Nanjing Agricultural University, National Engineering and Technology Center for InformationAgriculture, Nanjing, Jiangsu 210095, ChinaAbstract. A regional rice (Oryza sativa) grain yield prediction technique was proposed by inte-gration of ground-based and spaceborne remote sensing (RS) data with the rice growth model(RiceGrow) through a new particle swarm optimization (PSO) algorithm. Based on an initial-ization/parameterization strategy (calibration), two agronomic indicators, leaf area index (LAI)and leaf nitrogen accumulation (LNA) remotely sensed by field spectra and satellite images,were combined to serve as an external assimilation parameter and integrated with theRiceGrow model for inversion of three model management parameters, including sowingdate, sowing rate, and nitrogen rate. Rice grain yield was then predicted by inputting these opti-mized parameters into the reinitialized model. PSO was used for the parameterization andregionalization of the integrated model and compared with the shuffled complex evolution–University of Arizona (SCE-UA) optimization algorithm. The test results showed that LAItogether with LNA as the integrated parameter performed better than each alone for cropmodel parameter initialization. PSO also performed better than SCE-UA in terms of runningefficiency and assimilation results, indicating that PSO is a reliable optimization method forassimilating RS information and the crop growth model. The integrated model also hadimproved precision for predicting rice grain yield.