1"use strict";(self.webpackChunkNioLandReact=self.webpackChunkNioLandReact||[]).push([[4763],{4763:(e,t,o)=>{o.r(t),o.d(t,{default:()=>j});o(5043);var r=o(5475),s=o(3519),i=o(1072),n=o(8602),a=o(3107),l=o(7497),c=o(579);const d="'Inter', sans-serif",h="#28a745",f="coffee-pest-forecasts-feedback-loop",m="What makes a coffee pest forecast different from a hypothesis.",p="Most coffee pest models predict the conditions for an outbreak \u2014 they cannot tell you whether one actually happened. Closing that loop every season is what turns a prediction into a forecast, and what makes the prediction defensible across four sustainability disclosures.",y="POLICY & COMPLIANCE",g="27 APRIL 2026",u="2026-04-27",b={fontFamily:d,fontWeight:700,fontSize:"clamp(1.5rem, 2.4vw, 1.9rem)",letterSpacing:"-0.015em",color:"#000",marginTop:"2.5rem",marginBottom:"1rem"},x={fontFamily:d,fontWeight:400,fontSize:"1.1rem",lineHeight:1.7,color:"#222",marginBottom:"1.1rem"},w={color:h,textDecoration:"underline",textDecorationThickness:"1px",textUnderlineOffset:"3px"};const j=function(){const e={"@context":"https://schema.org","@type":"BlogPosting",headline:m,description:p,datePublished:u,dateModified:u,url:`https://www.weathersafe.co.uk/blog/${f}`,author:{"@type":"Organization",name:"Weathersafe",url:"https://www.weathersafe.co.uk"},publisher:{"@type":"Organization",name:"Weathersafe",url:"https://www.weathersafe.co.uk"},mainEntityOfPage:`https://www.weathersafe.co.uk/blog/${f}`,articleSection:y};return(0,c.jsxs)(a.A,{title:`${m} \u2014 Weathersafe`,description:p,canonical:`https://www.weathersafe.co.uk/blog/${f}`,ogType:"article",jsonLd:e,children:[(0,c.jsx)("section",{style:{paddingTop:"8rem",paddingBottom:"4rem"},children:(0,c.jsx)(s.A,{children:(0,c.jsx)(i.A,{className:"justify-content-center",children:(0,c.jsxs)(n.A,{lg:"8",children:[(0,c.jsx)(r.N_,{to:"/blog",style:{fontFamily:d,fontWeight:600,fontSize:"0.85rem",color:h,textDecoration:"none",letterSpacing:"0.04em"},children:"\u2190 Blog"}),(0,c.jsxs)("div",{style:{display:"flex",alignItems:"center",gap:"0.85rem",marginTop:"2rem",marginBottom:"1rem"},children:[(0,c.jsx)("span",{style:{fontFamily:d,fontWeight:600,fontSize:"0.75rem",letterSpacing:"0.06em",color:h,background:"rgba(40,167,69,0.1)",borderRadius:"999px",padding:"0.35rem 0.9rem"},children:y}),(0,c.jsx)("span",{style:{fontFamily:d,fontWeight:500,fontSize:"0.85rem",letterSpacing:"0.04em",color:"#888"},children:g})]}),(0,c.jsx)("h1",{style:{fontFamily:d,fontWeight:800,fontSize:"clamp(2.4rem, 5vw, 3.8rem)",lineHeight:1.1,letterSpacing:"-0.03em",color:"#000",marginBottom:"1.5rem"},children:m}),(0,c.jsx)("p",{style:{fontFamily:d,fontWeight:400,fontSize:"1.25rem",lineHeight:1.55,color:"#555",marginBottom:0},children:p})]})})})}),(0,c.jsx)("section",{style:{paddingBottom:"5rem"},children:(0,c.jsx)(s.A,{children:(0,c.jsx)(i.A,{className:"justify-content-center",children:(0,c.jsxs)(n.A,{lg:"8",children:[(0,c.jsx)("p",{style:x,children:"Coffee pest models predict outbreaks all the time. Most of them are never checked against what actually happened in the field. The result is a model that might be right, and might be wrong, with no mechanism for telling the difference."}),(0,c.jsx)("p",{style:{...x,fontWeight:700,color:"#000"},children:"The model writes the question. The field writes the answer."}),(0,c.jsx)("h2",{style:b,children:"Why most coffee pest predictions stay hypotheses"}),(0,c.jsx)("p",{style:x,children:"Most pest-pressure models in commercial use are trained on broad datasets \u2014 weather conditions correlated with outbreak occurrence in the published literature, generalised across geographies and varieties. They predict that conditions are favourable for coffee leaf rust, or unfavourable, or borderline. They do not check whether rust actually appeared in the field after the prediction. They cannot. The data does not flow back."}),(0,c.jsx)("p",{style:x,children:"The result is a vendor selling a forecast that nobody can audit. The buyer either trusts the model or does not. The farmer applies \u2014 or does not apply \u2014 intervention based on a prediction that has no way to learn from being wrong. False positives erode farmer trust over a season; false negatives lose the harvest."}),(0,c.jsx)("h2",{style:b,children:"What closes the loop"}),(0,c.jsx)("p",{style:x,children:"Closing the loop requires three things together: every prediction has to be checked against what subsequently happened in the field; every confirmed outbreak has to be recorded against the conditions that preceded it; and every miss \u2014 every outbreak the model failed to predict \u2014 has to be folded back into the next season's calibration. The miss is the most valuable data point in the system. It is the only one that tells the model where it is
1wrong."}),(0,c.jsxs)("p",{style:x,children:["Weathersafe has been recording confirmed outbreaks at coffee origin for ",(0,c.jsx)("strong",{children:"thirteen years"})," across mapped farms in Kenya and Rwanda. Each one is tied to the conditions that preceded it. Each one feeds the calibration of the next season's pest and disease models."]}),(0,c.jsx)("h2",{style:b,children:"What four frameworks see in better predictions"}),(0,c.jsx)("p",{style:x,children:"Better predictions translate into disclosure-grade evidence across four sustainability frameworks, not just yield protection:"}),(0,c.jsxs)("ul",{style:{paddingLeft:"1.25rem",marginBottom:"1.25rem"},children:[(0,c.jsxs)("li",{style:{...x,marginBottom:"0.5rem"},children:[(0,c.jsx)("strong",{children:"CSRD ESRS S2 (Workers in value chain)"})," \u2014 accurate prediction means intervention timed to actual pressure, not blanket spray. The number of application events the farmer is involved in drops to those genuinely needed."]}),(0,c.jsxs)("li",{style:{...x,marginBottom:"0.5rem"},children:[(0,c.jsx)("strong",{children:"CSRD ESRS E4 (Biodiversity)"})," \u2014 targeted intervention means less collateral pressure on non-target species. Beneficial insects, soil microbes and surrounding biodiversity are preserved when the prediction tells the farmer to act on one farm rather than ten."]}),(0,c.jsxs)("li",{style:{...x,marginBottom:"0.5rem"},children:[(0,c.jsx)("strong",{children:"SBTi FLAG (productivity intensity)"})," \u2014 outbreaks prevented are yield retained. Higher yield per hectare drops the per-kg coffee emission factor that flows into the buyer's Scope 3 inventory."]}),(0,c.jsxs)("li",{style:{...x,marginBottom:"0.5rem"},children:[(0,c.jsx)("strong",{children:"IFRS S2 / TCFD (physical climate risk)"})," \u2014 forward-looking pest pressure under climate scenarios is precisely the type of physical-risk disclosure that the buyer's anticipated financial effects section asks for. A calibrated forecast is a defensible input; an uncalibrated one is not."]})]}),(0,c.jsx)("h2",{style:b,children:"How Weathersafe runs the prediction layer"}),(0,c.jsxs)("p",{style:x,children:["At every washing station running GAP77, three data streams run continuously into the model: ",(0,c.jsx)(r.N_,{to:"/blog/why-coffee-farms-get-wrong-rain-forecast",style:w,children:"hyperlocal weather forecasts"})," at the farm-cluster level, satellite stress signals on every mapped farm, and ",(0,c.jsx)(r.N_,{to:"/blog/every-coffee-farm-every-two-weeks",style:w,children:"scout observations every two weeks per farm"}),". The model fuses these into pest and disease event templates \u2014 predictions of which farms are at elevated risk, when, for which pressure (coffee leaf rust, Antestia, leaf miner and others)."]}),(0,c.jsx)("p",{style:x,children:"When conditions for one of those events are met on a specific farm, the alert is recorded against that farm in the farmer database and an SMS goes to the farmer. If mitigation is applied to the field, the application is recorded against the same farm \u2014 alert, response and outcome all attributable to the same prediction. The buyer's compliance team can audit the prediction, the farmer notification, and the action taken."}),(0,c.jsx)("p",{style:x,children:"What makes the predictions different from generic published models is the calibration history. Every prediction can be traced back to similar conditions in the past, the recorded outcome on those specific farms, and the model's adjustment since."}),(0,c.jsxs)("p",{style:x,children:["An alert without pre-positioned response capacity is a notification with no follow-through; the bio-input stocks at the station are what turn the prediction into a recorded outcome, covered in ",(0,c.jsx)(r.N_,{to:"/blog/coffee-bio-input-stocks",style:w,children:"why coffee's pesticide footprint is a reduction lever"}),". The same calibration-against-ground-truth principle drives the cherry yield forecast, covered in ",(0,c.jsx)(r.N_,{to:"/blog/coffee-yield-forecast-13-years-ground-truth",style:w,children:"why coffee yield forecasts need 13 years of ground truth"}
1),". The integrated weather-plus-satellite-plus-scout architecture is in the ",(0,c.jsx)(r.N_,{to:"/methodology",style:w,children:"methodology"}),"."]})]})})})}),(0,c.jsx)(l.Cu,{className:"py-10 py-lg-90",style:{borderTop:"1px solid #f0f0f0",borderBottom:"1px solid #f0f0f0"},children:(0,c.jsx)(l.Cu.Content,{children:(0,c.jsx)(s.A,{children:(0,c.jsxs)(i.A,{className:"align-items-start",children:[(0,c.jsxs)(n.A,{lg:"5",style:{marginTop:"-0.75rem"},children:[(0,c.jsxs)("h2",{style:{fontFamily:d,fontWeight:800,fontSize:"clamp(2.2rem, 4vw, 3.5rem)",letterSpacing:"-0.03em",lineHeight:1.05,color:"#000",marginBottom:"2rem"},children:["Forecasts that",(0,c.jsx)("br",{}),"learn from misses."]}),(0,c.jsxs)("div",{style:{display:"flex",gap:"1rem",flexWrap:"wrap"},children:[(0,c.jsx)(r.N_,{to:"/contact-us-solution",style:{fontFamily:d,fontWeight:600,fontSize:"0.95rem",color:"#fff",background:h,borderRadius:"6px",padding:"0.75rem 1.75rem",textDecoration:"none"},children:"Talk to us"}),(0,c.jsx)(r.N_,{to:"/methodology",style:{fontFamily:d,fontWeight:600,fontSize:"0.95rem",color:"#000",background:"#fff",border:"1px solid #d0d0d0",borderRadius:"6px",padding:"0.75rem 1.75rem",textDecoration:"none"},children:"Read the methodology"})]})]}),(0,c.jsx)(n.A,{lg:"5",className:"offset-lg-2",children:(0,c.jsx)("p",{style:{fontFamily:d,fontWeight:400,fontSize:"1.55rem",lineHeight:1.4,color:"#333",marginBottom:0},children:"Thirteen years of confirmed outbreaks. Folded back into every season's calibration."})})]})})})})]})}}}]); 2//# sourceMappingURL=4763.cbc378f8.chunk.js.map
Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.