The question, kept as it was asked原样保留的那个问题
Five years of work went into the things that could be metered: renewable and carbon-free electricity, energy efficiency, low-carbon and recycled materials, waste, water recycling. Those are on the way. Land is not. Forest, wetland, soil and watershed are still "good to do" — not scoped, not defined, not in any supplier ask that has teeth.
And there is an honest reason for the hesitation. A forest is a system, not a contract. Money in does not equal carbon out. Some forests are not gaining at all; some are losing. So the topic circles the conference room, year after year, and never lands as an action.
This page does not close the question. It refuses to lose it.
过去五年，我们把力气放在"能计量"的事情上：绿电与无碳电力、能效、低碳与再生材料、废弃物、水回用。这些都在推进。土地没有。森林、湿地、土壤、流域，仍然停留在"做了很好"，没有界定范围，没有定义，也没有真正写进供应商要求。
犹豫是有真实原因的：森林是一个系统，不是一份合同。投入并不直接等于碳汇产出；有些森林并没有增长，甚至在退化。于是这个议题在会议室里绕了一年又一年，始终没有落成行动。
这一页不打算把问题关上，而是不让它被丢掉。
“You cannot get nature to sign a contract with you — so much effort in, so much carbon straight back. Is that the reason it stays in the conference room, year after year, and never becomes action?”
「你没办法让自然跟你签一份合同：投入多少，就直接还回来多少碳。这是不是就成了它年复一年留在会议室、始终没有变成行动的理由？」 The question this station was built around · 本站点的出发点
Small issue, or big issue? Let the data talk小问题，还是大问题？让数据说话
Ten numbers, each with a source at the bottom of this page. Read them in one breath and the "good to do" framing collapses.
十个数字，来源都列在本页末尾。一口气读完，"做了很好"这个定位就站不住了。
of value generation — over half of global GDP — is moderately or highly dependent on nature and its services.
约44万亿美元的经济价值创造，超过全球GDP的一半，中度或高度依赖自然及其服务。
WEF, Nature Risk Rising [1]deforestation, and 4.7 Mha/yr net forest loss, over 2015–2020 — a forest area larger than South Korea gone every two years.
2015—2020年间年均森林砍伐1,000万公顷，净损失470万公顷——大约每两年消失一个比韩国更大的森林面积。
FAO, Global Forest Resources Assessment [2]of global deforestation is driven by agricultural expansion — cropland ~50%, livestock grazing ~39%. This is a supply chain issue, not a conservation hobby.
全球约90%的森林砍伐由农业扩张驱动：耕地约50%，放牧约39%。这是供应链议题，不是环保爱好。
FAO remote sensing survey, 2022 [3]of primary tropical forest lost in 2024 — roughly double 2023 and the worst year on record, with fire overtaking agriculture as the leading cause.
2024年热带原始林损失约670万公顷，约为2023年的两倍，为有记录以来最严重的一年，火灾首次超过农业成为首要原因。
WRI / Global Forest Watch [4]of the cost-effective CO₂ mitigation needed through 2030 for a 2 °C pathway can come from natural climate solutions — up to 23.8 GtCO₂e/yr of potential.
在2℃路径下，到2030年所需的成本有效减排中，最高约37%可来自自然气候解决方案，潜力可达每年238亿吨CO₂e。
Griscom et al., PNAS 2017 [5]finance flowing to nature-based solutions each year, against roughly US$7 trillion a year of nature-negative public and private flows. A 35:1 gap.
每年流入自然解决方案的资金约2,000亿美元，而每年对自然产生负面影响的公共与私人资金约7万亿美元，差距约35倍。
UNEP, State of Finance for Nature [6]of Amazon forest loss in 2005–2015 was linked to mining — with impacts reaching up to 70 km beyond the lease boundary. Your mine's footprint is not its fence line.
2005—2015年亚马逊森林损失中约9%与采矿相关，影响可延伸至矿权边界之外70公里。矿山的足迹远不止那道围栏。
Sonter et al., Nature Communications 2017 [7]— the share of credits from 18 examined REDD+ projects that could be linked to additional emission reductions. Offsets are not a substitute for supply chain work.
在被评估的18个REDD+项目中，仅约6.1%的信用额可与真实的额外减排挂钩。抵消不能替代供应链本身的工作。
West et al., Science 2023 [8]is the minimum maximum fine under the EU Deforestation Regulation — of your annual EU-wide turnover, plus goods confiscation and market exclusion.
欧盟零毁林法规下的罚款上限不低于企业在欧盟年营业额的4%，另可没收货物并禁止进入市场。
Regulation (EU) 2023/1115 [9]The cut-off date. Land cleared after it makes the product ineligible for the EU market — no matter how good this year's certificate looks.
分界日。在此日期之后被清除的土地，其产出无法进入欧盟市场——无论当年的认证证书多漂亮。
EUDR cut-off · AFi cut-off dates [9][17]of degraded ecosystems under effective restoration by 2030, and all large companies disclosing nature impacts — the two GBF targets that will shape every future customer questionnaire.
到2030年，30%退化生态系统进入有效恢复，且所有大型企业披露其自然影响——昆明-蒙特利尔框架的这两项目标，将决定未来客户问卷的样子。
Kunming-Montreal GBF, Targets 2 & 15 [10]A factory's water risk is not measured in cubic metres. It is measured in the basin it sits in — where 25% of the world's population already faces extremely high water stress.
工厂的水风险不是用立方米衡量的，而是用它所在的流域衡量——全球已有约25%的人口生活在极高水压力地区。
WRI Aqueduct [18]The rules are arriving. Land is becoming market access规则正在落地：土地正在变成市场准入条件
If your product enters the EU or North America, the question is no longer "is it good to do". It is: can you show, plot by plot, that your supply chain did not convert a forest? Below is the high-level map — check dates at the source links, several are still moving.
如果产品进入欧盟或北美市场，问题已经不是"做了很好"，而是：你能否逐块地证明供应链没有毁林？下面是高层级地图——具体时间请以来源链接为准，部分仍在调整。
EU Deforestation Regulation — (EU) 2023/1115
Cattle, cocoa, coffee, oil palm, rubber, soy, wood and a long list of derived products. You must collect geolocation coordinates (polygons) of every plot of production, prove legality in the country of origin, file a due diligence statement, and show the land was not deforested after 31 Dec 2020. Operators are liable for their whole upstream chain.
牛、可可、咖啡、油棕、橡胶、大豆、木材及一长串衍生品。必须收集每一块生产地块的地理坐标（多边形），证明原产国合法性，提交尽职调查声明，并证明该土地在2020年12月31日之后未发生毁林。责任覆盖整条上游链。
Application phased from end-2025 for large/medium operators, 2026 for small — simplification still under discussion [9]CSRD & ESRS E4 / E3
Under double materiality you disclose land-use change, sites located in or near biodiversity-sensitive areas, ecosystem condition, and water withdrawal/consumption in water-stressed areas — with transition-plan logic, not a paragraph of intent.
在双重重要性下，需披露土地利用变化、位于或邻近生物多样性敏感区的场址、生态系统状况，以及在水压力地区的取水与耗水——需要转型计划的逻辑，而不是一段意愿声明。
Phased in; scope and timing revised by the 2025 omnibus package [11][12]Batteries Regulation — (EU) 2023/1542
Battery passport plus a supply chain due diligence policy covering cobalt, natural graphite, lithium and nickel — explicitly including environmental risk categories: land use, water, biodiversity, hazardous substances. This is the first EU product law that names land use in a minerals chain.
电池护照，加上覆盖钴、天然石墨、锂、镍的供应链尽职调查政策——明确纳入环境风险类别：土地利用、水、生物多样性、有害物质。这是欧盟首个在矿产链中点名"土地利用"的产品法规。
Due diligence obligations phasing in from 2025–2027; verify current dates [13]Corporate Sustainability Due Diligence Directive
Risk-based environmental and human rights due diligence across the chain of activities, with remediation duties. Scope and timing were amended in 2025 — treat it as direction of travel, not a fixed deadline.
对整条经营活动链条开展基于风险的环境与人权尽职调查，并承担补救义务。范围与时间在2025年被修订——把它当作方向，而非固定截止日。
Transposition timeline under revision [12]Lacey Act (2008 plant amendment)
It is already illegal to import plant or wood products harvested in violation of foreign law, with declaration requirements and strict liability — this is the North American deforestation hook that exists today, and it has been enforced with criminal penalties.
进口违反外国法律采伐的植物或木制品在美国已属违法，须申报并承担严格责任——这是北美当下已经存在的"毁林抓手"，并已有刑事处罚案例。
In force · declarations required for listed products [14]FOREST Act (proposed) & California SB 253 / SB 261
The FOREST Act would extend an illegal-deforestation import ban to cattle, soy, palm oil, cocoa, rubber and wood pulp — it is not law. California's climate disclosure laws already pull large companies into scope 3 and climate-risk reporting, which is where land risk surfaces first in the US.
FOREST法案拟将"非法毁林产品"进口禁令扩展至牛、大豆、棕榈油、可可、橡胶与木浆——目前尚未成为法律。加州气候披露法已把大型企业纳入范围三与气候风险报告，这通常是土地风险在美国最先显形的地方。
Bill status and California rulemaking both ongoing — check before quoting [15][16]SBTi FLAG + SBTN land & freshwater targets
If more than 20% of your emissions come from forest, land and agriculture, SBTi requires a FLAG target plus a public no-deforestation commitment with a cut-off date no later than 2025. SBTN goes further: no conversion of natural ecosystems, landscape engagement, and validated freshwater targets by basin — the first corporate targets have already been validated.
若企业超过20%的排放来自森林、土地与农业，SBTi要求设定FLAG目标，并公开承诺不毁林，分界日不晚于2025年。SBTN更进一步：不转换自然生态系统、参与所在景观治理、按流域设定并验证淡水目标——首批企业目标已通过验证。
Available now · first SBTN validations completed [19][20]IFC Performance Standard 6 & the Equator Principles
The closest thing to a hectare rule that already applies: mitigation hierarchy in order, no net loss of natural habitat, and net gain in critical habitat. If a new plant is project-financed, this is the standard your own capex is measured against — borrow its logic for sites that are not.
目前已经适用、最接近"公顷规则"的标准：按顺序执行缓解层级，自然栖息地无净损失，关键栖息地净增益。如果新厂使用项目融资，这就是衡量你自己资本支出的尺子——不使用融资的场址，也可以借用同样的逻辑。
In force for IFC/Equator-financed projects [21]TNFD (LEAP), GRI 101 Biodiversity 2024, CDP Forests & Water
All three now ask for location: which sites, which basins, which ecosystems, which upstream commodities. Once your customer adopts any of them, your site coordinates stop being optional.
三者现在都要求位置信息：哪些场址、哪些流域、哪些生态系统、哪些上游商品。一旦客户采用其中任何一个，场址坐标就不再是可选项。
Voluntary today, contractual tomorrow [22][23][24]The hectare question: a back-of-envelope you can run this week公顷问题：本周就能算起来的一把粗尺
The question was simple: if a factory occupies X hectares, and before that it was something else — how much land should be invested back, and where? There is no ISO number for this. But there is enough public data to make an honest first estimate, and an estimate you can defend is what unlocks a budget.
原始问题很朴素：如果一座工厂占用了X公顷土地，而它原本是别的东西——那么应该"还回去"多少土地，还在哪里？这件事没有ISO编号。但公开数据已经足够做出一个诚实的初估——而一个站得住的初估，正是打开预算的钥匙。
Scoping formula · 界定用的粗算式
- Aconverted — hectares of natural ecosystem on your site converted since the cut-off year (use 31 Dec 2020 to align with EUDR; for legacy sites, use the construction year and say so). 自分界年以来，场址上被转换的自然生态系统面积；与EUDR对齐可用2020年12月31日，老厂可用建厂年并明确说明。
- R — replacement ratio: 1× if the land was already cropland or brownfield · 2× natural forest, grassland or wetland · 3×+ peatland, mangrove, primary forest or IFC critical habitat. 补偿倍率：原为耕地或棕地取1倍；天然林、草地、湿地取2倍；泥炭地、红树林、原始林或IFC关键栖息地取3倍以上。
- Aimpervious — roofs plus paving, in hectares. This is the water half of the answer: rain that used to infiltrate now leaves as runoff. 屋面加硬化铺装面积（公顷）。这是答案里的"水"那一半：原本下渗的雨水，现在成了径流。
- W — watershed factor from your basin's Aqueduct water-stress score: 0.1 low · 0.3 high · 0.5 extremely high. 流域系数，取自Aqueduct水压力评分：低压力0.1，高压力0.3，极高压力0.5。
This is not an accounting standard and should never be reported as one. It is a scoping tool — a way to turn "good to do" into a number, a place and a line in next year's budget. Then replace it with SBTN, IFC PS6 or a landscape programme's own method. 这不是核算标准，也不应作为核算结果对外披露。它是界定工具：把"做了很好"变成一个数字、一个地点、和明年预算里的一行；之后再用SBTN、IFC PS6或所在景观项目的方法替换它。
Draw the fence line, then split it
Total site area, broken into built · paved · landscaped · retained natural. Most factories already have this in the land certificate and the site drawing; nobody has ever added it up as hectares.
先画出围墙，再拆分：总用地面积按建筑物、硬化地面、绿化、保留的自然区域分类。多数工厂在土地证和总平面图里已有这些数据，只是从没有人把它加总成"公顷"。
Ask what was there before — with images, not memory
Pull historic land cover for the site: ESA WorldCover (10 m), Google Earth historical imagery, Global Forest Watch tree-cover-loss layers, national land-use maps. Classify the "before" state. One afternoon per site, and it is free.
用影像而不是回忆确认"之前是什么"：ESA WorldCover（10米）、Google Earth历史影像、Global Forest Watch林损图层、国家土地利用图。判定转换前的地类。每个场址大约半天，而且免费。
Score the place, not only the hectares
Ten hectares of degraded cropland and ten hectares of peat swamp are not the same liability. Screen every site through IBAT (protected areas, Key Biodiversity Areas), WRI Aqueduct (basin stress) and the WWF Risk Filter. Rank sites — most companies find that fewer than 10% of sites carry most of the risk.
重要的是"地点"，不只是"公顷数"。十公顷退化耕地与十公顷泥炭沼泽，风险完全不同。用IBAT（保护区、关键生物多样性区域）、WRI Aqueduct（流域压力）和WWF风险筛选器逐一筛查并排序——多数企业会发现，不到10%的场址承担了绝大部分风险。
Run the mitigation hierarchy in order
Avoid (do not site the next plant on natural habitat) → minimise (footprint, paving, lighting, noise, water intake) → restore on site (permeable surfaces, native planting, retention wetland) → and only then offset elsewhere. Starting at step four is the single most common failure.
严格按缓解层级顺序执行：避免（下一座工厂不选址在自然栖息地）→ 减量（占地、硬化、照明、噪声、取水）→ 就地恢复（透水铺装、本地物种绿化、滞留湿地）→ 最后才是异地补偿。直接从第四步开始，是最常见的失败。
Put the payback in the same ecoregion and the same sub-basin
Like for like: same ecosystem type, same ecoregion, ideally within ~50 km or the same sub-basin, and connected to existing habitat rather than isolated. For water, replenish at least 100% of consumptive use in high-stress basins. A plantation in another country is not a payback.
在同一生态区、同一子流域内偿还：同类生态系统、同一生态区，最好在约50公里内或同一子流域，并与现有栖息地连通而非孤立。水方面，在高压力流域至少回补100%的耗水量。在另一个国家种一片人工林，不算偿还。
Make it durable, monitored, and keep the claims separate
Budget 20–30 years of management, not one planting day. Secure it legally (easement, covenant, protected-area partnership, community agreement with FPIC). Monitor survival and canopy annually by satellite plus ground plots. And do not also book those hectares as carbon offsets — one hectare, one claim.
要有持久性、可监测，并且口径分开：预算覆盖20—30年的管护，而不是一天的植树活动；用法律工具固定下来（地役权、保护性契约、保护区合作、经自由知情同意的社区协议）；每年用卫星加地面样地监测存活率与郁闭度。同时不要把这些公顷再计为碳抵消——一公顷，只做一次主张。
Where it usually breaks通常是在哪里失败的
A commitment with no cut-off date
"Zero deforestation" without a date and a defined ecosystem scope is unauditable. AFi exists precisely to fix this: define commodity, cut-off date, ecosystem coverage, and the unit of production.
没有分界日、没有界定生态系统范围的"零毁林"承诺，无法审计。Accountability Framework正是为解决这一点而存在：明确商品、分界日、覆盖的生态系统类型和生产单元。
Tier-1 vision
The land impact is almost never at your tier 1. It is four to six tiers away — at the farm, the plantation, the smelter, the mine. A supplier list that stops at the assembler tells you nothing about land.
土地影响几乎从不发生在一级供应商。它在四到六级之外：农场、种植园、冶炼厂、矿山。一份停在组装厂的供应商清单，关于土地什么也说明不了。
Certificates mistaken for traceability
Mass-balance and credit-based certification proves a system, not a plot. Regulators now ask for coordinates. Leading buyers verify with satellite monitoring on top of certificates, not instead of coordinates.
质量平衡与信用制认证证明的是体系，不是地块。监管方现在要坐标。领先买家在认证之外叠加卫星监测，而不是用认证替代坐标。
Cheap credits standing in for supply chain work
Only ~6% of credits from 18 examined REDD+ projects were linked to additional reductions. Buying avoided-deforestation credits while your own soy or nickel chain is unmapped converts a land problem into a reputation problem.
在被评估的18个REDD+项目中，仅约6%的信用与额外减排相关。在自己的大豆或镍供应链尚未测绘时购买"避免毁林"信用，只是把土地问题变成声誉问题。
Planting instead of growing
Tree-planting programmes have repeatedly failed to increase canopy or benefit local people — wrong species, no survival monitoring, monoculture, no tenure clarity. Count trees alive at year 5 and year 10, or do not count them.
大量植树项目并未真正增加郁闭度或让当地社区受益：树种选错、不监测存活、单一林、权属不清。请统计第5年和第10年仍存活的树木——否则不要统计。
Land taken without consent
Indigenous peoples and local communities hold or manage a large share of the world's remaining intact forest. Skipping FPIC and tenure due diligence is the fastest way to lose both the forest and the licence to operate.
原住民与当地社区持有或管理着全球剩余完整森林的很大一部分。跳过自由知情同意与权属尽调，是同时失去森林和经营许可的最快方式。
Carbon-only mathematics
A drained peatland keeps emitting for decades. A monoculture plantation can legally count as "forest". Water, soil, biodiversity and community outcomes do not convert into tonnes — report them alongside, not inside.
排干的泥炭地会持续排放数十年；单一人工林在某些定义下仍算"森林"。水、土壤、生物多样性与社区结果无法折算成吨——请并列报告，而不是塞进碳里。
Fire and permanence not budgeted
2024's record tropical forest loss was driven by fire. Any restoration or protection commitment without a fire, drought and buffer plan is a press release with a shelf life.
2024年创纪录的热带林损失主要由火灾驱动。任何恢复或保护承诺，如果没有火灾、干旱与缓冲储备计划，就只是一份有保质期的新闻稿。
One-year contracts for ten-year ecology
Landscape restoration works on 5–20 year horizons with several companies in the same basin. Procurement cycles of 12 months quietly guarantee failure before the first seedling goes in.
景观恢复的时间尺度是5—20年，且需要同一流域内多家企业共同参与。12个月一轮的采购周期，在第一株苗木落地前就已经注定了失败。
What actually slows it down — inside the company真正拖慢它的，是公司内部的这些事
There is no single number, so it never enters the operating review
Energy has kWh. Water has m³. Land has no unit everyone accepts — so it has no slide, no owner and no target. Unlock: publish a hectare baseline anyway (site area, converted area, basin stress) and let it be imperfect in public.
能源有kWh，水有m³，土地没有一个大家都接受的单位——于是它没有幻灯片、没有责任人、也没有目标。破局：先公布一个公顷基线（场址面积、被转换面积、流域压力），允许它在公开状态下不完美。
Restoration looks like cost with no revenue line
It is treated as donation, not as risk management or capex. Unlock: price it as market access (EUDR fines up to 4% of EU turnover, delisting, shipment holds) and as basin continuity for the plant's own water licence.
它被当作捐赠，而不是风险管理或资本支出。破局：用市场准入来定价（EUDR罚款可达欧盟营业额4%、下架、货物滞留），并用工厂自身取水许可的流域连续性来定价。
Legal caution: mapping past conversion feels like admitting liability
So the baseline never gets drawn. Unlock: frame it as a compliance readiness exercise with a stated cut-off date; regulators reward disclosure and remediation far more than silence.
于是基线永远画不出来。破局：把它定位为合规准备工作，并明确分界日；相比沉默，监管方对披露与整改的宽容度高得多。
Ownership is split three ways
EHS owns the fence line, procurement owns the supplier, sustainability owns the report — nobody owns the landscape. Unlock: name one owner per priority basin, not per function.
EHS管围墙内，采购管供应商，可持续管报告——没人负责"景观"。破局：按优先流域指定唯一责任人，而不是按职能划分。
Sub-tier data is not contracted, so it does not exist
Suppliers answer what they are asked. If mine and plantation origin is not in the RFQ, the purchase order and the audit scope, it will never appear in a questionnaire. Unlock: put origin disclosure in the commercial documents, with a phase-in schedule.
供应商只回答被问到的问题。如果矿山与种植园来源不写进询价、订单和审核范围，它就永远不会出现在问卷里。破局：把来源披露写进商务文件，并给出分阶段时间表。
Perfect measurement is used as the reason to wait
Nature is a system; the error bars will never close. Unlock: separate what must be exact (did conversion occur after the cut-off date — yes/no, from imagery) from what can be approximate (how much benefit the restoration delivers). The first is already binary and auditable today.
自然是系统，误差区间永远不会闭合。破局：把"必须精确的"与"可以近似的"分开：分界日后是否发生转换（影像可判定的是/否）必须精确；恢复带来多少效益可以近似。前者今天就是二元的、可审计的。
Follow one battery all the way down顺着一颗电池一路往下看
The question in the original note: if we only watch the battery model and the cell plant, what are we not watching? Six levels. Visibility falls at every step; land impact rises at every step.
原始笔记里的问题：如果只盯电池型号和电芯厂，我们没有盯住什么？一共六层。每往下一层，可见度下降，土地影响上升。
Pack & model
Fully visible: BOM, supplier of record, warranty data. Land impact here: effectively zero. This is where almost all of the reporting effort currently sits.
完全可见：BOM、备案供应商、质保数据。此处的土地影响基本为零——而目前几乎所有报告工作都集中在这一层。
Cell manufacturing
Well measured on energy, water and waste. Rarely measured: what the greenfield site was before it was an industrial park, and how much of the basin's water the park now takes.
能源、水、废弃物计量良好。很少被计量的是：这块新建场址在成为工业园之前是什么，以及园区现在取用了流域多少水。
Cathode & precursor materials
Partly visible through contracts. Blind spots: industrial-park siting on converted land, wastewater to rivers, tailings and captive coal power co-located with the plant.
通过合同部分可见。盲点：工业园选址占用被转换土地、废水排入河流、尾矿，以及与工厂配套的自备燃煤电力。
Smelting & chemical refining
Traceable to smelter/refiner lists via RMI's CMRT/EMRT templates — the most under-used tool in this whole chain. Land and water risks: HPAL tailings disposal, land clearing for industrial estates, coal power footprint.
可通过RMI的CMRT/EMRT模板追溯到冶炼/精炼厂清单——这是整条链上最被低估的工具。土地与水风险：HPAL尾矿处置、为工业园区清林、燃煤电力足迹。
Mining — where land actually changes
Nickel in Indonesia (Sulawesi, Halmahera): forest clearing inside concessions and around industrial parks. Cobalt in the DRC: land conversion and community relocation. Lithium brine in the Atacama: water balance and ground subsidence. Graphite, bauxite, copper: same story, different landscape.
印尼镍（苏拉威西、哈马黑拉）：矿权与工业园周边的林地清除。刚果（金）钴：土地转换与社区搬迁。阿塔卡马锂盐湖：水平衡与地面沉降。石墨、铝土矿、铜：同样的故事，不同的景观。
The level nobody's KPI covers
Access roads, in-migration, sediment into rivers, fisheries and downstream farms, cumulative effects of many concessions in one landscape — measured up to 70 km beyond mine boundaries in the Amazon. No supplier scorecard in general use contains this line.
进场道路、人口迁入、河流泥沙、渔业与下游农田、同一景观内多个矿权的累积影响——在亚马逊，影响被测到矿界之外70公里。目前通用的供应商评分卡里，没有这一行。
How to close it
Three moves, in this order: (1) require smelter- and mine-level origin disclosure using RMI CMRT/EMRT plus mine names and coordinates; (2) require IRMA or equivalent third-party site assessment for the named mines, with the report shared, not just the badge; (3) monitor those coordinates annually with satellite forest-loss alerts and treat a new alert as a quality incident with a 90-day response — the same discipline you already apply to a defect.
怎么补上
按顺序做三件事：（1）要求披露到冶炼厂与矿山层级——使用RMI的CMRT/EMRT，并附矿山名称与坐标；（2）对具名矿山要求IRMA或等效第三方现场评估，并共享报告本身，而不只是标识；（3）对这些坐标每年用卫星林损预警监测，把新预警当作质量事件处理，90天内响应——就像你已经在对待缺陷时所做的那样。
Benchmarks: who is moving, with money and numbers标杆：谁在真的投钱、并公开数字
Selected because they publish a measurable position and back it with capital or contractual requirements — not because their commitments are perfect. Figures move every reporting cycle: always take the current number from the source link before you quote it.
入选标准是：公开了可衡量的立场，并以资金或合同要求支撑——不是因为它们的承诺完美。数据每个报告周期都会变化：引用前请务必到来源链接核对最新数字。
Nestlé
Assesses deforestation-free status for its primary commodities (palm oil, soy, pulp & paper, meat, sugar) with satellite monitoring (Starling) over the whole supply base rather than certificates alone, and publishes the percentage verified. Commodity-level policies with cut-off dates, plus a multi-year, billion-scale regenerative agriculture programme.
对主要商品（棕榈油、大豆、纸浆与纸、肉类、糖）的"无毁林"状态，采用覆盖整个供应基础的卫星监测（Starling）而非仅凭认证，并公布已核实比例。各商品政策均设分界日，并配套数十亿规模、跨年度的再生农业计划。
Unilever
A €1 billion Climate & Nature Fund, deforestation-free targets across palm oil, paper & board, tea, soy and cocoa with published verification percentages, and — unusually — public palm oil mill and supplier lists, which is what makes external checking possible at all. Also invests in regenerative agriculture and water stewardship in sourcing landscapes.
设立10亿欧元气候与自然基金；对棕榈油、纸与纸板、茶、大豆、可可设定无毁林目标并公布核实比例；并且少见地公开棕榈油榨油厂与供应商清单——这正是外部核查得以进行的前提。同时在采购景观内投资再生农业与水管理。
Kering
Among the first companies with SBTN-validated land and freshwater targets, and runs a Regenerative Fund for Nature converting sourcing landscapes to regenerative practice at the million-hectare scale of ambition. Also publishes a monetised environmental profit & loss account — the rare case of land and water appearing in financial language.
属于首批通过SBTN验证土地与淡水目标的企业；设立"自然再生基金"，以百万公顷量级的目标推动采购景观转向再生农业实践。同时公布货币化的环境损益表——把土地与水放进财务语言的少数案例之一。
L'Oréal
A dedicated Fund for Nature Regeneration (impact investment, not donation) alongside zero-deforestation policies for palm derivatives with traceability pushed towards plantation level — the hardest tier in that commodity because the derivatives chain fragments so fast.
设立专门的自然再生基金（影响力投资，而非捐赠）；同时对棕榈衍生物实施零毁林政策，并把可追溯性向种植园层级推进——由于衍生物链条极度分散，这是该商品中最难的一层。
IKEA (Inter IKEA & Ingka)
Wood is ~a fifth of its material footprint, so it treats forest as core business: near-total FSC-certified or recycled wood, published country-level forest risk work, and direct ownership and management of forest land — vertical integration used as a traceability instrument rather than a slogan.
木材约占其材料足迹的五分之一，因此把森林当作核心业务：木材几乎全部为FSC认证或再生来源，公开国别森林风险工作，并直接拥有并经营林地——把纵向整合当作可追溯工具，而不是口号。
Apple
The Restore Fund puts corporate capital into working forests and restoration as an investment vehicle (initial US$200 million, later expanded) with third-party verified carbon and a stated intent to return capital. Whatever you think of forest carbon, the mechanism is the interesting part: a balance-sheet route into land, not a CSR line item.
Restore Fund把企业资本以投资工具的形式投入经营性森林与恢复项目（初期2亿美元，后续扩容），要求第三方核证碳并明确追求资本回报。无论如何看待森林碳，值得关注的是机制本身：一条通过资产负债表进入土地的路径，而不是CSR支出项。
BMW Group
Buys some raw materials directly from named mines to control origin, requires third-party mine-site assessment (IRMA) in its raw-material requirements, and took a public position against deep-sea mining until impacts are understood. Mine-level requirements in purchasing conditions are the practice most worth copying.
部分原材料直接向具名矿山采购以控制来源；在原材料要求中纳入第三方矿山现场评估（IRMA）；并在影响尚不明确前公开反对深海采矿。最值得借鉴的做法是：把矿山层级要求写进采购条件。
Volvo Cars & Polestar
Applied blockchain-based traceability to cobalt, then mica and nickel in battery chains, and publish product-level LCAs so the upstream share is visible. The lesson: traceability became possible only once it was a condition of the supply contract.
在电池链中对钴，随后是云母与镍应用基于区块链的可追溯；并发布产品级LCA，使上游占比可见。启示是：只有当可追溯成为供货合同条件时，它才真正做得到。
Holcim
The closest analogue to the factory-hectare question: quarries physically remove land, so it reports a biodiversity indicator across sites and rehabilitation plans per quarry, and joined the SBTN corporate engagement work. If your business converts land directly, this is the disclosure pattern to study.
与"工厂公顷"问题最接近的类比：采石场直接移除土地，因此其披露包含覆盖各场址的生物多样性指标与逐个矿场的复垦计划，并参与SBTN企业实践工作。若你的业务直接转换土地，这是值得研究的披露范式。
H&M Group
Publishes a supplier list reaching beyond tier 1 into fabric and yarn mills — the precondition for any land or water target — and has worked through SBTN's freshwater and land methods in its sourcing basins. Transparency first, target second: that order matters.
公开的供应商清单超越一级，延伸至面料与纱线厂——这是任何土地或水目标的前提；并在采购流域内实践SBTN的淡水与土地方法。先透明、后目标：顺序很重要。
What to put in the supplier ask — and what to leave out供应商要求里写什么，不写什么
Nine clauses. Each one is answerable with data a supplier can actually get, and each one maps to a rule or standard above — so it is a requirement, not a favour.
九条条款。每一条都可以用供应商真正拿得到的数据来回答，且都能对应到上文的某项法规或标准——因此它是要求，不是人情。
- Site coordinates and area. Latitude/longitude and total site area in hectares for every production site serving us, including sub-contracted sites. 场址坐标与面积。为我们供货的每一个生产场址（含分包场址）的经纬度与总用地面积（公顷）。
- Land-cover baseline. What the site was before construction, with the year and the evidence used (imagery, land certificate, national map). 土地覆盖基线。建厂前该场址是什么地类，注明年份与所用证据（影像、土地证、国家地图）。
- No conversion, with a cut-off date. No conversion of natural forest or other natural ecosystems on own or leased land after 31 December 2020, per AFi definitions. 不转换，并设分界日。依照AFi定义，自2020年12月31日之后，自有或租赁土地上不得转换天然林及其他自然生态系统。
- Commodity origin to plot level. For any EUDR-listed commodity in what you sell us, geolocation of the plots of production plus a legality statement. 商品来源到地块。供货中涉及EUDR清单商品的，提供生产地块地理位置与合法性声明。
- Sub-tier origin for minerals. Completed RMI CMRT/EMRT plus named smelters/refiners and, for battery and critical minerals, named mines with coordinates. 矿产的次层级来源。提交完整的RMI CMRT/EMRT，并列明冶炼/精炼厂；电池与关键矿产还需列明矿山名称与坐标。
- Water in context. Withdrawal and consumption by site, with the basin's water-stress classification and, in high-stress basins, a replenishment plan. 放在流域里的水。按场址报告取水与耗水，附流域水压力等级；高压力流域需提交回补计划。
- Sensitive-area screening. Declare any site or sourcing area inside or within 5 km of a protected area, Key Biodiversity Area, Ramsar wetland or peatland. 敏感区域筛查。申报位于保护区、关键生物多样性区域、拉姆萨湿地或泥炭地内部或5公里范围内的场址与采购区域。
- Monitoring and response. Accept annual satellite deforestation-alert screening of the declared coordinates and respond to any alert within 90 days with an investigation and corrective plan. 监测与响应。接受对已申报坐标的年度卫星毁林预警筛查；任何预警需在90天内完成调查并提交纠正计划。
- Rights and consent. Evidence of legal land tenure and, where Indigenous peoples or local communities are affected, documented free, prior and informed consent and a functioning grievance channel. 权利与同意。提供合法土地权属证据；涉及原住民或当地社区时，提供有记录的自由、事先与知情同意，以及有效运行的申诉渠道。
Do — start small, start measured
- Start with 10 sites or 3 commodities, not the whole footprint.从10个场址或3种商品开始，而不是整个足迹。
- Use the same cut-off date everywhere (31 Dec 2020) so data stays comparable.全公司统一分界日（2020年12月31日），保证数据可比。
- Publish the baseline before you publish the target.先公布基线，再公布目标。
- Buy long: 10-year restoration contracts, 20–30 year management budgets.买长期：10年恢复合同，20—30年管护预算。
- Restore in the same ecoregion and sub-basin as the impact.在与影响相同的生态区与子流域内恢复。
- Share cost with other buyers in the same landscape — nobody fixes a basin alone.与同一景观内的其他买家分担成本——没人能独自修好一个流域。
- Report hectares under improved management, survival rate at year 5, and basin water balance — three numbers, every year.每年报告三个数字：改善管理的公顷数、第5年存活率、流域水平衡。
- Treat a deforestation alert like a quality defect: root cause, corrective action, verification.把毁林预警当质量缺陷处理：根本原因、纠正措施、验证关闭。
Don't — the shortcuts that cost more later
- Don't wait for a perfect metric before spending the first euro.不要等到指标完美才花第一笔钱。
- Don't let cheap avoided-deforestation credits replace mapping your own chain.不要用廉价的"避免毁林"信用替代测绘自己的供应链。
- Don't count restored hectares twice — once as nature, once as carbon offset.不要把恢复的公顷重复计算——一次算自然，一次算碳抵消。
- Don't accept a certificate in place of coordinates.不要用一张证书代替坐标。
- Don't plant monocultures or exotic species for a photograph.不要为了拍照种单一林或外来树种。
- Don't stop the supplier map at tier 1.供应商测绘不要停在一级。
- Don't sign land commitments without checking tenure and consent.不要在未核查权属与同意的情况下签署土地承诺。
- Don't run it as a one-year project with a one-year budget.不要用一年的预算做一年的项目。
Sources — go to the raw material and check me资料来源：请回到原始材料核对
Links point to the issuing body or publisher. Regulatory dates, company percentages and annual forest-loss figures all change — take the current number from the source, then benchmark it against your own sites.
链接指向发布机构或出版方。法规时间、企业比例与年度林损数据都在变化——请以来源为准取最新数字，再与自己的场址做对标。
Scale & evidence · 规模与证据
- WEF — Nature Risk Rising世界经济论坛：自然风险上升（US$44tn / 超过半数GDP依赖自然）
- FAO — Global Forest Resources Assessment联合国粮农组织：全球森林资源评估（10 Mha/yr 砍伐、4.7 Mha/yr 净损失）
- FAO — Agricultural expansion drives almost 90% of global deforestation粮农组织：农业扩张驱动近90%的全球毁林
- WRI — Global trends in forest loss · Global Forest Watch dashboard世界资源研究所与Global Forest Watch：全球森林损失趋势与数据看板
- Griscom et al. — Natural climate solutions, PNAS 2017自然气候解决方案：可提供2030年前所需成本有效减排的最高约37%
- UNEP — State of Finance for Nature联合国环境规划署：自然融资状况（约2,000亿美元 vs 约7万亿美元）
- Sonter et al. — Mining drives extensive deforestation in the Brazilian Amazon, Nat. Comms 2017采矿驱动巴西亚马逊大范围毁林（9%、影响达70公里外）
- West et al. — Action needed to make carbon offsets from tropical forest conservation work, Science 2023热带森林保护碳抵消的有效性评估（约6.1%可关联额外减排）
- Sonter et al. — Renewable energy production will exacerbate mining threats to biodiversity, Nat. Comms 2020可再生能源扩张将加剧采矿对生物多样性的威胁
- Fleischman et al. — Pitfalls of tree planting, BioScience 2020植树造林的常见陷阱：为何需要以人为中心的自然解决方案
- Climate Rights International — reporting on nickel mining and forests in Halmahera, Indonesia关于印尼哈马黑拉镍矿与森林的调查报告
- IEA — Global Critical Minerals Outlook国际能源署：全球关键矿产展望（需求量级与集中度）
- Rights and Resources Initiative — land tenure and Indigenous lands data权利与资源倡议：土地权属与原住民土地数据
- Trase — commodity supply chains and deforestation exposureTrase：商品供应链与毁林暴露度数据
- Forest 500 — annual ranking of the most influential companies on forestsForest 500：对森林最具影响力企业的年度排名
Rules, standards & frameworks · 法规与标准
- European Commission — EU Deforestation Regulation (EU) 2023/1115欧盟委员会：零毁林产品法规（含分界日、地块坐标、罚则与实施时间）
- Accountability Framework — definitions, cut-off dates, verificationAccountability Framework：无毁林/无转换定义、分界日与核查方法
- European Commission — Corporate Sustainability Reporting (CSRD)欧盟委员会：企业可持续发展报告（CSRD）
- EFRAG — ESRS, including E3 Water and E4 Biodiversity & ecosystemsEFRAG：ESRS准则，含E3水与E4生物多样性与生态系统
- European Commission — Batteries Regulation (EU) 2023/1542欧盟电池法规：电池护照与原材料尽职调查（含土地利用与生物多样性风险）
- USDA APHIS — Lacey Act plant and wood product declarations美国农业部APHIS：《雷斯法案》植物与木制品申报要求
- US Congress — track the FOREST Act (proposed, not enacted)美国国会：查询FOREST法案进展（仍为提案，尚未立法）
- California Air Resources Board — SB 253 / SB 261 climate disclosure加州空气资源委员会：SB 253 / SB 261企业气候披露规则
- UK Environment Act 2021, Schedule 17 — forest risk commodities英国《2021环境法》附表17：森林风险商品（细则待出）
- SBTi — Forest, Land and Agriculture (FLAG) guidanceSBTi：森林、土地与农业（FLAG）指南（含不晚于2025年的分界日要求）
- Science Based Targets Network — land and freshwater target methodsSBTN：土地与淡水目标方法及首批验证企业
- IFC — Performance Standard 6: biodiversity, no net loss, net gainIFC绩效标准6：生物多样性、无净损失与净增益要求
- TNFD — LEAP approach and disclosure recommendationsTNFD：LEAP方法与自然相关财务披露建议
- GRI 101: Biodiversity 2024GRI 101 生物多样性（2024版）
- CDP — Forests and Water Security disclosureCDP：森林与水安全披露问卷
- CBD — Kunming-Montreal Global Biodiversity Framework, Target 15《昆明-蒙特利尔全球生物多样性框架》目标15（企业披露）
Free tools you can open today · 今天就能打开的免费工具
- WRI Aqueduct — basin water stress and risk mappingWRI Aqueduct：流域水压力与水风险地图
- ESA WorldCover — 10 m global land cover for your "before" baselineESA WorldCover：10米全球土地覆盖，用于建立"之前"的基线
- Global Forest Watch Pro — monitor your own supplier coordinates for forest-loss alertsGlobal Forest Watch Pro：对自有供应商坐标进行林损预警监测
- IBAT — protected areas, Key Biodiversity Areas, IUCN species screeningIBAT：保护区、关键生物多样性区域与IUCN物种筛查
- WWF Risk Filter — biodiversity and water risk screening by siteWWF风险筛选器：按场址筛查生物多样性与水风险
- Responsible Minerals Initiative — CMRT / EMRT reporting templates责任矿产倡议：CMRT / EMRT 尽责报告模板
- IRMA — Standard for Responsible Mining, mine-site assessmentsIRMA：负责任采矿标准与矿山现场评估
- CGF Forest Positive Coalition — shared supplier requirements by commodity消费品论坛"森林正向"联盟：按商品统一的供应商要求
- Nature Positive Initiative — the definition everyone is now converging onNature Positive Initiative：各方正在收敛的"自然向好"定义
Keep it open · 让它保持打开
Do you also feel this question is a question?你有时候，是不是也觉得这是一个问题？
This page is not an answer, and it is not finished. It is a question kept alive with enough data attached that it can survive a meeting. Take any piece of it — the hectare formula, the nine supplier clauses, one source link — into your next supplier review, and let something start, even a small piece. Then come back and correct it: go to the raw materials, benchmark your own sites, and change the numbers where you find better ones.
The measurement will keep improving. The forest will not wait for it.
这一页不是答案，也没有完成。它是一个被保存下来的问题，附上了足够的数据，好让它能在会议里活下来。请随意取用其中任何一块——公顷粗算式、九条供应商条款、一个来源链接——带进你的下一次供应商评审，让一件事情先开始，哪怕只是很小的一块。然后回来修正它：回到原始材料，对标自己的场址，遇到更好的数据就替换掉这里的数字。
计量方法会不断变好，森林不会等它。