Effects of forest age on albedo in boreal forests estimated from MODIS and Landsat albedo retrievals

- Linear unmixing was applied to estimate dependency of MODIS albedo on forest age.
- Exponential function described the decline of coniferous forest albedo with stand age.
- The functions differed between species, wavelength, and season.
- More detailed information on forest albedo could be obtained from Landsat data.

The relationship between boreal forest age and albedo in the visible, near-infrared and shortwave broadband was studied using MODIS and Landsat albedo retrievals and forest resource maps. The dependency of forest albedo on stand age was estimated from MODIS albedo retrievals using a linear unmixing approach combined with non-linear regression. Exponential function was appropriate to describe the decline of coniferous forest albedo with stand age in all seasons and the decline of deciduous broadleaved albedo in winter. The effect of stand age on albedo was stronger for spruce than for pine, and the variation in albedo as a function of stand age was larger in winter than in summer for all species and wavelength regions. When applied to the test data, the albedo values predicted using the estimated age dependent albedo functions produced somewhat better results than when using only the species specific mean albedos as predictors. The tendency of the albedo values predicted using the age dependent functions towards the mean of the observed values suggested that the models did not cover all variation in the species specific forest albedos. A larger variation in the predicted albedos was obtained when age dependent albedo functions were estimated using Landsat albedo estimates.