An integrated modelling framework for simulating regional-scale actor ...

ARTICLE IN PRESS +

ENSO2060_proof 6 March 2008 1/27

MODEL

Available online at www.sciencedirect.com

Environmental Modelling & Software xx (2008) 1e27 www.elsevier.com/locate/envsoft

OF

An integrated modelling framework for simulating regional-scale actor responses to global change in the water domain R. Barthel a,*, S. Janisch b, N. Schwarz c, A. Trifkovic a, D. Nickel a, C. Schulz d, W. Mauser e a Institute of Hydraulic Engineering, Universita¨t Stuttgart, Pfaffenwaldring 7a, D-70569 Stuttgart, Germany Institute of Computer Science, Ludwig-Maximilians-Universita¨t Munich, Oettingenstrasse 67, D-80538 Munich, Germany c Helmholtz Centre for Environmental ResearchdUFZ, Permoserstrasse 15, D-04318 Leipzig, Germany d Centre for Environmental Systems Research, University of Kassel, Kurt-Wolters-Strasse 3, D-34131 Kassel, Germany e Faculty for Geosciences, Ludwig-Maximilians-Universita¨t Munich, Luisenstrasse 37, D-80333 Munich, Germany

RO

b

DP

Received 16 November 2006; received in revised form 8 February 2008; accepted 14 February 2008

Abstract

CO RR

EC

TE

Within coupled hydrological simulation systems, taking socio-economic processes into account is still a challenging task. In particular, systems that aim at evaluating impacts of climatic change on large spatial and temporal scales cannot be based on the assumption that infrastructure, economy, demography and other human factors remain constant while physical boundary conditions change. Therefore, any meaningful simulation of possible future scenarios needs to enable socio-economic systems to react and to adapt to climatic changes. To achieve this it is necessary to simulate decision-making processes of the relevant actors in a way which is adequate for the scale, the catchment specific management problems to be investigated and finally the data availability. This contribution presents the DEEPACTOR approach for representing such human decision processes, which makes use of a multi-actor simulation framework and has similarities to agent-based approaches. This DEEPACTOR approach is embedded in DANUBIA, a coupled simulation system comprising 16 individual models to simulate Global Change impacts on the entire water cycle of the Upper Danube Catchment (Germany, 77,000 km2). The applicability of DANUBIA and in particular the DEEPACTOR approach for treating the socio-economic part of the water cycle in a process-based way is demonstrated by means of concrete simulation models of the water supply sector and of the domestic water users. Results from scenario simulations are used to demonstrate the capabilities and limitations of the approach. Ó 2008 Elsevier Ltd. All rights reserved. Keywords: Climate change; Water supply; Domestic water use; Integrated water resources management; Coupled simulation; Social simulation; Actors; Framework technology; Regional scale model

Software availability

UN

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57

The framework, the simulation models and the results discussed in this article rely, amongst others, on software developed by various project partners of GLOWA-Danube (www.glowa-danube.de) within the first two phases of the project (2001e2007). One of the main aims of the third project period (2007e2010) is to convert the complete system into an open source project which will be available * Corresponding author. Tel.: þ49 711 685 66601; fax: þ49 711 685 66600. E-mail address: [email protected] (R. Barthel).

to the public at the end of the third project phase (April 2010) at the latest. 1. Introduction Suppose a decision maker or, more generally, a stakeholder wants to know if the available water resources in a large, diverse catchment will still meet the water demands of various water users in 30 years’ time, assuming that the worst predictions of Global Climate Models (GCM) downscaled to regional conditions come true. More specifically, this decision maker wants to know where and when in the catchment problems are

1364-8152/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.envsoft.2008.02.004 Please cite this article in press as: Barthel, R. et al., An integrated modelling framework for simulating regional-scale actor responses to global change in the water domain, Environ. Model. Softw. (2008), doi:10.1016/j.envsoft.2008.02.004

58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114

ARTICLE IN PRESS +

MODEL

R. Barthel et al. / Environmental Modelling & Software xx (2008) 1e27

OF

social systems or concepts known from social sciences are used to form simplified conceptual representations of natural systems are appropriate (see Section 2). e Rely on a realistic representation of the options how actors (water suppliers and water consumers) may react to changes. e Allow for a differentiated treatment of actors, as the options and preferences of different actors can vary when reacting to change, and similar actors can react differently at different locations.

1.1. Scope and organisation of this article Without attempting to cover all dependencies shown in Fig. 1, this article presents an approach to include feedback and adaptation by simulating responses of the socio-economic framework to Global Change by using ‘actors’ to simulate human decision makers at various levels from individual households to communities and large companies. Note, that although our approach follows ideas from the field of agentbased social simulation we use the term ‘actor’ instead of ‘agent’ to avoid name clashes with the term ‘software agents’, which have a more specific meaning in the computer sciences (see discussion in Drogoul et al., 2002). Hereby we focus upon the domestic water consumption and the drinking water supply infrastructure. The developed approach will hence be referred to as the DEEPACTOR approach that is embedded in the DANUBIA simulation system (described in Section 4). To summarise, the approach presented here is

CO RR

EC

TE

DP

to be expected in order to plan counter measures and to allocate appropriate financial resources. In order to answer these questions, the decision maker needs tools to predict the development of (a) the state of the catchment’s natural water resources, (b) the water demands of the different consumers under a changed climate and (c) the state of the water supply infrastructure and its ability to provide for adequate water supply. All these predictions need to be spatially and temporally explicit, if the ‘where’ and the ‘when’ questions are to be addressed. Physical models (hydrological, hydraulic, groundwater etc.) are traditionally used to predict the spatially and temporally explicit state of water resources. The traditional prediction of demands relies on data driven methods such as trend extrapolation, regression analysis, time series analysis, rule and expert knowledge based systems, etc. However, all methods that neglect the existing feedbacks (see Fig. 1) between demand and supply and the various options of adaptation the consumers have, are unrealistic or even meaningless. As indicated in Fig. 1, Climate Change will affect not only the state of resources but also the ways and the magnitude of consumption (e.g. introduction of irrigation if the climate gets dryer). Changing consumption will in turn lead to an additional change of the resources state; the resources state might influence the attitude towards consumption or invoke legally defined restrictions for consumption. Consumers might then adapt their consumption to decreasing resources (save water) and/or water suppliers might increase resources by, for example, importing water from neighbouring catchments. Furthermore, the number of consumers might decrease e.g. because water intensive industries close down. Finally, it must be taken into account that different consumers will react differently to changes according to their possibilities, their preferences and their location. Therefore, an attempt to simulate such complex systems with their numerous mutual interdependences needs to:

RO

2

e Integrate the social and physical side of the water cycle within a catchment. We suggest that both sides should be simulated by the most appropriate state of the art models and are doubtful whether combined approaches where natural science modelling concepts are imposed to

Climate Change

Natural Resources: Groundwater, Surface Water, Water Imports

Political Change

Land Use Change

UN

115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171


Supply: S(D)

Demand: D(S)

Consumers / Actors: Water Supply Companies Households, Farmers, Industry, Tourism

Demographic Change

Fig. 1. Schematic view of the influences of changing outer boundary conditions (‘Global Change’) on the demand/supply relation in a catchment where demand is a function of supply and vice versa.

e Based on a fully coupled simulation system that allows for the parallel interactive simulation of physical and socioeconomic processes and their interactions. e Part of an integrated approach which considers the full water cycle from the clouds to the groundwater and its interaction with the socio-economic framework. e Applicable on the regional (¼large catchment) scale and targeted on the evaluation of long-term scenarios, mainly based on Global Climate scenarios but also in a wider sense on Global Change scenarios including socioeconomic change. It is therefore meant to show potential regional impacts of Global Change on larger spatial and temporal scales rather than to analyse human or institutional behaviour on an individual level. The central focus of this paper is the actual application of the DEEPACTOR approach to concrete management problems in the Upper Danube catchment. Before going into the details of DANUBIA and the DEEPACTOR approach, the problem which is described in this introduction section on a rather specific and conceptual level will be described from a more general view point in the context of integrated water management on the regional scale in Section 2. Section 2 thereby explains the context in which the DEEPACTOR approach is meant to be applied. In Section 3 we briefly discuss and compare approaches presented by other authors which follow partly or

Please cite this article in press as: Barthel, R. et al., An integrated modelling framework for simulating regional-scale actor responses to global change in the water domain, Environ. Model. Softw. (2008), doi:10.1016/j.envsoft.2008.02.004

172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228

ARTICLE IN PRESS +


MODEL


completely the same ideas. Section 4 describes the background of the GLOWA-Danube project within which the DEEPACTOR approach was developed, as this background is needed to understand both the approach itself and its application. Our approach is described and illustrated in Section 5 by the abstract features of the DEEPACTOR framework together with the concrete simulation models WaterSupply and Household. Section 6 presents the application of the DEEPACTOR approach with the selected model implementations in the Upper Danube catchment. The results of three climate scenario simulations in combination with three water management scenarios and the options for a validation of the approach are discussed. Finally, in Section 7 the main conclusions are summarised.

OF

average boundary conditions of natural systems (temperature, rainfall, percolation, etc.) were more or less constant. The widespread recognition of Climate Change in the last years has aroused awareness that this assumption might lead to dangerous mismanagement. Additionally environmental and social impacts of water management measures have become more evident and more pronounced in modern societies (Biswas, 1996; Arnell, 1998; IPCC, 2001; Bolwidt, 2005; Solomon et al., 2007). 5. Increased prediction capabilities: The availability of increasingly sophisticated optimisation models and the possibility to run simulation models for large and complex systems, together with a growing availability of observed data increases the applicability and predictability of modelling tools (e.g. Loucks et al., 1981; Wilson, 1999; Yurdusev, 2002). The changes of objectives and options listed above, which for a large part are combined in the European Water Framework Directive (European Commission, 2000), clearly demand new approaches for dealing with the management of water resources. Management must consider a multitude of aspects and objectives, larger spatial areas and, finally, the variability of natural and social boundary conditions. It is clear that such a multifaceted management requires the integration of expert knowledge from many disciplines: the effects of Global Climate Change on water and land resources can not be evaluated on the basis of one discipline alone. Therefore management approaches are required that integrate the relevant disciplines and make use of existing knowledge and models and that deal with management problems on the appropriate scale. A very comprehensive discussion of most of the aspects discussed in the above listed enumeration can be found in (Giupponi et al., 2006). Integrated assessment and management, multi-objective decision making and the involvement of human actors and stakeholders are topics which are extensively discussed in works of Tony Jakeman, Claudia Pahl-Wostl and their co-authors (Pahl-Wostl, 2002, 2005, 2007a,b; Jakeman and Letcher, 2003; Letcher et al., 2004, 2007; Jakeman et al., 2006).

DP

Management of water (and land1) resources has always been led by the two fundamental questions: What is needed and what is available? In the past decades, however, the manner of dealing with these basic questions has changed quite essentially. The driving factors for the respective changes in water management can be summarised as follows:

RO

2. Integrated water resources management and the role of human actors

CO RR

EC

TE

1. Sustainability and the long-term management perspective: Instead of aiming mainly at satisfying present human needs, management is additionally concerned with the welfare of future societies. Sustainability has become a key issue in planning the development and exploitation of resources (IUCN et al., 1980; UN, 1992). 2. The integration of multiple objectives of different decision makers and stakeholders: Whereas water management used to be led by economic considerations, it is now widely accepted that a range of economic, environmental and social impacts needs to be considered. Impacts of the changing state of an exploited resource on the state of dependent systems, e.g. water-dependent ecosystems, and on the quality of human life are taken into account (UNESCO, 1987; Bouwer, 2002). 3. Consideration of all scales from local to regional2 to global: A rapid technological evolution allows for the compensation of local deficits by transfer of water over large distances. At the same time human mobility and flexibility has increased so that water-related management may not focus on local aspects only (UNESCO, 1987; Do¨ll et al., 2003). 4. The recognition of Global Change3: Former management was often based on the assumption that the long-term

UN

229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285

3

1 In this article we will refer to water resources management only, even if the presented DEEPACTOR-approach and DANUBIA as a whole integrate water and land use management. 2 Regional in the context of this article: River basins >10,000 km2, countries, states, larger political entities. 3 Here understood as the sum of physical, social, political and economic changes that affect the environment at present and in the future on at least a regional scale.

2.1. Process-based simulation of human actors in the water sector Dealing with uncertainty of assumptions, models and data are a major concern of IWRM under conditions of Global (Climate) Change and a crucial aspect for the reliability and acceptance of model results (Bogardi and Kundzewicz, 2002; Brugnach et al, 2007; Krysanova et al., 2007). An accepted way of reducing uncertainty or showing the influence of uncertain processes on model results is by modelling the actual processes. Today, it is a common understanding that the less conceptual (i.e. derived through trend extrapolation, regression analysis, time series analysis, rule and expert knowledge based systems, fuzzy logic, etc.) a model is and the more realistic individual processes are represented, the better a model can be used in decision making and for stakeholder discussions


286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342

ARTICLE IN PRESS +


OF

in a process-oriented way. Simulation models that address the environmental impacts and socio-economic effects together in a fully integrated way were for example presented by Wu (1995), Kirshen et al. (1995), Watkins et al. (2004) and Yamout and El-Fadel (2005), however with a focus on smaller systems on the local scale. Athanasiadis et al. (2005) and Lo´pez-Paredes et al. (2005) provide regional and integrated models for water management, including both water suppliers or municipalities and area residents. Although implemented as agent-based models, both approaches do not explicitly represent the processes involved in the case of water scarcity on the supply and on the demand side. Moss et al. (2000) sketch a promising approach within the FIRMA project that explicitly includes policy makers and consumers. However, this approach seems to be not sufficiently specific with respect to the explicit representation of important variables such as water demand. The negotiation process simulated in the approach of Thoyer et al. (2001) and the model of Espinasse and Franchesquin (2005) focus on a small local scale. Berger et al. (2007) and Feuillette et al. (2003) describe integrated approaches to model water management which are not applicable to central European water management due to water resources being a major influence on farming and therefore on household income in both cases. In view of this brief discussion which of course cannot reflect all agent-based or integrated simulation approaches in the water sector, we are not aware of any approach that would fulfil the requirements explained in Section 2 in a problem context as exemplified in Section 1. After having shown in more detail what the DEEPACTOR approach actually is and does in the following sections we will come back to this in the conclusions in Section 7.

TE

DP

(Hauhs, 1990; Ewen and Parkin, 1996; Croke and Jakeman, 2001; Bathurst et al., 2004; Ewen et al., 2006; Jakeman et al., 2006). In water resources management, natural processes (e.g. groundwater flow) have long been treated as physical based as possible. Social processes on the other hand (e.g. changes of water demand per capita) have mainly be dealt with on a highly conceptual basis. Under conditions of Global Change, the use of conceptual approaches to model social processes is questionable, as they largely ignore that human decisions and behaviour are changed as a result of (Global) change (see Fig. 1). Human decisions made as a reaction to changing natural conditions (e.g. water consumption or the expansion of a water supply network) change the actors’ environment and form a new basis for future decisions (Pahl-Wostl, 2002, 2007a). Furthermore, human behaviour not only adapts to change but also undergoes an evolution that might be forced by external drivers or be random (Grothmann and Patt, 2005). It seems therefore judicious to extend the idea of process-based modelling to the socio-economic components of an integrated simulation system in an effort to reduce their uncertainty and to improve their transparency. Mechanisms need to be provided for social systems to respond to changes and for capturing the effects of these responses upon the environment. Process-based in that context means capturing the triggers, options, and temporal and spatial aspects of an actor’s reaction in a direct, transparent and realistic way.

EC

3. Related work: Comparison with other approaches

CO RR

Predicting future water availability and water demand has always been one of the central questions in water resources management. Accordingly, many approaches for simulating and predicting the state of resources and the evolution of demands, amongst them also many agent-based simulation approaches, exist (see below). The specific problem at hand (Section 1) requires (a) an integration of natural and socioeconomic processes, (b) the consideration of multiple actors with different options and preferences, (c) a regional scale focus, (d) distributed, spatially and temporarily explicit models. It is therefore not meaningful to compare the developed DEEPACTOR approach to approaches that deal with demand or supply predictions on a sectoral basis and approaches that are dedicated to describe small-scale systems in a discrete4 way or lumped regional models. As already pointed out there is a growing demand for integrated management and assessment of resources on a regional scale (UNESCO, 1987; GWP, 2000; Bouwer, 2002; Quinn et al., 2004) which has accordingly brought upon a number of regional integrated research projects and approaches (e.g. Scoccimarro et al., 1999; Gaiser et al., 2003, 2007; Krysanova et al., 2007; Rodgers et al., 2007). None of these projects has to our knowledge come up with a fully coupled system or model that represents responses of the socio-economic framework

UN

343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399

MODEL

RO

4


4 Discrete here means modelling systems on a technically or personally explicit level, i.e. real infrastructure, people or institutions.

4. Project background: GLOWA-Danube GLOWA-Danube (www.glowa-danube.de) is one of five projects within the GLOWA-programme (www.glowa.org). Within GLOWA, integrated approaches for carrying out Global Change research are compared using six catchments in different climates on different continents. The central aim of GLOWA is to provide an integrated approach for describing, modelling, and forecasting physical, social, economic, and political processes related to the hydrological cycle, in particular with regard to Global Change on the river basin scale, in order to meet the requirements of modern IWRM. In GLOWA-Danube, the Upper Danube watershed was selected as a representative regional-scale test catchment in the temperate mid-latitudes, covering an area of approx. 77,000 km2 (Fig. 2). GLOWA-Danube equally considers the influence of natural changes in the ecosystem, such as Climate Change, and social changes, e.g. changes in land use or water consumption. The central objective of GLOWA-Danube is the joint development and application of the Decision Support System (DSS) DANUBIA (Ludwig et al., 2003; Mauser and Strasser, 2005). The GLOWA-Danube consortium comprises a universitybased network of roughly 40 scientists from the fields of


400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456

ARTICLE IN PRESS +


MODEL


0

5

10

15

20

25

30

60

60

The Upper Danube 55 Basin

55

#

Germany Czech Republic #

Passau

Switzerland

Austria

#

DP

40

45

RO

45 Italy

50

OF

50

40

DANUBE

35

5

10

15

20

35

25

30

TE

0

EC

Fig. 2. Location of the Upper Danube Basin, the investigation area of GLOWA-Danube.

CO RR

Meteorology, Hydrology, Hydrogeology, Hydraulic Engineering, Plant Ecology, Environmental Psychology, Environmental Economics, Agricultural Economics, Glaciology, Remote Sensing, Tourism Research and Computer Sciences. The project started in January 2001 and will terminate in April 2010. The first phase of the project (2001e2004) was dedicated to the development of the DANUBIA prototype. In the second project phase (2004e2007), the consolidation and application of DANUBIA, i.e. scenario evaluation, stakeholder involvement, decision making, and practical water and land use management support were in the centre of the research activities. In addition the process-based simulations in the socio-economic sector were further developed. The third phase (2007e2010) has the main objectives of:

UN

457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513

5

(a) making DANUBIA practically available for the management of water and land resources in the Upper Danube catchment, and (b) making DANUBIA and all its components available and usable for scientists and end users under an open source licence.

4.1. The Upper Danube catchment The Upper Danube catchment is a heterogeneous, mountainous catchment with altitudes ranging from 287 to

4049 m a.s.l., precipitation between 650 ands >2000 mm/ a, evaporation between 450 and 550 mm/a, discharge between 150 and 1.600 mm/a, average annual temperature between 4.8 and þ9 C and approximately 10.8 million inhabitants. The last century showed a rapid transgression from a mainly agricultural society to a highly industrialised high-tech economy (mainly IT, automotive, chemistry). However, industrial activities are quite focussed in few urban agglomerations leaving vast parts of the country to traditional farming and tourist activities. In particular the latter are a major source of income (e.g. skiing in the Alps) and therefore a major concern of GLOWA-Danube (Sax, in press). The parallel evolution of a high tech industry on the one and the need for preserving the nature for agriculture and recreation purposes on the other hand forms a major source of conflicting interests. DANUBIA was developed as an instrument to help decision makers and stakeholders to solve such conflicts. Water resources management in the Upper Danube is complex, in part because the area extends over a number of countries (Fig. 2), but even more due to the very distributed nature of the water supply system. Well over a 2400 municipalities are supplied by roughly 2000 independent water supply companies, leading to a highly distributed scheme of supplier consumer relations (Barthel et al., 2005; Nickel et al., 2005). Water supply in the public sector is mainly (90%) groundwater based (Barthel and Trifkovic, 2007).


514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 Q3570

ARTICLE IN PRESS +


CO RR

EC

TE

DP

DANUBIA currently comprises a total of 16 disciplinary simulation models, which are coupled to each other in order to exchange data dynamically during runtime. As depicted in the UML 2.0 diagram (UML: see e.g. Rumbaugh et al., 2005) of Fig. 3, main components to group domain-related simulation models are used. Natural science simulation models address physical processes concerned with the natural water cycle and water availability whereas so-called actor models address demography, water consumption and supply infrastructure in the investigation area (compare component actor in Fig. 3) simulate the socio-economic aspects of the water cycle using an ‘agent-based’ social simulation framework. The system is used to evaluate the sustainability of future water resources management alternatives, and to evaluate consequences of IPCC derived climate scenarios for the period from 2000 to 2100. Table 1 summarises purpose and scale of the socio-economic simulation models of DANUBIA which can make use of the DEEPACTOR approach. To facilitate the smooth integration of the different simulation models, DANUBIA makes use of an object-oriented framework approach (D’Souza and Wills, 1999). Following this approach, the development of the individual simulation models is based on common modelling and implementation frameworks. In the following we briefly describe DANUBIA’s developer framework which provides the basic simulationrelated model features required for all model implementations of DANUBIA. In DANUBIA, the simulation area (Upper Danube Basin, see Fig. 2) is modelled by a fixed 2-dimensional grid comprising

cells of 1 1 km. Each cell can be identified by a unique ID and is modelled by an object that has a state determined by attributes, for example elevation. Apart from such general cell attributes common to all simulation models, there are also model-specific cell attributes, for example soil temperature in case of the simulation model Soil (Fig. 3). Following an object-oriented approach, the cells may also perform calculations in the form of functions computed over local values of the particular cell. For this reason we call the cells proxels, an acronym of ‘process pixel’. The values of the model-specific proxel attributes are subject to the data exchange between different coupled simulation models, which is formally specified by interfaces. A DANUBIA interface gathers the methods used for data exchange, one for each parameter, and is accompanied by a specification of allowed data ranges. For each simulation model, there is an interface specifying its required import data and an export interface specifying the data which is provided, i.e. computed by this model. The main purpose of these specifications is to ensure the consistent interconnection of the different simulation models of the project partners within DANUBIA during runtime. DANUBIA provides concepts and an implementation for the treatment of simulation time. On the one hand, any simulation model participating in a coupled simulation follows the same abstract simulation cycle of getting data, internal computation, provision of data, getting data and so forth. On the other hand, each model defines an individual local time step, such as 1 h, 1 day, 1 month and so forth, depending on the particular process to be modelled. Therefore, while a 1 h model runs 24 cycles, a concurrently executed 1 day

OF

4.2. The DANUBIA simulation system and framework

«component» Atmosphere

AtmoStations AtmoSat AtmoMM5 «component» Actor

«component» Landsurface

UN

571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627

MODEL

RO

6


Biological RadiationBalance Snow Soil Surface

«component» Rivernetwork RModel

Demography Economy Farming Household Tourism WaterSupply

«component» Groundwater GroundwaterFlow Fig. 3. Main components and interfaces of the coupled simulation system DANUBIA. Components comprise at least one simulation model (listed in the lower section of the boxes). Please cite this article in press as: Barthel, R. et al., An integrated modelling framework for simulating regional-scale actor responses to global change in the water domain, Environ. Model. Softw. (2008), doi:10.1016/j.envsoft.2008.02.004

628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684

ARTICLE IN PRESS +


MODEL

R. Barthel et al. / Environmental Modelling & Software xx (2008) 1e27 Table 1 Socio-economic models within DANUBIA: objectives, number of actors and actor types

Demographyb Economy Farming Tourism

Demographic changes Industrial water demand Agricultural water demand Water demand and development of tourism infrastructure (e.g. skiing resorts, golf courses, hotels) State of water supply systems, state of water resources, groundwater and surface water withdrawal, distribution of water to consumers (see Section 5.4) Domestic water demand, water related consumer satisfaction, use of water saving technology (see Section 5.5)

WaterSupply

Household

a

# Actor Types

w1350 w40000 w1050

1 28 8

w2000

2

w50000

5

Only the main water related objectives relevant in this article are listed. Currently not using the DEEPACTOR approach.

DP

b

# Actors

OF

Main objectivea: to predict the spatial (1 km2) and temporal (monthly) distribution of:

2006), MASON (Luke et al., 2004) or Swarm (Minar et al., 1996), the DEEPACTOR framework also provides an integration with the developer framework and the runtime environment of DANUBIA. Therefore concrete DEEPACTOR models may be integrated and coupled with any other simulation model of GLOWA-Danube (see Fig. 3) using the same basic concepts and mechanisms for coordinated exchange of spatially explicit data (via formally specified interfaces) as for any other simulation model of DANUBIA. Both the DANUBIA and the DEEPACTOR frameworks apply object-oriented techniques to provide the basic building blocks for model implementations in the form of interfaces, abstract base classes and predefined relationships. A concrete DEEPACTOR model may implement different specialisations using different concrete subclasses of the abstract base classes (examples are described in Sections 5.4 and 5.5), each of them implementing inherited abstract methods in a different, typespecific way. In the following, the most important static elements of the framework are explained. After that, possibilities for the implementation of model computations on the one hand and actor decisions on the other hand are described. It must be stressed that some of the provided framework features are optional. Therefore its concrete application, or maybe better, its reification, may well differ between different simulation models.

RO

Model

EC

TE

model runs one cycle only. In order to coordinate the simulation models in the coupled system such that mutual exclusion of data exchange with respect to individual time steps is taken into account, a coordination mechanism was developed, whose concepts, design, implementation and formal verification is detailed in Barth et al. (2004), Hennicker and Ludwig (2005) and Hennicker and Ludwig (2006) and will not be further elaborated here. 5. The DEEPACTOR approach

CO RR

In contrast to physical processes, social processes are often the result of complex considerations, which include individual possibilities and preferences along with characteristic behaviour (Parunak et al., 1998; Axtell, 1999). In order to better support the modelling and simulation of this kind of entity in DANUBIA, the DEEPACTOR framework was developed, providing a common conceptual and architectural basis for the modelling and implementation of the socio-economic simulation models in GLOWA-Danube. Following their concrete requirements, the framework applies an approach to agent-based simulation in social sciences (Davidsson, 2002; Macy and Willer, 2002; Gilbert and Troitzsch, 2005), which in turn is based upon agent concepts of (distributed) artificial intelligence (Weiss, 1999; Russell and Norvig, 2003). Deciding entities such as individuals, organisations, companies and so forth are explicitly modelled and simulated as ‘actors’. An actor observes its environment and selects plans to execute as a reaction to its observations (for a slightly different definition see Pahl-Wostl, 2005). Thereby different actors may have different courses-ofaction as well as varying preferences, represented by their individual plans and their type-specific decision procedure. Besides providing the modelling support for simple reactive actors up to more complex deciding entities, which would in principle also be supported by toolkits and libraries for agent-based social simulation such as Repast (North et al.,

UN

685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741

7

5.1. Static structure of model implementations

Fig. 4 shows a conceptual view of the interfaces, base classes and relationships that are relevant for the development of a DEEPACTOR model. The upper part shows the most important elements of the DANUBIA developer framework which are inherited by the DEEPACTOR framework shown in the lower part of the figure. Any DEEPACTOR model implementation (specialisation of AbstractActorModel) is a DANUBIA model (AbstractDanubiaModel) following the same core concepts as natural-science model implementations, i.e. it comprises a set of proxels representing the simulation area (AbstractProxel), specifies import and export interfaces for the data exchange with coupled simulation models (DanubiaInterface), defines a local timestep and finally proceeds in a cyclic fashion with the implementations of data import (getData), local computation (compute) and data export (provide). The cyclic execution is coordinated by the runtime environment of DANUBIA according to the particular model timestep. DEEPACTOR models extend the concepts of a DANUBIA model, amongst others, by a number of new methods that are executed cyclically. In order to integrate these extensions transparently for the runtime environment, the DEEPACTOR framework maps invocations of methods of the base class AbstractDanubiaModel to calls of methods of base classes AbstractActorModel, AbstractActor, AbstractPlan and AbstractAction as well as methods of concrete DEEPACTOR framework classes such as the implementation of sensors or the history. For example the sensors are triggered to update their data directly


742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798

ARTICLE IN PRESS +

MODEL


8

DanubiaInterface

*

AbstractDanubiaModel

AbstractProxel elevation ...

timestep DanubiaInterface

getData() compute() provide()

...

Danubia Developer Framework DeepActor Framework

getData() preCompute() postCompute() provide()

data events

AbstractPlan

RO

Sensor

* id location collaborators 1

query() options() filter() export()

TE

execPlans failedPlans

DP

AbstractActor

History

...

rating

* isActive()

*

...

OF

AbstractActorModel

computeRating()

* AbstractAction period isApplicable() execute()

EC

Fig. 4. Central interfaces and classes of the DANUBIA and the DEEPACTOR developer framework.

CO RR

after getData of the model implementation is executed, compute is implemented by an invocation of preCompute followed by calls to options and filter for all actors, and finally completed by an invocation of the model’s postCompute. Last but not least the actors’ histories are triggered to update their state just before the model implementation of provide is executed. An actor represents an entity capable of reacting to system changes in an individual way, i.e. it is capable of making decisions. The individuality is thereby achieved by individual properties and preferences of each actor object. Fig. 5 demonstrates schematically how the individual properties and preferences are assigned. It has to be noted that, since the simulation area in DANUBIA is represented by a set of 1 1 km proxel objects, an actor in the most cases does not explicitly represent a real person or organisation but rather an abstract, effective ‘average’ of real actors located on one or more proxels which in turn define the environment of this actor. To show an example of averaging actors, if 100 real households exist in a proxel (model cell) for which 30% of the households belong to the ‘conservative’ household actor type and 70% belong to the ‘progressive’ household actor type, then the corresponding proxel will host two actors. The domestic water demand on such a grid cell is then computed as a weighted average of the water demand of these two actor types (see Section 5.5).

UN

799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855


As depicted in Fig. 4, one of the basic actor properties is its location, usually defined by a set of proxel object IDs. Also an actor may define a number of collaborators, i.e. other actor objects of the same DEEPACTOR model that this actor needs to collaborate with. The concept of collaborators allows establishing or simulating social network aspects. These basic properties may be assigned in concrete actor types and additional properties such as an attribute to store the number of wells may be declared. System Changes / Boundary Conditions

Actor

Changes of water demand / availability

WSC

• Type

• Type: long distance water supply

• has properties

• has properties: No. of wells

• has options

• has options: drill new wells, buy/import water

• has preferences

• has preferences: use own resources, use groundwater

Fig. 5. Schematic diagram of an actor: Left: abstract actor, right: a specific water supply company (WSC) actor as used in the DANUBIA model ‘WaterSupply’ (see Section 5.4). The specific actor will react to system changes (e.g. changes of groundwater available for extraction) in a specific way depending on its properties, options and preferences.


856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912

ARTICLE IN PRESS +


MODEL


OF

for the succeeding decision in the filter step. This includes the recalculation of the rating of all loaded plans (computeRating). 3. filter. Implements the definite plan selection. After this step the set of actions of each selected plan is executed which usually results in the (re-)computation of some proxel or actor attributes. Afterwards, the final plan execution status is stored in the history. 4. export. The results of actors relevant for their collaborators are stored to be available via the actor sensor in the next time step.

RO

The decision process of actors is embedded within computations of a model’s ‘main’ class (specialisation of AbstractActorModel). After having imported data from other models (getData), the instance of this class may optionally prepare (preCompute) and post-process (postCompute) its actors’ decisions, allowing for example the aggregation of some fine-grained results before providing them to other DANUBIA simulation models (provide). Note that the possibility to pre- and post-process within the same simulated time step allows implementing a macro level of simulation which dynamically takes into account the results of the ‘agent-based’ part of the simulation model, as described in (Duboz et al., 2003), amongst others. The pre- and post processing steps of the two chosen model implementations discussed in the next sections will not be addressed in full detail here as they are based on rather simple, yet many calculations and were in principle described in previous publications (e.g. Barthel et al., 2005; Ernst et al., 2005).

CO RR

EC

TE

DP

The plans (AbstractPlan) of an actor represent its course-of-actions and its options (Fig. 5). A plan in turn refers to a set of actions (AbstractAction), each of them explicitly modelling some state modification associated with the execution of the particular plan. Individual type-specific preferences may then be encoded by the type-specific implementation of a plan selection algorithm within filter and options. Based on the DEEPACTOR framework, a whole variety of decision algorithms can be implemented, ranging from simple habits or heuristics to elaborate cognitive decisions. To simplify both priority-based decision procedures and decision algorithms based on the Multiattribute Utility Theory (Russell and Norvig, 2003), a plan may store a rating attribute. In the former case, this value may be fixed initially and kept constant during runtime. In the latter case a function (computeRating) may be implemented which dynamically computes a new rating value in each time step. Plans are activated and deactivated during the decision procedure of an actor. The actions associated with activated plans (isActive) are executed (execute). An activated plan may fail to be executed if its actions were not executable, e.g. due to time period constraints (period, isApplicable) or the like. A number of sensors (Sensor) allows an actor to observe its environment by directly reading raw data or by being notified about the occurrence of pre-defined events. An event is a fixed model-specific interpretation of raw data. In fact, sensors merely encapsulate the data import facilities of an actor, adding the possibility of qualitative interpretation using model-specific events. Currently the framework provides three types of sensors. The proxel sensor enables importing proxel values computed within other simulation models. The constraint sensor allows retrieving constraints, e.g. legal requirements needing to be observed and which may have an impact on the decisions to be made. The actor sensor enables importing data from other actor objects of the same model. The value of the attribute collaborators specifies exactly which actors these are. Finally, each actor disposes of a history (History), tracing successful and failed plan execution of previous time steps by storing the corresponding plan IDs (execPlans, failedPlans), therefore providing the possibility to implement actors with some form of learning capabilities. The history base class may also be subject to different extensions for different concrete actor type implementations.

UN

913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969

9

5.2. Decision procedure and model computation of model implementations An actor decides within each time step of the particular simulation model. The decision incorporates four fundamental methods (see Fig. 4 also): 1. query. The actors observe its environment via the three sensor types described above. 2. options. According to the actors’ characteristics and preferences available plans are loaded and made available

5.3. Interaction of water supply company and household actors The models Household and WaterSupply were chosen as examples from the 6 available model implementations (see Table 1 and Fig. 3), since the water supply system and the domestic users form the ‘core’ of water management in many regions. WaterSupply is a model of the water supply sector comprising water extraction, treatment and distribution. Household is a model for estimating the water usage of households under changing environmental conditions. Before describing the models, we will briefly introduce the general idea of how the two models interact. This can best be demonstrated for a situation of water scarcity. Within GLOWADanube ‘water shortage’ is not understood as a technical issue only, but also and foremost as a term related to sustainability of water resources and the good ecological status of natural systems. It is obvious that responses to water shortage are actor- and actor type-specific. Large water supply companies can increase their capacity by tapping new sources and by importing water from other areas while small companies usually only have the first option if they don’t collaborate with other companies. The response options of water supply companies are generally rather constrained by legal restrictions and economic considerations. Consumers in the domestic sector on the other hand may reduce their water demand by


970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026

ARTICLE IN PRESS +


OF

The calculation of the GQN flag is a moderately complex procedure that relies on the results of the DANUBIA models Soil, Rivernetwork and Groundwater (Fig. 3). In principle the GQN flag calculation is based on an analysis of the changes in groundwater recharge, groundwater level and river discharge (baseflow) with respect to reference conditions. The description of the approach would require a sound explanation of the aforementioned hydrological models and an introduction into the hydrological and hydrogeological characteristics of the catchment. Such a description is given in (Barthel and Trifkovic, 2007). DQN flags indicate the quantitative state of the drinking water resources at the disposal of the water suppliers. It is a water supplier evaluation of the quantitative changes in availability of drinking water resources that is further committed to the water users in terms of states from 1 to 4. Essentially, DQN flags represent the simple and standardised way to transfer information about the states of the resources and available supplies that in the real world occur through public media, governmental announcements or the like. Real world analogies for the DQN flag concept can mainly be found in Australia (e.g. Victorian Water, 2005). Fig. 6 shows the principles of data exchange between the models that provide the data for the flag calculation and the flag exchange between the models WaterSupply and Household.

CO RR

EC

TE

DP

reducing the irrigation of gardens or changing bathing and showering habits. Here, legal and economic constraints are much less important for the behaviour of individual households than for public water suppliers. Whereas certain households might adjust their habits only to an actual physical water shortage (no more water from the tap) others will obey much earlier to appeals to save water (see below). It is within this framework of different options and preferences where the interesting interaction between water suppliers and water consumers takes place. Both sides will react to system changes (here decreasing resources) in typical ways. However, only the water supply companies have the data and knowledge to assess the current and future situation whereas consumers rely on secondary information. Therefore the response of the two actors (suppliers and consumers) will take place in a stepwise and distributed way: First, water supply companies will try to make a prediction on the future development of the available resources. Minor problems will usually be overcome by a temporal increase of capacities (using strategic reserves etc.) without letting the public know. If it is foreseeable that resources and reserves together will not be enough to meet the demand and no significant replenishment of the resources can be expected before the crisis approaches, the public will be informed and encouraged to save water. This usually happens in different stages: early warnings (information on potential upcoming shortages), direct appeals to save water, restriction of certain water uses and finally periodical system shut downs. What exactly the coping and information strategy of a specific water supply company looks like and how a specific consumers reacts (i.e. if he follows the appeal and obeys the restrictions) depends on the individual options and preferences (see Sections 5.4 and 5.5). As described above, the interface between the two actors (supplier and consumer) is asymmetric. The suppliers have the full knowledge of the system whereas the individual consumers simply consume without being fully aware of how they influence the system. In DANUBIA, the Household model (consumers) sends the demands per model cell and time step to the WaterSupply model whereas the WaterSupply models provides ‘information’ on the state of the water supply system. This information is provided in a condensed form using so-called ‘flags’. Flags assume integer values from 1 (good) to 5 (catastrophic). The WaterSupply model calculates the flags at each time step based upon a set of physical parameters (see below). With respect to the models described hereafter (WaterSupply and Household), two different types of flags are of importance:

UN

1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083

MODEL

RO

10


1. The groundwater quantity (GQN) and quality (GQL) flags describe the system state of the groundwater resources in a defined zone. The flags are computed for a complete groundwater body, i.e. an array of proxels with similar hydrogeological properties. 2. The drinking water quantity (DQN) flag describes the state of the water supply system. A DQN flag is valid for the supply area of a water supply company.

5.4. The WaterSupply model The WaterSupply model simulates the provision of water from sources to consumers. Having a focus on the public drinking water supply of the Upper Danube Catchment, the sectoral goal of WaterSupply is to identify regions that could experience water stress in the future, resulting either from climate-induced or water-quality-related changes on the supply side or demand-side changes, e.g. quantitative, qualitative, spatial and temporal changes in water consumption. WaterSupply aims to demonstrate the ability of present water supply schemes to compensate such changes and to evaluate the effects of different interventions to solve or prevent problems arising as a result of change. In this context it is important to point out that not the technical infrastructure of individual supply companies is modelled, but rather the supply areas, the important groundwater and surface water bodies used, the degree of use, cooperation between companies, and long-distance water transportation schemes. The model is therefore neither a prediction tool of future water supply infrastructure nor a design tool on the company level. It can only serve for the evaluation of the water supply strategies on a catchment level. Within these described limitations, WaterSupply does foresee typical response mechanisms on a company basis to compensate for change, e.g. the change of preferences regarding resources used or a stronger reliance upon long-distance water supply. In order to achieve its goals, WaterSupply reads the quantitative state of water resources at pre-defined spatial and temporal units, interprets water availability according to sustainability requirements of predefined


1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140

ARTICLE IN PRESS +


MODEL


«component» Soil

WaterSupplyToGW WaterSupplyToHH

getGroundwaterWithdrawal() ...

«component» Groundwater

getDQNFlag() ...

GroundwaterToWS

«component» WaterSupply

getGroundwaterRecharge() ...

«component» Household

OF

HouseholdToWS getDrinkingWaterDemand() ...

RivernetworkToWS getRiverDischarge() ...

RO

«component» Rivernetwork

HouseholdToRN

getWasteWaterQuantity() ...

DP

Fig. 6. Some interface details for the coupled data exchange of natural science and socio-economic simulation models.

level(CommunityWSC), and regional WSC, representing all forms of grouped and joined suppliers that transfer water among communities (RegionalWSC) (Fig. 8). These actor types are equipped with different preferences and different course-of-actions in the form of plans and actions. Communities represent the consumers and are modelled explicitly because they are the legally responsible entities for water supply in Germany. By the summation of imported water demands from other socio-economic models (Table 1), a COM knows where and how much water is consumed. Furthermore, a COM knows from which WSC it is primarily served; typically these are local suppliers. In contrast, a WSC possesses information regarding extraction sites, water rights, raw water quality and collaborating WSC. Extraction sites are mainly groundwater wells where water rights (capacity constraints) are defined by site-specific extraction limits set by the water authorities. The basis of WaterSupply is the comparison of demands obtained by COM with the existing

EC

TE

natural parameters such as groundwater recharge or river water discharge and evaluates the ability of the present water supply schemes to comply with the dynamics of user demand. WaterSupply seeks then both to simulate typical behaviour of supply companies within the limitations specified above and to identify critical regions for which further, more extensive or fundamental adaptation of the water supply scheme will become necessary under changing boundary conditions.

CO RR

5.4.1. Static structure of the WaterSupply model WaterSupply is a strictly object-oriented simulation model that implements the decision making of water supply companies based on the DEEPACTOR framework. Fig. 7 provides a structural overview of the model implementation. At the core of the model is the distinction between water supply companies (WaterSupplyCompany, WSC) and communities (Community, COM). Furthermore, we distinguish between local WSC, which act on the community

has consumers

ConsumerProxel

UN

1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197

11

DeepWSC

Community

activates plan supply

WaterSupplyProxel

access

BusinessAsUsual

has sources

WaterSupplyCompany ExpandCapacities

SourceProxel

flags

has aquifiers

Zones

...

Fig. 7. Structural overview of the WaterSupply implementation. Please cite this article in press as: Barthel, R. et al., An integrated modelling framework for simulating regional-scale actor responses to global change in the water domain, Environ. Model. Softw. (2008), doi:10.1016/j.envsoft.2008.02.004

1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254

ARTICLE IN PRESS +

MODEL


12

1312 1313 1314 1315 BusinessAsUsual BusinessAsUsual DeepWSC DeepWaterSupply 1316 1317 timestep = MONTH 1318 CommunityWSC CalculateCapacityReq ExpandCapacities 1319 1320 RegionalWSC TapNewResources ExpandExistResources 1321 1322 CrisisManagement TapNewResources 1323 1324 EmergencyMeasures 1325 1326 Fig. 8. Specialisation of the DEEPACTOR framework’s base classes in the WaterSupply implementation. 1327 1328 1329 1330 and potential supply capacity evaluated for the current state of of the DeepWSC implementation of the actor methods 1331 the resources. The result of the above mentioned comparison specified by the DEEPACTOR framework (see Section 5 and 1332 triggers a ‘decision’ on behalf of the particular WSC actor. Fig. 4, AbstractActor). The single steps are implemented 1333 To ‘decide’ means here to select and execute an appropriate as described in the following: 1334 plan, consisting of a predefined set of actions with the aim 1335 of satisfying water demands either by exploiting existing 1. query. The necessity for change arises either from grow1336 capacities or by opening new ones. The selection of a plan ing demands, which surpass the present capacity, or from 1337 is not only dependent on demand and state of the resources a quantitative degradation of one or more of the present 1338 but also on the type of the WSC (local or regional), its location sources in use, which leads to a capacity reduction. These 1339 and its collaborators. are perceived by the WaterSupply actors via their sensors. 1340 The actors use the proxel sensor to import natural science 1341 data and the actor sensor to import data from collaborating 5.4.2. Decision procedure and model computation 1342 WSC. of the WaterSupply model 1343 2. options. According to the WSC type (CommuniDuring model initialisation, static model and actor 1344 tyWSC or RegionalWSC) and the characteristics attributes are set and dynamic attributes are initialised. Static 1345 (recourses at disposal, potential collaborators etc.), preferattributes that are not modified during the simulation are for 1346 ences and the available plans are loaded. example existing water sources, company supply areas or 1347 3. filter. The decision procedure to select, activate and transport links among companies. Source’s capacities, with1348 execute one of the available plans is depicted in the drawal values or the state of the resources need to be modified 1349 UML activity diagram in Fig. 9. The parameters during run time and are therefore examples for dynamically 1350 necessary for plan selection are supplier type, demand, usassigned attributes. 1351 able capacity, collaborators, number of potential sources, After initialisation, the model follows the common DANUBIA 1352 GQN flags of sources and potential sources. A detailed decomputation cycle of data import, computation and data export 1353 scription of this procedure can be found in (Barthel et al., with a time step of one month. In getData, the demands submitted for publication). After selection of a plan, its set Q3 1354 calculated and exported by the Actor models Household, 1355 of actions is executed which results in the modification of Farming, Economy and Tourism. Output data such as ground1356 the water supply characteristics such as an expansion of water level, river discharge or groundwater recharge is 1357 existing sources, tapping of new sources, or relieving or imported from the natural science models Groundwater, River1358 closing overused sources, thus increasing the demands network and Soil. The data imported from the natural science 1359 sent to a collaborating regional or long-distance supplier. models is used to assess the state of the supply side, i.e. the 1360 4. export. The results of WaterSupply actors relevant for quantitative and qualitative status of the groundwater and 1361 the collaborators of the WSC are stored to be available surface water resources (GQN flag). The output of the Actor 1362 via the actor sensor in the next time step. models is used to calculate the total drinking water demands 1363 that accrue at each time step (preCompute). Based on the 1364 Finally, the changes of the water supply structure and its comparison among the available supplies and demands needed 1365 view of the state of the resources are actualised (postComat each time step (preCompute), the decision-making ability 1366 pute) and made visible to the other DANUBIA simulation of the water suppliers is activated in order to respond to 1367 models (provide). The main results are the actual potential future problems and to make structural changes 1368 withdrawal quantities exported to the Rivernetwork and accordingly. In technical terms this translates to the execution AbstractActor

AbstractPlan

AbstractAction

CO RR

EC

TE

DP

RO

OF

AbstractActorModel

UN

1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311



ARTICLE IN PRESS +


MODEL


DeepWSC::filter check for collaborator [collaborator = 1]

[else]

check capacity [else]

[else]

[no. of sources with GQF 1,2 < 1]

[else]

select plan 1

check GQF 1

[no. sources with GQF 1,2 >= 1]

select plan 1

check GQF 1 potential

select plan 1

select plan 2

[total dem / total use cap < 0.9]

[else]

select plan 2

OF

check GQF 1

[else]

[no. pot res with GQF 1,2 >= 1]

RO

[total use cap / total init cap = 1]

check capacity index

check GQF 3

[no. sources with GQF 1,2 >= 3]

select plan 2

[else]

select plan 3

DP

select plan 4

TE

Fig. 9. The filter step in the decision procedure of the WaterSupply actors.

EC

Groundwater models as well as DQN flags exported to other Actor models (Fig. 6). 5.5. The Household model

CO RR

The main purpose of the Household model is to simulate the water use of domestic households under changing climatic conditions, showing plausible reactions of households to water scarcity or rising temperatures. Simulation results include water demand, wastewater and the water-related satisfaction of households. In order to calculate the water demand, ten water-usages such as showering, toilet flushing or laundry washing are distinguished. Within the Household model, Household actors decide upon these ten specific water uses. The model sums up the respective water demands and the overall demand is sent to the WaterSupply model (Fig. 6). The most important import parameters for the Household actors’ decision procedure are the drinking water price and the DQN flags, both provided by the model WaterSupply (Fig. 6), and the air temperature, computed within the component Atmosphere (Fig. 3). The four DQN flag levels are interpreted as shown in Table 2.

UN

1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425

13

Table 2 Interpretation of DQN flag values in the Household model DQN flag value

General interpretation by Household Actors

1 2

No problems reported Multiple reports in the local newspaper about potential water supply problems Public appeal to save water issued by the mayor Official restrictions for water use

3 4

The ‘deciding entity’ of the model is the household, not an individual person, because water-related behaviours such as the installation of water-use technologies or the frequency of laundry and dish washing are related to entire households. Households are categorised according to the sociological concept of lifestyles (Bourdieu, 1984). For the model implementation, the Sinus Sociovision lifestyle model (www.sociovision.com) was chosen. Sinus Sociovision divides the German population into ten so-called Sinus-MilieusÒ. Each milieu is described with general values, typical behavioural patterns and socio-demographic data. Microm, a marketing company cooperating with Sinus Sociovision, provides spatially explicit data for the SinusMilieus in Germany. Therefore, the simulation model uses a specific distribution of Sinus-Milieus for each proxel. According to empirical studies conducted within the project GLOWA-Danube, these ten Sinus-Milieus are aggregated to five clusters (Socially Leading Milieus, Post-Materialists, Mainstream, Traditional Milieus, and Hedonistic Milieus) whose characteristics provide the basis for the modelling of the five different Household actor types. Therefore, each inhabited grid cell hosts five Household actors, with each actor representing all households of that specific type. Furthermore, the model applies the framework concept of collaborators to simulate a social network between households: An artificial social network was generated to link similar household types. The specific Household actors take the behaviour of their collaborators into account when deciding upon buying water-related technologies such as rain harvesting systems.


1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482

ARTICLE IN PRESS +

MODEL


OF

At the beginning of a simulation, the Household model is initialised, providing the Household actors with values for habits, preferences, starting values for water-related technologies, and socio-demographic characteristics. For each time step, all necessary data from other DANUBIA models are imported (getData). Next, the Household model computes the water-related satisfaction and the water demand of the public sector (preCompute). Water-related satisfaction is based upon the level and number of DQN flags. When computing the water demand for the public sector, the model takes the population of the proxels into account. After preCompute, the decisions for all Household actors are calculated by the DHHActor implementation of the abstract framework methods inherited from AbstractActor (see Fig. 4): 1. query. Each Household actor perceives its local physical and social environment via its sensors, taking into account air temperature, drinking water price, DQN flags, and the behaviour of its collaborators. 2. options. Certain events can trigger a deliberate decision for normally habitual behaviours, e.g. a DQN flag on a proxel prompts Household actors on that proxel to make a conscious decision about their showering frequency with a possible reduction of water-use. 3. filter. In this step, every Household actor calculates its deliberate decisions and decides upon water-related technologies. For that, they use two different decision algorithms depending on actor type and technology (e.g. Post-Materialists and Social Leaders are highly motivated and therefore always use a deliberate decision, while Traditional actors only use this kind of decision if they think about buying a rain harvesting system, else they use a simple decision heuristic. Details on the installation of water-related technologies are given in Schwarz and Ernst (submitted for publication). Furthermore, actors may decide upon normally habitual behaviour if outer circumstances (as evaluated in options) trigger such a decision. For example, if actors decide upon shower frequency deliberately because of very high air temperatures or water scarcity, they evaluate all plans within the plan group ‘shower frequency’ (to shower twice a day, once

TE

DP

5.5.1. Static structure of the Household model The model implementation specialises the base classes of the DEEPACTOR framework as depicted in Fig. 4. The class DHHActor serves as a base class integrating all Household actors. Household actor types are determined dynamically according to their initially assigned profile. Each proxel contains an actor object for each of the household types existing in this proxel. The amount of households of the same type is a fraction value assigned to the respective actor object. It is computed using the spatial distribution of Sinus-Milieus and the total amount of households imported from the DANUBIA model Demography, which is part of the main component Actor in Fig. 3. Each Household actor has type-specific characteristics, which stem from various empirical studies undertaken within the project. For each water usage, several plans (ShowerHead, ShowerFreq, etc.) and their corresponding actions are implemented (Fig. 10). The plans have several characteristics, which are important for the decision making of Household actors. For example, characteristics for the decision regarding shower frequency are the amount of water used, the financial costs and the modernity of the plan, representing in how far the specific plan complies with modern hygienic habits such as taking a shower every day. Each plan has a corresponding action (ShowerHeadAction, ShowerFreqAction, etc.), which facilitates the computation of the drinking water demand.

RO

14

CO RR

EC

5.5.2. Decision procedure and model computation in the Household model Domestic water-use comprises of both habits and deliberate decision making The Household model addresses both kinds of behaviour which can be explained nicely using the example of showering: Shower frequency and length are normally a matter of habit and not of deliberate decision, while the installation of a new showerhead requires a conscious decision process. Habits are chosen in each time step if there are no extraordinary events such as high temperatures or water scarcity. At the same time, households decide consciously upon plans for water-related technologies in each time step. The UML activity diagram in Fig. 11 provides a high-level overview of the resulting decision procedure.

AbstractActorModel

UN

1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539


DHHModel

AbstractActor

AbstractPlan

AbstractAction

ShowerHead

ShowerHeadAction

ShowerFrequency

ShowerFrequAction

BathFrequency

BathFrequAction

DHHActor

timestep = MONTH

...

...

Fig. 10. Specialisation of the DEEPACTOR framework’s base classes in the Household implementation. Please cite this article in press as: Barthel, R. et al., An integrated modelling framework for simulating regional-scale actor responses to global change in the water domain, Environ. Model. Softw. (2008), doi:10.1016/j.envsoft.2008.02.004

1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596

ARTICLE IN PRESS +


MODEL


DHHActor::options + filter for each of 10 water uses: check events for "breaking the habit" or "(re)install innovation"

[habit]

[conscious decision]

[innovation]

a certain % of actors (re)installs its innovations

[water use]

get plan groups for current innovation

OF

get plan groups for current water use

perform habitual behaviour

calculate utility values

perform conscious decision

RO

calculate fraction values

Fig. 11. High-level description of options and filter in the decision procedure of the Household actors.

show possible reactions of domestic households to Climate Change.

DP

a day, every second day, once a week, or not at all). A Multiattribute Utility approach is used to evaluate these normally habitual behaviours. Situational circumstances (e.g. the level and number of DQN flags in the present and the last time steps) are taken into account by enhancing the importance of single factors entering the decision process (e.g. environmental concerns in the case of DQN flags). Each plan in the plan group executes its corresponding action, which assigns related consequences to the proxel proportionally to the calculated utility value. 4. export. This step allows every actor to make his behaviour or installation visible to his neighbours and friends (i.e. the collaborators).

The presentation of the scenario results shows representative clippings that display some of the principles and the capabilities of the developed methodology. Full scenario simulations of the DANUBIA system provide an extremely large amount of outputs in the range of 100 parameters which can each be presented in maps, time series, tables and so forth. The GLOWA-Danube online Atlas5 provides a platform for all interested in a wider facet of results.

TE

EC

CO RR

After these computation steps considering the behaviour of the specific Household actors, the main class DHHModel sums up the water demand of all Household actors and the public drinking water demand on a specific proxel to provide both the overall drinking water demand and the amount of waste water produced (postCompute). Finally, the Household model provides this data for other DANUBIA models via its export interface (provide).

6. Application of the DEEPACTOR approach

UN

1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653

15

Applicable results of WaterSupply and Household according to the principle aims of GLOWA-Danube and DANUBIA are outputs that can be used to identify and visualise water supply and consumption-related tendencies and developments in the Upper Danube Basin under conditions of Global Change, based on scenario calculations (2000e2100) i.e. to: predict the general situation of water supply in the basin (good, stable, endangered, unsafe), identify critical regions and the rate at which critical developments arise (slow, rapid), predict overuse and ecological risks, sketch middle-term trends in water consumption, and to

6.1. Climate scenarios used The results presented here were obtained by applying the DANUBIA system to three climate scenarios which are explained briefly in Table 3. The ‘business as usual scenario’ (Table 3) was created using a stochastic procedure to derive long time series of synthetic future climate data6 from measured historical records. To compile a future meteorological data set spanning the next 100 years, the procedure considers measured relations between temperature and rainfall, applies a random variation of temperature, overlays a trend (here 2.7 K/100a which corresponds to a IPCC B2 scenario), and selects the appropriate time slice from the given basic population of measurements (30 years of DWD (German Weather Service) recordings). It should noted that this methodology has the following disadvantages: reduced representation of auto-correlation, lacking consideration of potential changes in extreme values and an increase of the relative error of determination when approaching the edges of distributions. However, for the present purposes, the advantages clearly dominate: physical consistency of 5

http://www.glowa-danube.de/atlas, currently (August 2007) still restricted to members of the research consortium. 6 The full scale approach for generating regional climate data is a downscaling of GCM results using MM5 integrated in the Atmosphere component of DANUBIA (see Fig. 3) and detailed in Fru¨h et al. (2006).


1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 Q6

ARTICLE IN PRESS +

MODEL


16

Table 3 Climate scenarios used for simulation runs of the DANUBIA system Description

Comment

Business as usual scenario

IPCC B2 type scenario generated using a stochastic climate scenario generator (see below) Observed data from 1970 to 2003 (reference period), yet using starting conditions (physical models) as for the other scenarios An extremely dry ‘scenario’ generated by rearranging the respective 5 hottest and driest months from the 1970 to 2003 reference period

Temperature increase of 2.7 K/100a; only the first 35 years are used here 35 years; used for model validation; optimistic in the sense that all current GCM predictions are ‘warmer’ than the conditions in this period 35 years; used to force significant model reactions and to show the sensitivity of the system. Rather unrealistic and in the pessimistic range of current GCM results

Optimistic scenario (also validation period) Pessimisticdworst case scenario

OF

Scenario name

for more conscience use of water or capacity problems of the water supply structure. e Plan execution (choice of plans and actions to be taken) of the WaterSupply actors as an indication of the behaviour of the water supply sector as a consequence of Global Changes. The choice of plans can be interpreted as an identification of ‘bottlenecks’ of the system and a suggestion of the possible solution at the same time according to some predefined water management orientation.

DP

RO

meteorological input data, meteorological model inputs within a validated range, methodological consistency for ensemble simulations, possible validation versus baseline scenarios; consistent spatial resolution of input data. Fig. 12 shows a comparison of precipitation, which is the main relevant driver of the system for the three different scenarios described in Table 3. 6.2. Results of the WaterSupply model

Generally, the following outputs of WaterSupply (amongst others) can be used for further analysis:

In the following, examples for each of these three result types will be presented. Fig. 13 shows the spatial distribution of the GQN flag at the end the simulation period for all three climate scenarios. More interesting than the situation at one single time step is the temporal evolution of the flags during the simulation period. Unfortunately a combined presentation of spatially and temporarily distributed results is not possible in a printed medium. Therefore, Fig. 14 shows time series of spatially aggregated values of the GQN flag for the whole catchment from all three scenarios.

3000 2800 2600 2400 2200 2000 1800 1600 2005

UN

3200

CO RR

EC

TE

e Changes to the GQN flag in space and time, indicating changes in natural water availability as a consequence of Global Change. Changes for the worse indicate that ecological constraints or sustainability criteria are endangered. e Changes to the DQN flag in space and time, indicating changes in technical water availability as a consequence of selected water management scenarios. Changes for the worse indicate demand management measures, appeals

Precipiation [m3/s]

1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767


2010

2015

2020

2025

2030

2035

2040

Time P 'business as usual'

P 'b.a.u.' mov.av.5a

Linear (P 'business as usual')

Linear (P 'pessimistic')

P 'pessimistic'

P 'optimistic'

Linear (P 'optimistic')

Fig. 12. Comparison of precipitation (P) for three different climate scenarios (Table 3); annual sums and linear trends are shown for all scenarios, the moving average (5 years, central) is shown for the business as usual (b.a.u.) scenario only. Please cite this article in press as: Barthel, R. et al., An integrated modelling framework for simulating regional-scale actor responses to global change in the water domain, Environ. Model. Softw. (2008), doi:10.1016/j.envsoft.2008.02.004

1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824

ARTICLE IN PRESS +


MODEL


OF

RO

Fig. 13. Spatial distribution of the GQN flags for the whole catchment on July 2039 for the 3 climate scenarios: optimistic (left), business as usual (middle) and pessimistic (right).

of the model availability to implement different actor’s behaviour three exemplary behaviour modes of the WaterSupplyCompanies are shown in Table 4. Although each WSC actor may have different behaviour, for simplicity reasons three scenarios in which all WSC actors behave in the same mode are presented here. The DQN flags are obtained by comparing exerted water demands with the state of the available supplies and maximal withdrawal source capacities on the WSC basis. Beside technical limitations of available sources, the behaviour mode of WSC actors has

TE

DP

The GQN flags will not be discussed further as they form rather an input to the decision-making process of the actors than an output. With respect to the DEEPACTOR approach, the DQN flags are more interesting as they express different behaviour or preferences of WSC actors. The interpretation of changes in the state of the groundwater bodies can vary depending on the sensibility (i.e. do they care about the environment or not) of the Water Supply Companies and their willingness to communicate these changes to the water users (household in this case). For the purpose of the demonstration

1400

EC

4

CO RR

1200

Average GQN

3

800

600

400

200

2038

2037

2036

2035

2034

2033

2032

2031

2030

2029

2028

2027

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

0 2015

2014

2013

2012

2011

2009

2008

2007

2006

1

2010

UN

2

1000

Groundwater Recharge [m3/s]

1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881

17

Time Average GQN "Business as usual"

Average GQN "Optimistic"

Average GQN "Pessimistic"

GWR Buisness as usual MovAv 3a

GWR Optimistic MovAv 3a

GWR Pesimistic MovAv 3a

Fig. 14. The quantitative state of the groundwater resources expressed with the GQN flags as consequence of the changes in groundwater recharge (GWR, moving Averages 3 years, right y-axis, upper part of the figure) where 1 means ‘good’ state and 5 means ‘bad’ state or identified changes to the worse in the available quantity for all three climate scenarios (Table 3). Please cite this article in press as: Barthel, R. et al., An integrated modelling framework for simulating regional-scale actor responses to global change in the water domain, Environ. Model. Softw. (2008), doi:10.1016/j.envsoft.2008.02.004

1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938

ARTICLE IN PRESS +

Description

Ignorant

WSC actors disregard the GQN flag, do not communicate these changes to the water users and try to satisfy water demands by implementing technical measures for greater water extraction without paying any attention to sustainability and ecological issues WSC actors consider changes in natural state of the water resources as very important, communicate these directly to the users and try to identify management options in order to protect the sustainable state of the water resources WSC actors respect environmental aspects in general but behave at the same time pragmatic and economically oriented. It therefore can be seen as a middle (normal) course between the two extremes ‘ignorant’ and ‘sensitive’

Middle

6.3. Results of the Household model

DP

a prevailing role in the calculation of the DQN flags. Similar to the GQN flags (see above), the DQN flags are presented spatially distributed in snapshot maps from the end of the simulation period (Fig. 15) and temporarily distributed as time series of aggregated flag values (Fig. 16) for the three management scenarios. Looking at the average values of the DQN in Fig. 16, the ‘ignorant’ behaviour of the WSC causes reporting of constantly ‘good’ DQN flag values to the water users, which in turns provokes no actions from the user side. The ‘middle’ and the ‘sensitive’ behaviour of WSC report the changes in the state of the water resources to the water users, thereby demanding their participation in the identification of the sustainable management solutions. The behaviour of the WSC actor can be expressed through the selection of the different plans (see Fig. 8). Fig. 17 summarises the number of WSC that executed a certain plan over the simulation period for the three management scenarios (Table 4): Fig. 17 shows that most companies select the Plan 1 (BusinessAsUsual) most of the time. This is directly related

OF

Behaviour mode

to the good state of the water resources in much of the investigation area. It should be noted that the initially predefined water demand and provision of water is higher than the actual water demands exerted on water supply companies. Therefore, the WSC actors must adjust their capacities at the beginning of the simulation, either by selecting Plan 2 (‘ExpandCapacities’) or Plan 4 (‘CrisisManagement’). For the few WSC actors that do not have enough potential sources to expand, Plan 4 remains the only option. In these cases, the use of Plan 4 does not indicate the real water stress in the area but is a consequence of data unavailability regarding existing water supply capacities. These initial increases of the available capacity proved to be enough for the provision of enough water throughout the rest of the simulation period. The changes in the availability of the water resources are small enough to be satisfied from the existing technical capacities and only small upgrades have been identified as necessary (small variations in the selection of Plan 1 and Plan 2).

RO

Table 4 Behaviour modes of WSC actors used in the scenario simulations

Sensitive

MODEL


18

CO RR

EC

TE

The Household model also provides a large variety of result data on various aggregation levels, e.g. seasonal changes of water demand within the catchment showing reactions of households to air temperature; discontinuous changes of water demand due to DQN flags, stemming from both changing, formerly habitual waterrelated behaviour and an enhanced installation of watersaving technologies; continuous changes of water demand due to the installation of water-related innovations, leading to a slight reduction of water demanddthese changes can be tracked back to the various actor types showing different adoption of these innovations.

With the three chosen climate scenarios (Table 3) and the three different water supply management scenarios (Table 4), the number of scenarios (‘storylines’) from the viewpoint of the domestic water users represented by the Household model

UN

1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995


Fig. 15. Spatial distribution of the DQN flags for the whole catchment on July 2039 for the business as usual climate scenario (Table 3) and the three behaviour modes of water supply companies (Table 4): sensitive (left), middle (middle) and ignorant (right). Please cite this article in press as: Barthel, R. et al., An integrated modelling framework for simulating regional-scale actor responses to global change in the water domain, Environ. Model. Softw. (2008), doi:10.1016/j.envsoft.2008.02.004

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052

ARTICLE IN PRESS +


MODEL

R. Barthel et al. / Environmental Modelling & Software xx (2008) 1e27 4

2000 1900 1800 1700 1600

3

1500 1400 1200 1100

2

OF

1000

1

2038

2037

2036

2035

2034

0 2033

2023

DP

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

0

2028

100

2026

200

2025

300

2024

400

RO

500

2032

600

2031

700

2030

800

2029

900

2027

Number of DQN

1300

Average DQN

Time

DQN=3

DQN=2

DQN=1

DQN=4

Average DQN "sensitive"

Average DQN "middle"

TE

Average DQN "ignorant"

EC

Fig. 16. The quantitative state of the water supply system expressed with DQN flag as a consequence of the three water supply management scenarios (Table 4).The average of all the flag values for the catchment and the total number of flags of a category are shown for all three management scenarios.

2000 1900 1800 1700

CO RR

1600 1500 1400

Number of Plans

4

1300 1200 1100 1000 900 800 700

2

UN

600

3

500 400 300 200 100

Average DQN

1

0

2038

2037

2036

2035

2034

2033

2032

2031

2030

2029

2028

2027

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

0

2006

2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109

19

Time Plan 1

Plan 2

Average DQN "middle"

Plan 3

Plan 4

Average DQN "sensitive"

Average DQN "ignorant"

Fig. 17. Summary distribution of the Plans for the whole catchment over the whole simulation and average DQN flags for all three water supply management scenarios (Table 4). Please cite this article in press as: Barthel, R. et al., An integrated modelling framework for simulating regional-scale actor responses to global change in the water domain, Environ. Model. Softw. (2008), doi:10.1016/j.envsoft.2008.02.004

2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166

ARTICLE IN PRESS +

MODEL


OF

business water consumption are combined to one value in the official statistics. This issue is discussed later on in Section 6.4, which deals with the validation of the models. When exploring temporal dynamics, domestic water demand shows clearly seasonal changes in all scenario runs (Fig. 20), mostly due to increased showering with higher air temperatures. In addition, discontinuous variability in water demand can be detected. Differences in domestic drinking water demand between the scenarios ‘ignorant’ and ‘sensitive’ and between ‘ignorant’ and ‘middle’(see lower lines of Fig. 20) are reactions to the different scenarios of DQN flags provided by WaterSupply. This variability is due to the reaction of households to DQN flags: Households perceive flags and interpret them as newspaper articles, appeals to save water or even restrictions on water use (Table 2). According to their characteristics, households (1) reduce their water demand by constricting e.g. shower frequency, bathing or cleaning, (2) are further encouraged to buy water-saving technologies (see below). Over the years, the water demand decreases (see Fig. 20, upper line). This is the result of two factors. First, the simulated water scarcity results in some regions in a reduced drinking water demand because Household actors lower their water use in response to the DQN flag. The second and main reason is the installation of more water-efficient technologies (e.g. water-saving showerheads, toilet tanks, and washing machines). In the scenario runs, water-saving technologies further diffuse in the catchment area, e.g. rain harvesting systems almost double from about 6.4% to 12%

EC

12

CO RR

3

8

6

4

2

2035

2033

2031

2029

2027

2025

2023

2021

2019

2017

0 2015

2013

2011

2009

2007

1

UN

2

10

Air temperature (mean) [°C]

Average of DQN flag [ ]; Precipitation [(m3/s)/1000]

TE

DP

is nine. For that reason the following discussion of the results will be based on the business as usual scenario (Table 3), which is assumed to be the most consistent and likely climate scenario, only. In order to emphasise the interaction between the models WaterSupply and Household, population and water prices were kept constant for all simulation runs. The main driving forces for the Household model are the DQN flags from the WaterSupply model and the air temperature, which are shown in Fig. 18. According to the strategy chosen by the water supply companies, domestic Households are confronted with different levels of public awareness regarding water availability (in Fig. 18 shown as the average value of the DQN flag for the whole catchment). In the scenario ‘ignorant’, almost no communication of problems in water availability takes place, while in the scenario ‘sensitive’ nearly constant communication regarding these issues can be detected throughout the simulated area. In the scenario ‘middle’, which, from the three water supply management scenarios is probably the most likely one, the communication between WSC and Household remains in average on a very low level between ‘no problems reported’ and ‘newspaper reports’ (Table 2). A climax can be observed in the 2020ies, a period much dryer than average (see Fig. 12). The resulting domestic drinking water demand is spatially explicit, indicating high demands in Munich (agglomeration in the centre of the catchment) and a few larger cities (Fig. 19). These higher per capita demands in urban areas do not necessarily mean that people in big cities consume more water. In Germany, domestic water use and small

RO

20

2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223


Time DQN "ignorant" T mov. av. 3a

P mov. av. 3a DQN "sensitive"

Linear (T mov. av. 3a)

DQN "middle"

Linear (P mov. av. 3a)

Fig. 18. Main driving forces for the Household model (air temperature [T] and DQN flagsdadditionally precipitation [P] is shown as an indirect driver) for the three water supply company management scenarios (Table 4), all based on the business as usual climate scenario (Table 3). An ‘ignorant’ WSC behaviour causes a constant transmission of DQN flags with the value 1 (‘no problems reported’), i.e. all problems are solved on the WSC level (Table 4). For T and P moving averages and linear Trends are shown; P values were divided by 1000. Please cite this article in press as: Barthel, R. et al., An integrated modelling framework for simulating regional-scale actor responses to global change in the water domain, Environ. Model. Softw. (2008), doi:10.1016/j.envsoft.2008.02.004

2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280

ARTICLE IN PRESS +


MODEL


Domestic drinking water demand [m3/s] of households as simulated by the Household model (december 2036, scenario "ignorant"). uninhabited 0.001 - 0.005

TE

DP

RO

OF

0.005 - 0.01 more than 0.01

Fig. 19. Spatially explicit domestic drinking water demand as simulated by the Household model. As an exemplary result, the demand per proxel of December 2036 (scenario ‘ignorant’dTable 4) is depicted.

3

2

19 18 17 16 15

UN

Differences in DWD [m3/s]

DQN flags. Post-Materialists and Social Leaders always choose rain harvesting systems when they have the opportunity, therefore adoption cannot increase due to DQN flags. Hedonistic actors are the most responsive because they do

CO RR

EC

in 2036. Furthermore, households react to water scarcity when deciding upon their water use technologies: Example results for the diffusion of rain harvesting systems show (Fig. 21) that Household actors respond differently to

1

14 13 12

DWD in scenario "ignorant" (sum) [m3/s]

11 0

Difference "ignorant"-"sensitive"

DWD "ignorant"

Difference "ignorant"-"middle"

2035

2033

2031

2029

2027

2025

2023

2021

2019

2017

2015

2013

2011

2009

10

2007

2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337

21

DWD ign. mov. av. 12month

Fig. 20. Domestic drinking water demand (DWD) of households in the three scenario runs (Table 4). Please cite this article in press as: Barthel, R. et al., An integrated modelling framework for simulating regional-scale actor responses to global change in the water domain, Environ. Model. Softw. (2008), doi:10.1016/j.envsoft.2008.02.004

2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394

ARTICLE IN PRESS +

MODEL


22

OF tic

on

H ed

M ai

ns

tre

is

am

s

ad

Tr

Le

al

iti

ad

on

al

er

ts

lis

ia

ci

UN

CO RR

EC

TE

DP

RO

"

"

nt

ra

nt

ra

no

So

no

"ig

ig

er

"-

-"

m at

ve

e"

st

iti

dl

Po

s en

id "m

s

Differences in installed rain harvesting systems [%]

2452 2453 1 2454 2455 0.8 2456 2457 2458 0.6 0.92 2459 2460 0.4 2461 2462 0.2 2463 0.09 0.07 0 0 0.14 2464 0.01 0.01 0 0 0 2465 2466 2467 2468 2469 Fig. 21. Comparison of the number of rain harvesting system installations of all five household actor types (differences between installation numbers for the three Q5 2470 scenarios according to Table 4 are shown in %). Differences are rather low because installation of rain harvesting systems is only considered when building a new 2471 house or during a major reconstruction). 2472 2473 2 2474 of 1000e5000 km and are mainly based on random not install rain harvesting systems at all if there are no 2475 problems in water scarcity. sampling, domestic water consumption is collected on 2476 a community level annually and no distinction is made 2477 between domestic and small business consumption, etc.). 2478 6.4. Validation of the DEEPACTOR framework and (b) Much ‘socio-economic’ data are not available to the public 2479 the DEEPACTOR models or are very difficult to obtain. Data are not collected 2480 systematically on a countrywide level, but on a ‘problem 2481 The development of validation strategies is a research topic oriented’ basis, i.e. for certain areas and for certain periods 2482 of its own within GLOWA-Danube and will not be treated here in only. The results of such data collections may be stored at 2483 depth. Generally, within fully integrated and multidisciplinary various locations (private companies, authorities on 2484 systems such as DANUBIA, the objective of validation cannot different levels) and are subject to data protection. Finally, 2485 be the validation of single models or even single process much data are privately owned. 2486 descriptions alone but more the validation of the coupled 2487 system as a whole. Here we will not discuss the validation of With respect to the DEEPACTOR approach and its model 2488 natural science models, which follows in principle the tradiimplementations WaterSupply and Household described in 2489 tional ‘comparison observed-simulated’ approach,7 but the this article, data belonging to the categories described above is 2490 validation options for results of the DEEPACTOR models in the only interesting for the intermediate level of results. More inter2491 socio-economic sector. For the socio-economic models within esting would be data that can directly be related to the results 2492 the actor component (Fig. 3, Table 1) data that can be used which are relevant with respect to the main objectives of these 2493 for validation is generally less accurate, meaningful and relimodels, i.e. results that can directly be used for decision making 2494 able than equivalent data that can be used in the natural science or in stakeholder discussion. Interesting results of this category 2495 sector. Typical socio-economic model output variables would would be the DQN flag or the water-related satisfaction of 2496 for example be domestic water consumption (model Housedomestic consumers. As was shown in the previous sections, 2497 hold), household income (Economy), crop yields/prices (Farmsuch results are highly influenced by the preferences and options 2498 ing), population (Demography) overnight stays (Tourism) and that were assigned to the respective actors. Unfortunately little is 2499 groundwater withdrawal (WaterSupply) (see Table 1). Regardknown about the ‘behaviour’ of water supply companies or the 2500 ing observed values for these data two main problems exist: development of water consumption behaviour under conditions 2501 of Global Change. It is therefore very difficult to validate the 2502 (a) The spatial and temporal accuracy of the available observed respective model results in a traditional way. 2503 data is low and data are often unreliable (e.g. agricultural This general problem of socio-economic models that rely on 2504 statistics are on district level so that values relate to areas a large degree on assumption about actor responses to change is 2505 amplified in the Upper Danube Catchment because problems 2506 with water quantity have been almost unknown during the last 7 In fact, the validity of the natural science model output has a major impact 2507 2e3 decades and water availability problems from earlier on the validity and uncertainty of the results of the dependent socio-economic 2508 periods (e.g. in 1976, reported in BAYLFW, 1979) cannot be models, but to shorten the discussion we assume that the inputs are ‘correct’. "s

2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451



ARTICLE IN PRESS +


MODEL


OF

short-term, local water scarcity in parts of the catchment could be observed. Reports on these events form a valuable basis for the parameterisation and further discussion of models and approaches (BAYSTMUGV, 2003; BAYLFW, 1979, 1980, 1990; BUWAL, 2004; Eybl, 2004; LUBW, 2004). Finally, for the validation of ‘behaviour’, telephone interviews with 1026 randomly selected households were carried out. The interviews were based on the results of a questionnaire sent to 1000 households in a smaller test area earlier on. More than 2000 questionnaires were sent to WaterSupply companies (Nickel et al., 2005). The questionnaire results formed valuable information to develop the models but will not be further discussed here. 7. Conclusions

RO

The DEEPACTOR approach described in this article was developed within the framework of the large-scale research project GLOWA-Danube and is integrated in the very complex simulation system DANUBIA. Our conclusions must therefore very briefly address the project and the DANUBIA system as a whole before we comment on the DEEPACTOR approach and the respective DEEPACTOR model results.

DP

compared to the current situation because the infrastructure has been extended considerably since this time. As a result, many important outputs of DANUBIA, in particular from the socio-economic sector, cannot be validated in the classical sense on the basis of measured data. The only means for validation is to include expert knowledge and consumer experiences, e.g. the judgement from water supply company managers, local water authorities or consumer opinions collected using questionnaires (see e.g. Dow et al., 2007). In order to use such soft data, it is necessary to use transparent, process-based models (see Section 2). Only then can subjective and uncertain parameters of the calculation be meaningfully discussed and adjusted according to the advice of local experts. Again it must be pointed out that the socio-economic DEEPACTOR models’ outputs depend directly on the natural science models outputs and therefore inherit the respective models’ errors and uncertainties. The DANUBIA system as a whole relies to a large degree on the quality and uncertainty of the drivers, i.e. foremost the climate scenarios. The scenarios used should therefore be regarded as a variety of possible futures (see Brewer, 2007) and the simulation results therefore as a way to explore ranges of possible developments within the boundaries of such future scenarios. The results are not predictions and the DSS DANUBIA does not propose solutions but rather provides a basis for discussing potentially upcoming problems and the keys to avoid or manage them. The means of validation that can be applied are often indirect and not of numerical type. The first validation option is the comparison of intermediate results to observed values. In the case of the Household model, this would be for example the domestic drinking water demand (per community). In the case of the WaterSupply model, it would be the groundwater withdrawal (per source). The matches between modelled and observed groundwater withdrawal rates for example are reasonably good but with limitations that can be contributed to the weak data availability and the large scale of the models (see Barthel et al., 2005). Second, with respect to the flag values, where no measured values exist, model simulations for the reference period 1970 to 2003 (Table 3) can be compared to verbal descriptions of the state of groundwater resources and the water supply system in the literature. In the case of the upper Danube Catchment, such evaluations are rather general, yet valuable for the parameterisation and a comparison of the model results for ‘normal’ periods (e.g. BAYLFW, 1980, 1990, 1996; BAYSTMLU, 1994). For such ‘normal situations’ and ‘normal behaviour’ of the actors in the past, model results of the hydrological models and subsequently the WaterSupply model show a good status of groundwater resources and an optimal performance (i.e. no shortages) of the water supply system if a ‘normal’ behaviour of the water supply companies is assumed. On the other hand, if rather drastic climatic conditions are forced (see pessimistic scenario, Table 3) problems develop in exactly those regions with unsuitable hydrological conditions that are outlined by the administrative groundwater reports mentioned above. For such water scarcity situations, the extremely hot and dry years 1976 and 2003 can be used for reference. In those years ecological (rivers, wetlands) and economical (navigation, hydropower) damage and some

7.1. The integrative perspective of GLOWA-Danube

The hydrological cycle is a complex system, playing an important role for many human activities. In turn many human activities affect and change the hydrological cycle significantly. A main challenge is the fact that human activities adapt to change and can therefore not be represented as static, repetitive processes. What is needed is a certain degree of ability to react to changes by making decisions that are not predefined by trivial input/output functions. In GLOWA-Danube, the DEEPACTOR approach and its generic implementation, the DEEPACTOR framework, were developed to tackle this challenge. The framework provides a common conceptual and architectural basis for the development of the socio-economic simulation models in the coupled simulation system of DANUBIA. A very important aspect hereby is the integration of decision makers, stakeholders and experts. Models that provide multidisciplinary results and show the interdependencies of the relevant processes are helpful tools to foster this integration. Models in that context should not necessarily be seen as the tools that provide the management solutions but as means to raise awareness of how natural and social systems interact and what the social, environmental and economic costs of human interference with the natural system are. A main challenge thereby is to find the balance between an appropriate, realistic and accurate representation of multifaceted, complex natural and social systems on the one hand, and on the other hand to provide results that satisfy the demands of disciplinary experts as well as the demands of stakeholders and decision makers in all required fields.

TE

EC

CO RR

UN

2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565

23

7.2. The DEEPACTOR framework The DEEPACTOR framework as an object-oriented extension is a specialisation of the DANUBIA developer framework. Its


2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622

ARTICLE IN PRESS +


OF

(b) its method and mechanism for data exchange during runtime via formally specified interfaces and (c) its concept of simulation time with a consistent dynamic integration of model-specific time steps. As a consequence any DEEPACTOR model may be executed and coupled to any other DANUBIA simulation model completely transparent for the rest of the system. In fact, this integration is one of the reasons not to build on existing toolkits such as Repast (North et al., 2006), MASON (Luke et al., 2004) or Swarm (Minar et al., Q3 1996). It would be by far not trivial and even confusing to Q3 embed them into DANUBIA, since the particular base classes and modelling elements usually are more general, having their own notions and concepts of simulation area, time and data exchange, i.e. of concepts which are fixed in DANUBIA. Summarising, the DEEPACTOR framework and also the developer framework of DANUBIA can be seen to be more special cases of general frameworks developed elsewhere. Both have been designed and implemented with focus on the requirements of the simulation models of GLOWA-Danube. We belief that this approach is most promising since it seems to be an appropriate compromise between generality and specificsdit provides starting points without restricting to much and it restricts enough to allow for useful generic implementations on the framework level.

DP

design follows the specific requirements for an ‘agent-based’ social simulation approach as given by the socio-economic simulation models of GLOWA-Danube. Along this line, the framework provides a common generic architecture, a modelling framework with a multitude of basic, but mostly optional, features to be used within the implementation of the concrete simulation models. An important architectural aspect is the explicit modelling of actor, plan and action. Even though in the case of the Household model, the actors’ types are designated only implicitly by an assigned profile, there are explicit subclasses for plan and action types providing a direct representation of the conceptual course-of-actions. In the case of the WaterSupply model, there are two actor types explicitly distinguished by their available plans and their type-specific decision procedure. In both model applications the direct and explicit modelling of types (subclasses) provides both a clear specification of the current model structure and the possibility for incremental modular extensions in future model versions. The flexibility of the framework design is illustrated by the quite different model implementations Household and WaterSupply. For example, both simulation models make use of the basic actor features location and collaborators, but, in the case of the WaterSupply model, the actors are located on more than one proxel, whereas the Household actors are defined as one per type (milieu) and proxel object. The collaborators of WaterSupply actors model ‘real’ collaborators of a WSC for providing water, whereas collaborators of Household actors are used to simulate an artificial social network and the diffusion of water-related technologies. From a more general point of view, the application of the DEEPACTOR approach is advantageous for the water supply sector because it facilitates a flexible and realistic response to system changes. Scenarios can be defined easily by adjusting actor types and preferences; critical states can be identified without having to model the infrastructure of individual WSC explicitly. Moreover, the modular and extendable model structure enables participatory model design and simulation, as explained and discussed e.g. in Ramanath and Gilbert (2004), providing an effective tool for analysing effects of different course-of-actions of water supply companies with respect to response to Global Change scenarios. In contrast, the simulation model of water-related behaviour of domestic households does not aim to provide adjustable parameters in the context of a realistic and direct model of environmental circumstances. Rather, in this case, the application facilitates the bottom-up approach to social modelling. Based upon the data provided by Sinus Sociovision, the model enables the explanation of observable macro phenomena with a low resolution of type dependent decision algorithms and local rules for social network generation, both formulated on the proxel-level. Therefore, in contrast to a purely equation-based model, the model can be more easily communicated to stakeholders and is, in general, more comprehensible due to the conceptually and technically explicit model structure. Finally, the embedding into DANUBIA as an extension of an already existing abstract base class allows benefiting from the treatment of (a) the simulation area as a grid of proxel objects,

7.3. The WaterSupply and Household DEEPACTOR model implementations

TE

EC

CO RR

UN

2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679

MODEL

RO

24


The two models chosen can only partly represent the complex processes and structures that determine the patterns of water supply and consumption in a large heterogeneous domain such as the Upper Danube catchment. The current model implementations should be regarded as preliminary versions to test and show the capabilities of the approach. The models are not yet fully validated and not all options offered by the DEEPACTOR framework are used. In that respect, the greatest advantage of the approach lies in the fact that it is easily extendable in terms of plans, actions, preferences and so forth in a consistent way without the need of significant changes to the model structure. This was demonstrated for example by assigning different reaction schemes of water supply companies toward change state of groundwater resources. Our results have been discussed from two points of view. First, we evaluated and compared the application of the DEEPACTOR approach and the DEEPACTOR framework, respectively, by means of the models WaterSupply and Household. Second, we briefly presented domain-specific results of a scenario simulation, primarily to illustrate the functionality of the coupled models WaterSupply and Household within DANUBIA. In addition to the specific outcome of the models (DQN flags and domestic water demand), the chosen plans and actions of actors as underlying mechanisms were analysed. It was shown on hand of a very limited choice of results that both models provide a spatially and temporally explicit, process based way of simulation response of main actors in the water sector to Global Change. They show very clearly where and when undesirable developments might occur in a relatively simple


2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736

ARTICLE IN PRESS +


MODEL


way that allows for an easy detection and explanation of causeeeffect relations. Rather than forming black boxes, the processes leading to system changes are controlled by a limited set of intuitively understandable preferences, options and triggers of model actors that are simplified but still very similar to the actors in the real world. The results are therefore very well suited to form a basis for active participation of decision makers and stakeholders in coming to discussions of how we should prepare for possible developments which may be brought upon by Global Change.

and Stakeholders’ views and expertise will now be included in the system to improve its applicability. The final aim is to provide a system in the context of an open source project which is operational outside the research consortium at the end of the third project phase (April 2010). Uncited references Barthel et al., 2008; Belz et al., 2004; Sociovision, 2007.

RO

GLOWA-Danube is funded by the BMBF (Bundesministerium fu¨r Bildung und ForschungdGerman Federal Ministry of Education and Research). We would like to thank all governmental organisations, private companies and others who supported our work by providing data, models, advice or additional funding. We would like to thank our colleagues from the partner projects within GLOWA-Danube for the cooperation throughout the last six years. Our special thanks go to Rolf Hennicker and Andreas Ernst, who supported the present study directly as leaders of the sub-projects and contributed in a number of discussions to the concepts of the DEEPACTOR approach, and to Roman Seidl who kindly supported scenario runs of the Household model.

7.5. Outlook

CO RR

EC

TE

DP

The decision support system DANUBIA and the DEEPACTOR approach presented in this article are a compromise between the size and the complexity of heterogeneous natural systems in large catchments, the complexity of human behaviour, the high degree of inherent uncertainty in both natural systems and human society and the need to realistically and meaningfully evaluate the impact of Global Change on the environment and human welfare. The aim of DANUBIA is to describe the water cycle and its physical and socio-economic components as a whole and not so much to describe individual sectoral processes. Experts from different disciplines may therefore find the representation of their discipline over-simplified. DANUBIA was developed for use on a very high administrative level (governmental institutions on state, country or river basin level), where knowledge, data and financial resources to set up and run the required models are available. It was developed as a generic system that is transferable and reusable but not necessarily scalable. It can be applied everywhere, but an application must be based on a high-level political decision and respective financial resources to allow its implementation, since the volume of data and financial resources needed to parameterise the individual models are quite extensive. An application to smaller scale ‘test case studies’ is technically possible but largely meaningless since it contradicts the regional scope of the approach.

OF

Acknowledgements 7.4. Applying DANUBIA and the DEEPACTOR approach elsewhere

Within the last 7 years the multidisciplinary GLOWA-Danube consortium has developed the highly complex integrated simulation system DANUBIA to describe the water cycle of a regional scale river basin in all its aspects. DANUBIA has until now been developed and validated to a degree that allows its use by the research consortium to perform a large variety of scenario calculations. The next step is to use the software to tackle concrete management questions, defined by the main water authorities8 of the basin. A stakeholder process, mediated by IFOK,9 has now been initiated. Water authorities’

UN

2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793

25

8

The LfU Bayern, the state wide responsible authority for all water and environmental questions, is now a member of the research consortium. 9 http://www.ifok.de/en/home/.

References

Arnell, N.W., 1998. Climate change and water resources in Britain. Climatic Change 39, 83e110. Athanasiadis, I., Mentes, A., Mitkas, P., Mylopoulos, Y., 2005. A hybrid agentbased model for estimating residential water demand. Simulation 81 (3), 175e187. Axtell, R.L., 1999. Why agents? On the varied motivations for agent computing in the social sciences, in: Agent Simulation: Applications, Models and Tools. Barthel, R., Janisch, S., Nickel, D., Trifkovic, A. Using the multiactor approach in GLOWA-Danube to simulate decisions for the water supply sector under conditions of global climate change. Water Resources Management (submitted for publication). Barthel, R., Mauser, W., Braun, J., 2008. Integrated modelling of global change effects on the water cycle in the upper Danube catchment (Germany)dthe groundwater management perspective. In: Carillo, J.J., Ortega, M.A. (Eds.), Groundwater Flow Understanding from Local to Regional Scale, 12. International Association of Hydrogeologists, Selected Papers on Hydrogeology, pp. 47e72. Barthel, R., Nickel, D., Meleg, A., Trifkovic, A., Braun, J., 2005. Linking the physical and the socio-economic compartments of an integrated water and land use management model on a river basin scale using an objectoriented water supply model. Physics and Chemistry of the Earth 30 (6e7), 389e397. Barth, M., Hennicker, R., Kraus, A., Ludwig, M., 2004. DANUBIA: An integrative simulation system for global research in the upper Danube Basin. Cybernetics and Systems 35 (7e8), 639e666. Barthel, R., Trifkovic, A., 2007. Regional-scale assessment of groundwater resources quantity with respect to water supply issues and the ecological role of groundwater. Proceedings of the XXXV Congress of the International Association of Hydrogeologists (IAH), Lisbon, 2007. Bathurst, J.C., Ewen, J., Parkin, G., O’Connell, P.E., Cooper, J.D., 2004. Validation of catchment models for predicting land-use and climate change impacts. 3. Blind validation for internal and outlet responses. Journal of Hydrology 287, 74e94.


2794 2795 2796 2797 2798 2799 2800 2801 2802 Q4 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 Q1 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850

ARTICLE IN PRESS +


OF

European Commission, 2000. Directive 2000/60/EC of the European Parliament and of the Council establishing a framework for the Community action in the field of water policy. Ewen, J., Parkin, G., 1996. Validation of catchment models for predicting land-use and climate change impacts: 1. Method. Journal of Hydrology 175, 583e594. Ewen, J., O’Donnell, G., Burton, A., O’Connell, E., 2006. Errors and uncertainty in physically-based rainfall-runoff modelling of catchment change effects. Journal of Hydrology 330, 641e650. ¨ sterreich im Jahr 2003. Bundesministerium fu¨r Eybl, J., 2004. Trockenheit in O Land und Forstwirtschaft. Feuillette, S., Bousquet, F., Le Goulven, P., 2003. SINUSE: a multi-agent model to negotiate water demand management on a free access water table. Environmental Modelling & Software 18 (5), 413e427. Gaiser, T., Arau´jo, J.C., Frischkorn, H., Krol, M., 2003. Global Change and Regional Impacts: Water Availability and Vulnerability of Ecosystems and Society in Semi-Arid Northeast of Brazil. Springer, Heidelberg. Gaiser, T., Printz, A., Schwarz von Raumer, H.G., Go¨tzinger, J., Dukhovny, V.A., Barthel, R., Sorokin, A., Tuchin, A., Kiourtsidis, C., Ganoulis, I., Stahr, K., 2007. Development of a regional model for integrated management of water resources at the basin scale. Physics and Chemistry of the Earth, doi:10.1016/j.pce.2007.04.018 available online. Gilbert, N., Troitzsch, K.G., 2005. Simulation for the Social Scientist, second ed. Open University Press, Milton Keynes. Giupponi, C., Jakeman, A.J., Karssenberg, G., Hare, M.P. (Eds.), 2006. Sustainable Management of Water Resources: an Integrated Approach. Edward Elgar Publishing, Cheltenham, UK, p. 361. Grothmann, T., Patt, A., 2005. Adaptive capacity and human cognition: The process of individual adaptation to climate change. Global Environmental Change 15, 199e213. GWP (Global Water Partnership), 2000. Integrated Water Resources Management. GWP Technical Advisory Committee. TAC Background Papers No.4. GWP, Stockholm, Sweden, 67 pp. Hauhs, M., 1990. Ecosystem modelling: Science or technology? Journal of Hydrology 116 (1e4), 25e33. Hennicker, R., Ludwig, M., 2005. Property-driven development of a coordination model for distributed simulations, In: Formal Methods for Open Object-Based Distributed Systems (FMODS’05), 7th IFIP WG 6.1 International Conference, Athens, Greece. Lecture Notes in Computer Science, vol. 3535. Springer, Berlin, pp. 290e305. Hennicker, R., Ludwig, M., 2006. Design and implementation of a coordination model for distributed simulations. In: Mayr, H.C., Breu, R. (Eds.), Proc. Modellierung 2006 (MOD’06). Lecture Notes Informatics, vol. P82. Gesellschaft fu¨r Informatik, pp. 83e97. IPCC, 2001. Climate Change 2001: Synthesis Report. A Contribution of Working Groups I, II, and III to the Third Assessment Report of the Intergovernmental Panel on Climate Change (Ed. By Watson, R.T. and Core Writing Team). Tech. rep. Cambridge University Press, Cambridge, UK/New York, USA. IUCN, UNEP, WWF, 1980. World Conservation Strategy: Living Resource Conservation for Sustainable Development. International Union for the Conservation of Nature and Natural Resources, Gland, Switzerland. Jakeman, A.J., Letcher, R.A., 2003. Integrated assessment and modelling: features, principles and examples for catchment management. Environmental Modelling and Software 18, 491e501. Jakeman, A.J., Letcher, R.A., Norton, J.P., 2006. Ten iterative steps in development and evaluation of environmental models. Environmental Modelling and Software 21, 602e614. Kirshen, P., Raskin, P., Hansen, E., 1995. WEAP: A tool for sustainable water resources planning in the border region, in: Domenica, M. (Ed.), Integrated Water Resources Planning for the 21st Century, Proceedings of the 22nd Annual Conference of ASCE, American Society of Civil Engineers. June. Cambridge, MA, USA, pp. 7e11. Krysanova, V., Hattermann, F., Wechsung, F., 2007. Implications of complexity and uncertainty for integrated modelling and impact assessment in river basins. Environmental Modelling & Software 22 (5), 701e709.

CO RR

EC

TE

DP

BAYLFW, 1979. Die Trockenperiode 1976.dSchriftenreihe des Bayrischen Landesamtes fu¨r Wasserwirtschaft, Heft 12. BAYLFW, 1980. Das natu¨rliche Grundwasserdargebot in Bayern.dSchriftenreihe des Bayrischen Landesamtes fu¨r Wasserwirtschaft, Heft 13. BAYLFW, 1990. Bericht zur Grundwassererkundung in Bayern.dSchriftenreihe des Bayrischen Landesamtes fu¨r Wasserwirtschaft, Heft 23. BAYLFW, 1996. Die Grundwasserneubildung in Bayern berechnet aus den Niedrigwasserabflu¨ssen der oberirdischen Gewa¨sser.dInformationsberichte des Bayerischen Landesamtes fu¨r Wasserwirtschaft, Heft 5/96. BAYSTMLU, 1994. GrundwasserdMenge und Beschaffenheit des Grundwassers in Bayern. Schriftenreihe Wasserwirtschaft in Bayern Bayrisches Staatsministerium fu¨r Landesentwicklung und Umweltfragen. BAYSTMUGV, 2003. Trockenperiode 2003dAuswirkungen auf das Grundwasser. Bayerisches Staatsmininisterium fu¨r Umwelt. Gesundheit und Verbraucherschutz. Belz, J., Engel, H., Krahe, P., 2004. Das Niedrigwasser 2003 in Deutschlands Stromgebieten. Bundesanstalt fu¨r Gewa¨sserkunde. Berger, T., Birner, R., Diaz, J., McCarthy, N., Wittmer, H., 2007. Capturing the complexity of water uses and water users within a multi-agent framework. Water Resources Management 21 (1), 129e148. Biswas, A.K., 1996. Water Resources, Environmental Planning, Management and Development. McGraw-Hill, New York. Bogardi, J.J., Kundzewicz, Z.W., 2002. Risk, Reliability, Uncertainty and Robustness of Water Resource Systems. International Hydrology Series. Cambridge University Press, Cambridge, UK. Bolwidt, L.J. (Ed.), 2005. Selected Proceedings of the International Conference Climate Change: a Challenge or a Threat for Water Management? Held in Amsterdam, The Netherlands, 27e29 September 2004, ISBN: 184339491x., 149 pp. Bourdieu, P., 1984. Distinction: A Social Critique of the Judgement of Taste. Routledge, New York. Bouwer, H., 2002. Integrated water management for the 21st century: problems and solutions. Journal of Irrigation and Drainage Engineering 128 (4), 193e202. Brewer, G.D., 2007. Inventing the future: scenarios, imagination, mastery and control. Sustainability Science 2 (2), 159e177. Brugnach, M., Tagg, A., Keil, F., de Lange, W.J., 2007. Uncertainty Matters: Computer Models at the Science-Policy Interface. Water Resources Management 21, 1075e1090. BUWAL, 2004. Auswirkungen des Hitzesommers 2003 auf die Gewa¨sser. Bundesamt fu¨r Umwelt. Wald und Landschaft Schriftenreihe Umwelt Nr. 369. Croke, B.F.W., Jakeman, A.J., 2001. Predictions in catchment hydrology: an Australian perspective. Marine and Freshwater Research 52, 65e79. Davidsson, P., 2002. Agent Based Social Simulation: A Computer Science View. Journal of Artificial Societies and Social Simulation 5 (1). http:// jasss.soc.surrey.ac.uk/5/1/7.html. Do¨ll, P., Kaspar, F., Lehner, B., 2003. A global hydrological model for deriving water availability indicators: model tuning and validation. Journal of Hydrology 270 (1e2), 105e134. Dow, K., O’Connor, R.E., Yarnal, B., Carbone, G.J., Jocoy, C.L., 2007. Why worry? Community water system managers’ perceptions of climate vulnerability. Global Environmental Change 17, 228e237. Drogoul, A., Vanbergue, D., Meurisse, T., 2002. Multi-agent based simulation: where are the agents? In: Sichman, J.S., Bousquet, F., Davidsson, P. (Eds.), Multi-Agent-Based Simulation. Lecture Notes in Computer Science, vol. 2581. Springer, Berlin, pp. 1e15. D’Souza, D., Wills, A., 1999. Objects, Components and Frameworks with UML: The Catalysis Approach. Addison-Wesley, Reading, MA. Duboz, R., Amblard, F., Ramat, E., Deffuant, G., Preux, P., 2003. Individualbased model to enrich an aggregate model, in: Proceedings of Workshop Model-to-Model (M2M), Marseille, France, pp. 57e64. Ernst, A., Schulz, C., Schwarz, N., Janisch, S., 2005. Shallow and deep modeling of water use in a large, spatially explicit coupled simulation system, in: Representing Social Reality: Approaches and Results, Proceedings of 3rd Conference of the European Social Simulation Association (ESSA’05). Espinasse, B., Franchesquin, N., 2005. Multiagent modeling and simulation of hydraulic management of the Camargue. Simulation 81 (3), 201e221.

UN

2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907

MODEL

RO

26



2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964

ARTICLE IN PRESS +


MODEL


RO

OF

Ramanath, A.M., Gilbert, N., 2004. The design of participatory agent-based social simulations. Journal of Artificial Societies and Social Simulation 7 (4). http://jasss.soc.surrey.ac.uk/7/4/1.html. Rodgers, C., van de Giesen, N., Laube, W., Vlek, P.L.G., Youkhana, E., 2007. The GLOWA Volta Project: A framework for water resources decisionmaking and scientific capacity building in a transnational West African Basin. Water Resources Management 21, 295e313. Rumbaugh, J., Jacobson, I., Booch, G., 2005. The Unified Modeling Language Reference Manual, second ed. Pearson Education. Russell, S., Norvig, P., 2003. Artificial Intelligence: A Modern Approach, second ed. Prentice-Hall, Englewood Cliffs, NJ. Sax, M., in press. Water and tourismdsimulating the tourist water use in the upper Danube catchment area, in: Schumann, A., Pahlow, M., Bogardi, J.J., van der Zaag, P. (Eds), IWRM 2006, IAHS Red Book, Publ. 317. Schwarz, N., Ernst, A., submitted for publication. Agent-based Modeling of the diffusion of environmental innovationsdan empirical approach. Technological Forecasting and Social Change. Scoccimarro, M., Walkera, A., Dietricha, C., Schreidera, S., Jakeman, T., Rossb, H., 1999. A framework for integrated catchment assessment in northern Thailand. Environmental Modelling and Software 14 (6), 567e577. Sociovision, 2007. Website of Sinus Sociovision. last visited 2007/10/16. http://www.sociovision.com. Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L. (Eds.), 2007. Climate Change 2007: The Physical Science BasisContribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK/New York, NY, USA. Thoyer, S., Morardet, S., Rio, P., Simon, L., Goodhue, R., Rausser, G., 2001. A bargaining model to simulate negotiations between water users. Journal of Artificial Societies and Social Simulation 4 (2). http://jasss.soc.surrey.ac. uk/4/2/6.html. UN, 1992. Report of the United Nations Conference on Environment and Development, United Nations, 3e14. June, Rio De Janeiro, Brazil. UNESCO, 1987. Methodological Guidelines for the Integrated Environmental Evaluation of Water Resources Development, United Nations Environment Programme, Paris, France. Victorian Water, 2005. Victorian uniform drought water restriction guidelines (Final) September 2005. http://www.vicwater.org.au/uploads/Water%20 Restrictions/Final%20Version%20Uniform%20Drought%20Water%20Res triction%20Guidelines.pdf. Watkins, D., Kirby, K., Punnett, R., 2004. Water for the everglades: Application of the south Florida systems analysis model. Journal of Water Resources Planning and Management 130 (5), 359e366. Weiss, G. (Ed.), 1999. Multiagent Systems: A Modern Approach to Distributed Artificial Intelligence. MIT Press, Cambridge, MA, USA. Wilson, E.M., 1999. Preparing for the 21st Century, 26th Annual Water Resources Planning and Management Conference of ASCE, 6e9 June, Tempe, AZ, USA. Wu, C., 1995. Integrated water resources planning of quantity and quality in Taiwan, in: Domenica M. (Ed.), Integrated Water Resources Planning for the 21st Century, Proceedings of the 22nd Annual Conference of the American Society of Civil Engineers Cambridge, MA, USA, pp. 65e68. Yamout, G., El-Fadel, M., 2005. An optimization approach for multi-sectoral water supply management in the greater Beirut area. Water Resources Management 19, 791e812. Yurdusev, M.A., 2002. Environmental Impacts for Water Resources Planning. WIT Press, Southampton, UK.

CO RR

EC

TE

DP

Letcher, R.A., Croke, B.F., Jakeman, A.J., 2004. Model development for integrated assessment of water allocation options. Water Resources Research 40 (5), doi:10.1029/2003WR002933. Letcher, R.A., Croke, B.F.W., Jakeman, A.J., 2007. Integrated assessment modelling for water resource allocation and management: A generalised conceptual framework. Environmental Modelling & Software 22 (5), 733e742. Lo´pez-Paredes, A., Sauri, D., Gala´n, J., 2005. Urban Water Management with Artificial Societies of Agents: The FIRMABAR Simulator. Simulation 81 (3), 189e199. Loucks, D.P., Stedinger, J.R., Haith, D.A., 1981. Water Resource Systems Planning and Analysis. Prentice-Hall Inc., Englewood Cliffs, NJ. LUBW, 2004. Das Niedrigwasserjahr 2003. Landesanstalt fu¨r Umweltschutz Baden-Wu¨rttemberg. Ludwig, R., Mauser, W., Niemeyer, S., Colgan, A., Stolz, R., EscherVetter, H., Kuhn, M., Reichstein, M., Tenhunen, J., Kraus, A., Ludwig, M., Barth, M., Hennicker, R., 2003. Web-based modeling of water, energy and matter fluxes to support decision making in mesoscale catchmentsdthe integrative perspective of GLOWA-Danube. Physics and Chemistry of the Earth 28, 621e634. Luke, S., Cioffi-Revilla, C., Panait, L. and Sullivan K., 2004. ‘MASON: A New Multi-Agent Simulation Toolkit’. Proceedings of the 2004 SwarmFest Workshop. Macy, M.W., Willer, R., 2002. From factors to actors: computational sociology and agent-based modeling. Annual Review of Sociology 28, 143e166. Mauser, W., Strasser, U., 2005. Statusreport GLOWA-Danube: integrative techniques, Scenarios and strategies regarding global change of the water cycle. Project duration: 01.03.2004e28.02.2007, Report period: 01.03.2004e28.02.2005. Tech. rep., GLOWA-Danube. URL http://www. glowa-danube.de/PDF/reports. Minar, N., Burkhart, R., Langton, C., Askenazi, M., 1996. The Swarm simulation system: a toolkit for building multi-agent simulations. Working Paper 96-06-042. Santa Fe Institute, Santa Fe. Moss, S., Downing, T., Rouchier, J., 2000. Demonstrating the role of stakeholder participation: an agent based social simulation model of water demand policy and response. URL. http://cfpm.org/wscott/water-demand/ demand-pilot1.pdf. Nickel, D., Barthel, R., Schmid, C., Braun, J., 2005. Large-Scale water resources management within the framework of GLOWA-DanubedThe water supply model. Physics and Chemistry of the Earth 30 (6e7), 383e388. North, M.J., Collier, N.T., Vos, J.R., 2006. Experiences creating three implementations of the repast agent modeling toolkit. ACM Transactions on Modeling and Computer Simulation 16 (1), 1e25. Pahl-Wostl, C., 2002. Towards sustainability in the water sectordthe importance of human actors and processes of social learning. Aquatic Sciences 64, 394e411. Pahl-Wostl, C., 2005. Actor based analysis and modelling approaches. Integrated Assessment 5, 97e118. Pahl-Wostl, C., 2007a. Transition towards adaptive management of water facing climate and global change. Water Resources Management 21 (1), 49e62. Pahl-Wostl, C., 2007b. The implications of complexity for integrated resources management. Environmental Modelling and Software 22, 561e569. Parunak, H.V.D., Savit, R., Riolo, R.L., 1998. agent-based modeling vs. equation-based modeling: a case study and users’ guide. In: Sichman, J.S., Conte, R., Gilbert, N. (Eds.), Proceedings of Multi-Agent Systems and Agent-Based Simulation. Lecture Notes in Computer Science, vol. 1534. Springer, Berlin, pp. 10e25. Quinn, P.F., Hewett, C.J.M., Doyle, A., 2004. Integrated Water Management: A Multi-Scale Framework for Decision Support. In Improving the Balance Between Economic Agricultural Production and Environmental Quality through Enhanced Decision Making. OECD Publication.

UN

2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024

27


3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 Q23039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084

An integrated modelling framework for simulating regional-scale actor ...

An integrated modelling framework for simulating regional-scale actor ...

Suggest Documents

seamframe, a proposal for an integrated modelling framework for

AN ACTOR-BASED FRAMEWORK FOR NEGOTIATING MOBILE ...

AN ACTOR-BASED FRAMEWORK FOR NEGOTIATING ... - VSIS

Simulating an Integrated Revenue Management

A Framework for Modelling and Simulating Networks of Cells

An integrated modelling framework to support manufacturing system

Towards an integrated systems-based modelling framework ... - Spiral

An integrated modelling framework to support ... - Semantic Scholar

an integrated approach for modelling business ...

An Actor-based Framework for Negotiating Mobile Agents - VSIS

An Actor Based Framework for Managing Multimedia ...

An Actor-Based Framework for Heterogeneous ... - Semantic Scholar

An activity-based modelling framework for quantifying

OpenStructure: an integrated software framework for ... - BioMedSearch

An Integrated Framework for Biometrics Security

An Open Framework for Integrated Qualification ...

An Integrated Framework for Operational ... - Semantic Scholar

An Integrated Framework for Sensor-Based ...

An Integrated Framework for Feature Extraction

An Integrated Framework for RFID Adoption and

Modelling Strategic Actor Relationships for Business Process

An Integrated Obstacle Detection Framework for ...

An Integrated Scientific Experiment Framework for ... - ThinkMind

SICOSYS: An integrated framework for studing interconnection ...