Measurement and estimation of evapotranspiration in semiarid grassland during the summer season in southwest Siberia

Abstract

This study quantifies actual evapotranspiration ($ET_a$) for a period from June to September2016 measured by two weighable gravitation lysimeters in a semi-arid grassland insouthwest Siberia. As part of a crop rotation system, the first lysimeter was fallow butcovered with ruderal vegetation. The second lysimeter is permanently characterized bypristine steppe vegetation. In addition to $ET_a$ measurements, the referenceevapotranspiration ($ET_o$) is computed by a Penman-Monteith model. The estimates arerelated to the $ET_a$ records and the model is evaluated with regard to its performance in asemi-arid environment. The results indicated an $ET_a$ driven by energy but limited bywater. Within 115 days the total amounts of $ET_a$ ranged from 205 mm to 374.1 mm, anddaily values varied from 0.1 to 6.9 mm day-1. The large differences are caused by thedifferent vegetation cover of the lysimeters. Due to the high and dense canopy of thepristine steppe vegetation, the transpiration term was considerably higher compared tothe ruderal vegetation where soil evaporation took the major part. The daily $ET_a$ recordsdiffered on average by -91.1% to the $ET_o$ estimates. The statistical analyses yielded a lowcorrelation between $ET_a$ of the ruderal vegetation and $ET_o$ but an acceptable modelperformance for the pristine steppe. However, it was observed that ETa occasionallyexceeds $ET_o$, particularly after precipitation. Due to the high water availability and thesubsequent rise of $ET_a$, $ET_o$ was underestimated, whereas it was overestimated during dryperiods. Finally, the quality of the Penman-Monteith model varied substantially with thewater supply at the study site.