Imaging spectroscopy is frequently used to assess traits and functioning of vegetated ecosystems. Applied reflectance- and radiance-based approaches critically rely on accurate estimates of surface irradiance. Accurate retrievals of surface irradiance are, however, nontrivial and often error-prone, thus causing inaccurate estimates of vegetation information. We analyze the irradiance field surrounding an isolated tree using the 3-D radiative transfer model DART in high spatial (25 cm) and spectral (1 nm, 350-2500 nm) resolution. We validate modeled irradiance with in situ measurements and quantify the impact of erroneous surface irradiance estimates on the retrieval of vegetation indices. We observe the irradiance gradients in the cast shadows of <;560% in the blue spectral range, while this gradient decreases with increasing wavelength and becomes negligible in the near infrared (NIR). Furthermore, we quantify a vegetation-induced decrease in the irradiance of <;6% in the visible spectral region and an increase of <;7% in the NIR outside the cast shadow. Commonly employed vegetation indices are also affected by such brightening or darkening effects. Outside the cast shadow, indices sensitive to the relative content of chlorophyll (CHL) and carotenoids (CAR) show an overestimation of <;14%. The photochemical reflectance index shows an underestimation of <;5%. This paper provides first quantitative insight in high spatial and spectral resolution, on the impact of vegetation on its surrounding irradiance field. Findings highlight important implications for vegetation assessments and provide the fundamental base to advance retrievals of vegetation traits and functioning from imaging spectroscopy data.