To be usefully applicable the models need extensive calibration. Current calibration methods, however, do not take into account uncertainties in reference land-use data and uncertainties in the parameterization of land-use change models. As a result, uncertainty in land-use change predictions are mostly unknown.
Modeling urban dynamics through GIS-based cellular automata
The ASIMUD project aims to provide a solution to this issue by applying a particle filter data-assimilation framework to the calibration of land-use change models. The framework will use remote sensing derived land-use data at time steps that they are available in order to optimize the parameters in the model. The proposed calibration framework will be based on the comparison of spatial metrics derived from historic remote sensing images and land-use change simulation results.
Parameters used in the simulation model will be tuned so that the simulated urban growth patterns, as described by the metrics, match the patterns observed in the remote sensing imagery. It is expected that the approach will result in a quantification and reduction of the uncertainty in simulations of future land use. E-mail address: hans. Selection and peer-review under responsibility of Spatial Statistics doi Introduction In recent years a number of powerful high-resolution land-use change models have progressed beyond purely conceptual and theoretical considerations and aim at realistically representing geographical systems in terms of the processes modeled, the geographical detail attained, and the calibration and validation of the modeling outcomes .
This is certainly the case for cellular automata CA based land- use change models . These models are gradually becoming important instruments for the assessment of policies aimed at improved spatial planning and sustainable development  as well as scenario-analysis . Clearly, neither good science nor practical planning and policy making can be based on tools which produce questionable output; rather the tools must be robust and reliable, based on the best available scientific knowledge and data.
This raises to the highest level of importance the issue of the calibration and validation of the models . A proper historic calibration of land-use change models typically relies on the availability of time series of land-use maps . Since the production of the detailed land-use maps needed for land-use change modeling is still a costly and time-intensive manual process, time series are often lacking or contain inconsistencies in mapping methodologies, legends and scales .
- U bent hier.
- Bibliographic Information.
- Modelling Urban Dynamics!
- Computer Science > Social and Information Networks!
- Japs Draw First Blood - A Day Before Pearl Harbor (Year one - Pacific - Aerial Combat from Pearl Harbor to Guadalcanal).
This will result in measured land-use changes caused by mismatches in the mapping procedures rather than an indication of real changes in the land-use patterns of interest. Spaceborne remote sensing images have many advantages over the use of land-use maps for a historic calibration : data acquired by a remote sensing sensor is more consistent in time and the temporal coverage is much larger than the update frequency of land-use maps. The spatial resolution of MR imagery, m, is sufficient for modeling urban dynamics, typically done at 50 to m resolution.
Nevertheless, typical land-cover arrangements found in urban areas e. As such, spatial metrics describing the composition and spatial configuration of land cover reveal information about urban land use.
Urban Dynamic Model
Originally developed for landscape ecological research, spatial metrics have recently been shown to have considerable potential for the analysis of urban environments . In spite of all the research efforts on the development of remote sensing based land-use classification algorithms, the accurate classification of all land-use classes required by sophisticated land-use change models remains a challenge .
This hampers a direct comparison between land-use maps produced by remote sensing and simulated land-use maps using goodness of fit measures, such as fuzzy kappa , that are commonly used in the historic calibration procedure. Although land-use maps directly obtained from remote sensing data do not have the same level of thematic detail as the land-use maps conventionally used for historic calibration of land-use change models, results of the MAMUD project show that spatial metrics derived from frequently available and temporarily consistent remote sensing data are able to capture characteristic urban development patterns and thus can be used for calibration .
This leads to uncertainties in the prediction of future land use, which need to be quantified and reduced. The probabilistic framework will use a particle filter data-assimilation algorithm to calibrate model parameters by taking into account 1 uncertainties in input parameters that propagate through a land-use change model, and 2 uncertainties in reference data derived from remote sensing images. It is expected that the framework enables a quantification and reduction of the uncertainty in simulated future land use. This is one of the most elaborate and widely used cellular automata based land-use models currently available.
Next, details will be given on the calibration framework, developed in the MAMUD project, which uses spatial metrics derived from simulated land use and remote sensing data.
 Adaptive modeling of urban dynamics during ephemeral event via mobile phone traces
The sources of uncertainty in the land-use change model and the remote sensing interpretation chain will be discussed. Finally, the use of this uncertainty information in the probabilistic framework will be explained. The model explores the likely development of land use for periods of some thirty-fourty years into the future, given alternative planning and policy scenarios and socio-economic trends.
It progresses through time in yearly time steps and computes state changes for every cell in a regular grid measuring some m on the side.
Thus, every model grid cell represents at least one state variable in the model, meaning that the calibration of any model with the level of complexity of the MOLAND model is not trivial, rather requires time and effort. The task of the calibration is to ensure that the model behaves in a realistic manner and is able of generating observed spatial patterns.
It requires a reference land-use map, from which the actual map i. The comparison between the reconstructed and the actual map is performed by means of dedicated goodness-of-fit measures.
- Louisiana by the Numbers - Important and Curious numbers about Louisiana and her cities (States by the Numbers Book 18);
- Fake News: El Presidente in Espanol (sort of).
- You are here.
The calibration consists of four steps: 1 an initial set of parameters describing the neighbor influence functions attraction-repulsion parameters is fixed. Step 4 is repeated until the reconstructed map satisfactorily matches with the actual map. The poor availability of high quality and temporally consistent land-use maps often constrains the choice of the calibration period.
The remote sensing based calibration framework, explained next, can increase the amount of data for the historic calibration, resulting in better predictions of the land-use change model.
Calibration framework using remote sensing Fig. Remote sensing derived land-use maps can be compared with simulated land-use maps of the same year using spatial metrics. Calibration of the land-use change model is done by tuning parameters, aiming at a minimization of the difference between spatial metrics produced by model simulations and those calculated from the remote sensing-based maps.
This procedure can be applied each time a remote sensing image is available within the model calibration period. The remote sensing derived metric corresponds better with the reference scenario than with the extreme scenario. Uncertainty in the MOLAND land-use change model Spatially-dynamic modeling of land-use change involves uncertainty caused by attribute errors, positional errors, logical inconsistencies, incompleteness and temporal errors in the model and in the reference land-use maps used for initiation and calibration.
It is assumed that positional errors of the geo- referenced input maps are smaller than the resolution used in the model m for Dublin, m for Flanders. Logical consistency and completeness have been tested in previous applications of the model. Temporal accuracy is determined by the synchronization of temporal input data and the model time step. These uncertainties are assumed to be small and are ignored. Uncertainties in the reference land-use maps, however, can be important, but are difficult to quantify objectively.
Therefore, the only uncertainties that will be considered here are uncertainties in input parameters. First, the uncertain input parameters of the land-use change models need to be identified and quantified.
Vous êtes ici
This part of the project will provide predictions of future land-use accompanied by probability maps. Uncertainty in the remote sensing interpretation chain Remote sensing data analysis involves uncertainty caused by limitations of the data and the image interpretation methods used. Since uncertainties propagate through the processing chain, they will affect land-use maps inferred from remote sensing images and the derived land-use patterns, quantified by means of spatial metrics.
An important part of the project is to characterize error and uncertainty in the different steps of the land-use interpretation process, using ground-truth data and process-related uncertainty models based on classification approaches. This analysis will be carried out for metric-based classification approaches developed in the MAMUD project.
Furthermore, sensitivity analysis will be carried out to estimate the relative contribution of different steps in the processing chain on land-use uncertainty. This will provide information on the process components that contribute most to uncertainty in observed land-use patterns and derived spatial metrics, and may indicate which measures could be taken to reduce overall uncertainty in the land-use interpretation process and, consequently, in the calibration of the land-use change model.
It will also give information on the sensitivity of different spatial metrics to uncertainty in land-use pattern, which might be one of the criteria for selecting a particular spatial metric for calibration purposes. Probabilistic framework for land-use change models For error propagation modeling and particle filtering use will be made of a Python framework for spatio-temporal modeling  . The framework offers a combined interface for the task of model construction and optimization.
Modeling options are Monte Carlo simulation and data-assimilation methods such as the Particle Filter and the Ensemble Kalman filter. Computers, Environment and Urban Systems , 23 3 pp.
- Modeling urban dynamics through GIS-based cellular automata - UCL Discovery.
- Modeling Urban Dynamics.
- 3D Dynamic Representation for Urban Sprawl Modelling: Example of India’s Delhi-Mumbai corridor.
- Submission history.
- Modeling urban dynamics through GIS-based cellular automata!
In urban systems modeling, there are many elaborate dynamic models based on intricate decision processes whose simulation must be based on customized software if their space-time properties are to be explored effectively. In this paper we present a class of urban models whose dynamics are based on theories of development associated with cellular automata CA , whose data is fine-grained and whose simulation requires software which can handle an enormous array of spatial and temporal model outputs.
We first introduce the generic problem of modeling within GIS, noting relevant CA models before outlining a generalized model based on Xie's , A general model for cellular urban dynamics. We present ways in which land uses are structured through their life cycles, and ways in which existing urban activities spawn locations for new activities. We define various decision rules that embed distance and direction, density thresholds, and transition or mutation probabilities into the model's dynamic and we then outline the software designed to generate effective urban simulations consistent with GIS data inputs, outputs and related functionality.
Finally, we present a range of hypothetical urban simulations that illustrate the diversity of model types that can be handled within the framework as a preclude to more realistic applications which will be reported in later papers.
Related Modeling Urban Dynamics
Copyright 2019 - All Right Reserved