Regional water balance and matter flow
Final Report Abstract
Summary of results (1st and 2nd project phase) Model development to investigate hydrological fluxes for the Xilin river catchment was in the center of research for the 1st MAGIM project phase. A nested catchment approach was established to measure discharge as well as C and N loads. Additional sampling points were set up to monitor precipitation and groundwater quality. Data from the long term discharge gauging station at Xilinhote were obtained. However, it took more than 2 years to finally get P7 - Regional water balance and matter flow. 136 the data. In response to this delay we not only focused on the development of a robust model approach to simulate hydrological fluxes on the catchment scale, but also investigated field scale development of soil moisture distribution in a geostatistical approach in areas with differing grazing intensities. The Xilin river catchment covers approx. 3600 km². Water is the most limiting factor for plant growth with high evaporation and transpiration rates. Despite little winter precipitation, spring snowmelt creates an annual peak in discharge. During this period, frozen soil layers impede infiltration and retention of melt water in deeper soil layers. A secondary minor discharge peak occurs in summer due to relatively high summer rainfalls of a convective type with a high rainfall intensity and erosion potential. Despite first hopes, we were not able to come up with a sound SWAT model set up for the Xilin catchment. Initially, we first tested a variety of evapotranspiration schemes as evaporation and transpiration losses mainly close the water balance of the region. We should that the SWAT internal estimation of evapotranspiration was well in agreement with results obtained by field measurements using eddy covariance techniques (see P5 for details). However, the mismatch between observed and predicted discharge was especially apparent during spring snow melt events where the model completely failed. In addition, rainfall events in summer lead to dramatic over predictions of discharge. Results suggested that adjustments in the snow subroutines as well as changes in surface properties are needed. However, the results also underscored the importance of further precipitation observations throughout the catchment to better capture the spatial variation in the meteorological input, as well as additional consideration of the wind driven erosion, sublimation and redistribution of fallen snow. Further investigations of stream solutes and groundwater composition also indicated that groundwater contribution might play an essential role in the generation of runoff in the Xilin, a flow contribution that is underrepresented in SWAT. Apart from a lack in capturing relevant hydrological process the failure of SWAT can also be due to the semi-distributed set up of the model. With regard to the size of the catchment and the high evapotranspiration rates it is questionable whether precipitation that falls somewhere in the catchment is really connected to the river system. Precipitation input in the semi-distributed SWAT model however is treated in a way that the entire catchment area receives precipitation and that water that infiltrates will contribute to runoff. Based on these results, we decided that more basic knowledge on rainfall runoff generation processes in the Xilin catchment is needed. Therefore, in the 2nd project phase, we first focused on the P7 - Regional water balance and matter flow. 137 collection of specific data that can help to better understand runoff generation processes. One of the grand challenges in large, ungauged catchments is to quickly and at low costs gain insight into the catchment relevant hydrologic processes which are necessary in land use change oriented hydrological modeling to evaluate if the model of interest captures all relevant hydrological and biogeochemical processes. The research of the 2nd phase made use of environmental tracer to improve the runoff generating processes. Hydrochemical data were collected of the stream and of a set of 9 potential sources, e.g. groundwater wells, rain, tributaries. The data were used in a more advanced technique called end member mixing analysis (EMMA) to identify and quantify the runoff contributing sources. The basic assumption of EMMA is that the stream water is a mixture of sources with fixed chemical compositions, the mixing chemistry is a linear process and the tracers are conservative. EMMA includes the performance of principal component analysis (PCA) to determine the amount of end members needed to explain runoff in the system and identify the end members among potentially sampled source waters. An over determined set of equations based on a least squares procedure is then solved to calculate the contributions of the respective end members. Our findings indicate that the runoff can be explained by three interannually changing end members. In wetter years, such as 2006 and 2008, the runoff is mainly produced by shallow groundwater sources. Dryer years exhibit a dominance of deeper groundwater aquifers as sources of stream flow. Rain is in all years only of minor importance. However, the 3-year measurement program lacks the documentation of the full spectrum of hydrologic conditions, e.g. very wet years and very dry years and hence prompts the continued development of long-term measurement program. Despite the limitations of EMMA, which are the assumptions of temporal and spatial invariance of the end members and which probably contribute a large part to the uncertainty in our models, this work is one step forward in improving catchment understanding in the Xilin river basin. We then developed an automatic procedure that iteratively conducts EMMA, varying the tracer set size and composition. We found that this approach is highly complementary to the traditional approach applied earlier in the project phase. This approach identified the sand dune groundwater aquifer, a deep groundwater aquifer in combination with either a tributary or the headwater area as most important runoff sources. The combined knowledge from the studies in project phase 2 together with results from the 1st phase lead us to the development of a conceptual reservoir model (Fig 4.8.1) that will be implemented in the Catchment Model Framework (CMF) next. CMF is a hydrological tool box that facilitates the straight forward implementation of newly acquired process understanding P7 - Regional water balance and matter flow. 138 into a hydrological model. CMF allows testing hypotheses such as the contribution of different groundwater aquifers to runoff generation. Fig 4.8.1: A conceptual reservoir model of the upper Xilin watershed consisting of 5 zones: SD = sand dunes, Marsh = marshland, Grass = grassland, T1 = tributary and H = headwater. P, E and T depict precipitation, evaporation and transpiration, respectively. Precipitation, that falls as snow (e.g. Marshsnow), is modeled with an energy balance model. The brown boxes represent multiple layers of the unsaturated zone (U) which is modeled with the Richards Equation (RE). The arrows represent water flow between the various storages (e.g. G1-5 = groundwater storages 1-5), to the Xilin river (QX) and to the tributary (QT). The question marks (?) highlight the model processes to be tested as hypotheses. Apart from hydrometric data and hydrochemical information on water solutes, P7 also obtained long term data of C and N losses throughout both project phases. These data are currently evaluated to estimate the long term aquatic losses of C and N and to assess their relative importance to the regional C and N budget. Preliminary results show that aquatic losses in the order of 6 kg C ha-1 a-1 and 0.2 kg N ha-1 a-1 contribute only minor to the landscape scale C and N balance.
Publications
- (2007): Evaluation of evapotranspiration methods for model validation in a semi-arid watershed in northern China. Advances in Geosciences 11, 37-42
Schneider, K., Ketzer, B., Breuer, L., Vaché, K.B., Bernhofer, C., Frede H.G.
(See online at https://doi.org/10.5194/adgeo-11-37-2007) - (2008): Ambiguous effects of grazing intensity on surface soil moisture: a geostatistical case study from a steppe environment in Inner Mongolia, PR China. Journal of Arid Environments, 72, 1305-1319
Schneider, K., Huisman, J.A., Breuer, L., Frede, H.G.
(See online at https://doi.org/10.1016/j.jaridenv.2008.02.002) - (2008): Gauging the unauged basin: a topdown approach in a large semiarid watershed in China. Advanced Geosciences, 18, 3-8
Barthold, F.K., Sayama, T., Schneider, K., Breuer, L., Vaché, K.B., Frede, H.G., McDonnell, J.J.
(See online at https://doi.org/10.5194/adgeo-18-3-2008) - (2008): Temporal stability of soil moisture in various semi-arid steppe environments and its application in remote sensing. Journal of Hydrology, 359, 16-29
Schneider, K., Huisman, J.A., Breuer, L., Zhao, Y., Frede H.G.
(See online at https://doi.org/10.1016/j.jhydrol.2008.06.016) - (2010): Identification of geographic runoff sources in a data sparse region: hydrological processes and the limitations of tracer-based approaches. Hydrolocical Processes, 24 (16), 2313-2327
Barthold, F.K., Wu, J.K., Vache, K.B., Schneider, K., Fede, H.G., Breuer, L.
(See online at https://doi.org/10.1002/hyp.7678) - (2010): Spatial and temporal variation of soil moisture in dependence of multiple environmental parameters in semi-arid grasslands. Plant and Soil, 340 (1-2), 73-88
Schneider, K., Leopold, U., Gerschlauer, F., Barthold, F., Giese, M., Steffens, M., Hoffmann, C., Frede, H.G., Breuer L.
(See online at https://doi.org/10.1007/s11104-010-0692-8) - (2011): Spatial and temporal variation of soil moisture in dependence of multiple environmental parameters in semi-arid grasslands. Plant and Soil, 340 (1-2), 73-88
Schneider, K., Leopold, U., Gerschlauer, F., Barthold, F., Giese, M., Steffens, M., Hoffmann, C., Frede, H.G., Breuer, L.
(See online at https://doi.org/10.1007/s11104-010-0692-8)
