KIDS model in PCRaster

Adv. Geosci., 21, 125–130, 2009 www.adv-geosci.net/21/125/2009/ Advances in © Author(s) 2009. This work is distributed under Geosciences the Creative Commons Attribution 3.0 License. Hydrologic comparison between a lowland catchment (Kielstau, Germany) and a mountainous catchment (XitaoXi, China) using KIDS model in PCRaster X. Zhang, G. Hörmann, and N. Fohrer Ecology Centre, Department of Hydrology and Water Resources Management, Christian-Albrechts-University of Kiel, Olshausenstr. 75, 24118 Kiel, Germany Received: 23 January 2009 – Revised: 18 March 2009 – Accepted: 28 April 2009 – Published: 12 August 2009 Abstract. The KIDS model (Kielstau Discharge Simulation However, as modelling and data power has increased there model) is a simple rainfall-runoff model developed originally has been a concurrent debate in its disadvantage, e.g. the for the Kielstau catchment. To extend its range of applica- cost and time required in the collection of massive hydro- tion we applied it to a completely different catchment, the logical data. It is also argued that it can lead to more model XitaoXi catchment in China. Kielstau is a small (51 km2 ) uncertainty from the integration of more input data and the lowland basin in Northern Germany, with large proportion increasing number of model parameters, where it can affect of wetland area. And XitaoXi is a mesoscale (2271 km2 ) the model prediction (Gupta et al., 2005). In some cases, mountainous basin in the south of China. Both catchments simple model development like lumped models is sufficient differ greatly in size, topography, landuse, soil properties, in its own right (Silberstein, 2006; Li et al., 2009). They and weather conditions. We compared two catchments in are still used for various applications, including the study of these features and stress on the analysis how the specific hydrological processes (Bingeman et al., 2007), estimation catchment characteristics could guide the adaptation of KIDS of runoff and catchment water balance (Xu, 1999), and as- model and the parameter estimation for streamflow simula- sessment of land use and climate change impacts on runoff tion. The Nash and Sutcliffe coefficient was 0.73 for Kielstau (Akhtar et al., 2008). Because lumped models have relatively and 0.65 for XitaoXi. The results suggest that the applica- few parameters, they can easily be regionalised to predict tion of KIDS model may require adjustments according to runoff. The modelling result is not therefore strongly de- the specific physical background of the study basin. pending on how sophisticated the model is. Another question concerning model application is whether a model is unique for each environmental problem. Even for a perfect model system, the unique properties of a location lead to a very im- 1 Introduction portant identifiability problem to decide the “optimal” model structure and parameter sets (Beven, 2001). As a result, com- Recent years have seen a rapid development of various hy- puter models are needed that can easily be adapted to the drologic models. With the ever-growing technology in re- problem under study. The KIDS model used in this study is mote sensing, data telemetry and computing, model devel- such a flexible model, which is a simple rainfall-runoff con- opment is striving how best to represent the heterogeneous ceptual model with the potential to be adjusted from lumped characteristics of a watershed. Much of the growth in so- to distributed ones. It is programmed in the dynamic mod- phistication of hydrological modelling is attributable to the elling language PCRaster (Wesseling et al., 1996). It was de- digital revolution of distributed models and the availability of veloped for the streamflow simulation in the Kielstau catch- geospatial data through the last hundred years (Vieux, 2004). ment, which is a very flat region with large area of wetlands (Zhang et al., 2007). The model structure was adjusted with integration of wetland representation for a better simulation Correspondence to: X. Zhang result. To extend the range of the KIDS model application we ([email protected]) Published by Copernicus Publications on behalf of the European Geosciences Union. 126 X. Zhang et al.: Hydrologic comparison between lowland and mountainous catchment using KIDS model Fig. 1. Geographic location and GEM for the Kielstau (left) and the XitaoXi basin (right) applied it to a completely different catchment, the mesoscale ital elevation map for both river basins are shown in Fig. 1. mountainous XitaoXi watershed in southern China. The goal The Kielstau catchment is located in the region of Schleswig- of this study was to determine the applicability of KIDS Holstein, Northern Germany, and covers an area of about model for modelling streamflow by comparing these two dif- 50 km2 . The catchment has a rather flat relief, with maxi- ferent watersheds in the hydrologic characteristics and mod- mum elevation difference of 55m. Soils are mainly consist- eling results. Specific objectives of this study were to: (1) ing of Gleysol, Podsol and Luvisol, among which Gleysol analyse the unique features of the two catchments based on belongs to the major wetland soil types (Sponagel, 2005). the available data, (2) compare the simulation results of the Most of the land in this catchment is used for agriculture basic and adapted KIDS models, and (3) check the link be- (87%), forest and urban landuse share the remaining area. tween parameter estimation and hydrologic characteristics of Average annual precipitation is around 860 mm, and evapo- the selected study basin. ration around 400 mm (Schmidtke, 1999). A large fraction of wetland area and the near-surface groundwater level are ob- served in this region (Trepel, 2004), but there is no accurate 2 Methods mapping data for it. The interaction between surface water and groundwater is active, especially in the riparian wetland The analysis framework includes three steps. First it is area for this region (Springer, 2006). worth stressing the watershed and hydrometeorological char- The second study watershed XitaoXi is a 2271 km2 sized acteristics for both large and small watersheds, by account- mountainous basin located in the semitropical zone in South- ing for differences in topography, vegetation, soil properties, ern China. It is a sub-basin of the Taihu Lake. In the Xi- weather conditions and other important hydrologic features. taoXi region, 63.4% of land use is agriculture-used drought With these considerations the KIDS model is then adapted to area and commercial forest, 20% paddy rice land. Average both locations for river discharge simulation. In addition, the rainfall within the watershed is 1466 mm annually, and av- problems of parameter estimation of the hydrology model for erage evaporation from water surface ranges from 800 mm both catchments are discussed with focuses on the parameter to 900 mm annually. The spatio-temporal variations in pre- sensitivity and its link to the unique catchment characteris- cipitation distribution and evaporation value are statistically tics. significant (Gao, 2006). The dominant soil types are red soil and rocky soil. Since these soils tend to have limited water 2.1 Site description storage capacity, the highest fraction of the river discharge comes from surface runoff and interflow. The study was carried out in the Kielstau and XitaoXi wa- tersheds. Geographic location, catchment scale and the dig- Adv. Geosci., 21, 125–130, 2009 www.adv-geosci.net/21/125/2009/ X. Zhang et al.: Hydrologic comparison between lowland and mountainous catchment using KIDS model 127 2.2 Data collection 1GW is storage change of groundwater, 1GW = f (“gw factor”). Basic spatial data for the KIDS model included a digital ele- Flow direction is then determined based on DEM, and vation model (DEM) and meteorological data. Other inputs channel flow is modelled with fully dynamic runoff rout- like soils and land cover are important input parameters to ing using kinematic wave function. For more details see KIDS but optional, as it can be added as extended submod- Hörmann et al. (2007) and Zhang et al. (2007). els to the basic model structure. Based on long-term climatic In our previous study on model structure uncertainty data (1983–1999) from some nearby weather stations, there (Zhang et al., 2008), we cited the method to build up the was no considerable spatial variation around the Kielstau re- KIDS model ensembles in order to find out the “optimal” gion (Zhang, 2006). The climatic data for Kielstau catchment model structure for a specific location. Based on the re- was taken from the data set of the Flenburg station, 9km north sult, we take the basic KIDS model and the optimized model from Kielstau basin (German Weather Service, Deutscher structure for both basins to compare model simulation. For Wetterdienst, DWD). The DEM was provided by Landesver- the Kielstau basin it is model “DW” with consideration of the messungsamt Kiel, and the LANU – Landesamt für Natur influence of agricultural drainage and wetland fraction. The und Umwelt – has provided river discharge values from 1990 agricultural drainage is introduced as the water amount ex- to 1999 (at the official Soltfeld gauge station). Other spatial tracted from the available soil water decided with the new pa- data like land use, soil maps are from the BGR (Bundesamt rameter “drainage factor”. Wetland fraction (12%) is mod- für Geowissenschaften und Rohstoffe). Owing to the data elled based on the soil map as additional water storage layer limitation and the flat area, the model in Kielstau basin is set in the soil zone, which has unlimited water support for evap- up with a completely lumped distribution of precipitation and oration as its actual ET equals the potential ET. The modified evaporation (calculated with Penman-Monteith method). model for the XitaoXi basin is “LTG”, which is coupled with In XitaoXi catchment, precipitation data are available lateral flow process and groundwater outflow threshold, us- from seven stations within the watershed area, and evapo- ing spatial distribution of potential evaporation adjusted with ration data are from two among the seven stations. Con- landuse coefficients. As the behavior of rainwater tends to be sidering the spatial and temporal variation of the climate more affected by lateral flow on slope area, another param- in this mesoscale mountainous area, we used sub-basin dis- eter “lateral factor” is added to the XitaoXi model to gener- tributed rainfall and evaporation (measured with the Chinese ate lateral flow. The water amount recharged from ground- pan standard method). The discharge data set from the Heng- water to river base flow is here restrained by a groundwa- tangcun gauge station is available from 1978 to 1987. All ter outflow threshold, which represents limited influence on data including soil and land use were provided by the Ad- river discharge. Spatial distributed ETp adjusted with lan- ministrative Bureau of TaiHu Basin. duse coefficients referring to Gao et al. (2006), especially for drought area and paddy rice land of season changes. The ad- 2.3 The KIDS model and model adjustments justed KIDS models with submodels added accordingly, are The basic KIDS model is driven by meteorological input data expected to produce better simulations as presented further and simulates river discharge in given river basins as a dy- in the next section. namic function of spatial information. It is composed of one lumped soil layer and one groundwater aquifer, where the flow from soil to groundwater is calculated according 3 Results and discussion to Glugla (1969). Sub-surface flow is modelled as 1-D 3.1 Hydrometeorological comparison bucket flow and the groundwater layer as a linear storage. There are five important parameters contained in the ba- As mentioned above, the two catchments in this study dif- sic module: “Intercp max” (maximum interception amount fer greatly in catchment scale, topography, soil properties, of vegetation cover); “Inf factor” (water infiltration rate of landuse and weather conditions. The analysis of long-term upper soil layer); “SWC” (maximum soil water capacity); mean monthly precipitation shows a very distinct seasonal “soil gw flux” (water seepage rate from soil zone to ground- pattern in the XitaoXi basin, with 75% of rain falling be- water aquifer); “gw factor” (groundwater discharge rate to tween April and October (Monsoon). The average daily the river baseflow). Runoff is calculated on each grid cell: runoff of the XitaoXi river (35.09 m3 /s) is much higher than Runoff = P − ETa − I − 1S − 1GW that of the Kielstau stream (0.45 m3 /s). The difference is sig- nificant as well when considering the different discharge area Where P is precipitation and ETa is actual evapotranspi- of the selected gauge stations for both basins. The runoff rate ration; per unit area is 8.82 l/s/km2 for Kielstau, and 23.02 l/s/km2 I is interception, I = f (1“Intercp max”); for XitaoXi. The two watersheds also differ in streamflow 1S is storage change of soil water, 1S = f (“SWC”, response to summer rains. Runoff efficiency based on the “Inf factor”, “soil gw flux”); ratio of monthly stream flow to monthly precipitation (Wu www.adv-geosci.net/21/125/2009/ Adv. Geosci., 21, 125–130, 2009 128 X. Zhang et al.: Hydrologic comparison between lowland and mountainous catchment using KIDS model Table 1. Nash and Sutcliffe coefficient NS for 5 year period model the field. Moreover, the wetland (“W”) plays an important calibration and a 10 year period validation in the two watersheds. role in the local water cycle. It can increase the capacity of a watershed to impound surface runoff and to enhance evapo- Kielstau XitaoXi transpiration dramatically especially in summer and autumn Periods NS Periods NS seasons. The model performance shows that the influence of drainage and wetland fraction makes a great difference in Default 1990 to 1994 0.08 1979 to 1983 0.21 model efficiency. For the mountainous XitaoXi basin, lateral Calibration 1990 to 1994 0.70 1979 to 1983 0.61 flow (“L”) is required as one of the dominating processes in Validation 1990 to 1999 0.73 1979 to 1988 0.65 sloping area. Spatial distribution of evaporation (“T”) is ad- justed with empirical coefficients applied to various land use types. This is an alternative to the very limited evaporation data from only two weather stations within the large-scale catchment. Groundwater outflow threshold (“G”) is a simply set value to reduce water discharge to base flow. It indicates that the influence from the groundwater is limited. From the model performance assessment, the “LTG” model structure may better capture the hydrological mechanisms in the Xi- taoXi basin. Fig. 2. Seasonal patterns of long-term monthly runoff efficiency 3.3 Parameter calibration (monthly streamflow to monthly precipitation) in the 10-year data periods: 1990–1999 in Kielstau and 1979–1988 in XitaoXi In the adapted model versions, six parameters need to be de- termined by calibration using daily discharge observations. As introduced before, five parameters are same for both study and Johnston, 2008) is plotted in Fig. 2 for the two water- basins and one different: “drainage factor” for Kielstau and sheds. Runoff efficiencies ranged from 0.08 to 0.58 in Kiel- “lateral factor” for XitaoXi. Most of the parameters were stau basin and 0.22 to 0.58 for the XitaoXi catchment. The adjusted on a trial-and-error basis, modifying parameter val- larger variations in runoff efficiency of the Kielstau basin ues within reasonable limits and selecting final values with might be caused by high evapotranspiration in the wetland maximum model efficiency. area of the watershed and its capacity to impound surface The result of the parameter estimation is displayed in Fig- runoff or to deter the streamflow events. The Kielstau and ure 3. The flat response surface of parameters “intercp max” XitaoXi watersheds have quite different hydrologic regimes, and “soil gw flux” indicates low parameter sensitivity for thereby providing diverse data sets to test the adapted KIDS both catchments. The optimum value of “swc” is much model used in this study. higher for the Kielstau than for the XitaoXi, which corre- sponds to large water storage capacity of the loamy soils in 3.2 Model simulation comparison Kielstau. The sharp curves of the parameters “gw factor” and “drainage/lateral factor” for the Kielstau model suggest Runoff simulations were carried for the calibration period the influence of groundwater and drainage is important for from 1990 to 1994 for the Kielstau and from 1979 to 1983 for streamflow simulation, but they are negligible for the Xi- the XitaoXi with the basic KIDS model and its optimal model taoXi catchment. Parameter identification problems will be- version respectively. The resulting NS values (Nash and Sut- come easier to solve with more detailed information about cliffe, 1970) are listed in Table 1. The first results of the the catchments that may be helpful to understand the unique- simulation using the default basic KIDS model yielded a low ness of location and its hydrological processes. model efficiency of 0.08 for Kielstau and 0.2 for XitaoXi. With the adjusted model structure, the model efficiency im- proves significantly to an NS value of 0.73 for the validation 4 Conclusions period for model “DW” of the Kielstau basin and to 0.65 for model “LTG” of the XitaoXi basin. The result demon- Performance of KIDS model simulation was carried out for strates that the selected submodels describe the hydrology of a small flat and a large mountainous watershed in different the catchment fairly well. climates. The differences between the two hydrogeological For the lowland Kielstau basin, the lateral flow may not be regions are significant in many ways like catchment scale, distinct due to the low altitude variance. Instead, the drainage topography, geology, landuse, soil properties, and weather (“D”) reflects anthropogenic influence to some extent, with conditions. For the purpose of river runoff simulation, we the evidence of large proportion of agriculture use in the lo- analysed the unique features based on observations and exist- cal region and drainage pipes and ditches commonly seen in ing data base, discussed the possible model adjustments that Adv. Geosci., 21, 125–130, 2009 www.adv-geosci.net/21/125/2009/ X. Zhang et al.: Hydrologic comparison between lowland and mountainous catchment using KIDS model 129 Fig. 3. Nash-Sutcliffe model efficiency with different, variable parameters in Kielstau and XitaoXi Watershed. can improve simulations, and compared the results in model Edited by: B. Schmalz, K. Bieger, and N. Fohrer efficiency and parameter estimations. Overall, the simula- Reviewed by: H. Bormann and another anonymous referee tion provided satisfactory agreement between observed and simulated discharge. The validated simulation reaches a NS value of 0.73 for Kielstau and 0.65 for XitaoXi. It proved the general applicability and flexibility of KIDS basic model References and submodel ensembles with for specific features of study area. For the Kielstau catchment, lumped model is adequate Akhtar, M., Ahmad, N., and Booij, M. J.: The impact of cli- for this small region and better model performance can be mate change on the water resources of Hindukush–Karakorum– achieved when considering the influence of wetland. Based Himalaya region under different glacier coverage scenarios, J. on the long-term climatic data, it exhibited substantially dif- Hydrol., 355(1–4), 148–163. 2008. ferent flow trends in the Kielstau than in the XitaoXi catch- Bingeman, A. Kouwen, N., and Soulis, E. D.: Validation of the Hy- drological Processes in a Hydrological Model, J. Hydrol. Eng., ment, with lower runoff efficiency during summer months. 11(5), 451–463. 2006. It suggests an important storage function of wetland and Beven, K.: Uniqueness of place and the representation of hydrolog- groundwater, and representative of the wetland components ical processes, Hydrol. Earth Syst. Sci., 4, 203–213. 2000. was necessary as added submodel. In the XitaoXi water- Beven, K.: How far can we go in distributed hydrological mod- shed we observed a significant impact of adjusted evapotran- elling? Hydrol. Earth Syst. Sci., 5(1), 1–12. 2001. spiration for various landuse types and limited effects from Gao, J., Lu, G., Zhao, G., and Li, J.: Watershed data model: a case groundwater. Owing to the larger catchment scale and dis- study of Xitiaoxi sub-waterhed, Taihu Basin. Journal of Lake tinct heterogeneity in topographic characteristics, more ac- Sciences, 18(3), 312–33l, 2006. curate geospatial data and distributed modelling are crucial Glugla, G.: Berechnungsverfahren zur Ermittlung des ak- for more accurate and reliable hydrologic predictions. The tuellen Wassergehalts und Gravitationswasserabflusses im Bo- parameter calibrations in both case studies demonstrate the den, Albrecht-Thaer-Archiv, 13(4), 371–376. 1969. Gupta, H. V., Beven, K. J., and Wagener, T.: Model calibration and strong link between parameter estimation and the observed uncertainty estimation. In: Encyclopedia of hydrologic sciences, catchment features. This study stressed that, for simulating edited by: Anderson, M., Wiley, Chichester, UK, 2005 the hydrological behavior of a watershed, we should consider Hörmann, G., Zhang, X., and Fohrer, N.: Comparison of a simple the unique features of the place much more explicitly (Beven, and a spatially distributed hydrologic model for the simulation of 2000), and adapt the models to the local situation. a lowland catchment in Northern Germany, Ecol. Model., 209(1), 21–28. 2007. www.adv-geosci.net/21/125/2009/ Adv. Geosci., 21, 125–130, 2009 130 X. Zhang et al.: Hydrologic comparison between lowland and mountainous catchment using KIDS model Li, H., Zhang, Y., and Chiew, F. H. S.: Predicting runoff Wesseling, C. G., Karssenberg, D. J., Burrough, P. A., and Van in ungauged catchments by using Xinanjiang model Deursen, W. P. A.: Integrated dynamic environmental models in with MODIS leaf area index, J. Hydrol., 155–162, GIS: The development of a Dynamic Modelling language, Trans- doi:10.1016/j.jhydrol.2009.03.003, 2009. actions in GIS, 11, 4048, 1996. Nash, J. E. and Sutcliffe, J. V.: River flow forecasting through con- Wu, K. and Johnston, C. A.: Hydrologic comparison between a ceptual models. Part I: A discussion on principles, J. Hydrol., 10, forested and a wetland/lake dominated watershed using SWAT, 282290, 282–290, 1970. Hydrol. Process., 22, 1431–1442, 2008. Schmidtke, K.-D.: Land im Wind, Wetter und Klima in Schleswig- Xu, C. Y.: Estimation of parameters of a conceptual water balance Holstein. Wachholtz Verlag, 119 pp. 1999. model for ungauged catchments, Water Resour. Manage., 13(5), Silberstein R. P.: Hydrological models are so good, do we still need 353–368. 1999. data?, Environ. Model. Softw., 21, 1340–1352. 2006. Zhang, X.: Test and application of hydrological models with a spa- Sponagel, H., Grottenthaler, W., Hartmann, K. J., et al.: Bo- tial modelling language (PCRaster) for the discharge simulation denkundliche Kartieranleitung. 5. verbesserte und erweiterte Au- of a wetland dominated catchment in Northern Germany. Mas- flage, Hannover, Germany, 438 pp. 2005. ter Thesis, Christian-Albrechts-University of Kiel, Germany, 100 Springer, P.: Analyse der Interaktion zwischen Oberflächenwasser pp., 2006. und Grundwasser am Beispiel einer Flussniederung im Nord- Zhang, X., Hörmann, G., and Fohrer, N.: The Effects of Different deutschen Tiefland. Diplomarbeit im Fach Geographie, der Model Complexity on the Quality of Discharge Simulation for a Christian-Albrechts-Universität zu Kiel, 191 pp., 2006. Lowland Catchment in Northern Germany. Heft 20.07 “Einfluss Trepel, M.: Development and application of a GISbased peat- von Bewirtschaftung und Klima auf Wasser- und Stoffhaushalt land inventory for SchleswigHolstein (Germany), edited by: von Gewässern”, Band 2, Forum für Hydrologie und Wasserbe- Päivänen, J., Proceedings of the 12th International Peat Congress wirtschaftung, ISBN: 978-3-940173-04-1, 2007. Wise Use of Peatlands, Vol. 2, 931936. 2004. Zhang, X., Hörmann, G., and Fohrer, N.: An investigation of the Vieux, B. E.: Distributed hydrologic modelling for flood forecast- effects of model structure on model performance to reduce dis- ing. GIS and Remote Sensing in Hydrology, Water Resour. Env- charge simulation uncertainty in two catchments, Adv. Geosci., iron., IAHS Publ., 289, 1–10, 2004. 18, 31–35, 2008, http://www.adv-geosci.net/18/31/2008/. Adv. Geosci., 21, 125–130, 2009 www.adv-geosci.net/21/125/2009/