Controlling Factors of Root-Zone Soil Moisture Spectra in Tropical and Temperate Forests

Monday, 15 December 2014
Taro Nakai1, Gabriel George Katul2, Ayumi Kotani3, Yasunori Igarashi1, Takeshi Ohta3 and Tomo'omi Kumagai3, (1)HyARC Hydrospheric Atmospheric Research Center, Nagoya, Japan, (2)Duke Univ, Durham, NC, United States, (3)Nagoya University, Nagoya, Japan
Characteristics of root-zone soil moisture spectra in a subtropical monsoon forest in Thailand (Mae Moh) and two warm-temperate forests in the US (Duke) and Japan (Seto) were examined for time scales ranging from 30 minutes to multiple years. These forested areas have comparable maximum leaf area index but markedly different phase relations between evapotranspiration, net radiation, precipitation, and soil moisture. A hierarchy of models that sequentially introduce the spectrum of precipitation, net radiation, and nonlinearites in the damping originating from stomatal controls and drainage losses were used. If the precipitation is random, and the damping term by evapotranspiration and drainage is increased linearly with increasing soil moisture, the temporal variability of soil moisture simplifies to a first order Markov process commonly employed in the analysis of soil moisture in climate models. Its spectrum exhibits a Lorentz function with a white-noise behavior at low frequency and red-noise behavior at high frequency separated by a time-scale constant for intermediate frequencies. Such first order Markov process model with its time scale defined by the maximum wet surface evapotranspiration, soil porosity, and root-zone depth did not represent the observed soil moisture spectra at all three sites. Adding the effect of precipitation and net radiation variability were necessary for representing the actual soil moisture spectra. While the observed soil moisture spectra were satisfactorily reproduced by these additions, the relative importance of precipitation and net radiation to the soil moisture spectra differed between sites. The soil moisture memory, inferred from the observed soil moisture spectra (model decay time scale), was about 25-38 days, which was larger than that determined from maximum wet evapotranspiration and available pore space alone, except that these two time scales in Seto forest were nearly the same.