Improving Urban Climate Data Collection: Assessment and Redesign of Tsinghua’s Meteorological Sensor Network

Abstract

Urban sensor networks have the potential to improve urban modeling, design, andplanning. Network data, with higher spatial resolution than traditional large weatherstations, can provide insight into localized effects of materials, building heights, andgeometry on microclimate. However, there does not currently exist a standardizedprotocol for designing urban climate sensor networks. This thesis is a case study ofthe Tsinghua University Sensor Network (TUSN) in Beijing, China, which consistsof seven sensor nodes across the Tsinghua University campus. Using relative RMSE(rRMSE) and correlation coefficient analysis, existing TUSN data from January toAugust 2016 were assessed for redundancy, and two of the seven stations were determinedto be most redundant. ENVI-met urban microclimate simulations werevalidated with the existing data so that model results could be used for re-locatingthe two redundant stations. rRMSE analysis was applied to the campus simulationresults from the most accurate ENVI-met simulation, and used to select two newstation locations. In addition to the redesign of stationary sensors, mobile sensingbased on the existing campus bus route is proposed as a means by which to increaseresilience and spatial resolution for TUSN. Incorporating new sensor types and establishingconsistent data collection across network stations are also recommendedways of making the network data more useful for future research. Finally, samplemetadata that describes the redesigned network is presented, as documentation isconsidered a critical final step of the design process. The design protocol developedin this case study can be applied toward assessment and design of existing and newurban networks, and it can be improved upon by using more accurate microclimatemodels as the basis for analysis.