Detection of vegetation drying signals using diurnal variation of land surface temperature: Application to the 2018 East Asia heatwave

Abstract

Satellite-based vegetation monitoring provides important insights regarding spatiotemporal variations in vegetation growth from a regional to continental scale. Most current vegetation monitoring methodologies rely on spectral vegetation indices (VIs) observed by polar-orbiting satellites, which provide one or a few observations per day. This study proposes a new methodology based on diurnal changes in land surface temperatures (LSTs) using Japan's geostationary satellite, Himawari-8/Advanced Himawari Imager (AHI). AHI thermal infrared observation provides LSTs at 10-min frequencies and ∼ 2 km spatial resolution. The DTC parameters that summarize the diurnal cycle waveform were obtained by fitting a diurnal temperature cycle (DTC) model to the time-series LST information for each day. To clarify the applicability of DTC parameters in detecting vegetation drying under humid climates, DTC parameters from in situ LSTs observed at vegetation sites, as well as those from Himawari-8 LSTs, were evaluated for East Asia. Utilizing the record-breaking heat wave that occurred in East Asia in 2018 as a case study, the anomalies of DTC parameters from the Himawari-8 LSTs were compared with the drying signals indicated by VIs, latent heat fluxes (LE), and surface soil moisture (SM). The results of site-based and satellite-based analyses revealed that DTR (diurnal temperature range) correlates with the evaporative fraction (EF) and SM, whereas Tmax (daily maximum LST) correlates with LE and VIs. Regarding other temperature-related parameters, T0 (LST around sunrise), Ta (temperature rise during daytime), and δT (temperature fall during nighttime) are unstable in quantification by DTC model. Moreover, time-related parameters, such as tm (time reaching Tmax), are more sensitive to topographic slope and geometric conditions than surface thermal properties at humid sites in East Asia, although they correlate with EF and SM at a semi-arid site in Australia. Additionally, the spatial distribution of the DTR anomaly during the 2018 heat wave corresponds with the drying signals indicated as negative SM anomalies. Regions with large positive anomalies in Tmax and DTR correspond to area with visible damage to vegetation, as indicated by negative VI anomalies. Hence, combined Tmax and DTR potentially detects vegetation drying indetectable by VIs, thereby providing earlier and more detailed vegetation monitoring in both humid and semi-arid climates.