where \(\text{ETa}_{Sebal,i}\) is the SEBAL actual evapotranspiration for each day and \(\text{ETa}_{i}\) the actual evapotranspiration obtained by Eq. 13. For both semiarid zones the ET0 was determined by using Penman-Monteith FAO equation (Allen et al., 1998). In the Pinares forest, we used the seasonal value of the ratio ETa/ET0 obtained by Liu et al. (2017) for pines (Table 2). In the Caatinga forest (Table 2), ETa/ET0 was calculated from theon-site water balance (Teixeira, 2018).
The trend analysis was performed using Mann-Kendall method (Mann, 1945; Kendall, 1975): it is a nonparametric test to identify a trend in a series. The null hypothesis H0 for this test means that there is no trend in the series, whereas the three alternative hypotheses are a negative, non-null, or positive trend. A significance level of p = 0.05 was applied in this study. The Mann-Kendall test was employed with weather variables (rainfall, maximum and minimum temperature) and with the actual and potential evapotranspiration.
2.6 Temporal Stability Index
The Temporal Stability Index (TSI) was calculated as in Penna et al. (2013) and in Zhao et al. (2010) for each season (autumn, winter, spring and summer) in the Pinares forest. For the Caatinga forest, we calculated TSI for the rainy season and for the dry season, since the four seasons cannot be differentiated there. Eqs. 16, 17, and 18 were used to assess the TSI: