Seasonality of Sun-Induced Fluorescence in the Amazon River Basin and its Relation to Precipitation and Land Use

Abstract:

The influence of land cover change on photosynthesis and precipitation in the Amazon River basin are studied along deforested (D) and well-preserved (P) areas at the northeast (NE), northwest (NW), southeast (SE) and southwest (SW) regions, using CHIRPS precipitation data, and Sun Induced Fluorescence (SIF) data from Terrestrial Ecosystems Retrieval version 2 (SIFTER v2), during 2007-2016 at 0.5°spatial resolution. A total of 17 deforested areas, with their respective nearby preserved paired areas, were delimited based on forest loss accumulated up to 2016. Overall, results show that SIF over the entire Amazon River basin exhibits a strong decreasing long-term trend, which is even stronger in D than in P areas. The annual cycle of SIF in D areas exhibits higher values than in P areas during the wet season (December-May), but lower values during the dry season (June-November) (Fig.1a). The annual cycle of rainfall in D areas exhibits much lower values than in P areas from April to October (Fig. 1b). Using cross wavelet analysis, SIF and precipitation were found to covary positively at the annual scale in areas located south of the equator, where higher SIF values were recorded during the rainy season or periods of relatively higher precipitation. Based on the Smirnov-Kolmogorov test and the cross-wavelet analysis, significant differences in the SIF seasonal cycle of deforested areas compared to well preserved areas were found at the SE of the basin, with SIF being lower (higher) during the dry (rainy) season (P < 0.05), with a stronger power at the 12-month frequency band. We also used a time-lagged causal inference method, PCMCI, to perform diverse statistical tests to infer non-linear causalities, including linear partial correlations (ParCorr) and three types of nonlinear independence tests: GPDC, CMI, and PCMCIplus (Runge et al., 2019). Results of the PCMCI test show that deforestation increases the temporal nonlinear persistence of SIF and precipitation, but also the nonlinear causality between both variables in the Amazon, which deserves further investigation.

Reference

Runge, J., Bathiany, S., Bollt, E., et al. (2019). Inferring causation from time series in Earth system sciences, Nat. Commun., 10, 2553, https://doi.org/10.1038/s41467-019-10105-3.