Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Titre : Precipitation controls on carbon and water relations in two African ecosystems

Auteur : Räsänen, Matti

Université de soutenance : University of Helsinki,

Grade : Doctoral dissertation (article-based) // Doctoral Programme in Atmospheric Sciences 2020

Résumé
Understanding the interaction between precipitation and vegetation growth in water-limited ecosystems is vital for various livelihoods that depend on water resources. Precipitation is the primary driver of vegetation growth in dry ecosystems, while fog deposition is essential for the microclimate at dry coastal ecosystems and cloud forests. The analysis of soil moisture, which incorporates the action of climate, soil, and vegetation, is the key to understanding the carbon and water relations and the interaction between precipitation and vegetation.

This thesis examines the impacts of precipitation variability on carbon and water relations in African savannas and the similarities in rainfall and fog deposition. The ecosystem-scale transpiration was estimated from eddy covariance measurements based on annually fitted water use efficiency and optimality hypothesis. The soil moisture measurements were analyzed using a hierarchy of soil moisture models with precipitation, NDVI, and potential evapotranspiration (PET) variability. The statistics of fog and rainfall were analyzed using an analogy with self-organized criticality.

The annual evapotranspiration (ET) was comparable to the annual precipitation at the grazed savanna grassland. While the annual precipitation was highly variable, the estimated annual transpiration was nearly constant 55 % of ET. The transpiration (T) was reduced only during the drought year due to grass dieback-regrowth and possibly due to other changes in soil surface properties that enhanced evaporation. The annual net CO2 exchange (NEE) had large variation ranging from –58 (sink) to 198 (source) gC m-2 yr-1. The annual NEE was related to the maximum of remotely sensed vegetation index (NDVI), and the annual ecosystem respiration was strongly correlated with early season rainfall amount. The analysis of measured soil moisture across savannas showed that NDVI and PET adjustments to daily maximum ET are necessary for modeling depth averaged soil moisture. The soil moisture memory timescale, a rough measure of the time it takes for a soil column to forget the initial soil moisture state, was linearly related to daily mean precipitation intensity at semi-arid savannas.

Both rainfall and fog time series showed approximate power-law relations for dry period and event size distributions consistent with self-organized criticality prediction. The spectral exponents of the on-off time series of the fog and rainfall exhibited an approximate f(-0.8) scaling, but the on-off switching was not entirely independent from the amplitude intermittency in fog and rainfall.

The results show the role of short and long-term variability in precipitation and its consequences for the carbon and water cycle of semi-arid savannas with significant tree cover. These findings can be used to develop minimalist water balance models to understand how vegetation state affects water resources.

Présentation

Version intégrale (1,4 Mb)