#### 3.1. Characteristics of Vertical Profiles and Surface Conditions

Temporal evolution of vertical profiles of potential temperatures and moisture, computed by horizontal mean of scalar values, over the experimental sites is plotted in

Figure 2. The solid black line represents the initial data used to initialize the LES PALM model.

A cooler and drier condition was evident at early morning (08:00 LT) featuring a strong stability near the surface during the DRY for both sites. As an initial condition, there is a remaining and deeper NBL about 475 m height. The WET initialized thermal profiles by setting a CBL at 125 m. Because the stability at the surface formed at night condition is more intense during the dry periods, it is possible the CBL growth takes longer to start the diurnal convection. Another fact contributing to the difference in the initial profiles between seasons was the influence of higher humidity during the WET, which causes a strong stable stratification on both sites.

Due to the strong convection that takes place during the day, climatic elements such as rain can cease the growth of the CBL evolution and develop a new structure of mixed layer, keeping this pattern during the residual layer, even in the day after during the nocturnal boundary layer erosion. An example is the DRY case in the forest: at 08:00 LT, the profile presented a case in which two residual layers occur, one just above the NBL height at 450 m and the other one at the interface between PBL and the free atmosphere (around 2625 m).

The moisture profiles during DRY were drier and not so well mixed layer at high altitudes, with lower humidity in the CBL (except a relatively small neutral layer of 9.0 g kg^{−1}) in the initial profile. At late afternoon (14:00–17:00 LT), the moisture profile showed neutral layers of approximately 8.0 g kg^{−1}, with mixed layer height lower than those observed, possibly due to absence of large-scale horizontal advection. At the pasture site, there was greater presence of moisture at high altitudes; since convection contributes to the greater moisture through vertical transport near the surface, it does not show major differences between the observed and simulated altitudes of layers. With the lack of horizontal advection and the increase of moisture at the advent of the rainy season, mixing layers contain higher humidity on FOR and PAS sites, respectively, 12.0 and 13.5 g kg^{−1}.

In general, with respect to the observed profiles, temperature profiles showed differences of about 1.0 K at both sites in WET, while the moisture profile in the DRY ranged 1.0–1.5 g kg^{−1}.

Figure 3 shows the comparison of surface turbulent fluxes (sensible and latent heat) used as forcing boundary conditions. During the DRY period, the mean

$\overline{{w}^{\prime}{\theta}^{\prime}}$ and

$\overline{{w}^{\prime}{q}^{\prime}}$ (maximum of 448.4 W m

^{−2}) over FOR was higher (100.0 W m

^{−2} and 130.0 W m

^{−2}, respectively) than over the PAS. For WET, the mean

$\overline{{w}^{\prime}{q}^{\prime}}$ over the FOR was higher 337.3 W m

^{−2} than over PAS, since this period has more moisture and easily produces a larger flow of moisture. In general,

$\overline{{w}^{\prime}{q}^{\prime}}$ for WET was 40.0 W m

^{−2} greater than during the DRY period and both surface turbulent fluxes (

$\overline{{w}^{\prime}{\theta}^{\prime}}$ and

$\overline{{w}^{\prime}{q}^{\prime}}$) showed diurnal variation.

The profiles of the components of the sensible heat flux (

$\overline{{w}^{\prime}{\theta}^{\prime}}$), are: SGS (dashed line) and the resolved (line dash-dot) partitions of

$\overline{{w}^{\prime}{\theta}^{\prime}}$, and the sum of both terms, which is the total of

$\overline{{w}^{\prime}{\theta}^{\prime}}$ (solid line). In the component profiles, the SGS of

$\overline{{w}^{\prime}{\theta}^{\prime}}$ shows that the model was able to solve most of the vortices, except in locations where the presence of smaller vortices dominate. Analyzing the sensible heat flux profile of both sites and periods (DRY and WET) at 14:00 LT, the forest had the minimum negative heat flux in CBL height, while more intense thermal energy production on the surface was found in pasture. Other sensible heat flux characteristics are presented in

Figure 4 as profile time series.

#### 3.2. Sensible and Latent Heat Flux Time Evolution

Figure 5 shows the time evolution of horizontal average values of sensible heat flux (

$\overline{{w}^{\prime}{\theta}^{\prime}}$) profiles. During the DRY period (

Figure 5a), the

$\overline{{w}^{\prime}{\theta}^{\prime}}$ at forest is lower than the pasture and it becomes negative at the height 425 m at 12:30 LT. However, the most intense negative

$\overline{{w}^{\prime}{\theta}^{\prime}}$, of about −70.0 W m

^{−2}, persisted longer over FOR, between 13:00 LT and 14:30 LT. The

$\overline{{w}^{\prime}{\theta}^{\prime}}$ at pasture (

Figure 5b) showed high values (above 120 W m

^{−2}) between 11:30 and 14:00 LT. In the atmosphere, two vertical increase of positive

$\overline{{w}^{\prime}{\theta}^{\prime}}$ occurred at 11:30 and 14:30 LT, causing more intense convection and providing a rapid growth of CBL height at these times. These abrupt

$\overline{{w}^{\prime}{\theta}^{\prime}}$ growths with height, observed in the pasture, are connected to the residual layer from the previous day. The positive values of

$\overline{{w}^{\prime}{\theta}^{\prime}}$ has a maximum at approximately 1600 m from 14:30 to 15:30 LT.

During WET, the

$\overline{{w}^{\prime}{\theta}^{\prime}}$, compared to DRY, presented in the forest site stronger values, while on the EZ level, DRY had more contribution of

$\overline{{w}^{\prime}{\theta}^{\prime}}$. In the pasture (

Figure 5d),

$\overline{{w}^{\prime}{\theta}^{\prime}}$ are closer to DRY, even with the pronounced

$\overline{{w}^{\prime}{\theta}^{\prime}}$ variation due to clouds in WET. In WET, these values were very intense. Over forest (

Figure 5c), the maximum sensible heat flux presented energies of 174.6 W m

^{−2} at 11:30 LT in surface and −86.5 W m

^{−2} on the ZE (at 14:00 LT), occurring 190 min after the maximum at surface. The physical reason to this difference of time is that the vertical movements at 14:00 LT are more intense, transporting more heat from the surface to the inner of CBL and also with stronger convective upper penetration movements (free atmosphere introducing heat to inside CBL). To the pasture, after a decay of the

$\overline{{w}^{\prime}{\theta}^{\prime}}$ to 51.9 W m

^{−2} at 12:00 LT, a maximum occurred with 231.0 W m

^{−2}, just before another decay (to 20.8 W m

^{−2}). From 14:30 to 16:00 LT occurred a very strong diminution of energy emitted by the surface (with maximum value of 15.9 W m

^{−2}), possibly due to the cloudy atmosphere, due to a value of zero of the

$\overline{{w}^{\prime}{\theta}^{\prime}}$ in the EZ.

The

$\overline{{w}^{\prime}{q}^{\prime}}$ in the DRY forest (

Figure 6a) presented its maximum value at 13:30 LT with approximately 850.0 W m

^{−2}, just below the level of 1000.0 m. In the DRY pasture (

Figure 6b), the maximum flow also occurred at 13:30 LT, with a value of 900.0 W m

^{−2} at 1700.0 m. In the forest site, this flow proved to be more intense until the end of the day, while, in the pasture, shortly after the maximum, there was a reduction in the intensity of

$\overline{{w}^{\prime}{q}^{\prime}}$. In the WET period,

$\overline{{w}^{\prime}{q}^{\prime}}$ over the forest was approximately 500.0 W m

^{−2} between 13:00 and 14:00 LT, while at the top of the CBL its maximum was 687.5 W m

^{−2} at 14:00 LT. Starting at 15:30 LT, there was a considerable reduction of

$\overline{{w}^{\prime}{q}^{\prime}}$ until the end of the day. Over the WET pasture (

Figure 7d), the EZ presented its maximum at 14:00 LT, as did the forest (

Figure 6c), but with lower intensity (450.0 W m

^{−2}). After 14:00 LT,

$\overline{{w}^{\prime}{q}^{\prime}}$ decreased until 102.6 W m

^{-2} at the surface, caused by the strong reduction of flux in the entire CBL. After 15:30 LT,

$\overline{{w}^{\prime}{q}^{\prime}}$ increased again, but did not surpass 300.0 W m

^{−2} inside the CBL. This strong reduction of surface energy flux was also associated with the cloudy atmosphere.

#### 3.3. TKE Budget

Estimated dimensionless TKE budget profiles are presented in

Figure 7 for 14:00 LT, which represents the time of maximum convection activities. The terms are normalized by

${\left({w}_{\ast}\right)}^{3}/{z}_{i}$, in which

${w}_{\ast}$ represents the scale of convective velocity and

${z}_{i}$ the CBL height. The ratio has an order of 10

^{−3} m

^{2} s

^{−3}. The boundary layer height z

_{i} is defined as the height where the minimum negative value of heat flux is located, and the vertical scale of each profile is normalized by its own z

_{i} in each case.

Thermal or buoyancy production (TP) was very characteristic in both seasons, with a maximum at the surface and decreasing with height to the minimum at the top of the CBL (z

_{i} = 1.0). Over the FOR site, due to normalization with local conditions, during DRY (

Figure 7a) and WET (

Figure 7b) periods, more intense TP values occurred in the CBL height than in the PAS, in relation to surface values. However, during DRY, the characteristics of the profile were very distinct between the sites; while in the PAS the reduction of buoyancy production occurred quickly, at the FOR it extended up to about z = 1.5 z

_{i}. The energy at the top of CBL over the FOR interacts strongly with the entrainment zone in the DRY, showing the high buoyancy active contribution to the CBL on this surface.

The shear production (SP), in the DRY period, in the FOR surface had a very high order close to the surface with 2.2, decreasing until 0 by 0.3 z_{i} deep inside the CBL and returning to produce more shear with an order of 1.4. The maximum SP that occurred in the PAS was at the surface with only 0.4. In WET period, PAS was almost totally absent and FOR presented a weak intensity at the top of the CBL top. This difference shows the very intense convection that was produced during the DRY, due to very high wind velocity.

The turbulent transport (TT) with the increase of convection tends to increase in intensity, but it is important to remember that it is not responsible for creating TKE, just for distributing it. Over FOR in DRY, TT near the surface was about −1.0, rapidly crossing 0.0 in z = 0.3z_{i} determining the downward movement for the rest of the extension of the CBL. The downward maximum transport for this surface was 0.2 surpassing the CBL. With very strong ascendant but weak descendent movements a compensation occurred diminishing the ascendant layer. The PAS shows a transport near the surface of −0.7 and, after crossing the axis at approximately z = 0.5z_{i}, remained below 0.4 until the top. In WET, both sites had a very similar TT, with more intense ascendant transportation due to the diurnal convection.

The viscosity dissipation (VD) shows a maximum near the surface especially when the winds have high magnitude, decreasing just above until disappearing above the CBL. In FOR in the DRY period, dissipation reached −3.2 with an average of −0.6 within the boundary layer. In the PAS, the dissipation rate was lower with a maximum at the surface of −1.7 and an average of −0.4 inside the layer. However, in the WET period, VD was less intense than in the DRY, since the SP was also weaker.