Google 向 Mitti Labs 採購 100 萬筆稻田甲烷碳權，田區量測與農戶受益須分帳

計畫以衛星與田間資料監測間歇灌溉，企業預期到 2030 年涵蓋逾 10 萬公頃，並宣稱可減少甲烷與灌溉用水。

The programme uses satellite and field data to monitor alternate wetting and drying, with more than 100,000 hectares targeted by 2030 and claimed methane and water savings.

水稻連續淹水形成缺氧環境，是農業甲烷的重要來源；間歇灌溉能改變水位與土壤條件。

Continuous flooding creates anaerobic conditions that make rice a material agricultural methane source, while alternate wetting and drying changes water and soil conditions.

不同百年與二十年暖化係數會產生不同 CO2e 數字，且衛星判讀、現場採樣、產量與農戶權益都可能影響可交付量。

Different 100-year and 20-year warming metrics produce different CO2e figures, while satellite inference, field sampling, yield and farmer rights affect deliverability.

專案需在種植季前鎖定田區與基準，季中保存灌排水和缺測紀錄，核發前再由獨立機構查驗。

The project needs pre-season field and baseline controls, in-season irrigation and missing-data records, and independent checks before issuance.

可確認的是四年合約、100 萬筆採購量與逾 7 萬名農戶的計畫範圍，尚不能視為 100 萬噸已核發或已註銷。

The four-year contract, one-million-credit purchase and programme scope above 70,000 farmers are supported, but they do not show that one million tonnes have been issued or retired.

農業碳市場的品質將取決於田區級基準、缺測處理、暖化係數選擇、農戶同意與收益分配。

Agricultural carbon-market quality depends on field-level baselines, missing-data treatment, warming-metric choice, farmer consent and benefit sharing.

SNN editorial analysis：本案對台灣的食品與稻米進口供應鏈、企業碳權採購及農業金融最具借鏡。第一，食品與零售採購團隊應在供應商准入及每個種植季前，以田區編號連結農戶同意書、基準灌溉方式、作物年度與灌排水日誌，低甲烷商品宣稱不得只依專案彙總數字。第二，碳權買方須在每批付款或註銷前，把衛星影像、地面水位、氣體採樣、模型版本及核發序號對到同一田區，遇到雲遮、缺測或農法反轉時暫停該批主張。第三，農業金融機構應在年度覆核時比對農戶付款紀錄、實際節水、產量與所有權移轉資料，分開呈現碳減排與農戶受益。揭露前證據基礎設施在此是田區、稻米商品、碳權核發及農戶付款之間的連續履歷。這是 editorial analysis，不表示印度專案成果已發生在台灣。

SNN editorial analysis: The most relevant Taiwan users are food and rice-import supply chains, corporate carbon-credit buyers and agricultural lenders. First, before supplier admission and each crop season, food and retail procurement teams should connect every field identifier to farmer consent, baseline irrigation, crop year and irrigation logs; a low-methane product claim cannot rely only on project-level totals. Second, before each payment or retirement, carbon-credit buyers must map satellite images, field water levels, gas samples, model versions and issuance serials to the same plot, suspending the batch when cloud cover, missing measurements or practice reversal appears. Third, during annual review, agricultural lenders should compare farmer payment records, measured water use, yield and ownership-transfer data and report carbon reduction separately from farmer benefit. Here, Pre-Disclosure Evidence Infrastructure is the continuous record linking fields, rice products, credit issuance and farmer payments. This is editorial analysis and does not state that the Indian project outcomes occurred in Taiwan.

後續應追蹤實際納入田區、核發批次、量測不確定性、農戶付款與產量結果。

Future checks should cover enrolled fields, issued batches, measurement uncertainty, farmer payments and crop yields.