This research project dedicated to emerging techniques centred on the development of Spatial Decision Support Systems (SDSS) and application of state of the art models and model integration tools.
|Title: Geo-Spatial Modelling Informing Policy
Sponsor : EC Framework (FP7)
Call: People – Marie Curie Actions: International Outgoing Fellowships for Career Development (FP7-PEOPLE-2013-IOF)
Outgoing Host: National Center for Smart Growth Research, University of Maryland
EU Host: School of Architecture, Planning and Environmental Policy, University College Dublin
Scientific Advisors: Prof. Brendan Williams (UCD) and Prof. Gerrit Knaap (UMD)
Start / End Dates : 01-AUG-14 / 31-JUL-17
In scope of this fellowship Dr. Harutyun Shahumyan has moved from University College Dublin (UCD) in Ireland to University of Maryland (UMD) in US for two years and then has returned and build on his research at UCD for another year, for the purposes of enhancing his training in integrated modelling supporting smart and sustainable growth. The research training is achieved through GeoSInPo project dedicated to emerging techniques centred on the development of Spatial Decision Support Systems (SDSS) and application of state-of-the-art geospatial analytical tools, which are uniquely available within the scientific resource bases of the host institutions. This project is delivered through practical application of integrated modelling approaches in two study regions (1) Baltimore-Washington Region (BWR) in US during outgoing phase (2014-2016) and (2) Greater Dublin Region (GDR) in Ireland during return phase (2016-2017).
This research is supported by a Marie Skłodowska-Curie International Outgoing Fellowship within the 7th European Community Framework Program.
Policy makers are facing challenges of managing complex urban and environmental systems influenced by global factors. These factors include population growth, migration, recession, climate change as well as actions by local actors such as parties or companies who direct the development according to their own vested interest which may not conform with a broader public interest. One of the key challenges for a sustainable future is to manage population growth, with its resulting issues of transportation and food production and relevant land use changes with minimal adverse impacts on environment. Rapid urbanization, drastic increase of number of vehicles and growing agriculture can substantially impact on air and water quality in a region. Agricultural activities may degrade water quality through excessive soil loss or export of fertilizers and pesticides. Whereas, urbanization increases the impervious area resulting in higher flows and more contaminants that have a negative effect on stream water quality; while, growth in number of vehicles increases emissions negatively effecting on air quality in the region.
Confronted with such complexity, decision makers need adequate tools to better understand and evaluate the effects of policy interventions in urban regions. Such pressure already led to the development of numerous mathematical and geospatial models covering various discipline-specific areas. However, the interconnected character of human and natural systems requires an integrated decision making and therefore – integrated modeling linking different disciplines.
Developing a new state-of-the-art comprehensive model integrating various disciplines is an expensive process and can take years. Instead, coupling already existing and tested models for a region may offer an effective and a quick solution to support integrated decision making. This project develops a model coupling approach and demonstrates it through practical applications in two study regions: the Baltimore-Washington Region in USA and the Greater Dublin Region in Ireland.
The developed model coupling approach has been applied to couple various models, including transportation, land use, land cover, emissions and water quality models. The task was complicated since in spite of covering the same region all those models have been developed independently without any built-in method for linking to other models. Moreover, they have been developed in different programming languages, software environments and have various licensing restrictions, making their integration a challenging task.
The project explored diverse model coupling approaches and evaluated their applicability for the research needs. However, the existing couplers have been developed with different objectives and constraints in mind. None of the identified couplers were suitable for the defined integration task covering. Therefore, a loose model coupling approach was developed using specific model wrappers, which separate the implementation of the couplers from the models’ source codes. This gives a flexibility, which can help in terms of portability, performance and maintenance of the model codes. The approach is especially efficient when the models are developed in different programming languages, their source codes are not available or the licensing restrictions or limited resources are making other coupling approaches infeasible. It offers a viable solution to coupling such models without the need of change or even access to the sources codes.
The loose coupling approach was successfully applied for the Baltimore-Washington Region coupling five independently developed models: Simple Integrated Land Use Orchestrator (SILO), Maryland Statewide Transport Model (MSTM), Building Emission Model (BEM), Mobile Emission Model (MEM), Chesapeake Bay Land Change Model (CBLCM). The integrated suite is now being applied at the National Center for Smart Growth at the University of Maryland in collaboration with the USGS Eastern Geographic Science Center to simulate and explore alternative scenarios of the region for 2040.
The outputs from the integrated modelling suite include several useful socio-economic indicators, covering: population and employment, transport flow, land use, building and mobile emissions, and more. It is known that the changes in transportation, land use and human behaviour in general impact also on nutrient loading and water quality in a region. Translating the effect of socio-economic alterations into nutrient loading in Chesapeake Bay for example will help us to explore the changes in flow and nutrients loads into the Bay and design more effective public policies and restoration plans. Adding environmental models to the policy decision making process will help to assess how social-economic changes and policy decisions in the Baltimore-Washington Region ultimately impact water quality in the Chesapeake Bay improving policy decision making. Aiming to support improved policy analysis and decision making, the following environmental models were also explored for further enhancement of the integrated suite: Integrated Transport and Health Impact Modelling Tool (ITHIM), Hydrological Simulation Program – Fortran (HSPF) and Chesapeake Bay ROMS Community Model (ChesROMS).
Greater Dublin Region
In the meantime, for the Greater Dublin Region, the approach was applied to couple land use change model MOLAND with the Source Load Apportionment Model SLAM to estimate annual nutrient losses in case of different regional development scenarios.
The results of this two case studies appear very promising. The independently developed models smoothly link in a form which does not require the user to track the models while running applications of these models. Instead, the user can more easily focus on the analysis and results, avoiding the need for a detailed understanding of model structure and data file systems. The approach can be easily applied to other models for other regions.
Maryland Statewide Transportation Model (MSTM)
The MSTM is an advanced trip-based model developed in 2008 by the Maryland State High-way Administration in conjunction with the Parsons Brinkerhoff. It was de-signed to estimate the impacts of transportation investments, changes to land use develop-ment, and impacts from factors beyond state boundaries, particularly freight.
The model Input data include population and employment by model zones, highway and public transit networks, and data on travel behavior. The model outputs report traffic impacts on the overall system, corridors or individual links.
Simple Integrated Land Use Orchestrator (SILO)
Initially developed as a research project by Parsons Brinckerhoff for Minneapolis/St. Paul, SILO has been implemented for the state of Maryland. It micro-simulates household relocation, demographic changes and developers who add, upgrade or demolish dwellings. SILO is designed as a discrete choice model. Thus, every household, person and dwelling is treated as an individual object. Spatial decisions, such as relocation, development of new dwellings, etc., are modeled with Logit models. Other decisions, such as getting married, giving birth to a child, etc., are modeled by Markov models that apply transition probabilities.
SILO uses the Public Use Microdata Sample (PUMS) to create individual households and their dwellings. The MSTM provides the zone-to-zone travel time by auto and public transit. SILO generates a synthetic population with households, persons, dwellings and jobs for the base year 2000 and incrementally updates these dataset in one-year increments through 2040. Every year the MSTM runs, SILO provides updated socio-demographic data.
Mobile Emissions Model (MEM)
The MEM estimates transportation emissions by applying emission of the MOVES 2010 EPA model to MSTM-generated traffic flows (Welch, 2013). The MEM input data includes road network, vehicle trips, temperatures by month and hour for each county in the study area, humidity, average speed distribution, the vehicle miles of travel (VMT) on varied road types, fuel formulation and supply. MEM runs every time the MSTM has run.
Building Energy Consumption & Emissions Model (BEM)
The BEM estimates CO2 emissions and energy consumption from the built environment with-in Maryland. It uses the building, location and climate variables of each proper-ty to determine whether the structure is likely to combust fossil fuels on site. If the probability is greater than 50%, then the model calculates CO2 emissions from local combustion based on a set of related multipliers derived from a regression of the data from US Energy Information Administration’s residential (RECS) and commercial building (CBECS) energy consump-tion surveys. SILO provides the building stock for BEM. CO2 emissions and energy consumption are the primary outputs of the model.
Chesapeake Bay Land Change Model (CBLCM)
The CBLCM was developed by the USGS within the Chesapeake Bay Program. It uses a sto-chastic methodology to emulate residential urban land use development in Maryland over a se-ries of pre-defined time segments. It is as an independent cellular automata model that trans-lates exogenous county-level projections of population and employment to estimates of urban land demand and then spatially allocates that onto 30m-resolution raster cells. The locations of future growth are informed by data on protected lands, zoning, slopes, land cover, proximity to urban centers, and proximity to locations of recent job and housing growth. The model calculates a probability surface for growth locations, and allocates households and dwellings provided by SILO. CBLCM generates fine-grained patterns of residential urban growth across the study area.
Greater Dublin Region
The MOLAND land use model has been developed by the Research Institute for Knowledge Systems as part of an initiative of the EC Joint Research Centre. It generates alternative future scenarios informing urban planners and policy makers on the possible implications of their decisions in terms of land use change.
The MOLAND model is based on the GEONAMICA framework; and stands out from other urban models because of its capability of simulating 32 land uses, which adopt the CORINE land cover classification system. It comprises two dynamic sub-models with a common temporal increment of one year but working at different scales. At the macro scale, the model allocates regional population and jobs among the sub-regions. At micro scale, the provision for population and jobs is translated into demand for various land types using constrained cellular automaton (CA), which for the GDR consists of (i) a land use raster grid with 200m cell size and 23 classes, (ii) a set of factors influencing the direction of land use change such as suitability, zoning, accessibility, and neighbouring land uses, and (iii) transition rules determining the attraction and repulsion between land uses. The model also includes a stochastic parameter which insures the generation of realistic land use patterns.
The Source Load Apportionment Model (SLAM)
There are two broad approaches to load apportionment modelling, (i) load-orientated approaches which apportion origin based on measured in-stream loads, and (ii) source-orientated approaches where amounts of diffuse emissions are calculated using models typically based on export coefficients from catchments with similar characteristics. The SLAM framework developed at UCD Doogle Centre for Water Resources Research takes the latter approach which enables estimates of the relative contribution of sources of nitrogen (N) and phosphorus (P) to surface waters in catchments without in-stream monitoring data. In contrast to models such as SWAT, this approach allows the model to be applied throughout Ireland, independently of the availability of measured in-stream calibration data. It uses the best available Irish data and models to quantify annual nutrient losses from both point discharges from urban wastewater, industry and septic tank systems, and diffuse sources including various land uses.
The SLAM framework incorporates multiple national spatial datasets relating to nutrient emissions to surface water, including land use and physical characteristics of the sub-catchments. The agriculture (pasture & arable) and septic tank systems modules use spatial outputs from the Catchment Characterisation Tool (CCT) and SANICOSE models, respectively. The diffuse nutrient emissions from forestry, peatlands and urban areas were modelled based on export coefficients from the land cover classes, calculated as: Diffuse=Area × Export, where Area is the area of the land cover category (ha) and ?????? is the export coefficient for nitrogen or phosphorus (kg ha-1 yr-1).
The multi-year goal for this project implemented at the NCSG, is to develop, disseminate, and promote the implementation of a sustainable development strategy for the Baltimore-Washington region. The following areas of sustainability are explored and modeled using loosely coupled modeling system, first for one or more baseline scenarios, then for alternative more sustainable scenarios.:
- Economic Productivity
- Land Use
- Access to opportunity
- Air emissions, including greenhouse gases
- Nutrient loading into to water bodies
- Energy Consumption
The UEP project aims to better understand the link between development, land-use change and associated economic and environmental impacts within urban regions. The group is based at UCD (formerly UCD Urban Institute Ireland) and originated with academic partners in Trinity College Dublin and Maynooth University as well as a specialist GIS company, ERA-Maptech to undertake work funded by the EPA. It also included collaboration with the local authorities in the Greater Dublin Region.
The MOLAND land use modelling research was applied within the following fields:
- Air Quality
- Urban Transport
- Climage Change
- Urban Sprawl
- Green City