A hybrid model for optimizing fungicide application schedules.

Abstract

The present invention relates to fungal disease management. A computer-implemented method (200) is provided for determining disease progression that can be used to schedule fungicide applications in an agricultural field to improve fungal disease management. The method includes receiving (210) data including crop variety data, environmental data, crop management data, and location data for the agricultural field. The crop variety data relates to a crop grown or to be grown in the agricultural field. The environmental data is indicative of environmental conditions of the agricultural field. The crop management data is indicative of fungicide application history for the agricultural field. The method further includes applying (220) a machine learning model to the received data to determine disease progression time series data for the fungal disease, the machine learning model being trained to learn disease progression under conditions defined by the crop variety data, the environmental data, the crop management data, and the location data based on historical data collected from one or more agricultural fields. The method further comprises determining (230) a disease onset date for the fungal disease based on the determined time series data of disease progression.

Description

[Technical field]

[0001] The present invention relates to the management of fungal diseases. In particular, the present invention relates to a computer implemented method and apparatus, and computer program elements, for determining disease progression that can be used to schedule fungicide applications in agricultural fields. [Background technology]

[0002] In agriculture, fungicides are used. Plant fungal diseases are very diverse and affect all parts of the plant, i.e. the tip (e.g. common root rot, crown root rot, sclerotium wilt), the stem or sheath (e.g. eye spot, black rust), the leaves (e.g. black leaf blight, rust, powdery mildew), the ear (e.g. ergot), and the seeds (e.g. scab, carnal bunt). The general recommendation for the timing of fungicide application is before the start of sowing, which is the primary infection. However, the start date of the primary infection of crop diseases can vary from season to season, depending on the availability of the infective agent and weather conditions. Thus, the management of fungal diseases becomes more complicated. Summary of the Invention [Problem to be solved by the invention]

[0003] Improved management of fungal diseases may be required. [Means for solving the problem]

[0004] The object of the invention is solved by the subject matter of the independent claims, in which further embodiments are incorporated. It should be noted that the below-described aspects of the invention also apply to computer-implemented data processing methods and devices, computer program elements, and computer-readable media.

[0005] According to a first aspect of the present invention there is provided a computer implemented method for determining disease progression that can be used to schedule fungicide applications in an agricultural field, comprising the steps of: a) Data, which is: - Crop variety data on the crops grown or to be grown on the agricultural field; - Environmental data indicating the environmental conditions of agricultural fields; - Crop management data showing fungicide application history for agricultural fields; and - Agricultural field location data receiving data including; b) applying a machine learning model to the received data to determine time series data of disease progression of the fungal disease, the machine learning model being trained to learn disease progression under conditions defined by the crop variety data, the environmental data, the crop management data, and the location data based on historical data collected from one or more agricultural fields; and c) determining the disease onset date of fungal diseases based on the determined time series data of disease progression; A method is provided that includes:

[0006] The computer-implemented methods and apparatus described herein apply machine learning models to identify primary infection (i.e., the onset date of primary infection of a crop disease).

[0007] The machine learning model may be any suitable model capable of determining the time series data of disease progression of fungal disease. Examples of the machine learning model may include, but are not limited to, XGB regression model, Artificial Neural Network, Recurrent Neural Network, and Support Vector Regression.

[0008] The onset date of primary infection (also referred to as disease onset) of a crop disease can be used to determine disease progression curves. For example, a process-based model can be used to simulate the interaction between disease onset date and varietal susceptibility to determine secondary infection (described in more detail below).

[0009] To understand how plant diseases develop and how to manage disease to maximize the efficacy of fungicides, it is essential to determine disease development and disease progression curves to characterize disease progression over time. Based on such information, fungicide application schedules can be determined. In this way, application timing can be set to take into account changes in disease dynamics, thereby improving treatment. For example, application timing can be set based on predicted disease risk, allowing fungicides to be applied when they are most effective during the growing season. In particular, the determined fungicide application schedule can target optimal application periods to more reliably halt disease progression and is less likely to miss risk periods. Optimal application timing can have a significant effect on the overall reduction of fungicide use, and can also reduce agricultural damage, reducing losses in yield, quality, and profits.

[0010] According to an embodiment of the present invention, the disease onset date of a fungal disease is determined utilizing a change-point detection algorithm.

[0011] According to an embodiment of the present invention, multiple machine learning models are provided for two or more fungal diseases, each machine learning model being trained on a single disease.

[0012] In this way, each machine learning model is specifically trained to determine disease onset and disease progression curves to characterize the progression over time for a particular fungal disease.

[0013] According to an embodiment of the present invention, the machine learning model includes an Xtreme Gradient Boosting (XGB) regression model.

[0014] In some examples, the model choice may be an XGB regression model tuned with Randomized Search Cross Validation. XGB is a scalable tree boosting system, which may be thought of as gradient boosting with regularization. This may allow for parallel tree construction, cache-aware access, sparsity awareness, and weighted quantile sketches as part of the optimization and algorithmic enhancements of the system.

[0015] According to an embodiment of the present invention, the computer-implemented method further includes d) applying the process-based model to determine a fungal disease infection rate from the disease onset date onwards under conditions defined by the crop variety data, the environmental data, and the location data.

[0016] In other words, a process-based model is proposed to simulate secondary infection, the process includes susceptibility, exposure, infection, and removal. The infection rate takes into account environmental factors including temperature and humidity. The infection rate calculation also takes into account cultivar susceptibility, fungicide, and crop growth stage. The process-based model can be used to simulate the interaction between disease onset date, cultivar susceptibility, and fungicide application. The infection rate of fungal diseases can be preferably provided in the form of a disease progression curve that can be used to optimize fungicide schedules.

[0017] According to an embodiment of the present invention, the infection rate of a fungal disease is determined by further including a condition defined by the crop variety disease resistance level.

[0018] According to an embodiment of the invention, the infection rate of a fungal disease is determined by further including conditions defined by fungicide application data, which includes fungicide data for a fungicide product to be used and at least one planned application timing.

[0019] In other words, the process-based model can be used to simulate the interactions between disease onset date, cultivar susceptibility, and fungicide application. By analyzing the effect of fungicide application on disease progression curves such as the curve shown in Figure 25, it becomes possible to determine fungicide application schedules.

[0020] According to an embodiment of the present invention, the process-based model includes a susceptibility-exposure-infection-elimination (SEIR) model.

[0021] According to an embodiment of the present invention, the crop variety data includes one or more of the crop's developmental stage, days since planting, and the crop's variety disease resistance level.

[0022] According to an embodiment of the present invention, the environmental data includes one or more of air temperature, cloud cover, shortwave radiation, longwave radiation, ice accumulation duration, liquid accumulation duration, relative humidity, adjusted precipitation duration, snow accumulation duration, and wind speed.

[0023] According to an embodiment of the present invention, the location data includes latitude and longitude data.

[0024] According to an embodiment of the invention, the computer-implemented method further comprises e) determining a fungicide application schedule based on the determined disease progression.

[0025] In some examples, the fungicide application schedule may include spray application timing data.

[0026] In some examples, a fungicide application schedule can be provided to a farmer who then controls the applicator to apply the fungicide according to the spray application timing data.

[0027] In some examples, a configuration file is created based on the fungicide application schedule, which can be loaded into an applicator to configure the applicator to apply fungicide according to the spray application timing data.

[0028] According to an embodiment of the present invention, the computer implemented method further includes f) creating a configuration file that can be used to configure the applicator for fungicide spray application, preferably based on the fungicide spray schedule.

[0029] According to a second aspect of the present invention there is provided an apparatus for generating a disease progression usable for scheduling fungicide applications in an agricultural field, the apparatus comprising one or more processing units for generating an application scheme, the one or more processing units comprising instructions which, when executed on the one or more processing units, carry out the steps of the method according to any one of the preceding claims.

[0030] In general, the functionality of any one or more components of the apparatus may be implemented in any suitable computing environment, such as a personal computing environment, a time-sharing computing environment, a distributed computing environment, a cloud computing environment, and a cluster computing environment.

[0031] In some examples, the apparatus may be embodied as or in a device or apparatus such as a server, workstation, or mobile device.

[0032] In some examples, the apparatus may be implemented with or without a processor, and may also be implemented as a combination of dedicated hardware for performing some functions and a processor (e.g., one or more programmed microprocessors and associated circuitry) for performing other functions.

[0033] In some examples, the apparatus may also be implemented in a distributed manner, for example some or all of the units of the apparatus may be arranged as individual modules in a distributed architecture and connected by a suitable communication network.

[0034] According to a third aspect of the present invention there is provided a computer program element comprising instructions for causing an apparatus according to the second aspect to carry out the steps of the method according to the first aspect and any associated examples.

[0035] As used herein, the term "agricultural field" is understood to be any area where organisms, particularly crop plants, are produced, grown, sown, and / or planned to be produced, grown, or sown. The term "agricultural field" also includes horticultural fields, forestry fields, and fields for producing and / or growing aquatic organisms.

[0036] As used herein, the term "crop" refers to a plant that can be grown and harvested on a large scale for profit or livelihood. Examples of crops may include, but are not limited to, onion (Allium cepa), pineapple (Ananas comosus), peanut (Arachis hypogaea), asparagus (Asparagus officinalis), oats (Avena sativa), sugar beet (Beta vulgaris spec. altissima), beet (Beta vulgaris spec. rapa), rapeseed (Brassica napus var. napus), rutabaga (Brassica napus var. napobrassica), Brassica rapa var.silvestris, Brassica oleracea, Brassica nigra, Tea plant (Camellia sinensis), Safflower (Carthamus tinctorius), Pecan (Carya illinoinensis), Lemon (Citrus limon), Orange (Citrus sinensis), Coffee plant (Coffea arabica), (Coffee plant (Coffea canephora), Coffee plant (Coffea liberica), Cucumis sativus, Corngrass (Cynodon dactylon), Carrot (Daucus carota), Oil palm (Elaeis guineensis), Wild strawberry (Fragaria vesca), Soybean (Glycine max), Upland cotton (Gossypium hirsutum), (Gossypium arboreum, Gossypium herbaceum, Gossypium vitifolium), sunflower (Helianthus annuus), rubber tree (Hevea brasiliensis), barley (Hordeum vulgare), hops (Humulus lupulus), sweet potato (Ipomoea batatas), Chinese walnut (Juglans regia), lentil (Lens culinaris), flax (Linum usitatissimum), tomato (Lycopersicon lycopersicum), apple (Malus spec.), cassava (Manihot esculenta), alfalfa (Medicago sativa), banana (Musa spec.), tobacco (Nicotiana tabacum) (N. rustica), olive (Olea europaea), rice (Oryza sativa), lima bean (Phaseolus lunatus), kidney bean (Phaseolus vulgaris), Norway spruce (Picea abies), pine (Pinus spec.), pistachio (Pistacia vera), pea (Pisum sativum), wild cherry (Prunus avium), peach (Prunus persica), pear (Pyrus communis), apricot (Prunus armeniaca), sweet cherry (Prunus cerasus), almond (Prunus dulcis) and plum (Prunus domestica), redcurrant (Ribes sylvestre), castor bean (Ricinus communis), sugar cane (Saccharum officinarum), rye (Secale cereale), white mustard (Sinapis alba), potato (Solanum tuberosum), sorghum (Sorghum bicolor) (s. vulgare), cocoa (Theobroma cacao), cacao, red clover (Trifolium pratense), bread wheat (Triticum aestivum), triticale (Triticale), durum wheat (Triticum durum), broad bean (Vicia faba), European grape (Vitis vinifera), and corn (Zea mays). Most preferred crops are: peanut (Arachis hypogaea), sugar beet (Beta vulgaris spec. altissima), rapeseed (Brassica napus var.napus, Brassica oleracea, Lemon (Citrus limon), Orange (Citrus sinensis), Coffee plant (Coffea arabica), (Coffee plant (Coffea canephora), Coffee plant (Coffea liberica), Corngrass (Cynodon dactylon), Soybean (Glycine max), Cotton plant (Gossypium hirsutum), (Gossypium arboreum, Gossypium herbaceum, Gossypium vitifolium), Sunflower (Helianthus annuus), Barley (Hordeum vulgare), Walnut (Juglans regia), Lentil (Lens culinaris), flax (Linum usitatissimum), tomato (Lycopersicon lycopersicum), apple (Malus spec.), alfalfa (Medicago sativa), tobacco (Nicotiana tabacum) (N. rustica), olive (Olea europaea), rice (Oryza sativa), lima bean (Phaseolus lunatus), common bean (Phaseolus vulgaris), pistachio (Pistacia vera), pea (Pisum sativum), almond (Prunus dulcis), sugar cane (Saccharum officinarum), rye (Secale cereale), potato (Solanum tuberosum), sorghum (Sorghum bicolor) (s. vulgare) (s.vulgare), triticale, bread wheat, durum wheat, Vicia faba, Vitis vinifera, and corn. Preferred crops are cereals, maize, soybean, rice, rapeseed, cotton, potato, peanut, or perennial crops.

[0037] In a preferred embodiment of the invention, by way of example, the disease or plant disease may be caused by one or more of the following pathogens: Albugo species in ornamental plants, vegetables (e.g., A. candida) and sunflowers (e.g., A. tragopogonis) spp.) (white rust); Alternaria species on vegetables (e.g. A. dauci or A. porri), rapeseed (A. brassicicola or brassicae), sugar beet (A. tenuis), fruits (e.g. A. grandis), rice, soybean, potato, and tomato (e.g. A. solani, A. grandis, or A. alternata), tomato (e.g. A. solani or A. alternata), and wheat (e.g. A. triticina). spp. (Alternaria leaf spot); Aphanomyces spp. on sugar beet and vegetables; Ascochyta spp. on cereals and vegetables, e.g. A. tritici (anthracnose) on wheat and A. hordei on barley; Aureobasidium zeae (synonym Kapatiella zeae) on maize; Bipolaris spp. and Drechslera spp. (teleomorph: Cochliobolus spp.) spp.), for example brown spot (D. maydis) or brown spot (B. zeicola) on corn, for example spot disease (B. sorokiniana) on cereals, and B. oryzae (B.oryzae; Blumeria (formerly Erysiphe) graminis (powdery mildew) on cereals (e.g. wheat or barley); Botrytis cinerea (teleomorph: Botryotinia fuckeliana: gray mold) on fruits and berries (e.g. strawberries), vegetables (e.g. lettuce, carrots, celery, and cabbage); B. squamosa or gray rot (B. allii) on onion, rapeseed, ornamentals (e.g. B eliptica), grapes, forest plants, and wheat; Bremia lactucae on lettuce lactucae (downy mildew); Ceratocystis (syn. Ophiostoma) spp. (root rot or dieback) on deciduous and evergreen trees, e.g. C. ulmi (Dutch elm disease) on elms; Cercospora spp. (Cercospora leaf spot) on maize (e.g. gray spot: C. zeae-maydis), rice, sugar beet (e.g. C. beticola), sugar cane, vegetables, coffee, soybean (e.g. C. sojina or C. kikuchii) and rice; Cladobotryum (syn. Dactylium) spp. (Cercospora leaf spot) on mushrooms. mycophilum (formerly Dactylium dendroides, teleomorphs: Nectria albertinii, Nectria rosella synonymous with Hypomyces rosellus); tomato (e.g. C. fulvum: leaf mold) and Cladosporium species on cereals, e.g. C. herbarum (black spot) on wheat; Claviceps purpurea (ergot) on cereals; maize (C.Helminthosporium of Cochliobolus (anamorph: Bipolaris) on cereals (e.g., C. carbonum), cereals (e.g., C. sativus, anamorph: B. sorokiniana), and rice (e.g., C. miyabeanus, anamorph: H. oryzae). cotton (e.g., C. gossypii), maize (e.g., C. graminicola: anthracnose root rot), soft fruits, potatoes (e.g., C. coccodes: black spot), legumes (e.g., C. lindemuthianum), soybeans (e.g., C. tomentosa: corn spot), C. truncatum or C. gloeosporioides), vegetables (e.g., C. lagenarium or C. capsici), fruits (e.g., C. acutatum), coffee (e.g., C. coffeenum or C. kahawae) Colletotrichum (teleomorph: Glomerella) species (anthracnose), and C. gloeosporioides on various crops; Corticium species, e.g. C. sasakii (sheath blight) on rice; Corynespora cassiicola (leaf spot) on soybean, cotton, and ornamentals; Cycloconium species, e.g. C. oleaginum on olive trees; fruit trees, grapes (e.g. C. liriodendron), and other crops.liriodendri, teleomorph: Neonectria liriodendri, black leg disease) and Cylindrocarpon spp. on ornamental plants (e.g. fruit canker disease or young grapevine decline, teleomorph: Nectria spp. or Neonectria spp.); Dematophora (teleomorph: Rosellinia) on soybean; necatrix (root and stem rot disease); Diaporthe spp. on soybean spp.), e.g. D. phaseolorum (seedling damping off); (synonym of Helminthosporium, teleomorph: corn, cereals such as barley (e.g. D. teres, net blotch) and wheat (e.g. D. tritici-repentis: tan spot), rice, and turfgrass. Drechslera (synonym: Helminthosporium, teleomorph: Pyrenophora) species; Formitiporia (synonym: Phellinus) - punctata, F. mediterranea, Phaeomoniella - Chlamydospora Esca disease (canker, apoplexy) on grapes caused by Phaeoacremonium chlamydospora (formerly Phaeoacremonium chlamydosporum), Phaeoacremonium aleophilum, and / or Botryosphaeria obtusa; Elsinoe spp. on pome fruits (E. pyri), soft fruits (E. veneta: anthracnose) and grapes (E. ampelina: anthracnose).; Entyloma oryzae (leaf mildew) on rice; Epicoccum spp. (black mold) on wheat; Erysiphe spp. (powdery mildew) on sugar beet (E. betae), vegetables (e.g. E. pisi), e.g. cucurbits (e.g. E. cichoracearum), cabbage, rapeseed (e.g. E. cruciferarum); Eutypa lata (Eutypa canker or blight, anamorph: Cytosporina lata, Libertella blepharis) on fruit trees, grapes, and ornamentals. blepharis); Exserohilum (synonymous with Helminthosporium) species on corn; Fusarium (teleomorph: Gibberella) species on various plants (damage, root or stem rot), for example F. graminearum or F. culmorum on cereals (e.g. wheat or barley) (root rot, scab or red mold disease), F. oxysporum on tomato, orum, F. solani (synonymous with f. sp. glycines, now F. virguliforme) and F. tucumaniae and F. brasiliense, which cause sudden death syndrome on soybean, and F. verticillioides on maize; Gaeumannomyces graminis (damaging disease) on cereals (e.g. wheat or barley) and maize; Gibberella spp. on cereals (e.g. G. zeae) and rice (e.g. G. fujikuroi: bakanae disease); Glomerella cingulata on grapes, pome fruits, and other plants, and G. malariae on cotton.gossypii; grain stain complex on rice; Guignardia bidwellii (black rot) on grapes; Gymnosporangium spp. on roses and junipers, e.g. G. sabinae on pear (rust); Helminthosporium spp. (syn. Drechslera, teleomorph: Cochliobolus) on maize, cereals, potato, and rice; Hemileia spp., e.g. H. vastatrix on coffee (coffee leaf rust); Isariopsis clavispora on grapes. clavispora (synonymous with Cladosporium vitis); Macrophomina phaseolina (synonymous with phaseoli) on soybean and cotton (root and stem rot disease);. Microdochium (synonymous with Fusarium) nivale (pink snow mold) on cereals (e.g. wheat or barley); Microsphaera diffusa (powdery mildew) on soybean; Monilinia spp., e.g. M. laxa, M. fructicola, and M. fructigena on stone fruits and other Rosaceae (synonymous with Monilia spp.: flower and branch blight, brown rot); Mycosphaerella spp., e.g. M. graminicola on wheat (anamorph: Zymoseptoria spp. on cereals, bananas, soft fruits, and groundnuts); tritici, formerly Septoria tritici (Septoria leaf spot), or M. fijiensis (synonymous with Pseudocercospora fijiensis (Black Sigatoka disease)) on bananas and M. musicola, M. arachidicola (M. arachidis or Cercospora arachidis) on groundnuts. arachidis), M. berkeleyi, M. pisi on pea, and M. brassiciola on cruciferous plants; Peronospora spp. (downy mildew) on cabbage (e.g. P. brassicae), rapeseed (e.g. P. parasitica), onion (e.g. P. destructor), tobacco (P. tabacina), and soybean (e.g. P. manshurica); Phakopsora pachyrhizi and P. meibomiae (soybean rust) on soybean; e.g. grapes (e.g. P.Phialophora spp. on soybean (e.g. P. tracheiphila and P. tetraspora) and soybean (e.g. P. gregata: stem rot); Phoma lingam on rapeseed and cabbage (synonymous with Leptosphaeria biglobosa and L. maculans: root and stem rot), P. betae on sugar beet (root rot, spot and damping off), and P. zeae-maydis on maize (Phyllostica zeae); Phomopsis spp. on sunflower, grape (e.g. P. viticola: stem spot), and soybean (e.g. stem rot: P. phaseoli, teleomorph: Diaporthe phaseolorum); Physoderma maydis on maize maydis (brown spot disease); Phytophthora spp. (damage, root, leaf, fruit and stem rot diseases) on various plants, e.g., peppers and cucurbits (e.g., P. capsici), soybean (e.g., P. megasperma, synonymous with P. sojae), potato and tomato (e.g., P. infestans: leaf rot disease), and on deciduous trees (e.g., P. ramorum: oak death); Plasmodiophora brassicae (club root disease) on cabbage, rapeseed, radish, and other plants; Plasmopara spp. spp.), e.g. P. viticola on grapes (downy mildew of grapes) and P. halstedii on sunflowers; Rosaceae, hops, pome fruits, and soft fruits (e.g. P. leucotricha on apples) and Cucurbits (P.Podosphaera spp. (powdery mildew) on cereals such as barley and wheat (P. graminis) and sugar beet (P. betae) and the viral diseases transmitted thereby; Pseudocercosporella herpotrichoides (synonym Oculimacula yallundae, O. acuformis: eye spot, teleomorph: Tapesia yallundae) on cereals such as wheat or barley. yallundae); Pseudoperonospora (downy mildew) on various plants, e.g. P. cubensis on cucurbits or P. humili on hops; Pseudopezicula tracheiphila (red fireworks or "rotbrenner", anamorph: Phialophora) on grapes; Puccinia spp. on various plants spp. (rusts), e.g. P. triticina (brown rust or leaf rust), P. striiformis (stripe rust or yellow rust), P. hordei (stunt rust), P. graminis (stem rust or black rust) or P. recondita (brown rust or leaf rust) on cereals such as wheat, barley or rye, P. kuehnii (orange rust) on sugarcane and P. asparagi on asparagus; Pyrenopeziza spp., e.g. P. brassicae on rapeseed.brassicae; Pyrenophora (anamorph: Drechslera) tritici-repentis (tan spot) on wheat or P. teres (net blotch) on barley; Pyricularia spp., e.g. P. oryzae (teleomorph: Magnaporthe grisea: rice blast or leaf blast) on rice and P. grisea on turfgrass and cereals; Magnaporthe oryzae (panicle-neck blast) on rice. blast); Pythium spp. (seedling damping off) on turfgrass, rice, corn, wheat, cotton, rapeseed, sunflower, soybean, sugar beet, vegetables, and various other plants (e.g. P. ultimum or P. aphanidermatum), and P. oligandrum on mushrooms; Ramularia spp., e.g. R. collo-cygni (ramularia leaf spot, physiological leaf spot) on barley, R. areola (teleomorph: Mycosphaerella areola) on cotton, areola) and R. beticola on sugar beet; Rhizoctonia spp. on cotton, rice, potato, turfgrass, maize, rapeseed, potato, sugar beet, vegetables, and various other plants. For example, R. solani on soybean (root and stem rot), R. solani on rice (sheath blight), or R. cerealis on wheat or barley (Rhizoctonia spring blight); Rhizopus stolonifer on strawberry, carrot, cabbage, grape, and tomato (black mold, soft rot); Rhynchosporium secalis and R. commune on barley, rye, and triticale (fire blight); Sarocladium oryzae on rice. oryzae and S. attenuatum (sheath rot); Sclerotinia spp. (stalk rot or white mold) on vegetables (S. minor and S. sclerotiorum) and field crops, e.g., on rapeseed, sunflower (e.g., S. sclerotiorum), and soybean, S. rolfsii (synonym Athelia rolfsii) on soybean, peanut, vegetables, corn, cereals, and ornamentals; Septoria spp. on various plants, e.g., S. glycines (brown spot) on soybean, S. tritici (Zymoseptoria spp.) on wheat. tritici (syn., Septoria leaf spot) on cereals, and S. (syn., Stagonospora) nodorum (Stagonaspora leaf spot) on cereals; Uncinula (syn., Erysiphe) necator (powdery mildew, anamorph: Oidium tuckeri) on grapes; Setosphaeria spp. (black leaf blight) on maize (e.g., S. tulsicum) on corn.turcicum, synonym Helminthosporium turcicum, and in turfgrass; Sphacelotheca species (sooty mildew) on corn (e.g. S. reiliana: synonym Ustilago reiliana: smut), Sphacelotheca species (sooty mildew) on sorghum and sugarcane; Sphaerotheca fuliginea on cucurbits (synonym Podosphaera xanthii: powdery mildew); Spongospora subbrellanea on potato subterranea (powdery scab) and the viral diseases transmitted thereby; Stagonospora spp. on cereals, e.g. S. nodorum on wheat (Stagonospora spot, teleomorph: Leptosphaeria [syn. Phaeosphaeria] nodorum, syn. Septoria nodorum); Synchytrium endobioticum on potato (potato wart); Taphrina spp., e.g. T. deformans on peach (leaf curl) and T. pruni on plum (pocket plum); Thielaviopsis spp. on tobacco, pome fruit, vegetables, soybean, and cotton. spp. (black root rot), e.g. T. basicola (syn. Chalara elegans); Tilletia spp. (common or smut) on cereals, e.g. T. tritici (syn. T. caries, wheat smut) and T. controversa (syn. T.controversa (stunt smut); Trichoderma harzianum on mushrooms; Typhula incarnata (gray snow mold) on barley or wheat; Urocystis spp., for example U. occulta (striped sooty mold) on rye; legumes (for example U. appendiculatus, synonymous with U. phaseoli), sugar beet (for example U. betae or U. beticola), and legumes (for example U. vignae, U. pisi, U. viciae-fabae, and U. fabae). fabae; Ustilago spp. (corn smut) on cereals (e.g. U. nuda and U. avaenae), maize (e.g. U. maydis: maize smut) and sugarcane; Ustilaginoidea virens (rice smut) on rice; Venturia spp. (scab) on apple (e.g. V. inaequalis) and pear; and Verticillium spp. (scab) on various plants, e.g. fruit and ornamental plants, grapes, soft fruits, vegetables and field crops. spp. (damaging diseases), e.g. V. longisporum on rapeseed, V. dahliae on strawberry, rapeseed, potato, and tomato, and V. fungicola on mushrooms; Xanthomonas oryzae (bacterial leaf blight) on rice; Zymoseptoria tritici on cereals.

[0038] In a preferred embodiment of the invention, the disease is a plant disease caused by: Sheath blight (CORTSS), leaf blast (also known as rice blast; PYRIOR), ear blast (PYRPRO), bacterial leaf blight (XANTOR), and rice false rot (USTNVI).

[0039] These and other aspects of the invention will become apparent and elucidated with reference to the embodiments described by way of example in the following description and the accompanying drawings, in which: FIG. [Brief description of the drawings]

[0040] [Figure 1] 1 shows a block diagram of an exemplary apparatus for determining disease progression applicable to fungicide application schedules in agricultural fields. [Diagram 2] FIG. 1 shows a diagram of a disease model simulating primary and secondary infections. [Figure 3A-B] After data cleaning and processing, the distribution of observations considered for sheath blight (CORTSS), leaf blast (PYRIOR), ear blast (PYRPRO), bacterial leaf blight (XANTOR), and rice false rot (USTNVI) is shown (top left). [Figure 4] 1 shows an example of disease severity interpolation using exponential growth curves. [Diagram 5] The distribution of the number of observations considered for each disease across all countries is shown. [Figure 6] A comparative analysis of disease severity observed for test and untreated areas is presented. [Figure 7] The raw prediction and the smoothed raw prediction are shown. [Figure 8] The distribution of the difference in days between the observed and predicted primary infection dates for leaf blast (PYRIOR), neck blast (PYRPRO), and bacterial leaf blight (XANTOR) for all untreated areas in Japan is shown. [Figure 9] The distribution of true positives, false positives, true negatives, and false negatives is shown. [Figure 10]This shows the geospatial distribution of true positives, false positives, false negatives, and true negatives for predicting sheath blight occurrence in test areas in Japan. [Figure 11] The geospatial distribution of true positives, false positives, false negatives, and true negatives for predicting leaf blast occurrence in test areas in Japan is shown. [Figure 12] This shows the geospatial distribution of true positives, false positives, false negatives, and true negatives for predicting the occurrence of rice neck blast in test areas in Japan. [Figure 13] This shows the geospatial distribution of true positives, false positives, false negatives, and true negatives for predicting rice false-smut occurrence in test areas in Japan. [Figure 14] This shows the geospatial distribution of true positives, false positives, false negatives, and true negatives for predicting bacterial leaf blight occurrence in test areas in Japan. [Figure 15] Accuracy, precision, and recall are shown. [Figure 16] Shows feature importance for all models. [Figure 17] The geographical spatial distribution of predicted sheath blight occurrence is shown. [Figure 18] The geographical spatial distribution of predicted leaf blast disease occurrence is shown. [Figure 19] The geographical spatial distribution of predicted rice neck blast occurrence is shown. [Figure 20] The geographical spatial distribution of rice false-smut disease occurrence predictions is shown. [Figure 21] The geographical spatial distribution of bacterial leaf blight occurrence predictions is shown. [Figure 22] Figure 1 shows the geospatial distribution of maximum severity predicted by models trained on sheath blight (CORTSS), leaf blast (PYRIOR), ear blast (PYRPRO), bacterial leaf blight (XANTOR), and rice false rot (USTNVI) for all disease-occurring regions. [Diagram 23] 1 shows the effect of disease onset date on disease progression curves. [Figure 24] 1 shows the effect of cultivar disease resistance level on disease progression curves. [Diagram 25] Figure 2 shows the effect of fungicide application on disease progression curves. [Figure 26] 1 shows a fungal disease management system. [Figure 27] 1 shows a flow chart illustrating a computer-implemented method for determining disease progression that can be used to schedule fungicide applications in agricultural fields. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0041] It should be noted that these drawings are purely schematic and are not drawn to scale. In these drawings, elements corresponding to elements already described may be given the same reference numerals. Examples, embodiments, or any features, whether or not shown as non-limiting, should not be understood as limiting the claimed invention.

[0042] 1 shows a block diagram of an exemplary apparatus 10 for determining disease progression that can be used to schedule fungicide spraying in an agricultural field. The exemplary apparatus 10 includes an input unit 12, a primary infection determination unit 14, and an output unit 18. Optionally, as shown in FIG. 1, the apparatus 10 may further include a secondary infection determination unit 16, and a fungicide spraying schedule determination unit 20.

[0043] In general, the apparatus 10 may include various physical and / or logical components for communicating and manipulating information, which may be implemented as hardware components (e.g., computing devices, processors, logic devices), executable computer program instructions (e.g., firmware, software) that may be executed by the various hardware components, or any combination thereof, as desired for a given set of design parameters or performance constraints. While FIG. 1 may show a limited number of components as an example, it may be understood that more or fewer components may be employed for a given implementation. Furthermore, the functionality provided by one or more components of the apparatus 10 may be combined or separate. Furthermore, the functionality of any one or more components of the apparatus 10 may be implemented by any suitable computing environment, such as a personal computing environment, a time-sharing computing environment, a distributed computing environment, a cloud computing environment, and a cluster computing environment. An exemplary distributed computing environment is shown in FIG. 26.

[0044] In some implementations, the device 10 may be embodied as or in a device or apparatus, such as a server, a workstation, or a mobile device. The device 10 may include one or more microprocessors or computer processors, which execute appropriate software. The primary infection determination unit 14, the secondary infection determination unit 16, and the fungicide application schedule determination unit 20 of the device 10 may be embodied by one or more of these processors. The software may be downloaded and / or stored in a corresponding memory (e.g., a volatile memory such as RAM, or a non-volatile memory such as Flash). The software may include instructions that configure the one or more processors to perform the functions described herein.

[0045] It should be noted that the apparatus 10 may be implemented with or without a processor, and may also be implemented as a combination of dedicated hardware for performing some functions and a processor (e.g., one or more programmed microprocessors and associated circuits) for performing other functions. For example, the functional units of the apparatus 10 (e.g., the input unit 12, the primary infection determination unit 14, the secondary infection determination unit 16, the fungicide spray schedule determination unit 20, and the output unit 18) may be implemented in the form of programmable logic (e.g., as a field programmable gate array (FPGA)) in a device or apparatus. In general, each functional unit of the apparatus may be implemented in the form of a circuit.

[0046] In some implementations, the device 10 may also be implemented in a distributed manner. For example, some or all of the units of the device 10 may be arranged as separate modules in a distributed architecture and connected with an appropriate communication network, such as a 3rd Generation Partnership Project (3GPP) network, a Long Term Evolution (LTE) network, the Internet, a LAN (Local Area Network), a Wireless LAN (Local Area Network), a WAN (Wide Area Network), and the like.

[0047] In the following, we describe, as an example, a hybrid model for simulating crop disease epidemics and the effects of weather, variety, and fungicide on disease progression. This model can be used to optimize fungicide application timing. In this hybrid model, a machine learning-based (ML) model is trained to simulate primary infection (i.e., disease onset). To simulate secondary infection, a process-based model is developed, which includes susceptibility, exposure, infection, and removal. The infection rate takes into account environmental factors including temperature and humidity. The infection rate calculation also takes into account variety susceptibility, fungicide, and crop growth stage. We found that the ML model can simulate disease onset date within 5 days for 75% of the areas. The ML model can simulate disease occurrence with 85% accuracy. The process-based model can also be used to simulate the interactions between disease onset date, variety susceptibility, and fungicide application.

[0048] 1. Overall Model Architecture The progression of multimorbid diseases can be calculated using the logistic equation:

number

[0049] By integrating this logistic equation, we obtain the following terms:

number

[0050] In equation 2, y0 describes the disease level at the time of disease onset and is closely related to the pathogen life cycle.

[0051] Many pathogens (e.g., leaf blast) produce spores associated with sexual and asexual reproduction. The sexual spores often survive unfavourable and / or non-host periods and are responsible for the first infections of the season (primary infections). The asexual spores produced in the primary focus cause repeated (secondary) infections during the growing season of the crop. These two types of spores often have different requirements for environmental conditions and epidemiological characteristics. The occurrence and progression of epidemics caused by these fungi results from the sequence and simultaneous occurrence of two types of infection cycles (primary and secondary cycles) over a period of time during the growing season of the host plant.

[0052] The progression of this epidemic is as follows:

number

[0053] Other pathogens, such as Rhizoctonia solani, for which the primary spores of sheath blight are usually derived from viable forms such as spores, sclerotia, or hyphae. The progression of an epidemic caused by this pathogen is as follows:

number

[0054] Due to the complexity and uncertainty of the primary infection source of airborne diseases such as leaf blast, there is still no reliable mechanical model to simulate the onset and early severity of the disease.

[0055] To this end, an apparatus and computer-implemented method are provided that uses a trained machine learning model to simulate the absence of infection over an entire disease onset date or season. Optionally, after the onset date is predicted, the apparatus and computer-implemented method may use this onset date as a starting point and may use the machine model to simulate secondary infections and disease cycles.

[0056] For example, Figure 2 shows a diagram of a disease model simulating primary and secondary infections. State variables (shown as rectangular boxes) are linked by rate variables. Primary infections are simulated by a machine learning based model, for example via a primary infection determination unit 14 of the device 10 shown in Figure 1. Secondary infections can be simulated by a process based model, for example by a secondary infection determination unit 16 of the device 10 shown in Figure 1.

[0057] 2. Method 2.1 Test data and newly collected data from public sources Data may be obtained from public sources. For example, the field collected in the public data is a geographically or historically high-risk field in Japan. The field has experienced high disease outbreaks almost every year.

[0058] The publicly available data for non-treated areas in Japan consisted of information on the crop planting date, crop establishment growth stage, evaluation or observation, disease assessed, observed parameters (e.g., growth stage, infestation rate, date of first symptom observation, number of lesions, etc.), latitude and longitude, number of plants assessed, and evaluation score (i.e., growth stage value, or disease infestation). The data also included other metadata such as field name, field address, prefecture, number of replicates, and crop variety.

[0059] A total of 96 locations had disease data collected. The date range of historical disease data collected was from 2004 to 2019. The data was available for all rice diseases except brown spot disease (i.e., Scab, Sheath Blight, Leaf Blast, Panicle Blast, and Bacterial Leaf Blight). The data consisted of 21 columns and 4304 rows. Of the 4304 rows of observations created, 243 were for GS observations.

[0060] 2.2 Preparation of historical untreated data The public dataset for the non-treated plots required cleaning and transformation before it could be used to retrain and reassess the model performance for predicting disease severity. To clean the data, we removed rows from the data that had inaccurate data values ​​for "evaluation date". This included values ​​such as "early July". We also cleaned the "evaluation date" column by converting the Chinese character values ​​to their numerical equivalents. The cleaned values ​​were the numerical equivalents of "middle 10 days of the month", "first 10 days of the month", and "last 10 days of the month". We removed rows that had no transplantation date. Next, we also removed rows with inaccurate evaluation dates for the transplantation year. We compared the evaluation year with the transplantation year and removed rows with mismatched values. There were 76 cases where the evaluation year and transplantation year were mismatched. We then corrected rows that contained non-numeric values ​​in the evaluation score, such as "one-leaf stage".

[0061] Once the data was cleaned, we converted all disease information into disease severity. As there were many target values ​​for each disease type, we considered multiple values ​​for the "endpoint" for the various diseases. For PYRIOR (leaf blast), we used "severity (all plants)" and if this value was missing, we employed "number of lesions". For PYRPRO (ear neck blast), we considered "severity (all plants)". For USTNVI (rice false smut), we used "number of infected kernels" and if this value was missing, we considered "number of infected ears". For CORTSS (sheath blight), we considered "severity (all plants)" and for XANTOR (bacterial leaf blast), we considered "severity (all plants)". In some regions, there is only one value for the endpoint, so we use either of the values ​​for the "Endpoint" column.

[0062] To transform the data, we aggregated disease data for all regions (i.e., latitude and longitude) by season for all replicates. After this aggregation, we created 450 rows of disease severity for 85 unique regions.

[0063] The distribution of the number of scores across all 450 scores, by disease, can be found in Figure 3A (left panel). The number of observations for leaf blast is high compared to the other diseases.

[0064] Figures 3A and 3B show the distribution of observations considered for sheath blight (CORTSS), leaf blast (PYRIOR), ear blast (PYRPRO), bacterial leaf blight (XANTOR), and rice smut (USTNVI) after cleaning and processing of the data (top left). Distribution of the number of regions with first symptom observation date or disease start date for the cleaned and transformed untreated disease data for Japan (see the left panel of Figure 3A). Distribution of the number of regions relative to the total number of observations per region (see Figure 3B). For example, there are only three regions with at least three bacterial leaf blight observations over the entire season combined. Similarly, there are two regions with a total of six leaf blast observations.

[0065] Our final goal with data preprocessing was to convert the individual disease assessments into a daily continuous disease severity value. For this, we needed the disease start date and the starting severity of the disease. After evaluating the disease start date information, we would fit an exponential growth curve from the disease start date to the last assessment date. Unfortunately, we did not have the disease start date for all the regions. This complicates the process of converting the individual disease severity into a daily continuous form. Error! Reference not found (at the bottom) shows the number of regions per disease and their total number of regions. For example, from this plot, we can see that there is only one region with at least three observations of bacterial leaf blight over the entire season combined. Similarly, for a minimum of four observations, we have only two regions in the case of leaf blast.

[0066] The difference in areas with first symptom observation date or disease onset date is shown in Figure 3A (right side). We have 79 areas for leaf blast, 9 for panicle blast, and 8 for bacterial leaf blight. For the 2020 test data, we ideally used 5-10 observations of disease with a maximum gap of 14 days. Currently, we only considered areas with disease onset dates, leaving us with only 57 observations from a total of 278 cleaned and transformed observations. We appropriately use an "exponential growth function" for these observed disease severity.

[0067] We have invented an algorithm to convert the discrete disease severity into its continuous form. First, we take all disease assessment points with the index of this point starting from the disease start date. Then, we normalize the disease severity values ​​to maintain the increasing trend of the disease data. For example, if we have disease values ​​of [1, 19, 25, 6, 32] for six observations, we change these values ​​to [1, 19, 25, 25, 32]. The stationary of the descending curve maintains the exponential growth characteristics. Then, we normalize the disease severity values ​​according to the following equation: y = a*e -b*x +c (equation 5) Fit to.

[0068] By applying the above algorithm, we were able to convert the discrete disease severity into a continuous daily disease curve.

[0069] Figure 4 shows an example of disease severity interpolation using exponential growth curves. The black dots represent observed disease severity, the vertical axis represents observed disease onset date, and the curve is interpolated based on the algorithm.

[0070] 3. Primary Infection Decision Unit The input unit 12 is configured to receive data including crop variety data, environmental data, crop management data and agricultural field location data.

[0071] The crop variety data relates to crops grown or to be grown in an agricultural field. Exemplary crop variety data may include, but are not limited to, the stage of growth at a particular time, crop density (i.e., the number of crops present per unit area of ​​the field), and days since planting. The crop variety data may be obtained from field data of the agricultural field.

[0072] The environmental data is indicative of environmental conditions of the agricultural field. Exemplary environmental data may include, but are not limited to, air temperature, cloud cover, dew point, shortwave radiation, longwave radiation, ice accumulation period, liquid accumulation period, relative humidity, adjusted precipitation accumulation period, snow accumulation period, and wind speed. In some examples, the environmental data may be collected by sensors installed in the agricultural field. In some examples, the environmental data may be received from a weather forecasting service.

[0073] The crop management data is indicative of a fungicide application history of an agricultural field. The crop management data may be obtained from a data management system that stores a fungicide application history of an agricultural field.

[0074] The location data for the agricultural field may include latitude and longitude data for the agricultural field (eg, decimal degrees, negative values ​​if south or west), which may be obtained from field data for the agricultural field.

[0075] The primary infection determination unit 14 is configured to apply a machine learning model to the received data to determine time series data of disease progression of the fungal disease, the machine learning model being trained to learn disease progression under conditions defined by crop variety data, environmental data, and location data based on historical data collected from one or more agricultural fields.

[0076] The machine learning model may be any suitable model capable of determining the time series data of disease progression of fungal disease. Examples of the machine learning model may include, but are not limited to, XGBoost, Artificial Neural Network, Recurrent Neural Network, and Support Vector Regression.

[0077] The primary infection determining unit 14 may further be configured to determine a disease onset date of the fungal disease based on the determined time series data of disease progression. For example, the disease onset date of the fungal disease may be determined using a change-point detection algorithm.

[0078] This machine learning modeling is described in detail below.

[0079] 3.1 Feature Engineering The historical interpolated untreated disease data from Japan was linked with the Xarvio trial data from India, Japan, and China. Figure 5 shows the distribution of the number of observations considered for each disease across all countries. We can see consistency in the distribution of the number of observations for each disease across all countries. This is a result of recording the severity as zero for all diseases when no epidemic was observed on a given observation date.

[0080] Figure 6 captures a comparative analysis of observed disease severity for the test and non-treated areas. We observe a higher distribution of disease severity for all diseases in the historical non-treated field published data compared to disease severity in the test area. The non-treated published data had disease severity for all diseases, but the distributions captured'. Figure 6 (right side) is for leaf blast, neck blast, and bacterial leaf blight only. As disease onset dates for these diseases were observed, only these diseases are considered for modeling.

[0081] Data for all diseases were pre-processed and feature engineered. Numerical data was not adjusted since we planned to train the machine learning model using a non-linear tree-based algorithm. Tree-based models are scale insensitive. The features used to train, test and validate the models can be categorized into three subtypes, which are presented in Table 1. Growth stage data was simulated from a growth stage model developed in rice growth stage simulation model. Weather data was collected from ITERIS (ClearAG, n.d.) for all regions. The target of the model was disease severity / prevalence.

[0082] [Table 1]

[0083] The latitude and longitude data were divided by 10 to consider upper limits for modeling. This eliminated the possibility that the model would be overly sensitive or biased to these two features. Of the 500 regions in the database, we isolated 40 fields for validation of all models. The data of the remaining 460 fields was split into 80% training and 20% testing.

[0084] 3.2 Training the Machine Learning Model We trained individual models for each disease rather than one model for all diseases combined, which allows each individual model to gain a deeper understanding of the abstract relationships between features.

[0085] Any suitable machine learning model capable of determining the time series data of disease progression of fungal disease may be adopted. Examples of the machine learning model may include, but are not limited to, Xtreme Gradient Boosting (XGB) regression model, Artificial Neural Network, and Support Vector Regression. An exemplary machine learning model (XGB regression model) is described in detail below.

[0086] The model chosen was the XGB regression model trained with Randomized Search Cross Validation. XGB is a scalable tree boosting system, which can be thought of as gradient boosting with regularization. This allows for parallel tree construction, cache-aware access, sparsity awareness, and weighted quantile sketching as part of the optimization and algorithmic enhancements of the system.

[0087] The hyperparameter ranges passed to the tuning of XGB are described in Table 2. The most important hyperparameters were the max depth and the number of estimators. Tuning the hyperparameters gives the model the ability to search for the best fit. Each hyperparameter affects the model in a certain way. By tuning the learning rate, we prevent overfitting. In addition, the lower the learning rate, the more robust the model is at preventing overfitting. Max depth refers to the depth of the tree or estimator in the XGB model. Max depth sets the maximum number of nodes that can exist between the root and the furthest leaf. The lower the max depth, the less overfitting there will be. Minimum child weight enforces regularization at the split step. This is the minimum Hessian weight required to create a new node. The Hessian is the second derivative of the XGB model equation. Max delta step sets the maximum absolute value possible for the weights. This is useful when dealing with imbalanced classes or data. The number of estimators represents the total number of trees built over the entire XGB model. In general, the fewer the number of estimators, the better the overall fit. However, for more biased data, this may not be the case.

[0088] [Table 2]

[0089] 3.3 Detecting changes in disease onset To calculate disease onset dates, we cleaned the raw disease time curves predicted by the model and ran a change-point detection algorithm on the onset index of the curves.

[0090] To clean the predicted disease curve, we first stabilize the noise from the model by converting any predicted values ​​less than 0.3 to 0. We then decompose the curve using a seasonal decomposition process. The results are obtained by first applying a convolution filter to the data to estimate the trend. The trend is then saturated and propagated from the curve. This reduces the Gaussian noise from the raw predicted values. The decomposed curve is smoothed using a convolution based smoothing using tsmoothie, a Python wrapper. The smoothed curve is then passed to a slope-based onset detection algorithm. In this algorithm, we apply a 7-day rolling slope. The first point on this rolling slope curve with a value of 0.1 or greater is selected as the onset. The diagram below represents the action of disease onset detection.

[0091] Figure 7 shows the raw predictions, which have been smoothed and cleaned as curves. The black dotted curve is the actual severity. The vertical dotted line is the generated disease onset date. The vertical dotted line is for the predicted disease onset date, and the black vertical dotted line is for the observed onset of disease severity.

[0092] 3.4 Validation of Machine Learning Models Based on observed disease severity for the 2020 Xarvio test areas, we calculated and identified areas where correct predictions were made and areas where incorrect predictions were made over a given crop growing season. Based on the presence of disease onset, disease or no-disease scenarios were calculated. Disease curves where no disease onset was observed were tagged as disease-free areas and vice versa. The growing seasons of the crop were labeled as true positives (i.e. areas where disease was observed and the model also predicted disease), false positives (i.e. areas where disease was not observed but the model also predicted disease), false negatives (i.e. areas where disease occurred but the model did not predict disease), and true negatives (i.e. areas where disease was not observed and the model also did not predict any disease).

[0093] After training the machine learning model, we used untreated field data from Japan to confirm the validity of the model. In addition to examining the mean absolute error (MAE) and root mean square error (RMSE) on the test data for model validation, we also examine the difference in days between the disease start date in the disease curve predicted by the model and the disease start point in the observed disease severity curve for all regions for all seasons and each disease. We generate the difference in days between the disease start in the disease curve predicted by the model and the disease start point in the observed disease severity curve for all regions for all seasons and each disease, and we start with the prediction, then clean and smooth the predicted disease curve, and finally generate the change points (i.e., disease start point / day). We then compare the results of all untreated fields. Of the 74 disease severity time series for untreated fields, 20 fields were from the validation data (i.e., they were not used in training the model), and the remaining 54 disease severity time series were used in training.

[0094] Based on the predicted data for the test and non-treated areas, we calculated the overall accuracy, precision, and recall for all diseases from the following equations:

number

[0095] We also performed simulations for all rice-growing regions in Japan in 2019. There were a total of 2721 regions across Japan. We predicted disease severity for all regions, then cleaned and smoothed each predicted disease severity curve and passed it to a disease start point detection algorithm to detect the start point of the disease. Once we predicted the disease start point, we calculated the growth stage and days after planting at the disease onset date.

[0096] 4. Secondary Infection Determination Unit The optional secondary infection determination unit 16 is configured to apply the process-based model to determine the infection rate of the fungal disease from the disease onset date onwards under conditions defined by the crop variety data, the environmental data, the crop management data, and the location data.

[0097] In some instances, the infection rate of a fungal disease may be determined by further including a condition defined by the crop variety disease resistance level.

[0098] In some examples, the infection rate of a fungal disease is determined by further including conditions defined by fungicide application data, which includes fungicide data for a fungicide product to be used and at least one planned application timing.

[0099] The process-based disease model is described in detail below.

[0100] 4.1 Structure of the Process-Based Model The process-based disease model may include any suitable model that can predict how an infectious disease will progress and show the possible outcome of an epidemic in plants, and can help inform plant health interventions. An example of a process-based disease model is a compartment model that is formulated as a Markov chain. A classic compartment model in epidemiology is the SIR model, which can be used as a simple model to model epidemics. Several other types of compartment models are also employed. An example of a process-based disease model, the SEIR model, is described in detail below. However, it can be understood that any model that can model the epidemic of an infectious disease in plants can be implemented as a process-based disease model.

[0101] This structure is the SEIR model (susceptible-exposed-infected-cleared), which is widely used to model the epidemics of plant infectious diseases. The structure of this process-based model is based on concepts from Van der Plank, JE, 2013. Plant diseases: epidemics and control. Elsevier., using the system representation from Gonzalez-Dominguez, E., Fedele, G., Salinari, F., Rossi, V., 2020. A General Model for the Effect of Crop Management on Plant Disease Epidemics at Different Scales of Complexity. Agronomy 10., and translated into plant epidemiology by Zadoks, J., 1971. Systems analysis and the dynamics of epidemics. Phytopathology 61, 600-610.

[0102] The system considered is 1m 2of rice crop and simulate the epidemic over the crop growth period resulting from a growth stage model, which includes four state variables for the crop stand: health (H), latency (L), infection (I), and post-infection site (P).

[0103] The central component of the model is the daily infection rate y', t Is: y' t =RC i *INF*COFR Agg (Equation 9) (In the formula, y' t is the daily infection rate, RC i is the suitability of the environment, INF is the number of infected sites, and COFR is the correction factor for disease sites. RC i =RCOpt *RCGS*RCT*RCW*RCV (Equation 10) where RCOpt is the infection rate under favorable conditions, RCGS is the modifier for crop growth stage, RCT is the modifier for temperature, RCW is the modifier for leaf wetness, and RCV is the modifier for variety resistance.

number

[0104] 5.Results 5.1 Primary infection determination unit 5.1.1 Machine learning in predicting disease onset date The distribution of the difference in days between the predicted and observed onset dates for all 74 disease severity time series is presented in Figure 8, including the distribution of the difference in days between the observed and predicted primary infection dates for leaf blast (PYRIOR), neck blast (PYRPRO), and bacterial leaf blight (XANTOR) for all non-treated areas in Japan.

[0105] 5.1.2 Accuracy of the model in simulating disease onset The distribution of true positives, false positives, true negatives, and false negatives is presented in Table 9, which shows the distribution of true positives, false positives, true negatives, and false negatives for sheath blight (CORTSS), leaf blast (PYRIOR), ear neck blast (PYRPRO), bacterial leaf blight (XANTOR), and rice smut (USTNVI) for all non-tested areas in Japan.

[0106] A geospatial representation of true positives, false positives, true negatives, and false negatives per disease can be found in Figures 10-14. The left side of the figure confirms where we made accurate predictions by showing geospatial locations tagged as true positives and true negatives, while the plots on the right suggest where inaccurate predictions were made (i.e., where we predicted false positives and false negatives).

[0107] Specifically, Figure 10 shows the geospatial distribution of true positives, false positives, false negatives, and true negatives for sheath blight occurrence prediction in test areas in Japan. The crosses represent areas used for validation, and the circles represent areas used during training.

[0108] Figure 11 shows the geospatial distribution of true positives, false positives, false negatives, and true negatives for leaf blast occurrence prediction in the test areas in Japan. Crosses represent areas used for validation, and circles represent areas used during training.

[0109] Figure 12 shows the geospatial distribution of true positives, false positives, false negatives, and true negatives for predicting rice neck blast occurrence in the test areas in Japan. The crosses represent the areas used for validation, and the circles represent the areas used during training.

[0110] Figure 13 shows the geospatial distribution of true positives, false positives, false negatives, and true negatives for rice false-smut occurrence prediction in the test areas in Japan. The crosses represent the areas used for validation, and the circles represent the areas used during training.

[0111] Figure 14 shows the geospatial distribution of true positives, false positives, false negatives, and true negatives for predicting bacterial leaf blight occurrence in test areas in Japan. Crosses represent areas used for validation, and circles represent areas used during training.

[0112] The accuracy, precision, and recall are presented in Figure 15, which shows the accuracy, precision, and recall for sheath blight (CORTSS), leaf blast (PYRIOR), ear neck blast (PYRPRO), bacterial leaf blight (XANTOR), and rice smut (USTNVI) for all regions in the Xarvio trial in Japan. Based on the disease onset point in the predicted disease curve for each region, a disease or no disease scenario was calculated. This value was calculated for all trial regions in Japan. A total of 140 regions were considered to calculate the performance, of which 80 were used for training, 20 were used for testing the model, and 40 were used for validation.

[0113] 5.1.3 Feature Importance We calculated the importance of each feature used to train each model. The features utilized to train the models are tabulated in Table 1. The following two plant-related features were used to train each model: 12 weather-related features, and two region-related features.

[0114] The feature importance per model is shown in a heatmap in Figure 16, which shows the feature importance of machine learning models trained on sheath blight (CORTSS), leaf blast (PYRIOR), ear blast (PYRPRO), bacterial leaf blight (XANTOR), and rice false rot (USTNVI). Every column represents a model trained on a disease, and the rows represent all the features used to learn the model. The numbers in each cell represent the feature importance ratio in the model. The color also indicates the relative importance. The numbers in each cell represent the importance ratio for the model.

[0115] All the XGB models assign importance to days since planting, location index, growth stage, and weather-related features in various orders. The numbers in each cell represent the ratio of feature importance in the model. The color also indicates the relative importance. Features that may be relatively important for the model trained on sheath blight are shortwave radiation and cloud cover. For leaf blast, days since planting and dew point appear to have significant predictive power for disease severity. For panicle blast, wind speed and cloud cover are the most important features. However, the rice false rot model assigns higher importance to shortwave radiation and longitude. This phenomenon of assigning higher importance to location index (i.e., longitude) is a result of the disease severity used to train the model. The fields in Japan are well treated with fungicides, so the overall observed disease spread or disease severity is lower compared to fields in China and India. The model corrects for such distribution of disease severity by assigning higher priority to location index. Since the model could be more sensitive to location information (i.e., latitude and longitude), we reduced this bias by dividing the latitude and longitude values ​​by 10 to obtain an upper limit. For example, if the latitude value was 79.884, we converted this value to 8 instead of considering the actual floating point value. This reduces the degree of sensitivity that the model may assign to location information. In the case of bacterial leaf blight, the most important features are cloud cover, wind speed, and adjusted precipitation accumulation period. Important meteorological features such as dew point, temperature, and wind speed are related to overall leaf wetness.

[0116] 5.1.4 Simulation results for all cultivation areas Onset GS (i.e., growth at disease onset date) and onset days after planting are represented geospatially in Figures 17-21.

[0117] Figure 17 shows the geospatial distribution of predicted sheath blight disease incidence. The left panel of the figure represents the mapping of onset GS (i.e. the growth stage recorded at the disease onset date) and the left panel of the plot represents the onset days after planting (i.e. the difference in days between the crop establishment date and the disease onset date). The intensity of onset GS values ​​and onset days after planting were captured in a color map. The inset histograms represent the distribution of each of the variables considered. The grey areas represent the disease-free scenario and the dark green crossed areas represent areas where the disease was predicted after harvest.

[0118] Figure 18 shows the geospatial distribution of leaf blast outbreak predictions. The left panel of the figure represents the mapping of onset GS (i.e. the growth stage recorded at the disease onset date) and the left panel of the plot represents the onset days after planting (i.e. the difference in days between crop establishment date and disease onset date). The intensity of onset GS values ​​and onset days after planting were captured in a color map. The inset histograms represent the distribution of each of the variables considered. The grey areas represent the disease-free scenario and the dark green crossed areas represent areas where the disease was predicted after harvest.

[0119] Figure 29 shows the geospatial distribution of predicted neck blast disease occurrence. The left panel of the figure represents the mapping of onset GS (i.e. the growth stage recorded at the disease onset date) and the left panel of the plot represents the onset days after planting (i.e. the difference in days between the crop establishment date and the disease onset date). The intensity of onset GS values ​​and onset days after planting were captured in a color map. The inset histograms represent the distribution of each of the variables considered. The grey areas represent the disease-free scenario and the dark green crossed areas represent areas where the disease was predicted after harvest.

[0120] Figure 20 shows the geospatial distribution of rice false-smut incidence predictions. The left panel of the figure represents the mapping of onset GS (i.e. the growth stage recorded at the disease onset date) and the left panel of the plot represents the onset days after planting (i.e. the difference in days between the crop establishment date and the disease onset date). The intensity of onset GS values ​​and onset days after planting were captured in a color map. The inset histograms represent the distribution of each of the variables considered. The grey areas represent the disease-free scenario and the dark green crossed areas represent areas where the disease was predicted after harvest.

[0121] Figure 21 shows the geospatial distribution of bacterial leaf blight incidence predictions. The left panel of the figure represents the mapping of onset GS (i.e. the growth stage recorded at the disease onset date) and the left panel of the plot represents the onset days after planting (i.e. the difference in days between the crop establishment date and the disease onset date). The intensity of onset GS values ​​and onset days after planting were captured in a color map. The inset histograms represent the distribution of each of the variables considered. The grey areas represent the disease-free scenario and the dark green crossed areas represent areas where the disease was predicted after harvest.

[0122] The predicted maximum disease severity for all rice-growing regions in Japan is represented geospatially in Figure 22, which shows the regional spatial distribution of maximum disease severity predicted by models trained on sheath blight (CORTSS), leaf blast (PYRIOR), ear neck blast (PYRPRO), bacterial leaf blight (XANTOR), and rice false rot (USTNVI) for all disease-occurring regions.

[0123] For sheath blight, it can be observed in FIG. 17 that the onset growth stage (i.e., the growth stage recorded on the disease start date) is often close to 21. It can also be observed that this disease is commonly predicted in central Japan. For leaf blast, a geospatial representation of the onset growth stage and the number of days since planting for onset is depicted in 18. It can be observed that leaf blast is predicted in almost all regions of Japan and is usually predicted to develop in the early growth stages. For ear neck blast, it can be observed in FIG. 19 that the onset growth stage is often close to growth stages 70-80. Ear neck blast is predicted to develop later in the crop cycle. Rice smut is predicted to be more common in southern Japan and disease onset is also predicted to be later in the crop cycle (i.e., growth stages 70-80), similar to ear neck blast. However, the inventors predict that there are no areas with bacterial leaf blight epidemics in rice growing regions in Japan. The geospatial distribution of maximum disease severity (presented in Figure 22) suggests that the predicted maximum severity recorded for most diseases is in the range of 20-40.

[0124] 5.2 Secondary infection determination module 5.2.1 Process-based models for simulating secondary disease transmission FIG. 23 shows the effect of disease onset date on disease progression curves.

[0125] Figure 24 shows the effect of cultivar disease resistance level on the disease progression curve. The resistance levels are ranked herein from 1 to 9, with 1 being susceptible and 9 being resistant.

[0126] Figure 25 shows the effect of fungicide application on disease progression curves. Cur indicates treatment effectiveness, cur_pd indicates treatment protection days, era indicates herbicidal effectiveness, and era_pd indicates herbicidal protection days.

[0127] Predicting disease onset and disease progression curves to characterize disease progression over time is essential to understand how plant diseases develop and what disease control measures should be taken to achieve the best fungicide effectiveness.As discussed above, the inventors have proposed an ML-based model to simulate disease onset, and have used a process-based model to simulate disease progression curves and the effects of weather conditions, growth stages, varieties, and fungicide sprays.This ML has achieved high accuracy in simulating disease onset dates, and the process-based model can reflect the disease response to the most important factors.

[0128] With reference to Figure 1, the determined disease progression is provided via output unit 18. The determined disease progression can preferably be used to schedule fungicide applications in agricultural fields.

[0129] Optionally, as shown in FIG. 1, the device 10 may further include a fungicide application schedule determination module 20 configured to determine a fungicide application schedule based on the determined disease progression. The fungicide application schedule may be determined based on an analysis of the effect of fungicide application on a disease progression curve, such as the curve shown in FIG. 25. In this way, application timing is set taking into account changes in disease dynamics, thereby improving treatment. In other words, application timing is based on predicted disease risk, which allows fungicide to be applied during the growing season when it is most effective. In particular, the determined fungicide application schedule may target optimal application periods to more reliably halt disease progression and is less likely to miss risk periods. As a result, optimal application timing may have a significant effect on the overall reduction of fungicide use.

[0130] The fungicide application schedule may include spray application timing data. Based on the fungicide application schedule, a configuration file may be created. The configuration file may be usable to configure an applicator to apply the fungicide spray to an agricultural field in accordance with the application timing data.

[0131] 26 illustrates a fungal disease management system 100, which may be a cloud environment. As shown, the fungal disease management system 100 includes one or more data sources 110, a data analysis server 120, an electronic communication device 130, a network 140, and an applicator 150. In this example, the device 10 illustrated in FIG. 1 is embodied in the data analysis server 120, e.g., resides in the data analysis server 120 as software.

[0132] The data sources 110 in the illustrated example may include databases, applications, local files, or any combination thereof. The data sources 110 may include data obtained from one or more sources. For example, the data sources 110 may include crop variety data obtained from field data of an agricultural field, environmental data obtained from sensors deployed in the field and / or weather forecasting services, crop management data obtained from a data management system, and location data from the field data of the agricultural field.

[0133] The data analytics server 120 in the illustrated example may be a server that provides a web server to facilitate management of data. The data analytics server 120 may include a data extraction module (not shown) configured to identify data in the data sources 110 to be extracted, retrieve data from the data sources, and provide the retrieved data to the device 10, which processes the extracted data according to the methods described herein. The processed data is then provided to a data analysis application residing in the electronic communication device 130 so that the data can be displayed and manipulated by a user. In some examples, the device 10 may provide a fungicide application schedule that may be provided to the electronic communication device 130 to enable a farmer to configure the applicator 150 according to the fungicide application schedule. In some examples, the device 10 may provide a configuration profile that may be loaded into the applicator 150 to configure the applicator 150 to apply fungicide according to the determined application timing.

[0134] The electronic communication device 130 in the illustrated example may be a desktop, notebook, laptop, cell phone, smartphone, and / or PDA. The electronic communication device 130 may include a data analysis application, which may be a software application that allows a user to manipulate data extracted from the data sources 110 by the data analysis server 120 and select and specify actions to be performed on individual data. For example, the data analysis application may be a desktop application, a mobile application, or a web-based application. The data analysis application may include a user interface, such as an interactive interface, including but not limited to a GUI, a character user interface, and a touch screen interface. The software application may allow a user to access the data analysis server 120 to obtain information such as disease onset data, fungal disease infection rates since disease onset date, fungicide application schedules, and / or configuration files that can be used to configure the applicator 150.

[0135] The applicator 150 may be, for example, a ground robot with a variable rate applicator, an aerial applicator, or other variable rate applicator for applying a fungicide to agricultural land. In the example of FIG. 26, the applicator 150 may be a smart agricultural machine. The smart agricultural machine may be a smart applicator and includes a connectivity system 152. The connectivity system 152 may be configured to communicatively couple the smart agricultural machine 150 to a computing environment.

[0136] The network 140 of the illustrated example communicatively couples the data source 110, the data analysis server 120, the electronic communication device 130, and the spreader 150. In some examples, the network 140 may be the Internet. Alternatively, the network 140 may be any other type and any other number of networks. For example, the network 140 may be implemented by several local area networks connected to a wide area network. For example, the data source 110 may be associated with a first local area network, the data analysis server 120 may be associated with a second local area network, and the electronic communication device 130 may be associated with a third local area network. The first, second, and third local area networks may be connected to the wide area network. Of course, any other configurations and topologies may be utilized to implement the network 140, including any combination of wired networks, wireless networks, wide area networks, local area networks, etc.

[0137] Figure 27 shows a flow chart illustrating a computer-implemented method 200 for determining disease progression that can be used to schedule fungicide applications in an agricultural field. The method may be understood to emphasize the operation of the apparatus 10 described above. However, it will also be understood that the method steps described in Figure 27 do not necessarily relate to the architecture of the apparatus 10 described above in connection with Figure 1. More specifically, the method described below may be understood to be instructional in itself.

[0138] In block 210 (i.e., step a)), data is received, for example, by the example device 10 shown in Figure 1 or Figure 26. The received data includes crop variety data, environmental data, crop management data, and agricultural field location data.

[0139] The crop variety data relates to crops grown or to be grown in an agricultural field. Exemplary crop variety data may include, but are not limited to, the stage of growth at a particular time, crop density (i.e., the number of crops present per unit area of ​​the field), and days since planting. The crop variety data may be obtained from field data of the agricultural field.

[0140] The environmental data is indicative of environmental conditions of the agricultural field. Exemplary environmental data may include, but are not limited to, air temperature, cloud cover, dew point, shortwave radiation, longwave radiation, ice accumulation period, liquid accumulation period, relative humidity, adjusted precipitation accumulation period, snow accumulation period, and wind speed. In some examples, the environmental data may be collected by sensors installed in the agricultural field. In some examples, the environmental data may be received from a weather forecasting service.

[0141] The crop management data is indicative of a fungicide application history of an agricultural field. The crop management data may be obtained from a data management system that stores a fungicide application history of an agricultural field.

[0142] The location data for the agricultural field may include latitude and longitude data for the agricultural field (eg, decimal degrees, negative values ​​if south or west), which may be obtained from field data for the agricultural field.

[0143] In block 220 (i.e., step b)), a machine learning model is applied to the received data to determine a time series of disease progression of the fungal disease, the machine learning model being trained to learn disease progression under conditions defined by crop variety data, environmental data, crop management data, and location data based on historical data collected from one or more agricultural fields.

[0144] The machine learning model may be any suitable model capable of determining the time series data of disease progression of fungal disease. Examples of the machine learning model may include, but are not limited to, XGB regression model, Artificial Neural Network, and Support Vector Regression.

[0145] An exemplary machine learning model, the XGB regression model, is discussed in detail above and in particular in Section 2, "Primary Infection Decision Module."

[0146] In some examples, multiple machine learning models are provided for more than one fungal disease, with each machine learning model trained on a single disease.

[0147] In block 230 (i.e., step c)), a disease onset date of the fungal disease is determined based on the determined time series data of disease progression, for example, using a change-point detection algorithm.

[0148] As an optional step, in block 240 (i.e., step d)), a process-based model may be applied to determine the infection rate of the fungal disease from the disease onset date onward under conditions defined by the crop variety data, the environmental data, the crop management data, and the location data. The process-based model may include a SEIR model. In some examples, the infection rate of the fungal infection may be determined by further including conditions defined by the crop variety disease resistance level.

[0149] As a further optional step, in block 250 (i.e., step e)), a fungicide application schedule may be determined based on the determined disease progression. The fungicide application schedule may be determined based on an analysis of the effect of fungicide application on a disease progression curve, such as the curve shown in Figure 25. In this way, application timing is set to take into account changes in disease dynamics, thereby improving treatment.

[0150] As a further optional step, in block 260 (i.e., step f)), a configuration file is created based on the fungicide spray schedule, which configuration file is preferably usable to configure the sprayer for fungicide spray application.

[0151] It will be appreciated that the above operations may be performed in any suitable order (e.g., sequentially, simultaneously, or a combination thereof), subject to any particular ordering necessitated, for example, by input / output relationships, where applicable.

[0152] In another exemplary embodiment of the invention, a computer program or a computer program element is provided, characterized in that it is configured to execute the method steps of the method according to one of the previous embodiments on a suitable system. The computer program element may therefore be stored in a computing unit, which may also be part of an embodiment of the invention. This computing unit may be configured to carry out or direct the execution of the steps of the above-mentioned method. Furthermore, it may be configured to operate the components of the above-mentioned apparatus. The computing unit may be configured to operate automatically and / or to execute the instructions of a user. The computer program may be loaded into the working memory of a data processor. The data processor may thus be equipped to carry out the method of the invention.

[0153] This exemplary embodiment of the invention covers both computer programs that use the invention from the outset and computer programs that, through updates, transform existing programs into programs that use the invention.

[0154] Moreover, the computer program element may be capable of providing all the steps necessary to carry out the procedures of the exemplary embodiments of the method described above.

[0155] According to a further exemplary embodiment of the present invention, a computer readable medium, such as a CD-ROM, is presented, on which computer readable medium are stored computer program elements, which are described in the previous section.

[0156] The computer program may be stored on and / or distributed on a suitable medium, such as an optical storage medium or a solid-state medium supplied together with or as part of other hardware, but also in other forms, such as via the Internet or other wired or wireless telecommunications systems.

[0157] However, the computer program may also be presented via a network such as the World Wide Web and downloaded from such a network into the working memory of a data processor. According to a further exemplary embodiment of the invention, a medium is provided making available for downloading a computer program element, which computer program element is configured to carry out a method according to one of the above embodiments of the invention.

[0158] It should be noted that multiple embodiments of the present invention are described with reference to different subject matters. In particular, some embodiments are described with reference to method-type claims, while other embodiments are described with reference to device-type claims. However, from the above and the following description, a person skilled in the art can see that, unless otherwise noted, any combination of features belonging to one type of subject matter, as well as any combination between features related to different subject matters, is also considered to be disclosed in the present application. However, the combination of all features can obtain a higher synergistic effect than the mere sum of these features. While the present invention has been illustrated and described in detail in the drawings and the foregoing description, such illustration and description must be considered as illustrative or exemplary and not restrictive. The invention is not limited to the disclosed embodiments. Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the dependent claims. It should be understood that all definitions, as defined and used herein, take precedence over dictionary definitions, definitions in documents incorporated by reference, and / or ordinary meanings of the defined terms.

[0159] The indefinite articles "a" and "an," when used in the specification and claims, unless clearly indicated to the contrary, should be understood to mean "at least one." In other words, the indefinite articles "a" or "an" do not exclude a plurality.

[0160] The term "and / or" as used in the specification and claims should be understood to mean "either or both" of the elements so conjoined (i.e., elements that are conjunctively present in some cases and disjunctively present in other cases). Multiple elements listed with "and / or" should be interpreted in the same manner, i.e., as "one or more" of the elements so conjoined. Other elements may optionally be present other than the elements specifically identified in the "and / or" clause, whether related or unrelated to the elements specifically identified.

[0161] A single processor or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage. Any reference signs in the claims shall not be interpreted as limiting the scope.

Claims

1. A computer implementation method (200) for determining disease progression that can be used in the spraying schedule of fungicides in agricultural fields, a) Data, as follows: - Crop variety data relating to crops currently cultivated or planned to be cultivated in agricultural fields; - Environmental data showing the environmental conditions of the aforementioned agricultural field; - Crop management data showing the application history of fungicides to the aforementioned agricultural fields; and - Location data of the aforementioned agricultural field Receiving data including (210); b) Applying a machine learning model to the received data to determine time-series data of disease progression of a fungal disease, wherein the machine learning model is trained to learn disease progression under conditions defined by crop variety data, environmental data, crop management data, and location data, based on historical data collected from one or more agricultural fields (220); and c) Determining the onset date of the fungal disease based on the time-series data of disease progression determined above (230) A method that includes this.

2. The computer implementation method according to claim 1, wherein the disease onset date of the fungal disease is determined using a change point detection algorithm.

3. The computer implementation method according to claim 1, wherein multiple machine learning models are provided for two or more fungal diseases, and each machine learning model is trained for a single disease.

4. The computer implementation method according to claim 1, wherein the machine learning model includes an Xtreme Gradient Boosting (XGB) regression model.

5. d) Applying a process-based model to determine the infection rate of the fungal disease after the disease onset date under conditions defined by the crop variety data, the environmental data, the crop management data, and the location data (240) The computer implementation method according to claim 1, further comprising:

6. The computer implementation method according to claim 5, wherein the infection rate of the fungal disease is determined by further including conditions defined by the crop disease resistance level of the crop.

7. The computer implementation method according to claim 5, wherein the infection rate of the fungal disease is determined by further including conditions defined by fungicide application data, which includes fungicide data of the fungicide product to be used and at least one planned application timing.

8. The computer implementation method according to claim 5, wherein the process-based model includes a susceptibility-exposure-infection-removal (SEIR) model.

9. The computer implementation method according to claim 1, wherein the crop variety data includes one or more of the following: the growth stage of the crop, the number of days after planting, and the disease resistance level of the crop variety.

10. The computer implementation method according to claim 1, wherein the environmental data includes one or more of the following: temperature, cloud cover, shortwave radiation, longwave radiation, ice accumulation period, liquid accumulation period, relative humidity, adjusted precipitation accumulation period, snow cover period, and wind speed.

11. The computer implementation method according to claim 1, wherein the position data includes latitude and longitude data.

12. e) Determining a fungicide application schedule based on the disease progression determined above (250) The computer implementation method according to claim 1, further comprising:

13. f) Based on the fungicide application schedule, preferably create a configuration file that can be used to configure a sprayer for fungicide application (260) The computer implementation method according to claim 1, further comprising:

14. A device for generating disease progressions usable for fungicide application schedules in agricultural fields, the device comprising one or more processing units for generating application schemes, the one or more processing units comprising instructions for performing the steps of the method according to claim 1 when performed by the one or more processing units.

15. A computer program element comprising instructions for causing the apparatus described in claim 14 to perform the steps of the method described in claim 1.

Images