Particularly, forest areas exhibited significantly higher cumulative NPP compared to other land types, with the order of NPP values being Crops>Grass>Wetlands. The transformation of land cover within the study area resulted in a substantial reduction in NPP, amounting to a decrease of 29.07 GgC. Importantly, even in the absence of image data, this method can estimate the net primary productivity of vegetation while maintaining high accuracy. Crucially, this approach can support long-term and large-scale monitoring of vegetation carbon storage within urban agglomerations. Keywords: Urban vegetation, carbon storage, Net primary productivity, Improved temporal NDVI, High observation frequency