Monitoring agricultural drought in Australia using MTSAT-2 land surface temperature retrievals

• Published in: Remote sensing of environment 2020-01, Vol.236, p.111419, Article 111419
• Main Authors: Hu, Tian, Renzullo, Luigi J., van Dijk, Albert I.J.M., He, Jie, Tian, Siyuan, Xu, Zhihong, Zhou, Jun, Liu, Tengjiao, Liu, Qinhuo
• Format: Article
• Language: English
• Subjects: Agricultural drought; Australian wheatbelt; Correlation coefficient; Correlation coefficients; Crop yield; Detection; Drought; Drought index; Growing season; Land surface temperature; LST morning rise rate; Mathematical analysis; MTSAT-2; Normalized difference vegetative index; Precipitation; Precipitation anomalies; Remote sensing; Satellites; Soil moisture; Soils; Stomata; Sunrise; Surface temperature; Temperature rise; Time lag; TRI; Vegetation; Vegetation index; Wheat; Wheat yield
• ISSN: 0034-4257; 1879-0704
• DOI: 10.1016/j.rse.2019.111419

Drought indices based on thermal remote sensing have been developed and have merit for effective early warning of agricultural droughts, but approaches so far are relatively complex or sensitive to land surface temperature (LST) estimation uncertainties. Here, we propose the temperature rise index (TRI), a drought index that is comparatively robust and easy to calculate, as the anomaly of the intrinsic morning rise of LST. The underlying principle is that the rate of LST rise between 1.5 and 3.5 h after the sunrise is approximately linear and occurs more rapidly under dry conditions than under wet conditions over vegetated surfaces as a consequence of stomatal control. TRI during the growing seasons of 2010–2014 was calculated over the Australian wheatbelt from LST retrievals from the geostationary Multifunction Transport Satellite-2 (MTSAT-2) instrument. The calculated TRI was compared with indices based on precipitation integrated over 1-, 3- and 6-month time scales, on Soil Moisture and Ocean Salinity (SMOS) soil moisture derived from passive microwave remote sensing, and on vegetation condition (normalized difference vegetation index, NDVI) derived from optical remote sensing. The various indices were also compared to annual wheat yield over large areas. The correlation coefficient between TRI and precipitation anomaly that serves as an operational drought index in Australia was above 0.6 in general with 3-month integrative time scale for precipitation. TRI produced spatiotemporal dryness patterns that were very similar to those in soil moisture, but with more detail due to its finer resolution. A time lag of >1 month was found between TRI and observed vegetation condition, supporting the use of TRI in early warning. Among the compared drought indices, TRI explained the largest fraction (35%) of wheat yield variations. TRI correlations with wheat yields peaked higher and earlier by almost one month in comparison to other indices. We conclude that the thermal drought index proposed here shows considerable potential for use in drought early warning as an effective complement. •Land surface temperature morning rise at 1.5–3.5 h after sunrise is quasi-linear.•Vegetation canopy under water stress has a faster temperature morning rise.•Temperature rise index provides earlier drought monitoring than greenness indices.•Temperature rise index peaked higher and earlier by 1 month in correlation to yields.