Leaf phenology plays a critical role in understanding changes in carbon, water, and energy cycles in forest ecosystems, yet prognostic models for leaf phenology in tropical dry deciduous forests (TDDFs) remain limited due to a lack of observational data and insufficient mechanistic understanding. However, recent advances in satellite remote sensing offer increased observational data on leaf phenology. This study aims to develop a new prognostic model for leaf phenology in TDDFs. We first evaluated the leaf area index (LAI) derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) observations aboard the Terra and Aqua satellites, spanning 2001 to 2024. The satellite data revealed weaker seasonality and more complex interannual variability in comparison to temperate forests. Subsequently, we examined the MsTMIP (Multi-scale Synthesis and Terrestrial Model Intercomparison Project) and TRENDY (Trends in Net Land-Atmosphere Carbon Exchange) model simulations of LAI dynamics, finding notable differences in simulated LAI among several water-limited TDDFs. Some models struggled to capture seasonal deciduousness, while others varied in seasonal and interannual patterns. These results indicate that significant uncertainties persist in TDDF phenology modeling, as current models may not fully account for meteorological variability. Based on these findings, we are developing a new TDDF phenology scheme, initially implemented in our Vegetation Integrative Simulator for Trace gases (VISIT), with ongoing field data collection for model validation and optimization.