Or­chard­ing and the power of (more) data

The Orchardist - - Technology - By Dan Bloomer – LandWISE

The re­port cites dis­rup­tion due to global trends such as pop­u­la­tion, cli­mate change, biotech­nol­ogy and smart tech­nolo­gies and in­dus­try-spe­cific change ac­cel­er­a­tors in­clud­ing con­sumer pref­er­ences, emerg­ing tech­nolo­gies and chang­ing value chain and firm con­fig­u­ra­tions.

As a con­se­quence of the iden­ti­fied dis­rup­tion driv­ers, the re­port says the agri­cul­tural ecosys­tem faces new chal­lenges, from which three sig­nif­i­cant growth op­por­tu­ni­ties for cur­rent and fu­ture play­ers can be de­rived:

1. Im­prov­ing yield ef­fi­ciency

2. In­creas­ing sup­ply chain ef­fi­ciency

3. De­creas­ing com­plex­ity along farm­ers’ value chain. Much of the anal­y­sis fo­cuses on feed­ing the grow­ing pop­u­la­tion, and the big in­dus­tries of wheat, rice, corn and soy­beans. What does the dis­rup­tion mean for or­chard­ing?

I sug­gest the same trends and op­por­tu­ni­ties ap­ply. Per­haps in fruit and wine pro­duc­tion we are less in­ter­est­ing in feed­ing the world, but we cer­tainly want to in­crease pro­duc­tiv­ity, cut losses, in­crease value cap­ture and do it all with less com­plex­ity.

Many new tech fac­tors will con­trib­ute to im­prove­ments. For

now, let’s have a quick look at the in­for­ma­tion revo­lu­tion.

Joseph Aami­dor of Green­tech Me­dia com­pared in­dus­tries and the po­ten­tial ben­e­fits of the In­ter­net of Things IoT revo­lu­tion. He notes that com­pared to build­ing man­age­ment and man­u­fac­tur­ing, many farms are start­ing at a very “data-light” po­si­tion. This may change rapidly, with farm data plat­form OnFarm sug­gest­ing the av­er­age farm will gen­er­ate 4.1 mil­lion data points per day by 2050, up from

190,000 in 2014.

How many of you are gen­er­at­ing (much less stor­ing) 190,000 data points per day in 2017? How are you go­ing to col­lect and man­age your 4 mil­lion data points? Where are they go­ing to come from?

First of all, we ex­pect the res­o­lu­tion of data cap­ture to in­crease, both in time and space. So we’ll start col­lect­ing in­for­ma­tion about ev­ery plant ev­ery time we go through (or over) the or­chard.

Imag­ine a set of sen­sors on the trac­tor (yes, I think you’ll be us­ing them for a while) that cap­ture canopy, fruit and soil in­for­ma­tion as you mow or spray (and record what sprays went where at what rates). Cli­mate in­for­ma­tion too – the tem­per­a­ture, hu­mid­ity, wa­ter sta­tus and light around each plant.

Com­bine that with satel­lite and UAV maps, con­nected sen­sors de­liv­er­ing cli­mate, soil and plant data ev­ery few min­utes, labour records in plant by plant de­tail and mul­ti­ply that by blocks and sea­sons and years. New agri­cul­ture jobs will have ti­tles like sens­ing tech­ni­cian, GIS of­fi­cer and chief data sci­en­tist.

What is some of the in­for­ma­tion that would be most use­ful to you now? As you de­velop more in depth un­der­stand­ing, what in­for­ma­tion will you want to col­lect later? How will you do that?

Think about yield pre­dic­tion. That in­for­ma­tion is vi­tal for the grower to bal­ance crop loads and es­ti­mate yield, and for the ex­porters and sec­tors for plan­ning lo­gis­tics and mar­ket­ing. How good is cur­rent prac­tice?

Our pre­dic­tions are based on a rel­a­tively small amount of data cap­tured by peo­ple try­ing to ac­cu­rately count buds and shoots and flow­ers and fruits. We know from many tri­als just how vari­able the re­sults from this type of col­lec­tion are. But it is the best we have, so we ex­trap­o­late this to guess whole or­chard and whole pack­house and whole sec­tor pro­duc­tion. And we are of­ten far from the mark.

Emma Leonard of SPAA re­ported (“Pre­ci­sion Viti­cul­ture in the River­land”) that wine mak­ers want an au­to­mated sys­tem that through­out the sea­son pre­dicts fi­nal yield to at least 95% ac­cu­racy. This com­pares to cur­rent sys­tems that are at best 85% ac­cu­rate at berry count­ing and 80% at bud count­ing. Lo­cal ob­ser­va­tions in the viti­cul­ture, ki­wifruit and ap­ple sec­tors sug­gest even poorer es­ti­mates are com­mon.

If a ma­chine could count buds, flow­ers and fruits, and mea­sure the size of fruits as the sea­son pro­gresses, would these es­ti­mates be bet­ter? Given cur­rent ac­cu­racy, the ma­chine hardly needs to be per­fect! But is it fea­si­ble?

A num­ber of groups in New Zealand and around the world are try­ing to de­velop tech­nolo­gies to count buds, flow­ers and fruits in ap­ple and ki­wifruit or­chards and vine­yards. Notably, all those I have seen are un­der de­vel­op­ment. Count­ing or­chard stuff with a ma­chine is very, very dif­fi­cult.

Count­ing flow­ers in not easy for peo­ple, let alone a ma­chine. The vari­abil­ity of an or­chard en­vi­ron­ment is far more com­plex than any fac­tory set­ting.

There is a mas­sive in­vest­ment in new sen­sor and data tech­nolo­gies de­signed to help grow­ers col­lect and man­age data and turn it into man­age­ment in­for­ma­tion.

Newspapers in English

Newspapers from New Zealand

© PressReader. All rights reserved.