2f:T490,# Modeling a combined heatwave and blackout scenario in Houston In this computational notebook, we attempt to estimate the mortality impacts of a city-wide blackout in Houston which coincides with unusual heat. We apply a simplified and adapted version of the methodology used by [Stone et al., 2023](https://pubs.acs.org/doi/10.1021/acs.est.2c09588). The conceptual steps are as follows: 1. We estimate people's _individually experienced temperature_ (IET) throughout the day which is a combination of outdoor and indoor temperatures. This incorporates an indoor heat model for a typical Houston single-family home, as well as assumptions about how much time people typically spend outdoors. 2. We calculate a heat stress metric called _degree-hours above threshold_ (DHAT) and observe its relationship with outdoor temperatures before the hypothetical blackout happens. 3. Once the blackout happens, people's indoor heat exposure increases significantly. We calculate their new DHAT metric and translate it back into an "inflated" outdoor temperature. This new outdoor temperature is used as the input for an exposure-response curve from Gasparrini et al., 2015. 30:T585,getSample = ( outdoorDist, threshold ) => { return new Promise(resolve => { resolve(d3.range(0, 20000).map( i => { const pct = i/20000 const age = pct <= 0.216 ? '0-14' : ( pct <= 0.88 ? '15-64' : '65+') const acRandom = Math.random() const ac = acRandom <= 0.9 ? 'full' : (acRandom <= 0.99 ? 'inadequate' : 'none') const backOnlineAfter = Math.random() <= 0.5 ? '2day' : '5day' const outdoorWorker = age === '15-64' && Math.random() <= outdoorShare/100 const hoursOutside = weightedRandomSelection([0, 1, 2, 3, 4], outdoorDist) const ap = getActivityPattern(age, outdoorWorker, hoursOutside) const uhi = 0 const buildingType = weightedRandomSelection(['new-good', 'new-bad', 'old-good', 'old-bad'], [0.25, 0.25, 0.25, 0.25]) const ietsNoBlackout = reduceArray(extendedHours.map(row => { return getIETNoBlackout(row, ap.before, buildingType, ac, 0) }), 24) const dmatNoBlackout = reduceArray(extendedHours.map(row => { return Math.max(0, getIETNoBlackout(row, ap.before, buildingType, ac, 0) - threshold) }), 24) const dmatBlackout = reduceArray(heatwave.map(row => { return Math.max(0, getIETBlackout(row, ap.after, buildingType, backOnlineAfter, ac, 0) - threshold) }), 24) return { age, buildingType, ac, outdoorWorker, activityPattern : ap, dmatNoBlackout, dmatBlackout, ietsNoBlackout } })) }) }31:T778,## 1. Estimating IET We create a sample of 20,000 synthetic persons and calculate their individually experienced temperatures in different scenarios. - We first assign a random "activity pattern" reflecting what hours they spend indoors vs outdoors. For most persons, we assume they spend 2 hours or less outside¹. We assign these outdoor hours randomly but assume they can't occur between 11pm and 6am. The notable exception are outdoor workers, which we assume are 10% of the working-age population² and spend the hours from 7am to 3pm outside instead. - We assign people randomly to one of four building archetypes³ modeled by [Amir Baniassadi](https://www.marcusinstituteforaging.org/who-we-are/profiles/amir-baniassadi-phd). - We calculate IET hour by hour. If the person is outside, hourly IET equals airport temperature. If the person is inside and their AC is working, hourly IET equals the set point of their thermostat. If the person is inside and the power is down, hourly IET is taken from Baniassadi's model.

¹ The average American spends [roughly 2 hours per day](https://www.nature.com/articles/7500165) outside. During an extreme heat wave, people probably spend less time outdoors. These assumptions are for our pre-blackout baseline. Once the blackout occurs, we assume that people spend even less time outdoors.

² 8% of the working-age population in Harris County works in construction or agriculture. Other sectors with significant outdoor work include landscaping, security services and transport.

³All 4 building archetypes represent single-family detached homes (the most common housing type in Houston). However, they differ in building/insulation standards, amounts of shading around the building, and occupant behavior (whether windows are opened or not).

32:T598,sample = d3.range(0, 20000).map( i => { const pct = i/20000 const age = pct <= 0.216 ? '0-14' : ( pct <= 0.88 ? '15-64' : '65+') const acRandom = Math.random() const ac = acRandom <= 0.9 ? 'full' : (acRandom <= 0.99 ? 'inadequate' : 'none') const backOnlineAfter = Math.random() <= 0.5 ? '2day' : '5day' const outdoorWorker = age === '15-64' && Math.random() <= outdoorShare/100 const outdoorDist = outdoorTime === 'More' ? [0.15, 0.25, 0.4, 0.18, 0.02] : [ 0.3, 0.5, 0.1, 0.08, 0.02 ] const hoursOutside = weightedRandomSelection([0, 1, 2, 3, 4], outdoorDist) const ap = getActivityPattern(age, outdoorWorker, hoursOutside) const uhi = 0 const buildingType = weightedRandomSelection(['new-good', 'new-bad', 'old-good', 'old-bad'], [0.25, 0.25, 0.25, 0.25]) const ietsNoBlackout = reduceArray(extendedHours.map(row => { return getIETNoBlackout(row, ap.before, buildingType, ac, 0) }), 24) const dmatNoBlackout = reduceArray(extendedHours.map(row => { return Math.max(0, getIETNoBlackout(row, ap.before, buildingType, ac, 0) - threshold) }), 24) const dmatBlackout = reduceArray(heatwave.map(row => { return Math.max(0, getIETBlackout(row, ap.after, buildingType, backOnlineAfter, ac, 0) - threshold) }), 24) return { age, buildingType, ac, outdoorWorker, activityPattern : ap, dmatNoBlackout, dmatBlackout, ietsNoBlackout } })33:T4a7,## Limitations - One major difference between our analysis and Stone et al. is that they run an urban-scale climate model to capture varying temperatures across a city, and the urban heat island effect in particular. We use airport data from Bush Intercontinental for all outdoor heat exposure and as the only input to our indoor heat models. According to Drew Shindell, professor of earth sciences at Duke University, there is limited epidemiological evidence on how much higher the risk of heat-related deaths is in the hottest neighborhoods of a city. - There are several variables that could influence heat exposure which we don't account for at all, notably income, race and gender. We also don't account for possible correlations between factors like income, occupation and housing quality. For instance, it's possible that lower-income groups also have lower-quality housing, live in hotter neighborhoods and/or are more likely to work outdoors. - The analysis is based on air temperatures only. More recent environmental health research explores metrics that also capture humidity and other factors, such as heat index, wet-bulb globe temperature or universal thermal climate index.34:[{"id":"b4009288e47fc76f424a6e91695d8ac8571221c1caf6fced30d1dfc5edb2f8e0466e71d4ae37f343857a1da229ad9ed22e25c82e5df386323304de665f5dbdc0","href":"https://static.observableusercontent.com/files/b4009288e47fc76f424a6e91695d8ac8571221c1caf6fced30d1dfc5edb2f8e0466e71d4ae37f343857a1da229ad9ed22e25c82e5df386323304de665f5dbdc0","name":"airport.csv","type":"text/csv","lastModified":1723727005501.818,"size":4973885},{"id":"840e9221334efe22b035202f8a8d3a2f14caa2182def09c3a9889f147226b64e5385b09b3afb5e3fde8fc9c820e13055dc0bae05e6b1845238d1d4985d17b1a9","href":"https://static.observableusercontent.com/files/840e9221334efe22b035202f8a8d3a2f14caa2182def09c3a9889f147226b64e5385b09b3afb5e3fde8fc9c820e13055dc0bae05e6b1845238d1d4985d17b1a9","name":"rangeOfResults (2).json","type":"application/json","lastModified":1725313687196.069,"size":9735},{"id":"6287ccbba8eff9e32808119f4d0e8cf39aaab6a148dbf78503e3a591374f202e24995a1e03e408cf47baf649aa2a1c0f2c6feec43a33e6e38aa6b16b73480f14","href":"https://static.observableusercontent.com/files/6287ccbba8eff9e32808119f4d0e8cf39aaab6a148dbf78503e3a591374f202e24995a1e03e408cf47baf649aa2a1c0f2c6feec43a33e6e38aa6b16b73480f14","name":"amir_all.csv","type":"text/csv","lastModified":1724070127162.351,"size":250311},{"id":"5cd29701ce8c4afb3f681084605c953662c61bf6a86af80d714d265c3dd397cb316afc056507e629af96f34898a4442dc135405c51f26dbec89934ad288454e8","href":"https://static.observableusercontent.com/files/5cd29701ce8c4afb3f681084605c953662c61bf6a86af80d714d265c3dd397cb316afc056507e629af96f34898a4442dc135405c51f26dbec89934ad288454e8","name":"amir_no_outage.csv","type":"text/csv","lastModified":1723727005501.818,"size":126978},{"id":"2ec30eec30ec90dbb184caadfae0b349c69b5a588c307112858d095c09a6a3663bdc63a7784cbe9e2e150429a5a25e8a02c8903771ea1a2bb4672718a857f61d","href":"https://static.observableusercontent.com/files/2ec30eec30ec90dbb184caadfae0b349c69b5a588c307112858d095c09a6a3663bdc63a7784cbe9e2e150429a5a25e8a02c8903771ea1a2bb4672718a857f61d","name":"hstn.usa8506.csv","type":"text/csv","lastModified":1723727005501.818,"size":39137},{"id":"808d1328cf79f5c597d81e77b1ce02d14b963923b3af4bdbdbb40ba1da8a0924af2c5f99411aed8b5133381ba5aa780a9523c108c27fabf85b8dad1c4c634d49","href":"https://static.observableusercontent.com/files/808d1328cf79f5c597d81e77b1ce02d14b963923b3af4bdbdbb40ba1da8a0924af2c5f99411aed8b5133381ba5aa780a9523c108c27fabf85b8dad1c4c634d49","name":"indoor_amir - 2_days (1).csv","type":"text/csv","lastModified":1723727005501.818,"size":7357},{"id":"cd00fca649f64ce8c27c25cab1584f9e32e8361ae6ce6d049565dbcc94314064cf9a99ebab275559d211bbf58cc3b5e2efba706f58bd6d558c545bced1534b69","href":"https://static.observableusercontent.com/files/cd00fca649f64ce8c27c25cab1584f9e32e8361ae6ce6d049565dbcc94314064cf9a99ebab275559d211bbf58cc3b5e2efba706f58bd6d558c545bced1534b69","name":"indoor_amir - 5_days (1)@1.csv","type":"text/csv","lastModified":1723727005501.818,"size":7289}] 2d:{"title":"Modeling a combined heatwave and blackout scenario in Houston","cells":[{"id":0,"value":"$2f","pinned":false,"hidden":false,"mode":"md"},{"id":519,"value":"We estimate between **${q25}** and **${numberWithCommas(q75)}** deaths over the course of the event, with a median estimate of **${median}** deaths.","pinned":false,"hidden":false,"mode":"md"},{"id":2190,"value":"### Distribution of deaths under different assumptions\n\nThe below plot shows the number of modeled deaths for different combinations of parameters. We vary average time spent outdoors between 1.5 and 2 hours, the threshold for degree-hours above threshold between 80F and 85F, and vary the share of the population assigned to better or worse insulated homes.","pinned":false,"hidden":false,"mode":"md"},{"id":2170,"value":"Plot.plot({\n\n height: 180,\n style : { fontSize : 14 },\n \n marginBottom: 45,\n y : {\n ticks : false,\n domain: [ -1, 1 ],\n \n \n },\n x : {\n domain : [ 0, 4000 ],\n label: \"Total deaths\"\n },\n \n marks: [\n Plot.rect([[ q25, q75, -0.6, 0.6 ]], { x1: d => d[0], x2 : d => d[1], y1 : d => d[2], y2 : d => d[3], fill : 'orange', fillOpacity: 0.25}),\n Plot.ruleX([ median ], { y1: -0.6, y2 : 0.6, strokeWidth: 1.5 }),\n Plot.dot(rangeOfResults, {x: \"totalDeaths\", y: d => 0, fill: 'red', r: 7, fillOpacity : 0.35 }),\n \n ]\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":1129,"value":"## 0. Scenario setup\n\nWe use observed hourly airport temperatures from a historic heat wave event spanning 7 Aug 2011 to 11 Aug 2011. The 5-day mean temperature corresponds to the 99th percentile of all 5-day periods across Houston's warm season from 2000 to 2023.\n\nWe assume this heat wave coincides with a blackout across the entire Houston metro area. Half of the population get their power back after 2 days while the other half only get it back after 5 days.¹\n\n

¹ This is a simplified version of Stone et al.'s scenario which assumes a complete blackout for 2 days and gradual resumption of power over the following 3 days.

","pinned":false,"hidden":false,"mode":"md"},{"id":1159,"value":"Plot.plot({\n style : { fontSize : 14 },\n y: {ticks: 5, grid : true},\n marginLeft: 65,\n marginBottom: 45,\n marks: [\n Plot.axisX({ label : 'Warm season temperatures, 5-day averages' }),\n Plot.rectY(airportDailyRolling, Plot.binX({y: \"count\", interval: 0.25, fill: d => {\n return 'red'\n }, fillOpacity : arr => arr.map(row => row.date).indexOf('09/08/2011') >= 0 ? 1 : 0.25\n }, {x: \"smoothed\"})),\n Plot.ruleY([0])\n ]\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":2139,"value":"$30","pinned":false,"hidden":false,"mode":"ojs"},{"id":956,"value":"$31","pinned":false,"hidden":false,"mode":"md"},{"id":1990,"value":"$32","pinned":false,"hidden":false,"mode":"ojs"},{"id":1629,"value":"viewof outdoorShare = Inputs.range([0, 20], {label: \"Share of outdoor workers (%)\", step: 1, value : 10})","pinned":false,"hidden":false,"mode":"ojs"},{"id":2135,"value":"viewof outdoorTime = Inputs.radio([\"Less\", \"More\"], {label: \"Time spent outdoors\", value: \"More\"})","pinned":true,"hidden":false,"mode":"ojs"},{"id":1514,"value":"avgTimeOutside = ss.mean(sample.map(row => row.activityPattern.before.filter(v => !v).length))","pinned":false,"hidden":false,"mode":"ojs"},{"id":463,"value":"### Mean degree-hours above threshold without blackout: ${round(ss.mean(_.flatten(sample.map(row => row.dmatNoBlackout))), 1)} hours C","pinned":false,"hidden":false,"mode":"md"},{"id":473,"value":"### Mean degree-hours above threshold with blackout: ${round(ss.mean(_.flatten(sample.map(row => row.dmatBlackout))), 1)} hours C","pinned":false,"hidden":false,"mode":"md"},{"id":365,"value":"Plot.plot({\n y: {grid: true},\n // x: { domain : [ 20, 35 ] },\n style : { fontSize : 14 },\n marginLeft: 65,\n marginBottom: 45,\n marks: [\n\n Plot.axisX({ label : 'Degree-hours above threshold' }),\n Plot.rectY(ietDistNoBlackout, Plot.binX({y: \"count\"}, {x: \"value\", interval : 0.05, fill: \"blue\", fillOpacity : 0.3, })),\n Plot.rectY(ietDistBlackout, Plot.binX({y: \"count\"}, {x: \"value\", interval : 0.05, fill: \"red\", fillOpacity : 0.6, mixBlendMode : 'multiply' })),\n Plot.ruleY([0])\n ]\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":522,"value":"totalDeaths","pinned":false,"hidden":false,"mode":"ojs"},{"id":370,"value":"ietDistBlackout = _.flatten(sample.map(row => row.dmatBlackout)).map(v => {\n return {\n value : v\n }\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":367,"value":"ietDistNoBlackout = _.shuffle(_.flatten(sample.map(row => row.dmatNoBlackout)).map(v => {\n return {\n value : v\n }\n})).filter((d, i) => i % 5 === 0)","pinned":false,"hidden":false,"mode":"ojs"},{"id":505,"value":"## 2. Modeling the relationship between DHAT and airport temperatures\n\nWe group our synthetic persons into groups based on whether they have functioning AC before the blackout, whether they are outdoor workers and also by their building archetype. Then, for each group, we create a univariate linear model between their daily degree-hours above threshold and the city's airport temperature of that day.\n\nWe explore thresholds between 80F and 85F as the baseline for our heat stress metric. A threshold close to the thermostat set point of our population (74F) would lead to implausibly large health impacts. A threshold very far away from the set point would suggest that humans can endure a prolonged and significant departure from said set point with little health implications.\n\n**Our mortality results are highly sensitive to the choice of threshold**. We run the analysis with a range of thresholds, as well as varying other parameters, and report the bottom and top quartile as well as median of our results.","pinned":false,"hidden":false,"mode":"md"},{"id":90,"value":"Plot.plot({\n\n style : { fontSize : 14 },\n marginLeft: 40,\n marginBottom: 45,\n x: {grid: true},\n y: {grid: true},\n marks: \n\n [\n Plot.axisX({ label : `Degree-hours above threshold (${threshold}C)` }),\n Plot.axisY({ label : 'Airport temp (in degC)' }),\n\n Plot.dotX(plotData, { x : d => d.dmat, y : d => d.oat, fill: 'red', fillOpacity : 0.05 }),\n\n ...Object.keys(models).filter(k => k !== 'all').map( k => models[k])\n .map( model => {\n return d3.range(0, maxDmat, 0.1).map(v => {\n return { ietAvg : v, tavg : model(v) }\n })\n }).map( data => Plot.line(data, {x: \"ietAvg\", y: \"tavg\", stroke: \"darkred\", strokeOpacity : 0.7 }))\n\n ]\n \n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":1952,"value":"Using a threshold of ${ round(cToF(threshold), 1) } degF","pinned":false,"hidden":false,"mode":"md"},{"id":1884,"value":"viewof threshold = Inputs.range([25, 29], {label: \"Threshold for health impacts (degC)\", step: 0.1, value : 27.8 })","pinned":false,"hidden":false,"mode":"ojs"},{"id":1046,"value":"## 3. Calculating health impacts\n\nOnce the blackout occurs, average IETs are much higher, as people now spend the majority of the day in _un_-airconditioned spaces¹. The linear model from step 2 maps degree-hours above threshold back to higher airport temperatures. To know the mortality impact of these heightened airport temperatures, we need an exposure-response curve that extends beyond historically observed temperatures in Houston. We extrapolate an exposure-response curve from Gasparrini et al. by fitting a linear regression through the natural logarithm of the risk ratio.\n\nFor simplicity's sake, we assume that half of the population of Houston gets their power back after 48 hours, and the other half only gets it back after 5 days.\n\n

¹We assume that people sharply reduce their time spent outdoors and only a small fraction of outdoor workers (eg emergency responders) continue to work.

","pinned":false,"hidden":false,"mode":"md"},{"id":496,"value":"### Extrapolated exposure-response curve\nbased on Gasparrini et al., 2015","pinned":false,"hidden":false,"mode":"md"},{"id":185,"value":"Plot.plot({\n\n style : { fontSize : 15 },\n // x : { grid : true, domain : [ 25, 35 ] },\n // y : { grid : true, domain : [ 1, 1.2 ] },\n marginLeft : 50,\n marginTop : 35,\n marginBottom : 40,\n \n marks: [\n Plot.ruleY([1]),\n Plot.axisX({label: \"Airport temperature (degC)\"}),\n Plot.axisY({label: \"Risk ratio\"}),\n \n Plot.lineY(testData, {x: d => d['Airport temp'], y: d => d.rrExp, stroke : 'teal', strokeWidth: 2}),\n //Plot.dot(dots, {x: d => d.tavg, y: d => d.rr, stroke : 'black', strokeWidth: 1}),\n ]\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":2321,"value":"$33","pinned":false,"hidden":false,"mode":"md"},{"id":1110,"value":"### Below, we define the population of the Houston metro area as well as the average daily number of deaths.","pinned":false,"hidden":false,"mode":"md"},{"id":5,"value":"totalPop = 7510253","pinned":false,"hidden":false,"mode":"ojs"},{"id":126,"value":"dailyDeathsPer100k = 2.3","pinned":false,"hidden":false,"mode":"ojs"},{"id":1137,"value":"### Total deaths are calculated as (relative risk - 1)*(typical daily deaths) and summed up over the 5 days.","pinned":false,"hidden":false,"mode":"md"},{"id":1135,"value":"totalDeaths = round(sample.map(row => {\n\n const model = models[row.outdoorWorker + '_' + row.buildingType + '_' + row.ac]\n\n const erCurve = erFunction === 'Quad' ? erCurveQuad : erCurveExp\n \n return row.dmatBlackout.map( v => {\n return (erCurve(model(v))-1)*dailyDeathsPer100k/100000\n }).reduce(sum, 0)\n \n}).reduce(sum, 0)/20000*totalPop, 0)","pinned":false,"hidden":false,"mode":"ojs"},{"id":1142,"value":"### The functions below are utility functions used for the above steps.","pinned":false,"hidden":false,"mode":"md"},{"id":2166,"value":"getModels = groups => {\n \n return _(groups)\n .mapValues((arr, k) => {\n\n \n const pairs = _.flatten(arr.map(person => {\n return heatwaveDaysExtended.map((row, i) => {\n return [ row.tavg, person.dmatNoBlackout[i] ]\n })\n }))\n\n console.log(pairs)\n\n //return pairs\n\n return getLinearModel(pairs)\n \n })\n .valueOf()\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":734,"value":"models = {\n\n return _(groups)\n .mapValues((arr, k) => {\n\n if(true) {\n const pairs = _.flatten(arr.map(person => {\n return heatwaveDaysExtended.map((row, i) => {\n return [ row.tavg, person.dmatNoBlackout[i] ]\n })\n }))\n\n console.log(pairs)\n\n //return pairs\n\n return getLinearModel(pairs)\n }\n\n // const p = arr[0]\n \n // return t => {\n\n // return (t - intercept - intercepts.outdoorWorker[p.outdoorWorker] - intercepts.ac[p.ac] - intercepts.buildingType[p.buildingType])/slope\n \n // //return (t - intercept - intercepts.age[p.age] - intercepts.hasAc[p.hasAc])/(slopes.age[p.age] + slopes.hasAc[p.hasAc])\n // }\n\n \n \n })\n .valueOf()\n}\n ","pinned":false,"hidden":false,"mode":"ojs"},{"id":2160,"value":"getGroups = sample => {\n const out = _(sample).groupBy(row => {\n return row.outdoorWorker + '_' + row.buildingType + '_' + row.ac\n })\n .valueOf()\n\n out['all'] = sample\n return out\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":730,"value":"groups = {\n const out = _(sample).groupBy(row => {\n return row.outdoorWorker + '_' + row.buildingType + '_' + row.ac\n })\n .valueOf()\n\n out['all'] = sample\n return out\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":877,"value":"nonExtremeDays = reduceArray(nonExtremeHours.map(row => row.ta), 24).map( v => {\n return { tavg : v }\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":569,"value":"heatwaveDaysExtended = heatwaveDays.concat(nonExtremeDays)","pinned":true,"hidden":false,"mode":"ojs"},{"id":933,"value":"function getRandomHoursOutside(numberOfHours) {\n // Define the range of numbers\n const numbers = Array.from({ length: 24 }, (_, i) => i )\n .filter(v => v >= 6 && v <= 23)\n\n let randomNumbers = []\n\n for(let n = 0; n < numberOfHours; n++) {\n \n // Pick n unique random numbers\n const randomIndex = Math.floor(Math.random() * numbers.length);\n const randomNumber = numbers[randomIndex];\n randomNumbers.push(randomNumber)\n\n // Remove the picked number to ensure the next is unique\n numbers.splice(randomIndex, 1);\n\n }\n\n return d3.range(0, 24).map( v => randomNumbers.indexOf(v) < 0 )\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":44,"value":"getActivityPattern = (age, outdoorWorker, numberOfHours) => {\n\n const outdoor = [ 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1 ]\n\n const othersBeforeBlackout = getRandomHoursOutside(numberOfHours)\n const othersAfterBlackout = getRandomHoursOutside(1)\n\n return {\n before : outdoorWorker ? outdoor : othersBeforeBlackout,\n after : outdoorWorker ? (Math.random() <= 0.25 ? outdoor : othersAfterBlackout) : othersAfterBlackout\n }\n}","pinned":true,"hidden":false,"mode":"ojs"},{"id":1944,"value":"indoorBlackout = {\n const arr = (await FileAttachment('amir_all.csv').csv({typed : true}))\n .map((row, i) => {\n const startDate = new Date(Date.UTC(2011, 7, 1, 1, 0, 0, 0))\n \n const date = new Date(startDate.getTime() + i*60*60*1000)\n return {\n ...row, date\n }\n \n })\n\n const out = {}\n arr.forEach(row => {\n out[String(row.date.getTime())] = row\n })\n return out\n \n}","pinned":true,"hidden":false,"mode":"ojs"},{"id":1605,"value":"indoorNoBlackout = {\n const arr = (await FileAttachment('amir_no_outage.csv').csv({typed : true}))\n .map((row, i) => {\n const startDate = new Date(Date.UTC(2011, 7, 1, 1, 0, 0, 0))\n \n const date = new Date(startDate.getTime() + i*60*60*1000)\n return {\n ...row, date\n }\n \n })\n\n const out = {}\n arr.forEach(row => {\n out[String(row.date.getTime())] = row\n })\n return out\n \n}","pinned":true,"hidden":false,"mode":"ojs"},{"id":1635,"value":"getIETNoBlackout = (row, ap, buildingType, ac, uhi) => {\n\n const entry = indoorNoBlackout[row.time.getTime()]\n \n const acType = {\n 'full' : 'ac100',\n 'inadequate' : 'ac50',\n 'none' : 'noac'\n }[ac]\n \n const key = `${acType}-${buildingType}`\n\n if(ap[row.hour]) {\n const t = entry[key] || 23.333\n return t\n }\n else {\n return row.ta\n }\n \n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":1085,"value":"indoorSeries = {\n return {\n '2' : indoorData2,\n '5' : indoorData5\n }\n}","pinned":true,"hidden":false,"mode":"ojs"},{"id":1021,"value":"$${ moment().year(2011).dayOfYear(nonexDay).format('MMM D') }","pinned":false,"hidden":false,"mode":"md"},{"id":783,"value":"getIndoor = ((day, hour, buildingType, backOnlineAfter) => {\n\n const indoorData = indoorSeries[backOnlineAfter]\n \n if(day === 0 && hour === 0) {\n return indoorData.find(row => row.day === day && row.hour === 1)[buildingType]\n }\n return indoorData.find(row => row.day === day && row.hour === hour)[buildingType]\n \n})","pinned":true,"hidden":false,"mode":"ojs"},{"id":1088,"value":"indoorData2 = (await FileAttachment(\"indoor_amir - 2_days (1).csv\").csv())\n .map((row, i) => {\n\n const d = new Date(row.time.replace('2024', '2011'))\n const startDay = new Date('2011-08-07T00:00Z')\n const day = d.getUTCDay() - startDay.getUTCDay()\n\n const out = {};\n ['new-bad', 'new-good', 'old-bad', 'old-good', 'outdoors'].forEach( k => {\n out[k] = Number(row[k])\n })\n \n return {\n ...out, day, hour : d.getUTCHours()\n }\n })","pinned":true,"hidden":false,"mode":"ojs"},{"id":781,"value":"indoorData5 = (await FileAttachment(\"indoor_amir - 5_days (1)@1.csv\").csv())\n .map((row, i) => {\n\n const d = new Date(row.time.replace('2024', '2011'))\n const startDay = new Date('2011-08-07T00:00Z')\n const day = d.getUTCDay() - startDay.getUTCDay()\n\n const out = {};\n ['new-bad', 'new-good', 'old-bad', 'old-good', 'outdoors'].forEach( k => {\n out[k] = Number(row[k])\n })\n \n return {\n ...out, day, hour : d.getUTCHours()\n }\n })","pinned":true,"hidden":false,"mode":"ojs"},{"id":1939,"value":"getIETBlackout = (row, ap, buildingType, duration, ac, uhi) => {\n const entry = indoorBlackout[row.time.getTime()]\n \n const acType = {\n 'full' : 'ac100',\n 'inadequate' : 'ac50',\n 'none' : 'noac'\n }[ac]\n\n \n \n const key = `${acType}-${duration}-${buildingType}`\n\n if(ap[row.hour]) {\n const t = entry[key] || 23.333\n return t\n }\n else {\n return row.ta\n }\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":238,"value":"reduceArray = (array, segmentSize, sum) => {\n let reducedArray = [];\n for (let i = 0; i < array.length; i += segmentSize) {\n let segment = array.slice(i, i + segmentSize);\n let mean = ss.mean(segment);\n if(sum) {\n reducedArray.push(ss.sum(segment))\n }\n else {\n reducedArray.push(mean)\n }\n }\n return reducedArray;\n};","pinned":true,"hidden":false,"mode":"ojs"},{"id":888,"value":"extendedHours = [ ...heatwave, ...nonExtremeHours ]","pinned":true,"hidden":false,"mode":"ojs"},{"id":886,"value":"nonExtremeHours = airport.filter(row => {\n\n const startDate = new Date('2011-01-01T00:00Z')\n const endDate = new Date('2011-01-01T00:00Z')\n startDate.setUTCDate(startDate.getUTCDate() + nonexDay - 1)\n endDate.setUTCDate(endDate.getUTCDate() + nonexDay + 20 - 1)\n\n return row.time >= startDate && row.time <= endDate\n \n \n}).map( row => {\n return { ...row, hour : row.time.getUTCHours() }\n})","pinned":true,"hidden":false,"mode":"ojs"},{"id":2203,"value":"dots = _.flatten(heatwaveDaysExtended.map((row, i) => {\n\n const rrByGroup = Object.keys(groups).map(k => {\n const arr = groups[k]\n \n const rrMean = ss.mean(arr.map(person => {\n return erCurveExp(models[`${person.outdoorWorker}_${person.buildingType}_${person.ac}`](person.dmatNoBlackout[i]))\n }))\n\n return rrMean\n })\n\n return {\n tavg : row.tavg,\n rr : Object.keys(groups).map((k, i) => {\n return groups[k].length * rrByGroup[i]\n }).reduce(sum, 0)/Object.keys(groups).map(k => groups[k].length).reduce(sum, 0)\n }\n \n \n}))","pinned":false,"hidden":false,"mode":"ojs"},{"id":1929,"value":"riskByVariable = k => {\n return _(sample).groupBy(k)\n .mapValues(arr => {\n return ss.mean(arr.map(row => {\n const model = models[row.outdoorWorker + '_' + row.buildingType + '_' + row.ac]\n return row.dmatBlackout.map(v => erCurveExp(model(v)) - 1).reduce(sum, 0)\n }))\n })\n .valueOf()\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":1489,"value":"viewof erFunction = Inputs.radio([\"Quad\", \"Exp\"], {label: \"ER curve to use\", value: \"Exp\"})","pinned":false,"hidden":false,"mode":"ojs"},{"id":66,"value":"heatwave = airport.filter(row => {\n //return moment(row.time).isAfter(moment('2011-08-07', 'YYYY-MM-DD')) && moment(row.time).isBefore(moment('2011-08-12', 'YYYY-MM-DD'))\n\n return row.time >= new Date('2011-08-07T00:00:00Z') && row.time <= new Date('2011-08-11T23:59:00Z')\n \n}).map( row => {\n return { ...row, hour : row.time.getUTCHours() }\n})","pinned":true,"hidden":false,"mode":"ojs"},{"id":132,"value":"heatwaveDays = reduceArray(heatwave.map(row => row.ta), 24).map((v, i) => {\n return { tavg : v }\n})","pinned":true,"hidden":false,"mode":"ojs"},{"id":182,"value":"testData = d3.range(24, 60, 0.1)\n .map(v => {\n return { 'Airport temp' : v, 'rrQuad' : erCurveQuad(v), 'rrExp' : erCurveExp(v) }\n })","pinned":false,"hidden":false,"mode":"ojs"},{"id":11,"value":"function rollingAverage(arr, n) {\n if (n <= 0) {\n throw new Error(\"The number of values to average over must be greater than zero.\");\n }\n\n let result = [];\n let half = Math.floor(n / 2);\n\n for (let i = 0; i < arr.length; i++) {\n let sum = 0;\n let count = 0;\n\n for (let j = i - half; j <= i + half; j++) {\n if (j >= 0 && j < arr.length) {\n sum += arr[j];\n count++;\n }\n }\n\n result.push(sum / count);\n }\n\n return result;\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":23,"value":"ss = require('simple-statistics')","pinned":false,"hidden":false,"mode":"ojs"},{"id":349,"value":"getLinearModel = data => {\n const {m, b} = ss.linearRegression(data)\n return y => {\n return (y-b)/m\n }\n}","pinned":true,"hidden":false,"mode":"ojs"},{"id":1436,"value":"forR = _(sample).map(row => {\n return row.ietsNoBlackout.map((iet, i) => {\n return {\n age: row.age,\n buildingType : row.buildingType,\n ac : row.ac,\n outdoorWorker : row.outdoorWorker,\n day : i,\n iet,\n outdoorTemp : heatwaveDaysExtended[i].tavg\n }\n })\n \n}) .flatten()\n .valueOf()","pinned":false,"hidden":false,"mode":"ojs"},{"id":76,"value":"// QUADRATIC extrapolation of exposure-response curve. based on Shindell et al\nerCurveQuad = t => {\n const smt = 28.4 // summer mean temperature. OT, below, refers to \"optimal temperature\" (minimum mortality temperature from Gasparrini et al.)\n \n return t <= ot ? 1 : 1-0.0014*(smt-30.9)*((t-ot)**2) + 0.005*(smt-26.7)*(t-ot)\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":2030,"value":"maxDmat = ss.quantile(_.flattenDeep(sample.map(row => [row.dmatBlackout, row.dmatNoBlackout])), 0.98)","pinned":false,"hidden":false,"mode":"ojs"},{"id":1247,"value":"numberWithCommas = x => {\n\treturn x.toString().replace(/\\B(?=(\\d{3})+(?!\\d))/g, \",\")\n}","pinned":true,"hidden":false,"mode":"ojs"},{"id":377,"value":"viewof stdUHI = Inputs.range([0.05, 3], {label: \"Standard deviation of UHI effect\", step: 0.05, value : 0.1})","pinned":false,"hidden":false,"mode":"ojs"},{"id":26,"value":"moment = require('moment')","pinned":false,"hidden":false,"mode":"ojs"},{"id":34,"value":"fToC = t => (t-32)/1.8","pinned":false,"hidden":false,"mode":"ojs"},{"id":32,"value":"cToF = t => 32 + t*1.8","pinned":false,"hidden":false,"mode":"ojs"},{"id":137,"value":"sum = (a, b) => a + b","pinned":false,"hidden":false,"mode":"ojs"},{"id":315,"value":"function normal(mean, stdDev) {\n let u1 = Math.random();\n let u2 = Math.random();\n let z0 = Math.sqrt(-2.0 * Math.log(u1)) * Math.cos(2.0 * Math.PI * u2);\n return z0 * stdDev + mean;\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":460,"value":"round = (value, exp) => {\n\tif (typeof exp === 'undefined' || +exp === 0)\n\t\treturn Math.round(value);\n\n\tvalue = +value;\n\texp = +exp;\n\n\tif (isNaN(value) || !(typeof exp === 'number' && exp % 1 === 0))\n\t\treturn NaN;\n\n\tvalue = value.toString().split('e');\n\tvalue = Math.round(+(value[0] + 'e' + (value[1] ? (+value[1] + exp) : exp)));\n\n\tvalue = value.toString().split('e');\n\treturn +(value[0] + 'e' + (value[1] ? (+value[1] - exp) : -exp));\n}\n","pinned":false,"hidden":false,"mode":"ojs"},{"id":774,"value":"function weightedRandomSelection(values, weights) {\n // if (values.length !== 4 || weights.length !== 4) {\n // throw new Error('You must provide exactly 4 values and 4 weights.');\n // }\n\n const totalWeight = weights.reduce((sum, weight) => sum + weight, 0);\n if (Math.abs(totalWeight - 1) >= 0.00001) {\n throw new Error('The weights must add up to 1.');\n }\n\n const random = Math.random();\n let cumulativeWeight = 0;\n\n for (let i = 0; i < weights.length; i++) {\n cumulativeWeight += weights[i];\n if (random < cumulativeWeight) {\n return values[i];\n }\n }\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":1015,"value":"viewof nonexDay = Inputs.range([213, 240], {label: \"Day for nonextreme heat\", step: 1, value : 224 })","pinned":false,"hidden":false,"mode":"ojs"},{"id":961,"value":"### Data\n\nLoading hourly airport temperature readings from Bush Intercontinental Airport.","pinned":false,"hidden":false,"mode":"md"},{"id":768,"value":"airport = (await FileAttachment(\"airport.csv\").csv())\n .map(row => {\n return {\n ta : Number(row.ta),\n time : new Date(row.time),\n hour : (new Date(row.time)).getUTCHours()\n }\n \n })\n // .map(row => {\n // return { ...row, time : moment(row.time).toDate() }\n // })","pinned":false,"hidden":false,"mode":"ojs"},{"id":1374,"value":"_(airportDailyRolling).find(row => row.date === '09/08/2011')","pinned":true,"hidden":false,"mode":"ojs"},{"id":1450,"value":"ap2011.filter((row, i, arr) => {\n\n const row2 = arr[i + 6]\n if(!row2) { return }\n return row.smoothed >= 31.5 && row2.smoothed < 30\n})","pinned":true,"hidden":false,"mode":"ojs"},{"id":1448,"value":"ap2011 = airportDailyRolling.filter(row => row.date.endsWith('2011'))","pinned":true,"hidden":false,"mode":"ojs"},{"id":1163,"value":"airportDailyRolling = {\n const smoothed = rollingAverage(airportDaily.map(row => row.tavg), 5)\n return airportDaily.map((row, i) => {\n return { ...row, smoothed : smoothed[i] }\n })\n}","pinned":true,"hidden":false,"mode":"ojs"},{"id":2340,"value":"","pinned":true,"hidden":false,"mode":"ojs"},{"id":1151,"value":"airportDaily = _(airport)\n .filter(row => {\n const d = new Date(row.time)\n return d.getUTCMonth() >= 5 && d.getUTCMonth() <= 8 && d.getYear() <= 2023\n })\n .groupBy(row => {\n return (new Date(row.time)).toLocaleDateString('en-GB', {\n day: '2-digit',\n month: '2-digit',\n year: 'numeric'\n })\n })\n .mapValues((arr, dateStr) => {\n return {\n date : dateStr,\n tavg : ss.mean(arr.map(row => row.ta))\n }\n })\n.values()\n.valueOf()","pinned":true,"hidden":false,"mode":"ojs"},{"id":1496,"value":"erCurveExp(32)","pinned":true,"hidden":false,"mode":"ojs"},{"id":1816,"value":"modelInput = {\n const data = (await FileAttachment('hstn.usa8506.csv').csv({typed : true}))\n const modelInput = data.filter(row => row.tmean >= 29.6)\n return modelInput\n}","pinned":true,"hidden":false,"mode":"ojs"},{"id":1467,"value":"erCurveExp = {\n const data = (await FileAttachment('hstn.usa8506.csv').csv({typed : true}))\n const modelInput = data.filter(row => row.tmean >= 29.6)\n const model = ss.linearRegressionLine(ss.linearRegression(modelInput.map(row => [row.tmean, row.logRR])))\n return t => {\n return Math.max(1, Math.E**(model(t)))\n }\n}","pinned":true,"hidden":false,"mode":"ojs"},{"id":910,"value":"viewof stayPut = Inputs.radio([\"Yes\", \"No\"], {label: \"Do people switch to staying inside all day (stay-put scenario)?\", value: \"Yes\"})","pinned":false,"hidden":false,"mode":"ojs"},{"id":177,"value":"viewof ot = Inputs.range([22, 36], {label: \"optimal temp\", step: 0.1, value : 29.6})","pinned":false,"hidden":false,"mode":"ojs"},{"id":1534,"value":"plotData = _.flatten(sample.filter((_, i) => i % 50 === 0)\n .map(row => {\n return row.dmatNoBlackout.map((dmat, i) => {\n return { dmat, oat : heatwaveDaysExtended[i].tavg, hasAc : row.hasAc, outdoorWorker : row.outdoorWorker }\n })\n })).filter(row => row.dmat <= maxDmat)","pinned":false,"hidden":false,"mode":"ojs"},{"id":1500,"value":"test = (await FileAttachment('hstn.usa8506.csv').csv({typed : true})).reverse()","pinned":true,"hidden":false,"mode":"ojs"},{"id":1583,"value":"erCurveExp(40)","pinned":true,"hidden":false,"mode":"ojs"},{"id":2118,"value":"out = {\n return (await FileAttachment('amir_all.csv').csv({typed : true}))\n .map((row, i) => {\n const startDate = new Date(Date.UTC(2011, 7, 1, 1, 0, 0, 0))\n\n const dateStr = row.date.trim().split(/\\s+/g)[1]\n \n const date = new Date(startDate.getTime() + i*60*60*1000)\n return {\n ..._.pick(row, ['outdoor', 'ac100-5day-old-bad', 'ac100-5day-new-good']), date, dateStr\n }\n \n }).slice(100, 330)\n}","pinned":true,"hidden":false,"mode":"ojs"},{"id":2178,"value":"median = round(ss.quantile(rangeOfResults.map(row => row.totalDeaths), 0.5))","pinned":true,"hidden":false,"mode":"ojs"},{"id":2176,"value":"q75 = ss.quantile(rangeOfResults.map(row => row.totalDeaths), 0.75)","pinned":true,"hidden":false,"mode":"ojs"},{"id":2174,"value":"q25 = ss.quantile(rangeOfResults.map(row => row.totalDeaths), 0.25)","pinned":true,"hidden":false,"mode":"ojs"},{"id":2172,"value":"rangeOfResults = FileAttachment(\"rangeOfResults (2).json\").json()","pinned":false,"hidden":false,"mode":"ojs"},{"id":2333,"value":"heatwaveDays.map(row => {\n return (erCurveExp(row.tavg) - 1)*dailyDeathsPer100k/100000*totalPop\n}).reduce(sum, 0)","pinned":true,"hidden":false,"mode":"ojs"},{"id":2336,"value":"dailyDeathsPer100k/100000*totalPop*5","pinned":true,"hidden":false,"mode":"ojs"},{"id":2345,"value":"airport.filter(row => row.time >= new Date('2024-07-08'))","pinned":true,"hidden":false,"mode":"ojs"}],"files":"$W34","stdlib":"1","titling":"auto"}