Lake uxes and inuence of the soil moisture

It is well known that heterogeneous surfaces aect landscape-scale uxes. The mea-surements presented above (Section 3.1.1) have shown that the uxes over land and lake surfaces behave dierently, as do the uxes over wet and dry alpine steppe. Within the Nam Co 2009 experiment turbulent uxes were observed with two stations. While the EC station NamITP always measured uxes over land surface grass⁻, the measure-ments of Nam UBT originated from dierent surface types. Within the footprint of this station the surfaces grass⁺, grass⁻ and lake could be found, with a strong diurnal varia-tion within contribuvaria-tion of the underlaying surfaces to the measured ux due to a land lake circulation system (Figure 3.11). During midday wind was coming predominantly from the direction of the lake while in the morning, evening and night-time hours wind from the land surface dominated. Therefore NamUBT provided ux measurements over the land and water surface only for certain periods of the day. The land surface furthermore was made up of grass⁺and grass⁻. However, as the inuence of grass⁻was comparatively small during neutral and unstable stratication and only larger during stable conditions which occur mostly at night, when ux dierences between the two surface types were small, the inuence can be regarded as negligible. Therefore it is reasonable to relate the observations and parameter over the land surface at NamUBT to the wetter grass⁺surface and to discriminate between grass⁺at NamUBT, grass⁻ at NamITP and the lake only. Although it was possible to attribute the measured uxes to the a specic surface type due to the footprint analyzes a direct comparison of the uxes was nearly impossible since the strict pattern of the lake breeze was causing gaps in one or the other time series at NamUBT. In order to ll the gaps and to complete the ux time series for grass⁺ and lake a model was applied for each surface type and validated with the existing data. The model SEWAB was used in the case of the land surfaces and the model HM in the case of the lake surface (model descriptions: Chapter 2.2.5).

The validation of the model was done by attributing measured uxes to a single surface type with footprint modeling, a tool that relates measurements to specic

sur-faces. For each time step, the footprint approach provides the relative contribution of all involved surfaces to the measured uxes. Fluxes can therefore be selected to their underlying surface. Obviously the investigated surface types have dierences in characteristics, especially the land and the lake surface. It is therefore not surpris-ing to see this reected in the observation-based simulated mean uxes for the whole period (Figure 3.12). The close combination of turbulent ux observations and land surface modeling could show that the mean latent heat ux became more dominant with increasing soil moisture for the land surfaces, showing a mean dierences in latent heat ux between grass⁺ and grass⁻ of −33.5W m⁻².The evaporation over the small lake is even higher, due to its shallow water table resulting in comparatively high sur-face temperatures, the resulting dierence between grass⁺ and lake is −27.3W m⁻². The dierence in sensible heat ux between the surfaces is smaller with 24.0W m⁻²

grass⁻ grass⁺ lake

Mean fluxes in Wm−2

0 50 100 150 200 250

−R_sw R_lw Q_G Q_H Q_E

Figure 3.12. Mean uxes for the three surface types (grass⁻, grass⁺, lake) from observation-based simulations. (Land surfaces: SEWAB, Lake: HM). Net shortwave radiation (Rsw) and net longwave radiation (Rlw) for the lake surface are calculated as explained in Fig. 3.1. For land surface uxes, Qg represents the ground heat ux.

For the lake energy balance it sums up the energy uxes not accounted for, e.g. stor-age change in the water body and ux into the sediment. Error bars indicate 1.96 times the standard error of the mean, based on daily mean uxes; assuming normal distribution and statistical independence of daily mean uxes, the bars would corre-spond to the 95 % condence interval. From Biermann et al. (2014a, Appendix B).

3.3. Linking ux measurements and land surface modeling to estimate regional features

dierence between grass⁺ and grass⁻ and 22.3W m⁻² dierence between grass⁺ and lake.

Using foot analysis to resolve relative contribution of the involved surfaces, gives a great advantage when observations originate from more than one land use type.

This enables simulations to be connected to the observations by calculating a weighted mean from the model output according to the actual land cover contribution, as is shown with the footprint integrated simulations for lake and grass⁺ together with the EC observations at NamUBT in Figure 3.13 for 17^th July. For all wind directions the eddy-covariance measurements can be closely modeled by the footprint integrated simulation. This also holds for measurements with contributions from both surfaces,

−100

00:00 06:00 12:00 18:00 24:00

100 50 0 50

−2 Q in WmE 100 LU contribution in %

Figure 3.13. Source weight integrated modeled uxes at NamUBT exemplarily shown on 17^th July. Displayed are simulated uxes with SEWAB (dashed line) and HM (solid grey line) and integrated simulations (solid black line) according to contribu-tions of lake or land within the footprint. Observacontribu-tions (not energy balance corrected) are shown as black circles. The land use contribution in % is indicated by bar plot, with upwind situations from the land in green and upwind situations from the lake in blue. The time axis is displayed in Beijing standard time (CST), and mean local solar noon during the observation period is at 1400 CST. Modied after Biermann et al. (2014a, Appendix B).

seen in some events on this selected day. As this approach enables to resolve uxes explicitly to the underlying surface it improves the possibility to use these uxes to validate model output of special circulation patterns, compared to a tile approach which would resolve uxes originating from dierent surfaces according to their percentage of occurrence within a grid cell.