To predict climate accurately, climate models need to realistically represent how sunlight interacts with vegetation. Most models use fixed values for vegetation type, like “broadleaf forest” or “grass”, which do not capture the actual diversity in how green or thick leaves are. In this study, we improve how models represent leaf optical properties by using global maps of chlorophyll content and leaf thickness. These traits were used to calculate how much light leaves reflect and transmit across visible and near-infrared wavelengths using a hyperspectral radiative transfer model.

We then integrated these trait-based values into the Community Earth System Model (CESM), replacing the usual vegetation-type-based settings. The new approach led to noticeable regional changes in surface brightness (albedo), how much sunlight is absorbed, and the energy exchanged between land and atmosphere. In the tropics, the improved model better matched satellite data and revealed a shift in where plants take up carbon: tropical forests absorbed less, while boreal forests absorbed more. These results show that accounting for real-world variation in leaf traits can significantly affect climate simulations. Our work points to a scalable way to make climate models more accurate using data from satellites and plant trait databases.