Next Article in Journal
Simulation of Sentinel-2 Bottom of Atmosphere Reflectance Using Shadow Parameters on a Deciduous Forest in Thailand
Next Article in Special Issue
Rural–Urban Transition of Hanoi (Vietnam): Using Landsat Imagery to Map Its Recent Peri-Urbanization
Previous Article in Journal
Monitoring Forest Change in the Amazon Using Multi-Temporal Remote Sensing Data and Machine Learning Classification on Google Earth Engine
Previous Article in Special Issue
Emergency Department Overcrowding: A Retrospective Spatial Analysis and the Geocoding of Accesses. A Pilot Study in Rome
Peer-Review Record

Supporting Policy Design for the Diffusion of Cleaner Technologies: A Spatial Empirical Agent-Based Model

ISPRS Int. J. Geo-Inf. 2020, 9(10), 581;
Reviewer 1: Anonymous
Reviewer 2: Anonymous
ISPRS Int. J. Geo-Inf. 2020, 9(10), 581;
Received: 23 August 2020 / Revised: 21 September 2020 / Accepted: 28 September 2020 / Published: 1 October 2020
(This article belongs to the Special Issue Geo-Information Science in Planning and Development of Smart Cities)

Round 1

Reviewer 1 Report

The problem discussed in the article is important and current. The authors proper identified research gaps and proposed proprietary solutions, taking into account the relevant research results. Both the literature review and the research part are well described. The only objection I have for the components described in 3.2.1. Please consider what is the variable related to Attitude (for the adoption of PV) and what is the moderator / background factors (fugure 1) in this case

Author Response

Thank you very much for this comment, you are perfectly right. Originally, the paper explained in an oversimplified way which background factors influence the formation of the different beliefs (and, so, the attitude toward the behavior) and why we decided to adopt a random distribution (for opinion and environmental awareness) and a different approach for the innovation level. In lines 416-425 (of the manuscript with track changes), we tried to explain better all motivations.

Reviewer 2 Report

This work built an agent-based model to simulate the adoptive behavior of local residents toward photovoltaic systems. The model was further used to evaluate efficiency of several promotion policies. Overall, I feel this is a well designed and well written work, which deserves a publication. I have a few concerns for the authors to further improve their work.

  1. Though GIS is incorporated, I don’t see how this work contributes to the geographic information science. The role of GIS in this work is barely to measure distance between agents. I feel it fits better to a social computational/social network journal.
  2. Please provide more details on how the spatial-social networks were generated for every five years. Further, how the diffusion spread spatially over the study area. I suggest some image illustrations on the generated social network, which is the basis for the simulation, and a series of diffusion maps to show the spatial spreading process.
  3. The methodology is well described in detail. However, I think the authors should further edit a bunch of equations. For example, equations for several dynamic variables, such as those for attitude and social norm, do not have a time component. Do people interact every year or every five year?
  4. Line 403, ‘…are both calculated based on the mean values of …..’, but the equation 11 that follows does not indicate how the averaging process is done. I am confused.
  5. The simulation last 30 years. I wonder when the simulation went beyond 2019, does the model consider the population change from 2020 to 2030?
  6. The article looks a bit lengthy. The literature review might be cut to some extent. For example, the very detailed description on the relative agreement model can be shorten. I don’t see it being used heavily later. Figure 3 is not necessary.

Author Response


For what concerning the role of GIS, it is not true that it is limited to measure the distance between agents. On the contrary, it is fundamental for the creation of the precise configuration of the population in the neighborhood, as underlined in lines 300-304 (track changes file). Finally, we added some maps that show the diffusion of PV in the neighborhood in three different years of the simulation (Figure 11). The figure exemplifies the combined use of ABM and GIS for a dynamic simulation of the process under investigation. Analogously, similar maps have been produced for the other scenarios considered in the present study that have been useful for visualizing in a clear and easy way the results of the model.


Thank you very much for your suggestions. We added more explanation on the spatial-social network generation, as you can see in Table 6 and in the text (lines 391-395, track changes file). Moreover, we added an animated figure (Figure 5) to show the network creation. For what concerning the role of GIS, it is not true that it is limited to measure the distance between agents. 


Thank you for the comment, we explained better the interaction timeframe in Table 6 and in other parts of the text (see Answer 1)). People interact every year, but the network changes every 5 years. This does not mean that an agent changes his/her opinion, attitude and social norm after five years; in fact, during this timeframe, agents are influenced also by the neighborhood as a whole, and this influence can also change opinion, attitude and norm in the network. Moreover, every year a certain percentage of people increases their environmental awareness, following the data reported by ISTAT 2015. It means that the attitude and social norm of individuals can change also due to this increase of “ecologist”.

Concerning the equations, it is not necessary to add the time because in Netlogo the same calculation is repeated each time step (that is equal to 1 year)


Thank you for this comment. Actually, equation 11 is correct, because the social pressure is exerted by the agent j on the agent i. However, in our model, we have considered j as the mean of the attitudes of all agents in the network or in the neighborhood. For simplifying the comprehension, we changed a little bit the text and we modified the subscript (from j to j*)


At the moment, the model does not consider a change in the population characteristics in the entire timeframe. Firstly, for simplifying the model implementation. Secondly, for testing policies without any other changes in the people characteristics. In the text, the specification of this choice is at line 312-317. We also underlined the second reason that is not directly explained here.


To facilitate the reading process, we removed all non-essential parts in the text. As you can see, we eliminated some specific references in the literature review simply referring to the three main literature reviews on this topic. Then, we removed Figure 3. However, for completeness and clarity of the methodology applied, we preferred to maintain the relative agreement section as it is.

Back to TopTop