Wind effects on evapotranspiration

Simulated and measured ET increase with distance from the tree strip for westerly and north-westerly winds (Fig. 4.10). With north winds, ET is reduced only near the tree strips, while in the centre, ET is reduced neither for simulations nor for measurements.

The visual agreement between ET from simulations and measurements, contradicts with differences between the magnitude of ET. The median of the relative ET reduction from measurements is lower than from simulations in the lee of the tree strips for west and north-west wind, whereas at the windward site the median of the relative ET reduction from measurements and simulations agree.

We interpret the disagreement between ET from simulations and measurements for west and north-west winds as an effect of the unaccounted spatial variability of air temperature, relative humidity and incident radiation in the simulations. In ET derived from simulations we account only for a spatially varying wind velocity and assume that incident radiation, relative humidity and air temperature are the same across the whole agroforestry system.

In contrast to the derivation of ET from simulations, ET derived from the microclimate measurements within the horizontal transect are spatially variable. The assumption of equal meteorological conditions across the whole agroforestry system for simulations hold only for the windward site of the tree strips. There, the air undergoes better mixing due to increased wind velocity. This causes a faster removal of moist air, leading to a higher vapour pressure deficit and subsequently increased ET. Our interpretation agrees with McNaughton,1988, who showed that the water vapour pressure during the middle of the day was higher in the sheltered area behind the tree strips (the quiet zone) and lower in the unsheltered area at the windward site of the tree strip (the wake zone).

Model Measured

231 239 245 251 260 269 275 281 289

(a)

231 239 245 251 260 269 275 281 289

(b)

231 239 245 251 260 269 275 281 289

(c)

x (m)

Relative ET reduction (%)

Figure 4.10: Relative reduction of evapotranspiration for west wind (a), north-west wind (b) and north wind (c) at a domain height of 2 m above ground, at y=320 m and at the westerly 48 m wide crop field. The time series correspond to the respective model simulation with a tree height of 2 m. The data from the model simulation represent the mean (solid line) and the standard deviation (ribbons) over the last 15 minutes of the 30 minute simulation time. Boxplots correspond to relative reductions of potential evapotranspiration derived from measurements at the site, filtered for wind direction and wind velocities ≥1 m s⁻¹ at the reference site. For northerly winds (Fig. (c)) we included wind directions from the south as well. The green dashed boxes indicate the tree strip locations.

4.3.2.2 Areal differences between evapotranspiration

Evapotranspiration is reduced within regions of reduced wind velocity for all different combinations of tree heights and wind directions by maximum ≈ -60 mm, as shown by the difference between simulated potential ET over the agroforestry systems and the monoculture system without trees (Fig. 4.11). This reduction is due to the linear dependency of ET0 on wind velocity (Eq. (4.1)).

However, we found the strongest reduction in ET for west and north-west winds, due to the strongest reduction in wind velocity inside the tree strips and over the crop fields between tree strips (Fig. 4.11(a)-(c) for west wind and (d)-(f) for north-west wind). For northerly winds the reduction in ET correspond to the region of the tree strips, whereas in between tree strips ET is only slightly reduced (Fig. 4.11(g)-(i)). Assuming an annual sum of ET of approximately 400 mm (typical for the site (Markwitz et al., 2020)) a difference in ET of maximum 40 mm in between the 8 m tall tree strips for northerly winds would account for 10 % of the annual sum at the specific location. Hence, even for northerly winds 10 % more soil water would be available for crop growth, assuming that wind velocity is the only controlling factor of ET. Previous studies found that the presence of tree strips leads to a wetter surface within the protected area and a reduced soil drying rate, compared to an unprotected area (Blacket al., 1988).

Nevertheless, if the protection of soils for ET losses is of interest winds parallel to the tree strips promote higher ET. To reduce the wind velocity and preventing increased water losses via ET Böhm et al., 2014 suggested to establish tree strips at the edges of the agroforestry system.

0

Figure 4.11: Difference between simulated potential ET over the agroforestry systems and the reference case without trees in a x-y plane for west wind (a)-(c), north-west wind (d)-(f) and north wind (g)-(i) at a domain height of 2 m for tree heights of 2, 5 and 8 m. The wind directions are indicated by the black bold arrows.

4.3.2.3 Cumulative evapotranspiration

The cumulative sum of area averaged potential ET (from now on only ET) for 2016 show a linear increase throughout the vegetation period until October and stays constant until the end of the year for all wind directions and tree heights (Fig. 4.12). ET was reduced for all agroforestry systems relative to the monoculture system without trees (Table4.2).

We observed a decrease in the annual sum of ET with increasing tree height between minimum 5 % and maximum 8 % for westerly and north-westerly winds. The reduction in ET correspond to a stronger wind velocity reduction for westerly and north-westerly winds. With northerly winds we observed a reduction of 3 % independent of tree height.

The potential evapotranspiration adjusted by the crop coefficient follows the same annual cycle, but changes the magnitude of the annual sum (Fig. 4.12). After correction, the annual sum of evapotranspiration amounts to roughly 60 % of the potential evapotranspiration.

The ratio∑︀

ET𝑎𝑐𝑡/∑︀

ET0 is low compared to those observed for a short rotation coppice system in the Czech-Moravian highlands between 62 % and 91 % (Fischer et al.,2013).

The reduction of ET for the adjusted potential evapotranspiration is slightly lower than

for the not-adjusted ET, but in the same order of magnitude.

However, the reduction in system-scale ET is relatively small, compared with the change in ET on plot-scale. The mean reduction in ET across a whole agroforestry system relative to a monoculture system was maximum≈8 %, whereas ET inside the agroforestry system on plot scale was reduced by maximum≈15 % (Fig. 4.10and4.11). In a recent independent study we observed differences between annual sums of actual ET over AF and MC between 1 % and 17 % across five agroforestry and five monoculture systems (Markwitzet al., 2020), without a clear trend on how differences are distributed (higher ET over AF or MC).During the evapotranspiration calculations we considered only the effect of spatially varying wind velocity on evapotranspiration, whereas the effect of air temperature, incident radiation, soil moisture and the plant physiological properties of crops and trees were not. If the parameter would have been considered, differences in evapotranspiration were expected to be higher.

Table 4.2: Reduction of the area averaged annual sum of potential evapotranspiration over agroforestry relative to the monoculture system without trees at a height of 2 m above ground, for tree heights of 2, 5 and 8 m and westerly, north-westerly and northerly winds. The data are representative for c = 0.14 (poplar) and c = 0.2 (black locust). The corrected potential evapotranspiration corresponds to the product of the potential ET and the crop coefficient.

hhhhhhhhh

hhhhhhhhh Wind direction

Tree height (m) 2 5 8 ET0 uncorrected

W 5.2 6.4 7.4

NW 6.0 6.6 7.9

N 2.6 3.1 3.0

ET0 corrected

W 4.8 5.9 3.8

NW 5.5 6.1 7.3

N 2.4 2.9 2.8

Tree height ^{2 m}

Figure 4.12: Cumulative evapotranspiration for a westerly wind direction, (a), a north-westerly wind direction, (b), and a northerly wind direction, (c). Each sub-plot shows the cumulative sum of evapotranspiration for tree heights of 2, 5 and 8 m for the year 2016. The change in evapotranspiration according to the standard deviation of wind velocity is included as ribbons around the lines. Solid lines correspond to the potential evapotranspiration and dashed lines correspond to the potential evapotranspiration corrected by the crop coefficient, indicated by a cc in the legend.