Abstract 110

Abstract ID: 110

Global observations highlight regions where vegetation can enhance S2S predictability

Lead Author: Bethan L. Harris
UK Centre for Ecology & Hydrology/National Centre for Earth Observation, United Kingdom

Keywords: land-atmosphere interactions, vegetation, Earth Observation

Abstract: The land surface is a key source of predictability for forecasts at the subseasonal-to-seasonal (S2S; 2 weeks to 2 months) timescale, since variables such as root zone soil moisture and leaf area vary more slowly than the atmospheric state. Previous work has mostly focused on the predictability gained from realistic soil moisture initialisations. Considering observable land surface variables, vegetation shows more persistent changes than surface soil moisture following subseasonal rainfall events, and therefore has the potential to provide predictability at longer lead times. We therefore perform the first investigation of vegetation feedbacks onto near-surface air temperatures using global daily data, to ascertain in which regions and seasons these feedbacks can provide S2S predictability. We use daily datasets of Vegetation Optical Depth (VOD, from the VODCA X-band product) and 2m temperature (from ERA5) at 0.25° horizontal resolution, and compute lagged correlations to identify where spatial structures in VOD anomalies are associated with similar structure in 2m temperature anomalies. Using daily data allows us to investigate how the correlations decay as a function of lead time within the S2S timescale. At zero lag, water-limited regions exhibit negative correlations, indicating that an increase in vegetation water content is associated with increased evapotranspiration, leading to cooler near-surface air temperatures. We find extensive regions in the semi-arid tropics and sub-tropics where at certain times of year VOD anomaly patterns are anti-correlated with temperature patterns 2 weeks ahead. These periods tend to occur outside of the wettest time of year. In some regions, e.g. southern Africa in MAM, predictability of temperature from VOD anomalies extends to lags of 30 days, suggesting that incorporating vegetation variability can improve S2S forecasting. We develop a model for the strength and persistence of vegetation feedbacks to near-surface temperatures based on seasonal cycles of rainfall and vegetation.

Christopher M. Taylor (UK Centre for Ecology & Hydrology/National Centre for Earth Observation)