一年前，總部位于北京的巨頭中國化工斥資430億美元收購了瑞士作物保護(hù)和種子生產(chǎn)商先正達(dá)，為中國公司最大一筆海外收購。很快中國化工就定下目標(biāo)，首先要利用先正達(dá)的數(shù)字技術(shù)實(shí)現(xiàn)農(nóng)業(yè)轉(zhuǎn)型，推動投資和增長。“中國化工目光長遠(yuǎn)，一直在大幅增加研發(fā)投入，投入增幅最大的是突破性數(shù)字技術(shù)領(lǐng)域。”先正達(dá)首席執(zhí)行官方華德告訴《財(cái)富》雜志。

先正達(dá)主要向農(nóng)民提供兩種數(shù)字服務(wù)，打包在AgriEdge產(chǎn)品中，產(chǎn)品里還包括一套種子和作物保護(hù)產(chǎn)品。第一項(xiàng)服務(wù)是FarmShots，是使用衛(wèi)星和無人機(jī)精確定位病蟲害的軟件。第二項(xiàng)是農(nóng)場管理系統(tǒng)，農(nóng)民可以創(chuàng)建損益表（P＆L），追蹤在玉米或大豆等領(lǐng)域收入支出的詳細(xì)流水。衛(wèi)星每天向農(nóng)民發(fā)送照片，拍攝各塊田地；平均一張照片可拍到數(shù)百英畝。而且照片在電腦，iPad或智能手機(jī)上可顯示為“熱力圖”，用不同顏色標(biāo)注生病和健康的作物。

衛(wèi)星軟件可過濾純土壤區(qū)域，只顯示作物。根據(jù)植物反射不同亮度的光，農(nóng)民的屏幕上會以不同顏色體現(xiàn)。“健康的植物可以反射更多陽光產(chǎn)生鮮艷的綠色圖像，具體因過濾器而異。”先正達(dá)數(shù)字農(nóng)業(yè)部門主管丹·伯德特表示。相比之下，紅色信號代表危險(xiǎn)，伯德特指著一片馬鈴薯田表示，紅色說明缺乏灌溉或肥料。黃色說明作物發(fā)育不良，例如過度施用氮肥導(dǎo)致玉米作物根部燒毀。

衛(wèi)星圖像的優(yōu)勢在于可在Gator 4×4上突出顯示問題區(qū)域，而之前農(nóng)民需要一周或更長時(shí)間才能發(fā)現(xiàn)。日常拍攝的照片只能顯示大片區(qū)域的顏色，無法實(shí)現(xiàn)葉片或莖的近距離拍攝，農(nóng)民也難以識別植物染上的特定疾病。為了實(shí)現(xiàn)圖像放大，先正達(dá)將衛(wèi)星與無人機(jī)技術(shù)結(jié)合。一旦熱力圖在農(nóng)民GPS上顯示問題區(qū)域，就可派遣無人機(jī)前往拍攝，發(fā)回高分辨率的樹葉或秸稈圖像。農(nóng)民根據(jù)經(jīng)驗(yàn)可迅速發(fā)現(xiàn)灰葉病等常見疾病，灰葉病是一種影響玉米生長的真菌感染，確認(rèn)后農(nóng)民前往現(xiàn)場用對應(yīng)的殺真菌劑噴灑患病作物。

但有一些農(nóng)害即使是經(jīng)驗(yàn)豐富的農(nóng)民也很難確認(rèn)。為此，先正達(dá)率先推動無人機(jī)技術(shù)的巨大進(jìn)步：可精確識別銹病、病毒或傳染病的軟件，并提供治療手段的詳細(xì)說明。先正達(dá)已組建七位退休植物病理學(xué)家的團(tuán)隊(duì)，編制囊括大多數(shù)植物病害的龐大圖像庫。無人機(jī)軟件通過機(jī)器學(xué)習(xí)技術(shù)，將目錄里的照片與無人機(jī)拍攝的樹葉或莖照片匹配。預(yù)計(jì)先正達(dá)將于2019年推出首款無人機(jī)診斷病蟲害的產(chǎn)品。對于使用無人機(jī)和衛(wèi)星圖像有難度的小農(nóng)戶，先正達(dá)將提供智能手機(jī)應(yīng)用程序。農(nóng)民拍攝患病作物后，應(yīng)用程序發(fā)回完整的診斷和治療計(jì)劃。

FarmShots應(yīng)與農(nóng)場管理系統(tǒng)結(jié)合使用，該系統(tǒng)是金融工具，可以精確顯示何處出現(xiàn)虧損以及具體原因。農(nóng)場管理系統(tǒng)有兩個(gè)主要部分。第一個(gè)是新款拖拉機(jī)軟件，可跟蹤農(nóng)場各部門使用的種子、肥料、水、除草劑和殺蟲劑的數(shù)量。先正達(dá)與拖拉機(jī)制造商約翰·迪爾、Case和紐荷蘭品牌的制造商CNH Global都簽有數(shù)據(jù)協(xié)議。廠商提供的拖拉機(jī)都配備類似iPad的電腦，可監(jiān)測拖拉機(jī)或噴灑機(jī)使用的種子、化肥和除草劑使用情況。過去，農(nóng)民僅能監(jiān)控幾種作物，覆蓋10,000多英畝土地。“現(xiàn)在每塊地都能當(dāng)成單獨(dú)的工廠。”伯德特表示。用上GPS后，拖拉機(jī)可準(zhǔn)確顯示機(jī)器種植、收獲或播種的位置。機(jī)器上的傳感器記錄農(nóng)場各區(qū)域的用料，并將數(shù)據(jù)發(fā)送至農(nóng)民手上裝有先正達(dá)FMS軟件的損益表中。

數(shù)據(jù)詳細(xì)列出哪些地塊和作物產(chǎn)生的投資回報(bào)較高。先正達(dá)向《財(cái)富》雜志提供伊利諾伊州兩處農(nóng)地的損益表，解釋相關(guān)工具如何應(yīng)用。一個(gè)損益表中包括117英畝的大豆，只是規(guī)模更大的農(nóng)場一部分。去年的種植季里，從5月播種到10月收獲，該產(chǎn)區(qū)產(chǎn)量為7,000蒲式耳（1蒲式耳約為35.24升——譯注），每蒲式耳售價(jià)10.24美元，總收入為71,708美元。拖拉機(jī)軟件記錄了三塊投入，即種子、肥料和作物保護(hù)產(chǎn)品的使用情況。材料加上勞動力成本為29,591美元。成本大頭是租金，為41,120美元。通過在損益表上列出所有數(shù)據(jù)，農(nóng)民發(fā)現(xiàn)這塊117英畝的土地只賺了998美元，基本上沒賺錢。

顯然，種植大豆的農(nóng)民需要重大調(diào)整。畢竟市場價(jià)格沒法控制。使用不同的種子或肥料提高產(chǎn)量沒準(zhǔn)行得通。先正達(dá)軟件將該地區(qū)的每英畝大豆產(chǎn)量與同一農(nóng)場的其他地區(qū)，以及該地區(qū)其他匿名地塊比較，顯示產(chǎn)量最高的農(nóng)場使用的產(chǎn)品。軟件還可顯示類似土地的租金。

農(nóng)民為每塊地選擇利潤最高的作物至關(guān)重要。附近農(nóng)場的數(shù)據(jù)顯示，如果農(nóng)民放棄大豆改種玉米會更好。259英畝種植玉米的土地利潤數(shù)字好得多。

所以，僅看去年221,147美元收入顯得規(guī)模大得多，但部分原因是農(nóng)場規(guī)模超過兩倍。還是看每英畝收入更有意義。盡管玉米價(jià)格為每蒲式耳3.60美元，僅為大豆價(jià)格的三分之一，玉米農(nóng)戶卻能彌補(bǔ)豐欠收成差異。玉米農(nóng)戶每英畝銷售額854美元，比大豆農(nóng)戶高出39.5％。每英畝收入的材料和人工成本與大豆農(nóng)場基本相同。除了每英畝產(chǎn)量更高從而收入更多，最主要的因素是租金。兩名農(nóng)民每英畝租金均為350美元。但是玉米農(nóng)戶負(fù)擔(dān)得起，種大豆的鄰居卻無力承受：該租金僅占玉米農(nóng)戶每英畝收入的41％，卻占大豆農(nóng)戶收入的57％。玉米農(nóng)戶扣除所有開支后每英畝可獲153美元收入，而大豆農(nóng)戶基本剩不下什么錢。

明細(xì)中還列出大豆農(nóng)戶的選擇，可以向土地主聲明種大豆無法支撐每英畝350美元地租，爭取降低租金。也可以研究是不是因?yàn)楦偁帉κ质褂貌煌N子或肥料所以產(chǎn)量更高，從而實(shí)現(xiàn)盈利。或者更簡單，放棄大豆改種玉米。

農(nóng)民們現(xiàn)在越發(fā)理解套利來源來自各種利潤中心的道理。“我告訴農(nóng)民，要真正管好農(nóng)場就要深入了解每塊地的數(shù)字。”伯德特說。現(xiàn)在AgriEdge是先正達(dá)增長最快的業(yè)務(wù)，在121億美元總銷售額里占5億美元。該產(chǎn)品現(xiàn)在服務(wù)于1900萬英畝土地，超過美國農(nóng)田總數(shù)的2％，而AgriEdge占有的市場每年增長超過25％。說起尋找處女地賺錢，大數(shù)據(jù)當(dāng)屬第一。（財(cái)富中文網(wǎng)）

譯者：Pessy

審校：夏林

?

A year ago, Bejing-based colossus ChemChina purchased crop protection and seed producer Syngenta of Switzerland for $43 billion, setting the all-time record for the largest overseas acquisition by a Chinese company. ChemChina quickly targeted Syngenta’s drive to transform farming through digital technology as the paramount area for investment and growth. “ChemChina is taking the long view by substantially raising spending on R&D, and by far the biggest increases are going to groundbreaking digital technology,” Erik Fyrwald, Syngenta’s CEO, tells Fortune.

Syngenta is providing two main digital services to farmers, packaged together under a product called AgriEdge, which also includes a suite of its seed and crop protection products. The first is FarmShots, a software offering that uses satellites and drones to pinpoint sick crops in the field. The second is its Farm Management System enabling farmers to create profit and loss statements (P&L) tracking every dollar spent and earned in each field of corn or soybeans. The satellites send farmers one photo each day of individual sections of their fields; an average shot covers several hundred acres. The photos are displayed on PCs, iPads, or smartphones as “heat maps” that identify ailing and healthy sections of plants using color codes.

The satellite software filters out the areas of soil, displaying only the foliage. The amount of light the plants generate helps establish the colors displayed on the farmer’s screen. “Healthy plants reflect a lot more sunlight, and depending on the filter, produce a bright green image,” says Dan Burdett, Syngenta’s chief of digital agriculture. By contrast, a patch of red signals danger, spotting a potato field, say, that’s starved for irrigation or fertilizer. Yellow can show that plants are stunted, as from over-application of nitrogen fertilizer that burns the roots of corn crops.

The advantage of satellite images is that they spotlight problem areas that could take a farmer a week or more to pinpoint scouting the spread on a Gator 4×4. Still, the daily photos produce only broad areas of color. They don’t yield up-close shots of leaves or stalks that might help the farmer identify the specific disease afflicting the plant. To zoom in, Syngenta teams the satellite with drone technology. Once the heat map shows the problem areas on the farmer’s GPS, he can dispatch a drone that sends back high-resolution shots of the leaves or stalks. Through experience, the farmer can frequently identify such common maladies as gray leaf, a fungal infection that damages corn, and then go to the site and spray the endangered crops with the correct fungicide.

Many diseases, however, are hard for even veteran farmers to diagnose with certainty. So Syngenta is pioneering the next giant step in drone technology: Software that identifies the precise rust, virus, or infestation, and sends instructions detailing the best treatment. Today, Syngenta is deploying a team of seven retired plant pathologists to compile a giant library of images displaying most plant diseases. The drone software uses machine learning to match a photo from that catalogue with the shot of a leaf or stalk taken by the drone. Syngenta expects to launch the first product enabling drones to diagnose plant diseases in 2019. For small farmers who don’t have access to the full package of drone and satellite images, Syngenta will offer a smartphone application. The farmer takes a shot of the diseased plant, and the app sends back a full diagnosis and plan for care.

FarmShots partners with the Farm Management System, a financial tool that highlights precisely where and why farmers are making or losing money. FMS comes in two main parts. The first is new tractor software that tracks the amounts of seed, fertilizer, water, herbicides and pesticides being used in each part of the farm. Syngenta has data agreements tractor manufacturers John Deere and CNH Global, maker of Case and New Holland brands. Their tractors are equipped with an iPad-like computer that quantifies all seed, fertilizer and herbicides that a tractor or tank sprayer is applying. In the past, farmers measured those inputs only for entire properties often spread over several crops and 10,000 or more acres. “Now, we’re treating every individual field as a separate factory,” says Burdett. Using GPS, the tractor shows exactly where the machines are planting, harvesting, or seeding. Their sensors record all the materials used in each area of the farm, and send the data to farmer’s P&L, which runs on the Syngenta’s FMS software.

That data provides a detailed breakdown of which fields and crops are generating an acceptable ROI. Syngenta provided Fortune with P&Ls from fields in two Illinois farms that illustrate how these tools spotlight how these tools are used. One P&L covers a 117-acre field growing soybeans that’s a section of a much larger farm. In last year’s growing season, stretching from planting in May to harvesting in October, that field produced 7,000 bushels that it sold for $10.24 each, for total revenue of $71,708. Its tractors recorded the use of three inputs, seed, fertilizer, and crop protection products. The cost of those materials, plus labor, was $29,591. The big cost was rent, amounting to $41,120. By tabulating all of that data on his FMS P&L, the farmer learned that it had earned just $998, or virtually nothing, on those 117 acres.

Clearly, the soybean farmer needs to make big changes. He can’t influence the market price. But a solution might be raising his yield by using different seed or fertilizer. Syngenta software “benchmarks” the field’s soybean production per acre compared with other areas on the same farm, and unnamed competing spreads in the region, and also shows the products the best farms are using to achieve superior yields. It also shows rental rates for similar properties.

It’s crucial for farmers to choose the most profitable crops for each field. And the numbers from a nearby farm show that the soybean farmer might be better off replanting his field with corn. Here are the far better numbers on the 259 acres this farmer has planted with corn.

Unsurprisingly, his revenues––$221,147 last year––are much bigger, in part because his farm is more than twice as large. It’s the per acre figures that tell the story. Even though corn prices at $3.60 per bushel were one-third those of soybeans, the corn farmer more than made up for the difference in bountiful production. His sales were $854 per acre, 39.5% higher than the soybean farmer’s. Material and labor cost per dollar of revenues per acre were the same as the soybean farmer’s. Besides the superior yield that generated much bigger revenues per acre, the big factor was rent. The two farmers paid the same amount, $350 per acre. But the corn producer could afford it and his soybean neighbor couldn’t: That figure represented only 41% of the corn producer’s revenue per acre, versus 57% for the soybean farmer. The corn purveyor earned a sizable $153 per acre after all expenses, compared to a profit drought for the soybean farmer.

The breakdown arrays the soybean farmer’s best options. He can negotiate with his landlord to lower the rent, arguing that soybeans won’t sustain a charge of $350 per acre. He can examine whether the different seed or fertilizer used by rivals generate the much higher yield needed to make the field profitable. Or he can replace soybeans with corn.

Farmers are getting comfortable viewing their spreads as a patchwork of profit centers. “I tell farmers that to be better farm managers, you need to know your numbers on a field by field basis,” says Burdett. AgriEdge is now Syngenta’s fastest growing business, accounting for $500 million of its $12.1 billion in sales. The product now serves 19 million acres. That’s more than 2% of all the farmland in the U.S. and AgriEdge’s domain is growing at over 25% a year. Nothing harvests cash from the back forty like big data.