Harnessing Crowdsourced Data and Prevalent Technologiesfor Atmospheric Research

Noam DAVID

Instituteof Industrial Science,The University of Tokyo,Tokyo 153-8505,Japan

ABSTRACT The knowledge garnered in environmental science takes a crucial part in informing decision-making in various f ields,including agriculture,transportation,energy,public health and safety,and more.Understanding the basic processes in each of these f ields relies greatly on progress being made in conceptual,observational and technological approaches.However,existing instrumentsfor environmental observationsareoftenlimited asaresultof technical and practical constraints.Current technologies,including remote sensing systems and ground-level measuring means,may su f f er from obstacles such as low spatial representativity or a lack of precision when measuring near ground-level.These constraints often limit the ability to carry out extensive meteorological observations and,as a result,the capacity to deepen the existing understanding of atmospheric phenomenaand processes.Multi-system informaticsand sensing technology havebecomeincreasingly distributed as they are embedded into our environment.As they become more widely deployed,these technologies create unprecedented data streams with extraordinary levels of coverage and immediacy,providing a growing opportunity to complement traditional observation techniques using the large volumes of data created.Commercial microwave links that comprise the data transfer infrastructure of cellular communication networks are an example of these types of systems.This viewpoint letter brief ly reviewsvariousworkson thesubject and presentsaspectsconcerning theadded valuethat may beobtained asaresult of theintegration of thesenew means,which are becoming available for the f irst time in this era,for studying and monitoring atmospheric phenomena.

Key words:atmospheric science,IoT(Internet of Things),crowdsourced data,commercial microwavelinks

1. The opportunity: using existing data for studying and monitoring atmospheric phenomena

Over the past decade,the Internet of Things(IoT)and smart devices have become increasingly common as part of the technological infrastructure that surrounds us.The f low of data generated by these systems is characterized by enormousgranularity,availability and coverage.As a result,new opportunities are being opened to utilize the newly available information for various needs and,in particular,for atmospheric research.If we consider the data generated by these means,we may notice that many produce measurements with high environmental value.To name some examples—surveillancecamerasthatoperateinthevisiblelightspectrum are positioned in a vast number of locations.Previous works have shown that they can be used for monitoring the temporal patterns of fn ie atmospheric particulate matter(Wong et al.,2007).Lab experiments have indicated a direct link between the speed of movement of car wipers and rainfall intensity,meaning advanced vehicles that store these data can,in essence,be used as moving rain gauges(Rabiei et al.,2013).Kawamura et al.(2017)revealed a novel technique for monitoring atmospheric humidity using terrestrial broadcasting waves,based on propagation delays due to water vapor.Data shared as open source from social networks have been found to bepotentially e f fectivein improving automatic weather observations.Indeed,for the most part,the initial weather observation isconducted automatically by dedicated sensors;however,some weather conditions are still better detected by the human eye.On the other hand,millions of“human observers”around the world use applications such as Twitter,which allows them to report publicly on subjects that are relevant to them,and in particular on weather phenomena(Cox and Plale,2011).As was recently reviewed by Price et al.(2018),in 2020 there will be more than 20 billion smartphones carried by the public worldwide.These mobile devices are equipped with sensors that can be used for environmental monitoring on a multisource basis.Recent works indicate the ability to obtain atmospheric temperature information for the urban canopy layer(Overeem et al.,2013a),to measure atmospheric pressure(Mass and Madaus,2014;McNicholasand Mass,2018a),or to study atmospheric tides(Price et al.,2018).Additional studies point to the potential of using any camera-enabled smart mobile device to monitor air quality(Pan et al.,2017).Given the comprehensive coverage of these new“virtual sensors”from all land locations across the whole globe,this low-cost solution introduces a wide range of possibilities that previously could not be offered through existing technologies.

2.Data from cellular communication infrastructureas an exampleof an environmental monitoring tool

A key exampleof utilizing existing datasourcesfor atmospheric monitoring is the use of measurements acquired by commercial microwave links(CMLs)that comprise the infrastructure for wireless data transport between cellular base stations.This technology has essentially become a valuable weather monitoring tool over the past decade(e.g.,Overeem et al.,2013b;Alpert et al.,2016;Gosset et al.,2016;Davıd and Gao,2017;Chwalaand Kunstmann,2019).Weather phenomena and atmospheric conditions interfere with the electromagnetic waves transmitted by CMLs.Thus,these networksprovide,in essence,an already-existing environmental monitoring facility.The various works done in the f ield indicate the ability to detect and map rainfall(e.g.,Messer et al.,2006;Berne and Uijlenhoet,2007;Chwala et al.,2012;Overeem et al.,2013b;Fencl et al.,2015),monitor fog(e.g.David etal.,2015),atmospheric moisture(David etal.,2009,2011;Chwala et al.,2014;Alpert and Rubin,2018)and dew(Harel et al.,2015;David et al.,2016).Recent research revealed the ability of these radio links to indirectly detect air pollution conditions(David and Gao,2016).

3.On added valueof this novel approach

Indeed,the new data available from these various means(smartphones,social networks,etc.),and particularly from CMLs,can provide observations with considerable spatial coverage and with minimal cost.However,the accuracy of each“sensor”is lower than that of a dedicated instrument.This being the case,is it possible to produce signif icant information compared to that derived from specialized tools?It can be assumed that these“virtual sensors”are not a substitute for conventional monitoring means,whenever those exist in the f ield.The correct approach,then,is to consider thesenewly availablesourcesof dataascomplementary measures to dedicated measurements and as a substitute during the many cases in which conventional monitoring tools are unavailable.However,the data acquired by prevalent technologies,even when taken alone,often holds enormous potential.In order to demonstratetheadded valuewhich liesin IoT dataand prevalent technologies,let usfocuson CMLsas an example of such a system.Atmospheric moisture is more poorly characterized than wind or even precipitation,due to thedi f fi culty inobservingthehumidity f ield.Therefore,questionssuchasthemagnitudeof small-scalevariability of moisture in the boundary layer,and its e f f ect on convection initiation,are still unanswered(Weckwerth et al.,2004).As a result,the ability to predict convective precipitation,on the storm scale,islimited.However,for signif icantly improving convection initiation measurements,one will need moisture measurementsatmeso-γresolutionwithaccuraciesof up to1 g m−3(Fabry,2006).Notably,such atypeof observationscan beacquired using CMLs(David etal.,2019).High-resolution precipitation distribution mapscan begenerated using CMLs,and thereforetherelationship between pollutant wash-o f f and rainfall providesan opportunity to potentially acquireimportant spatial information about air quality,as discussed in recent research(David and Gao,2016).Moreover,liquid water content(LWC)constitutesamajor parameter in fog research.Fog LWC changes in space,altitude,and over time,and is dependent on surfaceand atmospheric conditions(Gultepeet al.,2007).However,conventional sensorsfor acquiring LWC estimates are limited in the spatial range they can cover,and in their availability.It has been shown that CMLs are able to providefog LWCestimatesacrosslargespatial regionswhere dedicated sensorsare nonexistent.Indeed,theavailability of various spectral channels from satellites provides the possibility to observe clouds,aerosols,the Earth surface,and in particular,fog(Lensky and Rosenfeld,2008;Michael et al.,2018).However,CMLs have also been found to have potential advantagesfor detectingfogunder challengingconditions wheresatelliteretrievalsarelimited,e.g.,when high-altitude clouds cover the fog as observed from the satellite vantage point(David,2018).Alternatively,theability to monitor rainfall in areas where radars su f f er from clutter e f f ects(Goldshtein et al.,2009)or are blocked by complex topography(David et al.,2013),hasalso been demonstrated.

4.Summary

The possibilities for monitoring environmental phenomena via new observational powers are many,the available information vast,and the cost minimal,since such“opportunistic sensors”are already deployed in the f ield.As a result,this means of monitoring the environment is becoming advantageousfor atmospheric research.Notably,thesenewly available“virtual sensors”open thedoorsto thepossibility of assimilating their measurements into high-resolution numerical prediction models,which could lead to improvements in the forecasting capabilities that exist today(Kawamura et al.,2017;Madaus and Mass,2017;McNicholas and Mass,2018b;David et al.,2019).In apractical sense,thisnovel approach could lay the groundwork for developing new earlywarning systemsagainst natural hazardsand generating avariety of products required for a wide range of f ields.Thus,the overall potential contribution to public health and safety may beinvaluable.

Acknowledgements. Iwish to express my deepest gratitudeto Professor Yoshihide SEKIMOTO and his research team for fruitful discussions and for hosting me in their laboratory as a Visiting Scientist at the Institute of Industrial Science of the University of Tokyo,Japan,during 2018–19.