Status Report VI  
Reporting Period: 01.01.99 – 30.06.99

Partial Report ‘Large-Scale Hydrological Modelling’    

Potsdam-Institute for Climate Impact Research

Dipl.-Hydr. A. Güntner
Dr. A. Bronstert

1. Project objectives

Objective of the project is the quantification of physical availability of water on a regional scale in the semi-arid Northeast of Brazil. A large-scale hydrological model working in a spatially and temporarily distributed mode is to be developed. Various components of the hydrological cycle which control water availability are to be determined within this model, i.e. evapotranspiration, soil moisture, river runoff and storage in reservoirs and groundwater. Specific semi-arid conditions, sparse data availability, the large spatial scale and the goal to cope with the effects of land-use or climatic changes on water resources are important factors to be taken into account during the development of adequate modelling concepts.

Main objective of the reporting period was in extension of the first model version developed in 1998 the improvement of model conceptualisations particularly with respect to runoff generation and corresponding model programming. For forthcoming scaling-related examinations and sensitivity studies, appropriate methods and data have to be prepared. No modifications in relation to the original proposal were made during the reporting period.

2. Progress and results

2.1 Work progress: scientific and technical description

2.1.1 Model development

The large-scale water balance model developed in the last reporting period was continuously extended and modified with focus on the improvement of runoff generation conceptualisations.

A hierarchical scheme for structuring river basins (Fig. 1) was developed and implemented into the model. Corresponding model programming was performed during the reporting period. Key features of the approach are:

The hierarchical subdivision of the landscape implements 5 hierarchy levels, ranging from the river basin or municipio scale down to the point scale of the soil profile (Fig. 1). Thus, the approach copes with the scale problem of large-scale models (i.e. they cover a very wide range of relevant spatial scales) in the sense that the important processes at different scales can be readily conceptualised at the specific hierarchy level where they are relevant. Similarly, the necessary information needed for their parameterisation can be provided adequately at this specific level. For instance (see Fig. 1), lateral flow between highland and sloping areas is represented at the terrain component level with mean slope gradient as required parameter, whereas (vertical) percolation in the soil is represented at the profile level given hydraulic conductivity as a parameter. This hierarchical scheme is closely related to the FAO SOTER structure, implemented into the SPICE information system by the working group of Soil Sciences. This structure was extended for the present hydrological application in close co-operation of both working groups (see below).

The scheme developed here corresponds to a top-down approach, i.e., starting from a large spatial scale (river basin or municipality), the landscape is successively subdivided into smaller modelling units. Thus, the approach is flexible in the sense that (1) spatial subdivision can be confined to those features of which the spatial variability is considered to be relevant for describing the dominant processes and (2) the level to which the landscape is subdivided into areas of which the location in the landscape is geographically explicitly specified can be varied according to the resolution of available data. For a model application at the regional scale of the states of Piauí and Ceará, for instance, this corresponds to the landscape unit level (Fig.1), whereas for the local scale of the focus area Picos geographically explicit data are available down to the terrain component level.

The approach is able to cope with sub-scale variability by using areal fractions and distribution functions of relevant model parameters. At the soil/vegetation component level, for instance, sub-scale variability of soil types within a terrain component is represented by the distribution of corresponding soil parameters, e.g. field capacity (Fig. 1).

The main focus of many modelling approaches in hydrological models is on describing vertical hydrological processes. The representation of lateral in large-scale water balance models, however, is often rather crude and may refer to runoff routing in the river network only, though later flow being also very important for spatial variability of soil moisture, for instance. In the approach developed here, the principle of delimiting and parameterising the modelling units was removed from the classical vertical-based scheme. Modelling units (at the landscape unit level, Fig. 1) are defined as terrain patches with similar characteristics referring to lateral flow. Heterogeneity of the land surface with respect to vertical processes is represented as sub-scale variability at the profile level (Fig. 1). Thus, delineation of modelling units at the different units of the hierarchical approach is closely based on topographic and geomorphologic information. Landscape units (Fig. 1) are areas with similar characteristics with respect to major landform, general lithology, soil associations, toposequences and consequently with respect to the connectivity of sub-areas for lateral water flow. Thus, they are assumed to have similar behaviour with regard to spatial redistribution of water on lateral flow paths within them. These landscape units correspond to the so-called SOTER units within the SOTER structure as used by the working group of Soil Science.

The approach was closely developed with respect to available data at the regional scale of the entire area of Piauí and Ceará, thus ensuring the large-scale application of the model with its parameters being to the utmost derived from existing physiographic information. These are the geomorphologic map of the project Radambrasil, digitised during the last reporting period, as well as landscape and soil information derived from the EMPARABA soil surveys and implemented into the SPICE information system by the working group of Soil Sciences (see also below).

Further features of model development in the reporting period where the implementation of a module for the generation of infiltration-excess overland flow (a multi-layer Green-Ampt approach) and a volume class based approach for calculating the water balance of small reservoirs. The latter includes also a simplified cascade order of their location in the river basin (reservoirs of a smaller volume class are located upstream of reservoirs of a large volume class). As the explicit volume of each of these hundreds of small reservoirs and their location is unknown, but only the number of reservoirs in volume classes per municipality, this approach was implemented as an approximation to capture the non-linear relationships between storage volume, surface area, evaporation, relative actual storage volume and spill-over of reservoirs of different size.

Verification tests of the new model code and all its components were performed, resulting in reasonable results (e.g. a closed water balance).

2.1.2 Extension of the SOTER data base

The hierarchical scheme described above is closely related to the SOTER data base structure, which is implemented in the SPICE information system of the Soil Sciences working at University of Hohenheim. During the reporting period, the SOTER data base was extended in close co-operation with the Soil Sciences group in the following respect:

The SOTER data base structure, which was until then only available for the focus area Picos, was extended for the area of Ceará in a first basic version, including the principle soil components (see also chapter 2.2).

The SOTER data base was extended for hydrological applications by defining upstream-downstream relationships between SOTER-units (corresponding to landscape units in the hydrological model) and between terrain components. An appropriate classification scheme for these relationships was developed, resulting in two new features in the SOTER data base, i.e. (1) the position of SOTER-units in the landscape and (2) the position of the terrain components within the SOTER-unit. In the hydrological model, the lateral connectivity of modelling units is defined by this way, e.g., runoff generated in a terrain component with higher position in the landscape unit (e.g. highlands) is routed into the terrain component with lower position within the landscape unit (e.g. slopes). The corresponding information to fill these two features in the data base were derived from the supplement books of the EMPRABA soil maps and from topographic maps, the river network and the digital terrain model. In the reporting period, this was completed for the area of Ceará.

2.1.3 Additional data preparation and analysis

Geomorphologic maps. Mapping units of the geomorphologic maps of the project Radambrasil which were digitised in the last reporting period include a qualitative classification of the depth of valley dissections. By comparative analysis with topographic maps (1:100000), quantitative values were derived for each of these depth classes in this reporting period. Combining with the existent information of the mean spacing of valley dissections included in the maps, the range of mean slopes were derived for each mapping unit. These data serve for parameterisation of the hydrological model.

Representative basin Tauá. During his visit at PIK, Dr. E. Cadier (see annual report 1998) provided meteorological and hydrological data (partly in analogue form) of small representative basins in Tauá. These data were transformed in adequate formats for forthcoming scaling studies. For three nested catchments (<100 km²) an approximately ten year period of precipitation and runoff data was compiled.

2.1.4 Development of an ARCVIEW-Interface

With the objective of quick visualisation of spatial simulation results and a user-friendly program control, the development of an interface between the hydrological model (a FORTRAN program) and the GIS ARCVIEW was started in the reporting period. Written as an AVENUE-script, this interface allows to select the desired study area, to start the hydrological model, to select required simulation results and the time period of interest and to visualise them as separate views. A prototype version of this interface was compiled.

2.1.5 Rainfall disaggregation - Cascade model

Various modifications of the cascade-based disaggregation scheme for temporal rainfall disaggregation were tested during the reporting period. The latest high-resolution data of the PIK climate stations in Brazil were included into analysis. Synchronisation of model approaches and program codes were performed with working partners at Kyushu University / Japan. Model performance improved slightly due to the modifications and due to the broader data base which reduces the statistical scatter during calibration.

2.2 Partnership

Close co-operation existed in the reporting period with the working group of Soil Sciences at the University of Hohenheim (Gaiser). The concept for the extension of the SOTER data base structure to match the modelling scheme of the hydrological model was elaborated. Also with the assistance of J. Krywkow (working group climate), the data base was extended in a first version for the area of Ceará, being basically responsible for this task the University of Hohenheim, however. The most important features from a hydrological point of view, e.g. dominant soil types, where included in this first version. More detailed information will be included later.

During his travel to Ceará in March 1999, A. Güntner participated at a common one-week field trip with M. Hauschild and P. Döll (working group Large-scale water management model, University of Kassel), as well as with Prof. J. Carlos Araujo (working group Water at Universidade Federal do Ceará), including the collection of water quality samples for the working group Water Resources (Hydroisotop). Linkages and open questions between the different working groups were discussed. With Prof. Araujo, the possible linkage of the erosion and reservoir sedimentation model developed at UFC with the large-scale hydrological model and the integrated model was discussed and a first concept sketched out.

At the bilateral WAVES-workshop in April 1999 in Fortaleza, the large-scale hydrological model was presented by A. Güntner, concepts and data requirements were discussed with members of different WAVES working groups of the Brazilian side. Particularly interest was on the derivation of vegetation/land-use information for parameterisation of the evapotranspiration module in the hydrological model. It became clear, however, that suitable data applicable at the required large scale are hardly achievable. Co-operation with COGERH at Fortaleza was combined during visits of A. Güntner and M. Krol in April 1999. COGERH has very valuable knowledge and data in the field of water resources of Ceará, parts of which were planned to be provided for WAVES in exchange for an implementation of the large-scale hydrological model at COGERH.

2.3 Exploitation and dissemination of results

The preliminary results have been presented to the German partners of the project.

International publications:

Bronstert, A., Güntner, A., Jaeger, A., Krol, M., Krywkow , J. (1999): Großräumige hydrologische Parametrisierung und Modellierung als Teil der integrierten Modellierung. In: Fohrer & Döll (Hrsg.) (1999): Proceedings des 2. Workshops: „Modellierung des Wasser- und Stofftransports in großen Einzugsgebieten" 19/20. November 1998 in Rauischholzhausen bei Gießen. Kassel University Press (in press).

Bronstert, A., Jaeger, A., Güntner, A., Hauschild, M., Döll, P., Krol, M. (1999): Integrated modelling of water availability and water use in the semi-arid Northeast of Brazil. Physics and Chemistry of the Earth (in press).

Güntner, A., Olsson, J. (1999): Application of a cascade model for rainfall disaggregation in semi-arid tropics. In: Geophysical Research Abstracts, 1(2), 291, 24th EGS General Assembly, The Hague, The Netherlands.

3. Project management structure

The management will be realised as planned by the Steering Committee, the Management Group and direct contacts between the involved scientists.

4. Outlook

In the next steps, the modified water balance model has to validated mainly by comparison with measured runoff data. The application will primarily focus on the area of Ceará, as soil data are currently available only for this state. For Piauí, the soil information has to be compiled with the responsibility of the Soil Science group at University of Hohenheim. Model applications will also focus on the representative basins in Tauá, which allow a close assessment of the new runoff generation scheme due to their small size and comparatively good data availability.

A. Bronstert will present the concept of large scale hydrological modelling and A. Güntner will present the related approaches and technical details at the XXI IUGG General Assembly in July in Birmingham, UK. In October, A. Güntner will give a summary contribution on the water related activities in WAVES at the German Day of the Tropics in Berlin. Results of the model application will be presented by A. Güntner at the International Conference on Hydrological and geo-chemical processes in large-scale river basins in Manaus in November. A. Bronstert will go to Brazil for a technical visit in September, A. Güntner will visit Fortaleza and Manaus in November.

5. Administrative aspects

5.1 Financial situation

The project keeps to its financial plan.

5.2 Staff situation

A small temporal contract was given to Mr. M. Kersten to develop a specific data-interface between the large scale hydrological model and the GIS Arc-View. No other changes were performed.

Fig.1 Hierarchy for structuring river basins in the large-scale water balance model