Multiscale analysis of evapotranspiration and carbon assimilation for a long time series of micrometeorological observation in a typical semi-arid Mediterranean ecosystem

Abstract:

The evapotranspiration (ET) process is a key term of soil water balance. In the Mediterranean climates ET represents the main loss term, that could return up to 70 % of annual precipitation to the atmosphere. Due to the high seasonal and annual variability of precipitation typical of these ecosystems, ET may be 90% of annual precipitation. Considering that in the Mediterranean areas most of the available water for drinking purpose and for agriculture depends on the water stored in the artificial basins during the rainy period, the quantification of ET and its dynamics is of great importance.

ET exhibits a temporal pattern that varies from seconds to decades, and it is mainly dependent as well as by precipitation, also by its guiding factors (e.g. soil water moisture, solar radiation and vapor pressure deficit). Hence, identify the main factors that influence ET becomes fundamental to understand its temporal variability, and is needed when modeling ET over different timescales. The case study is the Orroli site in Sardinia (Italy), a semi-arid Mediterranean ecosystem, which is an experimental site for the ALTOS European project of PRIMA MED program, for which are available eddy covariance measurements of sensible heat (H), latent heat (LE) fluxes, and soil moisture, radiation, air temperature and air humidity measurements, over 15 years.

Based on the half-hour time series, the meteorological measurements were considered into investigation, and their variability has been detected at different time scale, from seconds to year. The conventional Pearson correlation coefficient between ET and its guiding factors has been estimated and showed the main influence of soil moisture and vapor pressure deficit on the ET process, which also varies on different timescales.

Furthermore, the orthonormal wavelet transformation (a spectral analysis methodology), was used to investigate the timescale variability of ET in the frequency domain and identify the role of its guiding factors. The ET spectral density has significant peaks at the daily, seasonal and annual timescales. In particular, the variability of the ET spectral density exhibits two order magnitude more than the daily variability. The wavelet cospectra of ET and its guiding factors showed that their interaction is strongest for the seasonal and the annual timescales.