Water dynamics (NA)

In terms of water dynamics, the weathering profile aquifer together with the crystalline rock aquitard was evaluated based on static water level measurements. The valley sediments aquifer was evaluated based on dynamic water levels measured in piezometers (Fig. 5.7 p.64). Due to the thicknesses and spa- tial extends of the aquifers, it is assumed that regional groundwater flow takes place in the weathering profile and the underlying crystalline rock, while a local flow system is found in the valley sediments.

Both systems might interact and both influence the wetland. Most pumping wells with groundwater level data are screened in the crystalline rock aquitard, which acts as an aquifer locally and is hydrau- lically connected to the weathering profile aquifer. Hence, in this section, the aquifer system composed of weathering profile and underlying crystalline rock will be referred to as regional aquifer.

7.2.1 Groundwater flow in the regional aquifer (NA)

Static groundwater levels in the regional aquifer ranged between 0.4 and 32.8 m below surface (Fig. 7.11). They were shallowest in valleys and deepest on hilltops and slopes. Calculated isolines of hydrau- lic head show that groundwater flows roughly from northwest to southeast following the regional relief gradient (Fig. 7.11). On a smaller scale, groundwater flows from the hilltops to the valleys. Thus, it is concluded that groundwater flow follows the relief on different scales. Isolines showed effluent condi- tions within the Nasirye Catchment, indicating that wetland and stream in the valley drain the ground- water flow system. However, isolines of hydraulic heads are uncertain, because static groundwater lev- els at different points were not measured simultaneously. Furthermore, uncertainties in hydraulic heads arise from uncertainties in elevation data gained from the DEM. Nevertheless, interpretation of the flow system was more reliable compared to the Ifakara study site, because hydraulic gradients were steeper.

Fig. 7.11: Piezometric map of groundwater in the weathering profile aquifer and crystalline rock aquitard (Aqp2NA and

Aqf1NA), showing isolines of hydraulic heads and point data of static groundwater levels. Data sources: earthex-

plorer.usgs.gov (elevation), GTK Consortium (2012) (streams), Gabiri (unpublished) (Nasirye Stream) (NA).

7.2.2 Groundwater flow in the valley sediment aquifer (NA)

Dynamic groundwater levels in the sedimentary valley aquifer ranged between one meter below surface and one meter above surface (Fig. 7.12). Hence, hydrographs showed a shallow water table within the wetland throughout the year. Groundwater levels were always located in the clay layer, confirming permanently confined conditions. Measured groundwater levels above surface were artesian and did

not represent flooding. In general, groundwater levels in this aquifer were variable and water table fluc- tuations were often rapid.

In order to compare rainy and dry season flow patterns, two extreme points of time were determined representing dry season (low flow conditions) and rainy season (high flow conditions). These points of time were set to 03/01/15 (dry season) and 12/01/15 (rainy season) (Fig. 7.12).

Fig. 7.12: Hydrographs of the piezometers at the three transects. Daily precipitation and the two representative points of time for dry (03/01/15) and rainy seasons (12/01/15) are shown above (NA).

Only few groundwater level time series showed high correlations with each other (Appendix B-BNA1). Hence, grouping based on correlation analyses was unsuccessful.Considering a threshold of 0.95, solely PZ06NA and PZ07NA, as well as PZ25NA and PZ26NA, and PZ46SNA and PZ46LNA were grouped together, indi- cating rather similar behaviors in same transects than in same distances to the stream. Lowering the threshold did not give satisfying groups as well. This shows that groundwater level fluctuations were highly diverse within the wetland.

Groundwater levels were highest after periods of high precipitation and lowest after periods of low or without precipitation (Fig. 7.12). In rainy seasons, groundwater levels rose with each precipitation event, while they fell continuously during dry seasons. Comparing water levels at different wetland positions,

piezometers at center and middle showed rapid responses to rainfall events and high amplitudes (Fig. 7.12). In contrast, piezometers at fringe displayed lower amplitudes and smoother behaviors.

Elevation data gained from SRTM DEM were too uncertain to determine groundwater flow directions in the flat valley sediments. But, a rough major groundwater flow direction from northwest to southeast draining the valley bottom in dry season as well as in rainy season was observed in the valley sediments, according to groundwater flow in the weathering profile aquifer. Based on the higher resolved DEM of Asiimwe (2015), groundwater flow was evaluated on a smaller scale at transect C. The DEM was devel- oped based on differential GNSS (Global Navigation Satellite System) technology (Asiimwe 2015). This method usually shows height accuracies at centimeter level (Paar et al. 2014). However, interpolation of points leads to higher uncertainties at some points (Lee et al. 2005). As the used DEM showed a spatial resolution of 5 m, interpolation uncertainties were assumed to be small. In this study, they were set to 5 cm. Uncertainty of hydraulic heads was calculated using propagation of uncertainties (Taylor 1997), resulting in an uncertainty of 10 cm.

At transect C, groundwater levels were measured in six positions. On the right side of Nasirye Stream, levels were measured in two different depths using long and short piezometers. The long piezometers are screened in the connected sand and gravel lenses and the short piezometers are screened in the upper sand lenses, which are separated from the connected sand and gravel lenses by clay.

Long piezometers revealed a constant flow direction from the slope to the central valley throughout the whole year (Fig. 7.13). However, during rainy season, absolute hydraulic heads were higher compared to dry season. Time series of hydraulic heads at transect C showed that the hydraulic gradient did never change, and hydraulic heads stayed always highest at fringe and lowest at center. This indicates groundwater recharge at the hilltops and slopes flowing to the valleys. While PZ40LNA showed artesian conditions during rainy and dry season, the other piezometers at transect C were solely artesian during rainy season. During rainy season, the stream level was almost at the same height as hydraulic heads of groundwater. However, as the aquifer is confined and the base of the stream bed is located within the impermeable clay layer, an interaction between shallow groundwater and stream water at transect C is not likely. It is rather assumed that stream level as well as groundwater recharge are correlated to pre- cipitation within Nasirye Catchment.

Groundwater levels of the upper sand lenses, measured in short piezometers, showed different hydrau- lic heads compared to long piezometers (Fig. 7.3, Fig. 7.4, Fig. 7.5). At fringe (PZ40NA), the long piezome- ter displayed significantly higher hydraulic heads than the short one (Fig. 7.14). This indicates an up- wards directed pressure gradient, which was already indicated by confined conditions of the aquifer. Hydraulic heads of long and short piezometers showed similar values at middle (PZ42NA). At center (PZ46NA), the long piezometer showed lower hydraulic heads than the short one. However, those differ- ences were smaller compared to fringe. At middle and center, differences between hydraulic heads of long and short piezometers were within the range of uncertainties and could thus not be evaluated.

Fig. 7.13: Hydraulic heads of long piezometers at transect C during dry (01.03.15) and rainy (01.12.15) seasons and stream water level during rainy season (01.12.15).

Fig. 7.14: Hydraulic heads of piezometers at transect C in the central field trials center, middle, and fringe (NA).