COS 223-4 - Assessing and modeling diurnal temperature buffering and evapotranspiration dynamics in forest restoration using ECOSTRESS thermal imaging

Background/Question/Methods

Restoration of temperate forests alters the diurnal (24-hour) dynamics of land surface temperature (LST) and evapotranspiration (ET) through increases in biomass, diversity, and complexity. By how much, and when, LST and ET changes with restoration is however relatively unknown. The change in thermal buffering over time could be used to assess the pace and trajectory of restoration by managers. The thermal buffering provided by restored forest can also be quantified as an ecosystem service as it impacts wildlife habitat and human wellbeing. The ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) is the first space-based instrument that allows for a study of diurnal LST and ET dynamics at a moderate spatial resolution (70 × 70 m). We modeled the LST and ET dynamics of two groups of forest restoration sites (43 sites total, 2 to 75 ha in size) in Southern Ontario, Canada, initially restored from mainly agriculture 2007-2019. We included all useable LST (n=29) and ET (n=9) ECOSTRESS image products from June 1st to September 30^th, 2020. Forest restoration sites were compared by age and against agriculture (pre-restoration state), mature forest (post-restoration state), and suburban residential sites (competing land use).

Results/Conclusions

The ability of forest restoration sites to buffer temperatures was highest around 14:00 local time (EDT). At that time, restoration sites had a significantly lower mean LST (4.4 – 7.4 ℃) compared to residential and agricultural sites, and significantly higher mean LST (1.6 – 2.9 ℃) compared to mature forest sites at both groups of sites. Relative diurnal LST variability of forest and restoration sites was significantly lower (0.9 – 2.9 ℃) compared to agriculture and residential sites (3.3 – 5.2 ℃). Daytime LST decreased significantly by 0.1 ℃ (3.1 %), per year for one of the groups of restoration sites as per a linear mixed effects regression model. The model forecasted a 32 year timeline for restoration to reach the same buffering as mature forests. In characterizing ET dynamics of a subset of sites, we found that more recently restored sites had a significantly higher overall ET than older ones and that daytime relative instantaneous ET decreased by year since restoration. The variation explained by the ET model was however low. Our study provides insights into how diurnal forest ecosystem energy conversion and storage dynamics changes with ecosystem changes due to forest restoration.