Every tonne you will ever cut is first a tonne you can name, locate and calculate. This is a hands-on map of Scope 1, Scope 2 and the fifteen doors of Scope 3 — for the people who actually run the factory, the fleet and the purchase orders. Open a node. Open the node inside it. Keep going until the number stops being abstract.你未来减掉的每一吨，都必须先是一吨你能说出名字、找到位置、算得出来的排放。这是一张可以动手点的地图：范围一、范围二，以及范围三的十五扇门——写给真正在开工厂、跑车队、下采购单的人。点开一个节点，再点开里面的节点，一直点到那个数字不再抽象为止。
11.4×Average company's value-chain (Scope 3) emissions versus its own operations — CDP disclosure data.披露企业的价值链（范围三）排放，平均是自身运营排放的 11.4 倍 —— CDP 数据。
15Scope 3 categories. Upstream 1–8, downstream 9–15. Most companies are material in three or four of them.范围三共 15 个类别：上游 1–8，下游 9–15。多数企业真正重大的只有其中三四个。
67%Minimum share of Scope 3 that a near-term SBTi target must cover — the screening line to aim at.SBTi 近期目标必须覆盖的范围三比例下限 —— 摸底盘查要瞄准的那条线。
40%If Scope 3 is more than 40% of Scopes 1+2+3, you cannot call it optional. That is the trigger.若范围三超过范围一+二+三总量的 40%，它就不再是"可选项"。这是触发线。
01
What counts, and whose is it?什么算数？算谁的？
A scope is not a category of pollution — it is a question of control. Who owns the flame, who signs the electricity bill, who is upstream and downstream of you. Get that right and everything else is arithmetic."范围"不是污染的种类，而是控制权的问题：火焰归谁、电费单谁签、谁在你的上游和下游。先把归属分清楚，剩下的都只是算术。
Scope 1Direct直接排放
Sources you own or control. If it burns, reacts or leaks inside your fence line, it is yours.你拥有或控制的排放源。凡是在你围墙之内燃烧、反应、泄漏的，都算你的。
EN / 例
Gas boiler, kiln, oven, backup diesel generator, company trucks and forklifts, refrigerant leaking from a chiller, SF₆ from switchgear, CO₂ released by calcining limestone.燃气锅炉、窑炉、烘箱、柴油备用发电机、自有卡车与叉车、冷水机组制冷剂泄漏、开关柜 SF₆、石灰石煅烧释放的 CO₂。
NOT / 不含
Purchased electricity (Scope 2). A contracted haulier's truck (Scope 3.4/3.9). An employee's own car (3.6/3.7).外购电力（范围二）；外包承运商的卡车（3.4/3.9）；员工自己的车（3.6/3.7）。
Scope 2Purchased energy外购能源
Electricity, steam, heat and cooling you buy. Burned somewhere else, consumed by you. Report both methods — this is not optional.你买进来的电、蒸汽、热、冷。别人烧的，你用的。两种方法都要报告 —— 这不是可选项。
LOCATION
Grid average factor of where you physically sit. Answers: "what does this place emit?"基于地点法：用你所在电网的平均因子。回答的是"这片电网有多脏"。
MARKET
Contractual instruments: PPAs, green tariffs, certificates (I-REC / GO / 绿证), else residual mix. Answers: "what did you choose to buy?"基于市场法：绿电合约 PPA、绿色电价、绿证（I-REC / GO / 中国绿证），无合约则用剩余电力组合。回答的是"你选择买了什么"。
TRAP
Transmission & distribution losses of that electricity are not Scope 2 — they are Scope 3, category 3.这部分电力的输配电损失不属于范围二 —— 它在范围三第 3 类。
Scope 3Value chain价值链
Everything else that happens because your company exists: what you buy, how it travels, how your product is used, how it dies. 15 categories, and for most companies this is 70–95% of the total.因为你的公司存在而发生的其余一切：你买什么、怎么运、产品被怎么用、最后怎么废弃。共 15 类，对多数企业占总量的 70–95%。
RULE
One activity, one place. Your supplier's Scope 1 is your Scope 3.1. Double counting across companies is normal and expected — inside your own inventory it is an error.一项活动只放一处。供应商的范围一就是你的范围三第 1 类。不同企业之间重复计算是正常的；在你自己的清单内重复，就是错误。
EXCLUDEDWhat is deliberately NOT in any scope哪些东西刻意不进任何范围
BIOGENIC
CO₂ from burning biomass/biofuel is reported outside the scopes as a separate line; the CH₄ and N₂O from that same combustion stay inside Scope 1.生物质/生物燃料燃烧产生的 CO₂ 在范围之外单独列报；但同一燃烧过程的 CH₄ 与 N₂O 仍留在范围一内。
OFFSETS
Purchased credits and removals are never subtracted from a scope total. Report gross first, then any credits separately.购买的碳信用与碳移除，永远不能从范围总量中扣减。先报总量，再单列信用。
MONTREAL
CFCs and HCFCs are controlled under the Montreal Protocol, not the Kyoto basket — report separately, still fix the leak.CFCs / HCFCs 属《蒙特利尔议定书》管控，不在京都七气之列 —— 单独报告，但泄漏照样要修。
AVOIDED
"Our product saves customers X tonnes" is an avoided-emissions claim. Useful story, separate page, never inside the inventory."我们的产品帮客户少排 X 吨"属于避免排放，是好故事，但必须单独呈现，绝不进清单。
THE SEVEN GASES — click one七种气体 —— 点开看看
CO₂ · GWP 1
CH₄ + 2O₂ → CO₂ + 2H₂O
Combustion and calcination. 16 g of methane makes 44 g of CO₂ — a factor of 2.75. Boring, abundant, and usually 80–95% of your tonnes.来自燃烧与煅烧。16 克甲烷生成 44 克 CO₂ —— 放大 2.75 倍。平淡、量大，通常占你 80–95% 的吨数。
CH₄ · GWP 27–30
Gas leaks, landfill, wastewater, agriculture. One leaking tonne ≈ 28 tonnes of CO₂e over 100 years. Short-lived, so fixing it pays back fast in real climate terms.天然气泄漏、填埋场、废水、农业。泄漏 1 吨 ≈ 100 年尺度上的 28 吨 CO₂e。寿命短，因此修复它在真实气候意义上回报极快。
Refrigeration, chillers, foam, aerosols. R-404A has GWP 3,922: one kilogram leaked = 3.9 tCO₂e. Switching to R-449A (1,397) or CO₂ (1) is one of the cheapest tonne-cuts on any site.制冷、冷水机组、发泡、气雾剂。R-404A 的 GWP 为 3,922：泄漏 1 公斤 = 3.9 吨 CO₂e。改用 R-449A（1,397）或 CO₂（1），是厂区里最便宜的减吨手段之一。
PFCs · CF₄ ≈ 7,380
Aluminium smelting anode effects, semiconductor etching. Very long-lived — effectively permanent on human timescales.电解铝阳极效应、半导体刻蚀。寿命极长 —— 在人类尺度上几乎是永久的。
SF₆ · 23,500–25,200
High-voltage switchgear insulation. Counted by mass balance: purchases + inventory change − disposal. Nameplate charge × leak rate is the fallback.高压开关柜绝缘气体。用物料平衡核算：采购量 + 库存变化 − 处置量。没有数据时退而用铭牌充装量 × 泄漏率。
NF₃ · ≈ 17,400
Display and semiconductor chamber cleaning. Added to the Kyoto basket in 2013 — check whether your 2010-era template even has a row for it.面板与半导体腔室清洗。2013 年才被纳入京都气体清单 —— 检查一下你那套 2010 年的模板里有没有这一行。
02
Follow the flow. Find the hotspot.跟着流走，找到热点。
Emissions move like material does: in from suppliers, through your site, out to customers. Walk the same route your product takes. Three or four nodes will hold nearly all the tonnes — the rest is rounding.排放像物料一样流动：从供应商流进来，穿过你的厂区，流向客户。沿着产品走的同一条路线走一遍。三四个节点就会装下几乎全部的吨数，其余都是零头。
= typically top-3 by tonnes＝ 通常吨数前三
01 · Upstream01 · 上游
3.1Purchased goods采购商品与服务30–70%
Hides in
Steel, aluminium, resin, textiles, chemicals, packaging, contract manufacturing, cloud services. Your AP ledger is the map.钢、铝、树脂、纺织、化工品、包装、代工、云服务。你的应付账款台账就是地图。
Count as
Best: kg of material × supplier-specific factor. Good: kg × industry average. Start: spend × EEIO factor (kgCO₂e per $).最好：材料公斤数 × 供应商实测因子；较好：公斤数 × 行业平均因子；起步：采购金额 × 投入产出因子（kgCO₂e/元）。
Move
Recycled aluminium ≈ 0.5 vs primary 8–16 kgCO₂e/kg (~95% lower). EAF steel ~0.4–0.7 vs BF-BOF ~2.0–2.3 kgCO₂e/kg. Material choice beats supplier lecturing.再生铝约 0.5，原生铝 8–16 kgCO₂e/kg（低约 95%）。电炉钢约 0.4–0.7，长流程钢约 2.0–2.3 kgCO₂e/kg。选材料，胜过对供应商说教。
3.2Capital goods资本商品2–10%
Hides in
New machines, moulds, vehicles, the new building. Fully counted in the year of acquisition — no depreciation across years.新设备、模具、车辆、新厂房。在购置当年全额计入，不按折旧年限摊。
Move
A capex year can spike your Scope 3 by double digits. Design-life and reuse decisions live here.大额资本开支年份可能让范围三跳升两位数百分比。设备寿命与再利用的决策就在这一类。
3.3Fuel & energy upstream燃料与能源上游2–8%
Hides in
Well-to-tank of your fuels + grid transmission & distribution losses (typically 4–8% of purchased kWh).燃料的"油井到油箱"上游 + 电网输配损失（通常为购电量的 4–8%）。
Count as
Same kWh and litres you already have, multiplied by a WTT factor. Cheapest category to close — one afternoon.用你已有的千瓦时和升数，乘以 WTT 因子即可。最容易补齐的一类 —— 一个下午的事。
Air freight is roughly 20–50× sea per tonne-km. One expedited shipment can outweigh a year of office electricity.按吨公里计，空运约为海运的 20–50 倍。一次加急空运，可能超过办公楼一整年的用电排放。
Small tonnes, enormous visibility. Every employee sees the flight and the bin; nobody sees the kiln.吨数小，能见度极高。每个员工都看得见航班和垃圾桶，却看不见窑炉。
Use it
Report them, but do not let them absorb the engineering budget. Use them for culture; spend capex where the mass is.照报，但别让它们吃掉工程预算。用它们做文化，把资本开支投到质量真正所在的地方。
02 · Our own site02 · 自有厂区
S1Boilers, kilns, ovens锅炉、窑炉、烘箱3–30%
CH₄ + 2O₂ → CO₂ + 2H₂O · 1 m³ gas ≈ 2.0 kgCO₂e
Data
Gas meter, fuel oil deliveries, coal weighbridge, LPG cylinders. Twelve bills, one number.燃气表、燃油入库单、煤的地磅记录、液化气钢瓶。十二张单据，一个数字。
Move
Economiser + burner tuning: 3–8% of fuel. Steam trap survey: 2–5%. Heat recovery on flue gas: 5–15%. Below 150 °C, a heat pump at COP 3 beats a gas boiler wherever the grid is under ~400 gCO₂/kWh.省煤器 + 燃烧器调优：省燃料 3–8%；疏水阀普查：2–5%；烟气余热回收：5–15%。150 ℃ 以下，只要电网低于约 400 gCO₂/kWh，COP 3 的热泵就优于燃气锅炉。
These CO₂ molecules come from the reaction, not the fuel. Clean electricity cannot touch them. Clinker alone emits ≈ 0.54 tCO₂ per tonne, about 60% of cement's footprint.这些 CO₂ 来自反应本身，而不是燃料。再干净的电也消不掉。仅熟料煅烧就排放约 0.54 tCO₂/吨，约占水泥总排放的 60%。
Move
Substitute the material (clinker factor, SCMs), change the route (H₂-DRI, EAF), or capture it. Nothing else works.替换材料（降低熟料系数、掺辅材）、改变路线（氢基直接还原、电炉）、或直接捕集。别无他法。
S1Fugitive: refrigerants, SF₆逸散：制冷剂、SF₆0–10%
Hides in
Chillers, cold rooms, process cooling, AC, switchgear. Almost never metered — found in the maintenance top-up log.冷水机组、冷库、工艺冷却、空调、开关柜。几乎从不计量 —— 数据藏在维修补加记录里。
Count as
kg refrigerant recharged during the year × GWP. That top-up is the leak.全年补加的制冷剂公斤数 × GWP。补加量就是泄漏量。
Move
100 kg of R-404A top-up = 392 tCO₂e. Leak detection + retrofit to R-449A cuts that by ~65%; CO₂ (R-744) by ~99.97%. Days of work, hundreds of tonnes.补加 100 kg R-404A = 392 吨 CO₂e。加装检漏 + 改用 R-449A 可降约 65%，改用 CO₂（R-744）可降约 99.97%。几天的工作量，几百吨。
S2Purchased electricity外购电力5–35%
Split it
Motors & drives ~50–60% of industrial electricity · compressed air ~10% · lighting 5–15% · HVAC 10–20%. Sub-metering turns guesses into decisions.电机与传动约占工业用电 50–60%，压缩空气约 10%，照明 5–15%，暖通 10–20%。装了分表，猜测才会变成决策。
Move
Compressed-air leak repair: 20–30% of air energy. VSDs on pumps/fans: 20–50% of that motor's energy. LED + controls: 40–60% of lighting. Then buy clean supply — a credible PPA can take market-based Scope 2 to near zero.压缩空气堵漏：省该系统 20–30%；泵/风机加变频：省该电机 20–50%；LED＋控制：省照明 40–60%。然后再买绿电 —— 一份可信的 PPA 能把市场法范围二压到接近零。
S1S2Fleet, forklifts, offices车队、叉车、办公楼0.5–4%
Data
Fuel cards (litres, not kilometres — litres are measured, kilometres are estimated). Office electricity from the landlord bill or the sub-meter.加油卡（用升数，不用公里数 —— 升是实测的，公里是估的）。办公用电取自业主账单或分表。
Move
Electrify forklifts first: indoor, short-cycle, high-utilisation. Moves tonnes from Scope 1 into a Scope 2 you can decarbonise with a contract.优先电动化叉车：室内、短循环、高利用率。它把吨数从范围一挪进范围二 —— 而范围二可以用一纸合同去脱碳。
03 · Downstream03 · 下游
3.9Outbound distribution下游运输与配送1–8%
Count as
t·km by mode, plus refrigerated warehousing and last-mile. Chilled distribution carries its own refrigerant leak.按运输方式的吨公里，加上冷库仓储与最后一公里。冷链配送还自带制冷剂泄漏。
Move
Load factor and packaging density are the quiet levers: +10% fill rate ≈ −9% freight emissions per unit.装载率和包装密度是沉默的杠杆：装载率 +10% ≈ 单位货物运输排放 −9%。
3.10Processing of sold products售出产品的加工0–20%
Applies to
Intermediate goods only: resin pellets, sheet steel, chemicals, components. If your customer transforms it, that energy is yours to report.仅适用于中间产品：树脂粒料、钢板、化学品、零部件。只要客户还要再加工，那部分能耗就要由你报告。
3.11Use of sold products售出产品的使用0–90%
Who
Anything that consumes energy or fuel in use: motors, pumps, appliances, vehicles, boilers, electronics. Also anything containing fuel or a GHG that is released.凡是使用中要耗能耗油的产品：电机、水泵、家电、车辆、锅炉、电子产品；以及任何含有燃料或会释放温室气体的产品。
Formula
units × power(kW) × hours/yr × lifetime(yr) × grid EF
Move
A 5% efficiency gain in the product usually cuts more tonnes than everything you can do inside the factory combined. Design is your biggest decarbonisation lever, and it sits with engineering, not with EHS.产品效率提升 5%，减掉的吨数通常比你在厂内能做的所有事加起来还多。设计才是最大的减碳杠杆 —— 而它握在研发手里，不在 EHS 手里。
3.12End of life产品废弃处置0.5–5%
landfilled organics → CH₄ (GWP ≈ 28)
Count as
Product mass by material × regional treatment mix (landfill / incineration / recycling) × factor.产品各材料质量 × 当地处置结构（填埋/焚烧/回收）× 因子。
Move
Mono-material and separable design changes this number and category 3.1 at the same time. One decision, two categories.单一材质、易拆解的设计，能同时改变这一类和第 3.1 类的数字。一个决定，两处受益。
Depends on your consolidation approach (equity share / financial control / operational control). Leased assets are Scope 1&2 under some approaches and Scope 3.8 under others — declare which you use, then be consistent for years.取决于你的合并口径（股权比例 / 财务控制 / 运营控制）。同一处租赁资产，在某些口径下是范围一二，在另一些口径下是范围三第 8 类 —— 先声明口径，然后连续多年保持一致。
Finance
For banks and investors, category 15 is typically >95% of the total. Use the PCAF standard.对银行与投资机构，第 15 类通常占总量 >95%。请使用 PCAF 标准核算。
You cannot manage the tonne you have never located.你从未定位过的那一吨，永远管不住。
03
Fifteen doors, and which to open first十五扇门，先开哪一扇
Every category, one line each, with the honest question: does this one matter for you? Open a door.十五个类别，每个一句话，外加一个诚实的问题：这一类对你真的重要吗？点开一扇门。
01Purchased goods & services采购的商品和服务
Cradle-to-gate of everything you buy that isn't capital. Almost always material. Start with the top 20 suppliers by spend — they are usually 70–80% of the spend.你采购的所有非资本类物品的"摇篮到大门"排放。几乎必然重大。从采购金额前 20 的供应商开始 —— 通常占总采购额的 70–80%。
02Capital goods资本商品
Machines, buildings, vehicles — counted fully in the purchase year. Material if you are expanding capacity.设备、厂房、车辆 —— 购置当年全额计入。若正在扩产，通常重大。
03Fuel & energy upstream燃料与能源相关活动
Well-to-tank + grid T&D losses. Small but nearly free to calculate — no excuse to skip.燃料上游 + 电网输配损失。量不大，但计算几乎不费力 —— 没有理由不算。
04Upstream transport上游运输和配送
Freight you pay for, inbound. Material for distributors, retailers and anyone using air freight.你付费的入厂运输。对经销、零售以及任何用空运的企业都重大。
05Waste in operations运营中产生的废弃物
Usually <1% — unless you send organics to landfill or handle hazardous waste. Report it; don't over-invest in it.通常 <1% —— 除非你把有机废物填埋或处置危废。照报，但别过度投入。
06Business travel商务差旅
Material only for consulting, audit, sales-heavy firms. High visibility, low mass for manufacturers.仅对咨询、审计、销售驱动型企业重大。对制造业而言：曝光度高，质量低。
07Employee commuting员工通勤
Survey-based, plus home working. Great engagement tool, rarely a lever. Include remote-work energy since 2022 guidance.基于问卷，含居家办公用能。是很好的员工参与工具，但很少是减排杠杆。2022 年后的指南要求纳入居家办公。
08Upstream leased assets上游租赁资产
Only if not already in your Scope 1/2 under your chosen consolidation approach. Check for double counting.仅在你所选合并口径下未计入范围一二时才填。注意别重复计算。
09Downstream transport下游运输和配送
Distribution you do not pay for, plus retail and warehousing. Material for FMCG and cold chain.不由你付费的配送，加上零售与仓储环节。对快消和冷链重大。
10Processing of sold products售出中间产品的加工
Only for intermediate goods. If you sell finished products, this is zero — say so, don't leave it blank.仅适用于中间产品。若你卖的是成品，此项为零 —— 请写"零"，不要留空。
11Use of sold products售出产品的使用
The giant for equipment, vehicles, appliances, electronics: 60–90% of total. Direct use-phase energy is mandatory; indirect is optional but disclose the choice.对设备、车辆、家电、电子产品而言是巨无霸：占总量 60–90%。直接使用阶段能耗为必报，间接为选报，但要披露你的选择。
12End-of-life treatment售出产品的废弃处置
Mass × regional treatment mix. Small tonnes, big design signal — and increasingly a regulatory one.质量 × 区域处置结构。吨数不大，但对设计的信号很强 —— 而且越来越受法规关注。
13Downstream leased assets下游租赁资产
You own it, someone else operates it. Material for property owners and equipment lessors.你拥有、别人运营的资产。对业主和设备出租方重大。
14Franchises特许经营
Franchisee operations. Material for food, retail and hospitality brands — often larger than the parent's own footprint.加盟商的运营排放。对餐饮、零售、酒店品牌重大 —— 常常大于总部自身的足迹。
15Investments投资
Financed emissions. For a bank, this is the footprint — >95%. Use PCAF attribution factors.投融资排放。对银行来说，这就是它的足迹 —— >95%。使用 PCAF 的归因因子。
04
Calculate it, one step at a time一步一步，把它算出来
Four real calculations, revealed step by step. Click Next step and watch a bill turn into a tonne. Every emission factor is a placeholder for the one from your country and your year — the sources are in section 08.四个真实算例，一步一步展开。点击下一步，看一张账单如何变成一吨碳。所有排放因子都只是占位 —— 请替换成你所在国家、你所报告年份的因子，来源见第 08 节。
Scope 1The gas boiler燃气锅炉
stationary combustion · one site · one year固定源燃烧 · 一个厂区 · 一个年度
01
Find the meter, not the estimate.找计量表，不要找估算值。
Twelve gas invoices for one boiler house: 100,000 m³ of natural gas in the reporting year.一个锅炉房、十二张燃气发票：报告年度共 100,000 m³ 天然气。
02
Check the unit and the basis.核对单位与热值口径。
m³ or kWh? Gross (HHV) or net (NCV)? A factor on the wrong basis is a silent 10% error.单位是 m³ 还是 kWh？高位热值还是低位热值？口径不匹配，会带来悄无声息的约 10% 误差。
03
Pick the factor.选取排放因子。
2.02 kgCO₂e per m³ (UK DEFRA-style, gross CV). It already contains CO₂ plus small amounts of CH₄ and N₂O converted with GWP₁₀₀.2.02 kgCO₂e/m³（英国 DEFRA 口径，高位热值）。该因子已包含 CO₂ 以及按 GWP₁₀₀ 折算的少量 CH₄ 与 N₂O。
≈ 1,055 MWh of gas → ≈ 0.19 tCO₂e per MWh. Compare with last year, and with tonnes per unit of product. If it moved >15% without a reason, the data is wrong, not the world.≈ 1,055 MWh 天然气 → 约 0.19 tCO₂e/MWh。与去年、与单位产品排放强度对比。若无缘由波动 >15%，那是数据错了，不是世界变了。
06
Where the tonne actually came from.这一吨究竟从哪来。
CH₄ + 2O₂ → CO₂ + 2H₂O | 16 g CH₄ → 44 g CO₂ (× 2.75)
Burn 1 kg of methane, create 2.75 kg of CO₂. The carbon was never destroyed — it changed address.烧掉 1 公斤甲烷，生成 2.75 公斤 CO₂。碳从未消失，只是换了个住址。
Scope 2One meter, two answers一块电表，两个答案
location-based vs market-based · both must be reported基于地点法 vs 基于市场法 · 两者都必须报告
01
The activity data.活动数据。
8,000,000 kWh purchased in the year, from the utility invoices.全年外购电量 8,000,000 kWh，取自电费发票。
02
Location-based: where you physically are.地点法：你实际在哪。
8,000,000 kWh × 0.55 kgCO₂e/kWh = 4,400 tCO₂e
Use the published regional/national grid average for the reporting year (eGRID, IEA, or the national grid factor).使用报告年度公布的区域/全国电网平均因子（eGRID、IEA 或本国电网因子）。
03
Market-based: what you contracted for.市场法：你签了什么合同。
3,000,000 kWh covered by a PPA with retired certificates → 0 gCO₂e/kWh. The remaining 5,000,000 kWh take the residual mix, which is dirtier than the grid average: 0.62.其中 3,000,000 kWh 有绿电合约并注销了绿证 → 0 gCO₂e/kWh；剩余 5,000,000 kWh 使用剩余电力组合因子（比电网平均更脏）：0.62。
Location-based 4,400 t · Market-based 3,100 t. The gap of 1,300 t is the value of your contracts — a 30% cut on paper that only becomes real when the certificates are additional, local and time-matched.地点法 4,400 吨，市场法 3,100 吨。1,300 吨的差额就是你合同的价值 —— 纸面上减 30%，只有当绿证具备额外性、同区域、同时段匹配时，才是真实的。
06
Don't forget the tail.别忘了尾巴。
Grid T&D losses on these 8 GWh (≈4–8%) belong in Scope 3.3, not here. One more line, five minutes.这 8 GWh 的电网输配损失（约 4–8%）属于范围三第 3 类，不在这里。多一行，五分钟。
Scope 3.1Climbing the data-quality ladder爬上数据质量的阶梯
same purchase · three methods · three different numbers同一笔采购 · 三种方法 · 三个不同的数
01
Start where the data already exists: spend.从已有的数据开始：金额。
Steel components purchased last year: $4,000,000. Nobody weighed anything, but finance has every invoice.去年采购钢制零部件 400 万美元。没人称过重量，但财务有每一张发票。
02
Spend-based (screening).支出法（摸底）。
$4,000,000 × 1.05 kgCO₂e/$ = 4,200 tCO₂e
EEIO factors (e.g. EPA supply-chain factors). Accuracy ±50%, but it tells you the order of magnitude today. That is enough to prioritise.采用投入产出因子（如 EPA 供应链因子）。精度约 ±50%，但今天就能告诉你数量级。用于排序，足够了。
03
Activity-based (better).活动数据法（更好）。
1,800 t steel × 2.10 kgCO₂e/kg = 3,780 tCO₂e
Now the number responds to engineering — lighter parts, better yield, different alloy — instead of to price inflation.此时这个数会随工程改进而变化 —— 轻量化、提高成材率、换合金 —— 而不再随物价波动。
04
Supplier-specific (best).供应商实测法（最好）。
1,800 t × 0.92 kgCO₂e/kg (EAF, verified) = 1,656 tCO₂e
The same steel, the same tonnage — 61% fewer emissions, purely because you asked which furnace it came from.同样的钢、同样的吨位 —— 排放少 61%，仅仅因为你问了一句"这是哪种炉子炼的"。
05
Read the difference correctly.正确解读这个差异。
You did not "reduce" 2,544 tonnes — you discovered them. Restate the baseline, document the method change, and never mix methods inside one category in one year.你并没有"减少"2,544 吨 —— 你只是发现了它们。请重述基准年、记录方法变更，并且绝不在同一年、同一类别内混用两种方法。
Scope 3.11The number that dwarfs the factory让工厂显得渺小的那个数
use of sold products · an industrial motor maker售出产品的使用 · 一家工业电机厂
Apply the grid factor of the markets you sell into.套用销售市场的电网因子。
30,000 kWh × 0.45 kgCO₂e/kWh = 13.5 tCO₂e per motor
04
Multiply by the fleet.乘以全部销量。
13.5 t × 200,000 = 2,700,000 tCO₂e
Two hundred times a typical factory's Scope 1+2. This single line usually is the company's footprint.约为典型工厂范围一+二的两百倍。这一行，往往就是这家公司的碳足迹。
05
Now feel the lever.现在感受一下杠杆。
A 5% efficiency improvement (IE3 → IE4 class) removes 135,000 tCO₂e from one year's sales. No solar array on your roof will ever do that. Decarbonisation, for a product company, is a design department activity.效率提升 5%（IE3 → IE4）就能从一年的销量中抹去 135,000 吨 CO₂e。屋顶上的光伏永远做不到这一点。对产品型企业而言，减碳是研发部门的工作。
05
Must include, or safe to wait?必须算，还是可以先放一放？
Tick what you have actually screened — not what you intend to. The bar shows roughly how much of a typical discrete manufacturer's footprint you can now speak for.勾选你已经摸过底的项，而不是打算摸的。进度条显示：以一家典型离散制造企业为参照，你现在能对多少比例的足迹说得清楚。
Tick the boxes above to see how much of the footprint you can already speak for.勾选上方选项，看看你已经能对多少比例的足迹说得清楚。
MUSTWhen a category is not optional什么情况下这一类不能不算
SIZE
It is a significant share of your total, or your screening says it could be. Threshold in practice: anything plausibly >5%.它占总量比例显著，或摸底显示可能显著。实务门槛：任何可能 >5% 的项。
INFLUENCE
You can actually change it — through specification, supplier choice or product design.你确实能改变它 —— 通过规格书、供应商选择或产品设计。
RISK
Carbon price, CBAM, customer contract clauses, or regulation touches it.涉及碳价、CBAM、客户合同条款或法规要求。
STAKEHOLDER
Customers, investors or your own staff already ask about it.客户、投资人或员工已经在问这件事。
OUTSOURCED
It used to be your Scope 1 and you outsourced it. Outsourcing is not abatement.它曾经是你的范围一，后来外包了。外包不是减排。
THE LINE
SBTi: if Scope 3 > 40% of total, a Scope 3 target is required, and it must cover ≥ 67% of Scope 3.SBTi 规则：若范围三 > 总量的 40%，必须设定范围三目标，且目标须覆盖范围三的 ≥ 67%。
LATERWhat can honestly wait哪些确实可以晚一点
RULE
Categories that are individually <1% and collectively <5% of the total, where you have no operational influence.单项 <1% 且 合计 <5%，同时你没有运营影响力的类别。
STILL
Disclose the exclusion and the reason. An honest "screened, 0.4%, excluded" is credible. A blank cell is not.仍需披露"排除"及理由。写"已摸底、占 0.4%、暂不细算"是可信的；留空格不是。
TYPICAL
For a factory: commuting, office waste, upstream leased assets, franchises. Screen once a year, then move on.对工厂而言常见的有：通勤、办公废弃物、上游租赁资产、特许经营。每年摸一次底即可，然后往前走。
NEVER
Never exclude something just because the data is hard. Hard data + big tonnes = that is your project.绝不要因为"数据难拿"而排除。数据难 + 吨数大 = 那正是你的项目。
06
What "good" looks like"做得好"长什么样
First: where your tonnes should be, by sector. Then: how mature your accounting is, from starting to outstanding.先看：按行业，你的吨数应该集中在哪里。再看：你的核算成熟度，从起步到卓越。
Sector行业Dominant主导项Typical share & what it means典型占比与含义
Cement, lime, steel水泥、石灰、钢铁Scope 1 · 60–90%Process chemistry dominates. Clinker ≈ 0.54 tCO₂/t. Only material substitution, route change or capture moves it.工艺化学反应主导。熟料约 0.54 tCO₂/吨。只有换材料、换工艺路线或捕集才有用。
Machinery, appliances, vehicles, electronics机械、家电、汽车、电子Cat 11 · 60–90%The use phase is the footprint. Your R&D roadmap is your climate plan.使用阶段就是足迹本身。你的研发路线图就是你的气候方案。
Apparel, food, FMCG服装、食品、快消Cat 1 · 60–85%Raw materials and agriculture. Supplier data and material choice beat everything on site.原料与农业阶段。供应商数据与选材，胜过厂内所有努力。
Retail, distribution零售、分销Cat 1 + 4/9Scope 1+2 is often <5% of total. Assortment decisions carry the tonnes.范围一+二常 <5%。真正扛吨数的是选品决策。
Software, services软件、服务业Cat 1 + Scope 2Cloud, purchased services, offices. Scope 2 10–30% — clean power contracts do most of the work.云服务、外购服务、办公。范围二占 10–30% —— 绿电合约能解决大半。
Banks, investors银行、投资机构Cat 15 · >95%Financed emissions. Everything else is a rounding error. Use PCAF.投融资排放。其余全是零头。请使用 PCAF。
Cadence节奏One spreadsheet a year一年一张表Quarterly review with operations与运营部门季度复盘Monthly hotspot dashboard owned by plant managers月度热点看板，由厂长负责
07
If you do this, you get this做这件事，能拿到这个数
Ranked by tonnes per unit of effort. Percentages are typical industry ranges — use them to argue for the study, not to report the result.按"每分力气减多少吨"排序。百分比为行业典型区间 —— 用它去争取立项，不要用它去报告结果。
Lever措施Typical cut典型降幅Where it lands · effort影响哪一块 · 投入
Design the product 10% more efficient把产品效率提升 10%−10% of Cat 11的第 11 类Scope 3.11 · years, but the largest absolute tonnage for equipment makers范围三 3.11 · 以年计，但对设备制造商是绝对吨数最大的一项
Contract clean electricity (PPA / green tariff, certificates retired)签订绿电（PPA / 绿色电价，绿证注销）up to −100% market-based S2市场法范围二最高 −100%Scope 2 · months; typically 5–30% of a manufacturer's total范围二 · 以月计；通常占制造企业总量的 5–30%
Heat: recovery, tuning, then heat pumps below 150 °C热能：余热回收、燃烧调优，150℃ 以下改热泵−30 → −70% of that heat该部分热能Scope 1 → 2 · quarters to years; only wins where the grid is <400 gCO₂/kWh范围一 → 二 · 季度到年；仅在电网 <400 gCO₂/kWh 时划算
Refrigerant: fix leaks, retrofit to low-GWP制冷剂：堵漏、改用低 GWP 工质up to −99% of that leak该泄漏项最高 −99%Scope 1 · weeks. Cheapest tonne on most sites范围一 · 以周计。多数厂区最便宜的一吨
Compressed air leaks + VSDs on pumps and fans压缩空气堵漏 + 泵与风机变频−2 → −8% of site electricity厂区总用电Scope 2 · weeks; payback usually under 2 years范围二 · 以周计；回收期通常 <2 年
Freight: eliminate air, shift road → rail/sea, raise fill rate运输：取消空运，公路转铁路/海运，提高装载率−20 → −60% of Cat 4+9的第 4+9 类Scope 3 · months; needs planning discipline, not investment范围三 · 以月计；靠计划纪律，不靠投资
Supplier engagement: primary data + their own targets供应商协同：一手数据 + 供应商自身目标−10 → −30% of Cat 1 over 5 yrs的第 1 类，5 年内Scope 3.1 · years; put carbon into the sourcing scorecard or nothing happens范围三 3.1 · 以年计；不把碳写进采购评分卡，就什么都不会发生
LED + lighting controlsLED + 照明控制−40 → −60% of lighting照明用电Scope 2 · weeks; ≈5–15% of site electricity. Good first win, small mass范围二 · 以周计；照明约占厂区用电 5–15%。开局好项目，但体量小
08
Where the numbers come from这些数字从哪来
Never publish a factor you cannot cite. Replace every number on this page with the one for your country and your reporting year.引用不出来源的因子，就不要发布。请把本页每一个数字替换成你所在国家、你报告年份的版本。