Urban Green-space Pattern Optimization for Mitigation of Overland flow



Completion Date:

Associate Prof. Min WANG, Tongji University
Prof. Harald Zepp, Ruhr Universität Bochum

Building Strong Ecological Security Patterns through Elevating Green Infrastructure’s Level of Ecosystem Services (China National R&D Program No. 2017YFC0505705)

April 2020


Due to the global climate change, cities are more frequently stuck with flood and water-logging nowadays, but their own rainwater regulation capacity has been weakened, since urbanization has greatly changed the urban water cycles. Urban green space (UGS) is the primary permeable landcover type, which can assist urban drainage system in preventing uncontrollable overland-flows, and its increased quantities have been proved to result in strengthened water-regulation capacity and lowered logging risks in many studies. However, there has been little discussion about the contribution of UGS. It remains unclear how UGS are spatially distributed could help cities strengthen the capacity of overland-flow mitigation.

Research Questions

In response to the research gap, my dissertation aims to explore the key features of UGS spatial pattern that associate with its service of overland-flow regulation and therefore to provide theoretical support for optimization of UGS spatial pattern in planning practices. For completion of the research goal, three main research questions are addressed as follows:

(I) How can we quantify the UGS’s service of overland-flow mitigation?

(II) What features of UGS spatial pattern have association with the water overland-flow mitigation?

(III) How to optimize UGS spatial pattern following the confirmed UGS pattern features?


The study area is located in the central urban area of Kunshan, Jiangsu Province, China. Firstly, spatial distribution of overland-flows is finely simulated using ArcGIS as well as its hydro-tool plugins. Then Unit Hydrograph of each catchment within the study area is drawn and exports its peak, peak time lag, duration, which can be integrally used to measure the overland-flow mitigation service; meanwhile a total of 10 pattern feature indices can be applied to delineate UGS pattern from the perspectives of the scale, the shape, and the spatial distribution. Finally, with correlation and regression analysis will the association between the pattern and the service be revealed.

Methodology Flowchart


The study drew some maim conclusions as follows:

① UGS Overland-flow mitigation service involves multiple dimensions of impact on the urban water cycle, thus it is more reasonable to quantify the service from three different aspects with the help of Unit Hydrograph, that is, flow peak, peak lag time, and runoff duration. A catchment that consumes good overland-flow mitigation service manifests itself with relatively lower flow peak, later peak lag time, and longer flow duration.

② Flow peak is negatively correlated with greening rate (GR), mean patch index(MPI), shape index(SI),integral index of connectivity(IIC), and positively correlated with patch density(PD), mean Euclidean nearest-neighbor distance(mENN). The multiple linear regression model composed of GR, MPI, shape index and mENN has the best performance compared with others. It is revealed that the efficiency of flow peak reduction decreases when greening rate has exceeded 30%; Besides, with the increase of the adjacent distance of UGS would flow peak rise faster.

③ Peak lag time is negatively correlated with edge density (ED), shape index(SI), related circumscribing circle(RCC), patch density(PD), mean Euclidean Nearest-Neighbor Distance(mENN) and integral index of connectivity(IIC), while positively related to greening rate(GR), largest patch index(LPI) and mean patch index(MPI). The multiple linear regression model composed of MPI, RCC, and division index (DI) has the best performance. When the average size of UGS is small, overland peak would delay with the increase of MPI; otherwise, the marginal effect would sharply shrink. The situation is similar when the shape index increases to a certain degree.

④ Runoff duration is negatively correlated with its edge density (ED), shape index (SI), related circumscribing circle (RCC), patch density (PD), mean Euclidean Nearest-Neighbor Distance(mENN) and integral index of connectivity (IIC), while positively related to greening rate (GR), largest patch index (LPI) and mean patch index (MPI). The multiplex linear regression model composed of LPI, MPI, mENN and IIC is the best performance. There is likely a quadratic function relationship between duration and IIC. The results revealed that the improved connectivity of UGS could prolong runoff duration only when it could exceed a threshold.

⑤ Based on the revealed association between overland-flow mitigation of UGS and the UGS pattern features, I proposed some guidelines for UGS pattern optimization, whose proves goes through investigation, simulation, diagnosis, strategy, and feedback.

Results of Overland-flow Simulation
Fine Simulation of Overland-flow

Selective Results of Curve Fitting

Selective Results of 5 Types of Unit Hydrograph


Strategies For Physical Environment Based On Water Ecosystem Services Overall Performance: A Case Study of Tian’ao Water Sensitive Rural Area In Shengsi

Tongji University, Shanghai, China
Group work with Jieqiong WANG, Shiying PARK, Junwen GE


Addressing the concept of water sensitive rural design, we aimed to explore the efficient physical forms of rural areas that impact the overall performance of water ecosystem services(WESSOP).

Through the literature review and case studies, we proposed a conceptual model which consists of 18 key environmental factors relating with four water ecosystem services, which can be categorized under involved elements into the water, bank, biotope, and anthropogenic intervention. Besides, WESSOP improvement should follow three sequential targets: the first is water ecology, which is the foundation of the holistic aquatic ecosystem. Then water habitat comes second, which highlights the promotion of ecosystem quality by serving more biodiversity. When both targets above can be satisfied, we are allowed to address water landscape for bonus benefits for society.

Taking Tian’ao Village in Shengsi, China as a case, we exemplified how the conceptual model could guide the planning and design of rural waterfront environment. The water body has been classified into three types according to the proposed functions(i.e., ecology, habitat, or landscape), for which specific design strategies have been proposed considering the related physical environment factors.

18 Environmental Factors Relating with Water Ecosystem services
Conceptual Model of Water Sensitive Rural Design
The pyramid represents the sequential  targets in the water sensitive rural design. The lower a water ecosystem service stands in the pyramid, the more prior it should be taken into account in the environment-friendly design. The target sequence follows as grey water interception, runoff management, purification, freshwater provision, aquatic flora and fauna diversity conservation, enhancing rural aesthetics, and recreational opportunities provision.

Assessing Affective Experience of In-situ Environmental Walk Via Wearable Biosensors For Evidence-Based Design

Tongji University, Shanghai, China
Group work with Zheng Chen, Sebastian Schulz, Wen Yang, Xiaofan He

In environmental psychology research, the most commonly used methods are phenomenological interviews and psychometric scales. Recently, with the development of wearable bio-sensing devices, a new approach based on bio-sensing data is becoming possible.

In this study, we examined the feasibility of using wearable biosensors to document affective experience during in-situ walk. An eight-channelled Procomp multi-bio-sensing devices (EKG, EEG, skin conductance, temperature, facial EMG, respiration) were used, in addition with a GPS tracker, to measure the in situ physiological affective responses to environmental stimuli.

This pilot experiment revealed consistent results between bio-sensing measures and two traditional methods, phenomenological interviews and psychological Likert scale rating, which indicated that mobile bio-sensing could be a promising method in measuring in-situ affective responses to environmental stimuli as well as diagnosing potential environmental stressor.

This new bio-sensory method, as exemplified in the research, could help identify negative stressful stimuli and provide evidence to support design strategies.

(1) Electrocardiogram(ECG);
(2) Electroencephalogram(EEG);
(3) Facial Electromyography(EMG);
(4) Skin conductance and skin temperature;
(5) Respiration;
(6) Signal amplifier.

Bio-sensory Data Analysis
With the help of signal filtering, heart rate from ECG and facial EMG were used to predict whether one feels pleasant, while the response from SC and several indicators from ECG were used to predict how much one feels aroused by the stimuli.
Environmental experience map based on bio-sensory signal computation
A georeferenced heat map was produced based on the biosensing dataset. A total of eight hotspots were identified. Among them, negative feelings revealed higher magnitudes(i.e., Location 2, 3 and 4), indicating stronger physiological arousals.
Design strategies based on multi-source diagnosis
Diagnosis indicated  location 3 triggered most negative emotions. Through a post-exposure participant interview, the negative emotional reactions were found out to mainly result from parking cars on the side of the walk. Therefore, we proposed to separate the pedestrian sidewalk from parking lots, which could initiate positive environmental perception, meanwhile maintaining the parking function.

TOWARD CLIMATE CHANGE: Supply-Demand Assessment And Planning Response of Urban Green Space Ecosystem Service For Overland Flow Regulation In Bochum, Germany

Ruhr-Universität, Bochum, Germany
Independent work, supervised by Prof. Harald Zepp
Proceedings of 4th International  Digital Landscape Architecture Conference

As a critical permeable landcover in the built environment, urban green space(UGS) can mitigate overland flow and decrease urban flooding risks. However, few studies have addressed the spatial disparity of the supply and demand for the overland-flow regulating ecosystem service(ORES).

In this research, an assessment framework was proposed to evaluate the demand, supply and budget of the ORES respectively based on the fine hydrological simulation of the pathways and catchments of overland-flows. The results of the disparity of supply and demand were illustrated using the method of Unit Hydrography.

The methodology was applied in the city Bochum, Germany and succeeded in identifying the areas with large demand meanwhile insufficient supply of ORES. With comparing the composition and the spatial distribution of UGS in each catchment, it was found out that approximately 40% of UGS was the threshold for a catchment to obtain enough service budget and that UGS sufficiently overlapping with overland-flow paths are able to provide ORES effectively. Some UGS development strategies and planning guidelines were consequently proposed to help practitioners arrange and allocate UGS efficiently.

Methodology Flowchart
Service Demand
Service Supply
Service Budget

Qk: Service Budget
Mk : Service Demand
Nk: Service Supply
k : Catchment k
Ak: Total area of Unit k
Ai : Total area of  non-green landcover in Unit k
Aj : Total area of  green landcover in Unit k
N: Precipitation of designed storm
Ψ: Runoff coefficient


Service Budget Relating to UGS Composition
Service budget in some catchments do not have a positive correlation with the UGS percentage and turn out to be unusually high or low. The downscaling analysis suggests that it could be due to the UGS composition. 
Service Budget Relating to UGS Spatial Distribution
UGS could contribute to the decreased waterlogging risk when it overlaps with overland flowpaths sufficiently meanwhile is evenly distributed along with their up- and downstream.
Service budget and social tradeoffs
Assign each catchment with a value according to their service budget and population density, respectively. Then identify four status by multiplying the assigned value, which leads to different development strategies.