Drought is a frequent and potential natural phenomenon, that directly affects social and economic development, industrial and agricultural production, urban and rural water supply, people's life and ecological environment (Diffenbaugh et al. 2005). The occurrence of drought is long and destructive, and scholars have been concerned about this phenomenon for a long time (Svoboda et al. 2002). Drought has recently become a worldwide concern (Dai 2011).
Solar-induced chlorophyll fluorescence (SIF) can reflect changes in photosynthesis (McFarlane et al. 1980). SIF can be used to characterize the response of plants to leaf and canopy water content (Daumard et al. 2010; Grace et al. 2007), and it is closely related to the photosynthesis of vegetation (McFarlane et al. 1980; Guanter et al. 2014). Moreover, SIF has been proved to have a close relationship with gross primary productivity (GPP) and drought (Liu et al. 2019; Migliavacca et al. 2017).
SIF is a kind of long-wave signal emitted by vegetation with 600–800 nm range after absorbing energy under the sunlight (Baker 2008; Lichtenthaler et al. 1986). The growing number of SIF datasets can be obtained from ground-based, airborne, and satellites (Liu et al. 2019). The satellites platforms and instruments include Greenhouse Gases Observing SATellite (GOSAT) (Frankenberg et al. 2011; J Joiner et al. 2011), Global Ozone Monitoring Experiment-2 (GOME-2) (J Joiner et al. 2013; Köhler, Guanter, and Joiner 2015), Scanning Imaging Absorption Spectrometer for Atmospheric CHartographY (SCIAMACHY) (Joanna Joiner et al. 2016; J Joiner et al. 2012; Köhler, Guanter, and Joiner 2015), the Orbiting Carbon Observatory-2 (OCO-2) (Frankenberg et al. 2014) etc. Using the datasets obtained by these satellites or sensors, various inversion algorithms have been proposed (Duveiller et al. 2019; Menghao Ji, Bohui Tang, and Zhaoliang Li 2019). Two kinds of inversion algorithms are recently used, namely, the inversion algorithm based on the physical model (Frankenberg et al. 2011; J Joiner et al. 2012) and data-driven algorithms (J Joiner et al. 2013; Köhler, Guanter, and Joiner 2015). The global SIF dataset we obtained has a coarser spatial resolution. To achieve global 0.05° products, downscaling methods were used (Duveiller and Cescatti 2016). Those studies either based on fine spatial resolution reflectivity from other satellite or filling vacancies from OCO-2 data (Duveiller et al. 2019; Gentine and Alemohammad 2018; Li and Xiao 2019; Yu et al. 2019; Y. Zhang et al. 2018). However, to study the carbon cycle and crop growth in regional scale, SIF products with higher resolution were needed.
The random forest (RF) downscaled method was used in this investigation and the downscaled SIF anomaly index was used to monitor and analysis drought. This study aims to: (1) introduce RF-based downscaled method and obtain the SIF data with 1km spatial resolution; (2) verify the SIF result after downscaling; (3) monitor and analysis drought using 1km spatial resolution downscaling SIF.