New simulation technologies applied for creating hydrogeological model of Noginsk District,

Series - Computer Science Boundary Field Problems and Computer Simulation - 42nd thematic issue________________________________________________________________________________________________________2000

New simulation technologies applied for creating hydrogeological modelof Noginsk District, Russia

A. Spalvins, J. Slangens, R. Janbickis & I. LaceE. Gosk

Environment Modelling Centre, Riga Technical University, LatviaGeological Survey of Denmark and Greenland, Denmark

L. Loukiantchikova

All-Russian Research Institute for Hydrogeology and Engineering Geology, Russia

Keywords: hydrogeological models, geological data interpolation

ABSTRACT: Since 1993, the team of the Environment Modelling Centre (EMC) of the Riga Technical Uni-versity is being active in the field of developing methodologies and special software for creating hydro-geological models (HM). In this paper, these tools applied together with commercial ones are described, onthe example of complicated HM created for the Noginsk District, Russia.1 INTRODUCTION

This paper shows applications of special and commercial tools used recently by the EMC team for creatingNoginsk HM. Practical knowledge combined with scientific approach suggested useful tools (see Table 1)and methodologies reducing HM errors (Spalvins et al., 1999a, Spalvins et al., 1999b, Spalvins et al., 2000a,Spalvins et al., 2000b). Processes of creating HM and results of modelling are shown by the scheme of Fig.1taken from (Loukiantchikova et al., 2000).

To consider main problems of creating HM, some mathematics is needful. The xyz-grid of HM is built of(h*h*hz)-sized blocks (h is the block plane size; hz is a variable block height). They constitute a rectangular(2s + 1) – tiered xy-layer system where s + 1 and s are, accordingly, the number of aquifers and interjacentaquitards (for Noginsk HM, s = 6, see Fig.1). Its four vertical sides compose the shell of the HM grid. Theground surface (relief) and the lower side of the model are its geometrical top and bottom, respectively.

The vector ϕ of the piezometric head is the numerical solution of a boundary field problem approximated,in nodes of the HM grid, by the following algebraic equation system:

A ϕ = β – G ψ, A = Axy + Az – G

(1)

where the matrices Axy, Az, and G represent, correspondingly, horizontal links axy of aquifers (arranged inxy-planes), vertical ties az originated by aquitards, and elements connecting “free” nodes of the grid with theones where the piezometric boundary conditions ψ are specified; in Noginsk HM, the ψ -distribution existson the whole HM surface: top + bottom + shell, and the source vector β contains only rates of the ground-water withdrawal via wells. The elements axy, az (or gxy, gz of G) are computed, as follows:

axy = k m, az = h2 k / m, k ≥ 0, mi = hzi = zi-1 – zi ≥ 0, i = 1, 2,…, 2s + 1

(2)

where zi-1 and zi are the elevation distributions of the top and bottom surfaces of the i-th geological layer; z0represents the ground surface map with the hydrograpical network (lakes, rivers, etc.) included; m = hz, k are,accordingly, elements of the digital m, k-maps of computed thickness and permeability distributions of lay-ers. The set of the zi –maps states the full geometry of HM. If, within some area, hzi = 0 then the i-th layer isdiscontinuous. Noginsk HM includes 7 discontinuous layers, and this feature makes creating of this modelextremely difficult (Gosk et al., 1999).

SCIENTIFIC PROCEEDINGS OF RIGA TECHNICAL UNIVERSITY

Series - Computer Science Boundary Field Problems and Computer Simulation - 42nd thematic issue________________________________________________________________________________________________________2000

2 GENERAL SCHEME. SOFTWARE APPLIED

According to Fig.1, the process of creating Noginsk HM conditionally contains the following stages:- scanning hard copies of necessary maps , in order to convert them into electronic images;

- digitizing electronic images of maps supporting geological z-surfaces of (2); obtaining the full set of

these surfaces; this is the most difficult and time-consuming part of the HM formulation process;

- building k-maps, β and ψ -distributions; applying them for HM calibration purposes; the modelling

system REMO (Spalvins et al., 1996) is applied to accomplish this final stage, when all calibratedHM results become available for further utilization;

- the contaminant mass transport is modelled by the Groundwater Vistas (GV) system (Environmental

simulations, 1997);. the HM data for GV are prepared by REMO.

All the above stages are interdependent, and they must be repeated many times, to meet HM calibrationtargets, which are multifunctional (Spalvins et al., 2000a).

Commercial and special (developed by EMC) software tools were applied, to create Noginsk HM. Thefollowing two commercial programs were used:

- the GV system, to investigate the contaminant mass transport in groundwater;

- the SURFER program (Golden Software, 1997) for supporting graphics, digitizing of electronic im-ages, and performing grid mathematics.

It was necessary to develop special tools, listed in Table 1, because the available commercial ones (forexample, MODFLOW (McDonald & Harbaugh, 1988) and interpolation modules of SURFER) turned out tobe helpless, when faced with practical cases insisting on investigation of complex hydrogeological systems(Atruskievics et al., 1994). Nowadays, the situation has not improved much, hence even the modern model-ling programs (like GV) are mostly still based on numerous outdated principles – survivals from the era ofslow computers (the zone scheme used for saving memory (Spalvins & Janbickis, 2000); β -conditions ap-plied, as infiltration flows on the HM top (Spalvins, 2000); unnecessarily strong human involvement (Spal-vins et al., 1999a), etc).

Table 1. Programs and methods developed by the EMC team for creating HM.

Nr12345678910111213

Name of program or method, its activity, (reference)Program REMO, modelling and calibration of (1), (Spalvins et al., 1996)Program GDI, creating interpolated data surfaces for (1), (Spalvins &Slangens, 1994)

Program CRP, preparing line data and joining them with the HM grid,(Slangens & Spalvins, 2000)

Program CRS, preparing line data of geological sections, (Lace & Spalvins,2000)

ProgramINTβϕ, interpolates β and ϕvalues from irregularly located wells tonodes of the HM grid and backwards, (Lace et al., 1995)

ProgramINTβ, interpolates β values from irregularly located wells to nodes ofthe HM grid, (Gosk et al., 1999)

Program GVR, creates images of vertical cross sections of HM results alongany line chosen, (Spalvins et al., 1996)

Program NFCG, solves the system (1), (Janbicka et al., 1993)

Method of using the ground surface elevation map, as the ψ–distribution, onthe HM top, (Spalvins, 2000)

Method of using the HM shell, as an interpolation device of the ψ –distribu-tion, (Spalvins, et al., 2000)

Method of tuning zone converter modules, (Spalvins & Janbickis, 2000)Method of avoiding unconfined aquifers in HM, (Spalvins, 2000)

Method of masking HM results located within areas (hz=0) of non-existentlayers, (Gosk, et al., 2000)

Related programsGV, SURFERREMO, GV,SURFERGDICRPREMOGVREMOREMOREMO, GVREMO, GVGV

REMO, GV

REMO, SURFER

SCIENTIFIC PROCEEDINGS OF RIGA TECHNICAL UNIVERSITY

Series - Computer Science Boundary Field Problems and Computer Simulation - 42nd thematic issue________________________________________________________________________________________________________2000

These special tools are being used mainly for preparing various input maps and distributions of boundaryconditions β, ψ needed to create HM of (1). In Table 1, the tools are conditionally sorted into programs andmethods. The method represents a serviceable algorithm, which can be utilized by any affined program, eventhe commercial one. The column “Related programs” gives the code of connected program- the user of theresults provided by the program (or method) considered. For example, the REMO results may be exported tothe GV program, or graphed by SURFER.

It follows from Table 1, that all tools, but Nr 5 and Nr 11, provide data for, or support the REMO pro-gram. REMO is a tested novel tool, it is able to accomplish tasks of utmost complexity.

Due to resourceful methods of mathematical physics embodied, the GDI program can provide compli-cated interpolated surfaces (including the interlinked z-surfaces of discontinuous layers) by applying minimalsets of input data (Spalvins et al., 2000).

The remarkable GDI-results are also due to extensive use of various line data (isolines, geologicalboundaries and profiles of sections, long lines of rivers, etc), as the initial ones for GDI. The CRP programcomputes them, at intersections (crospoints) of lines (carriers) with the HM grid. The novel version of CRPinsures high quality both of coordinates and values of data, at the crospoints (Slangens & Spalvins, 2000).The INTβϕ program considerably improves accuracy of REMO results, as follows:

- boundary β-values of irregularly located wells are interpolated to four nearest nodes of the HM grid;

(GV roughly moves the β-value to one nearest node!);

- to increase credibility of HM calibration, ϕ-values are computed, at randomly sited observation wells.The INTβ program considerably improves the β-distribution for the GV code by applying the REMOinterpolated β–values, as the formal production wells at nodes (Gosk et al., 1999).

It was necessary to develop the CRV program, because the GV system could provide sections for HM re-sults only along the HM grid lines. Moreover, these sections are misleading, even erroneous (Spalvins et al.,2000a).

As the REMO solver, the NFCG program has never failed. It is a robust, reliable tool. No adjustments ofits regimes are required from a modeller.

The methods Nr 9 and Nr10 of Table 1 present a breakthrough in modelling of hydrogeological processes.Due to them, reliability of HM results has raised drastically.

To prevent damage of data files imported into the GV system, the method of tuning its zone converterwas developed.

The method Nr 12 blocks inevitable ruin of (1), during its solution process by the GV system, at caseswhen HM includes discontinuous layers. In REMO, no regimes of unconfined aquifers are allowed.

In GV, no masking of HM results, located within areas of non-existent layers, is possible. The methodNr 13 solves this problem for REMO and SURFER.

It follows from the above text that, in order to create Noginsk HM, the almost full set of tools developedby EMC has been applied. Even some new ones have been made and used, to accomplish this very difficulttask.

3 CONCLUSIONS

To obtain reliable HM results for complex practical cases, novel tools have been developed and used by theEMC team. They support the following general ideas:

- model formulation, solving, and post-processing errors can be subdued if special software and meth-odology are focused on this purpose;

- orientation on teamwork, involving specialists from overlapping science fields, when complex HM is

to be created;

- withholding involvement of people in routine processes which can be run much safer by properly de-signed software;

- applying different, highly effective special tools for creating complex HM, because no single system

can cope with such a task.

We hope that our knowledge reported will be of interest for modellers creating complex hydrogeologicalmodels.

SCIENTIFIC PROCEEDINGS OF RIGA TECHNICAL UNIVERSITY

Series - Computer Science Boundary Field Problems and Computer Simulation - 42nd thematic issue________________________________________________________________________________________________________2000

REFERENCES

Spalvins, A., Slangens, J., Janbickis, R. & Lace, I. 1999a. Reducing of Model Formulation Errors as an Ef-fective Remedy for Improving Simulation Results. Proc. of International Conference on \"Calibration andReliability in Groundwater Modelling, ModelCARE'99\1999. Zurich::161-166.

Spalvins, A., Slangens, J., Janbickis, R., Lace, I. & Macans A. 1999b.Rational Preprocessing of Initial DataEnables to Create Credible Hydrogeological Models. Environment Simulation. Riga: 57-67. (BoundaryField Problems and Computer Simulations, 41st issue).

Spalvins, A., Slangens, J., Janbickis, R. Lace, I. & Gosk, E. 2000. Methods for Improving Verity ofGroundwater Modelling. Proc. of 16 th IMACS World Congress 2000, Lausanne, Switzerland, 21-25August. 6 pages on CDROM, ISBN 3-9522075-1-9.

Spalvins, A., Lace, I., Slangens, J., Janbickis, R. 2000. Improving Verity of Hydrogeological Models due toHeuristic Human Skills Applied within Man-Computer Systems. Proc. of the International conference on“Simulation, Gaming, Training and Business Process Reengineering in Operations”, 8-9 September2000. Riga: 266-270.

Loukiantchikova, L., Gosk, E., Spalvins, A., Janbickis, R. & Lace, I. 2000. Development of the Hydro-geological Model for Investigating the Impact of Contaminant Sources in the Noginsk District, Russia.Proc. of International Conference on \" Groundwater Research - Groundwater'2000\ Copenhagen, Den-mark, 6-8 June 2000. Rotterdam: Balkema, 109-110.

Gosk, E., Spalvins, A.& Vartanyan, G. (eds.). 1999. Hydrogeological and contamination transport models forNoginsk district, Moscow region. Report on sub-contractor agreement within the project: Groundwatercontamination and remediation in Noginsk district, Moscow region. Riga - Moscow.

Spalvins, A., Janbickis, R. Slangens, J., Gosk, E., Lace, I., Viksne, Z., Atruskievics, J., Levina, N. & Tol-stovs, I. 1996. Hydrogeological Model “Large Riga”. Atlas of maps. Riga – Copenhagen: (BoundaryField Problems and Computers, 37th issue).

Environmental Simulations, Inc. 1997. Groundwater Vistas. Guide to using.Golden Software, Inc. 1997. Surfer 6.0 for Windows. User's manual.

Spalvins, A. & Slangens, J. 1994. Numerical interpolation of geological environment data. Proc. of Latvian -Danish Seminar on “Groundwater and Geothermal Energy” (2). Riga – Copenhagen: 181-196. (Bound-ary Field Problems and Computers, 35th issue).

Slangens, J. & Spalvins, A. 2000. Reliable program for preparing line data of hydrogeological models (inthis book).

Lace, I. & Spalvins, A. 2000. Incorporating geological sections in hydrogeological models (in this book).Lace, I., Spalvins, A.& Slangens, J. 1995. Algorithms for accounting groundwater discharge in the regionalhydrogeological model and interpolation of simulation results at observation wells. Proc. of InternationalSeminar on \"Environment Modelling\" (1). Riga – Copenhagen: 201-216.(Boundary Field Problems andComputers, 36th issue).

Janbicka, A., Spalvins, A.& Slangens, J. 1993. Implementation of the Nested Factorization Conjugate Gradi-ent Method on a Spatial Grid. Numerical simulation for hygrogeology. Riga: 35-41.(Boundary FieldProblems and Computers, 34th issue).

Spalvins, A. 2000. Landscape elevation map as reliable boundary conditions for hydrogeological models (inthis book).

Spalvins, A., Janbickis, R. & Slangens, J. 2000c. Boundary shells of hydrogeological models as interpolationdevices (in this book).

Spalvins, A. & Janbickis, R. 2000. Misfortunes of zone scheme applied for storing hydrogeological data (inthis book).

Atruskievics, J., Janbickis, R., Krutofal, T., Lace, I., Levina, N., Slangens, J., Spalvins, A. & Viksne, Z.1994. Computer Based Regional Hydrogeological Model \"Large Riga\". Proc. of Latvian - Danish Semi-nar on “Groundwater and Geothermal Energy” (2). Riga – Copenhagen: 203-224. (Boundary FieldProblems and Computers, 35th issue).

McDonald, M. & Harbaugh, A. 1988. A Modular Three-Dimensional Finite-Difference Ground-Water FlowModel. - U. S. Geological Survey. Open File Report Washington: 83-875.

SCIENTIFIC PROCEEDINGS OF RIGA TECHNICAL UNIVERSITY

Series - Computer Science Boundary Field Problems and Computer Simulation - 42nd thematic issue________________________________________________________________________________________________________2000

Aivars Spalvins, Dr.sc.ing.Janis Slangens, Dr.sc.ing.Romans Janbickis, M.sc.ing.Inta Lace, M.sc.ing.

Riga Technical University, Environment Modelling CentreAddress: 1/4 Meza Str., Riga, LV-1048, LatviaPhone: +371 7089518; E-mail: emc@egle.cs.rtu.lvEdmnd Gosk, Project Manager,

Geological Survey of Denmark and Greenland

Address: 8 Thoravej, 2400 Copenhagen NV, DenmarkPhone: +45 38142000; E-mail: eg@geus.dk

Ludmila Loukiantchikova, Dr., Hydrogeologist

All-Russian Research Institute for Hydrogeology and Engineering GeologyAddress: 142452 Moscow Region, Noginsk District, Zeleny -Village, RussiaPhone: +7 095 5212000; E-mail: gvartany@online.ru

Spalviņš A., Šlangens J., Janbickis R., Lace I., Gosk E., Loukiantchikova L. Jaunas modelēšanas tehnoloģijas,kuras izmantotas Noginskas rajona hidroģeoloģiskā modeļa būvei Krievijā.

Kopš 1993. gada Rīgas Tehniskās universitātes Vides Modelēšanas centrs ir aktīvi izstrādājis metodikas un speciālasprogrammatūras hidroģeoloģisko modeļu izveidošanai. Šajā rakstā ir parādīts, kā šie speciālie līdzekļi un komercpro-grammatūras tika izmantoti komplicēta HM izveidošanai Noginskas rajonam Krievijā.

Спалвинь А., Янбицкий Р., Шланген Я., Лаце И. Госк Э., Лукянчикова Л. Новые технологии дляпостроения гидрогеологической модели для Ногинской области, Россия.

Начиная с 1993. года Центр Моделирования Среды Рижского Технического Университета активноразрабатывал методики и специальные программные средства для построения гидрогеологических моделей(ГМ). В этой статье показано как эти специальные и коммерческие средства используются совместно дляпостроения сложной ГМ Ногинского района, Россия.