Search

Abstract Litterfall load is crucial in maintaining ecosystem health, controlling wildfires, and estimating carbon stock in arid regions. However, there is a lack of spatiotemporal analysis of litterfall in arid riparian forests. This study aims to estimate Litterfall load using a BP neural network based on vegetation indices from Landsat 5 and 8 satellite images, litterfall inventory data, slope, and distance to major river tributaries. It also aims to analyze the spatiotemporal distribution pattern of litter in the research area by estimating and analyzing the spatiotemporal pattern of litterfall along the desert riparian forests of the lower Qarqan and Tarim Rivers from 2001 to 2021. The results show that the initiation of the ecological water transfer project has facilitated the decomposition of litterfall, leading to an initial decline. Subsequently, the vegetation gradually recovered, leading to an increase in leaf litter input. Since 2001, litterfall initially decreased until reaching its lowest value of 4.39 × 109 kg in 2005, followed by a subsequent increase, reaching its highest value of 12.5 × 109 kg in 2021. The study concludes that ecological water conveyance promotes both the decomposition and increase of litterfall. Initially, it accelerates litterfall decomposition, while later stages foster an increase in Litterfall load. Meanwhile, due to the ecological water transfer project and the higher vegetation cover along the Tarim River compared to the Qarqan River, the Tarim River basin experiences higher average Litterfall load and variation.