Topographically-Aware Optimization of the CASA Model for High-Resolution NPP Estimation in Mountain\u2013Plain Transition Zones

The accurate quantification of terrestrial net primary productivity (NPP) is critical for understanding the global carbon cycle. While the Carnegie\u2013Ames\u2013Stanford Approach (CASA) model is widely used for this purpose, its application in complex terrain is often hampered by conventional parameterizations that overlook the critical modulating effects of terrain-induced illumination and phenological dynamics. This oversight leads to systematic biases, including an overestimation of NPP in plains and an underestimation in mountainous regions. To address this, we propose a refined CASA framework that synergistically optimizes three core components. First, we generate high-fidelity, monthly 30-meter NDVI\/kNDVI time series using a Gaussian Filter\u2013Savitzky\u2013Golay (GF\u2013SG) fusion algorithm to enhance the derivation of the fraction of absorbed photosynthetically active radiation (FPAR). Second, we recalibrate the temperature stress factor () by introducing a novel phenological optimum temperature () derived from the Phenofit model. Third, we refine the maximum light-use efficiency () by incorporating a slope\u2013aspect illumination coefficient, F(s,a), to account for topographic shading. In a case study of Mianzhu City, China, a pronounced mountain\u2013plain transition zone, our evaluation of six progressive CASA schemes demonstrates that the GF\u2013SG fusion outperforms the Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM), achieving a higher correlation with Landsat NDVI (0.74 vs. 0.65). The comprehensive optimization scheme (CASA_5) achieved the highest accuracy (R\u00b2 = 0.70), effectively mitigating the classic topographic biases and more faithfully reproducing NPP\u2019s interannual variability, seasonal dynamics, and spatial gradients. Applied from 2000 to 2020, the model reveals a post-2008 earthquake \u201cdegradation\u2013recovery\u2013enhancement\u201d trajectory, identifying forests and cultivated vegetation as primary carbon sinks. This topographically-aware framework provides a robust tool for monitoring carbon fluxes in complex landscapes and offers critical support for ecological assessment and carbon budgeting.