Methods and systems for use in mapping cover crops based on remote data

Abstract

Systems and methods are provided for use in mapping presence or absence of cover crops in fields based in remote data. One example computer-implemented method includes accessing, by a computing device, an image of one or more fields, the image including multiple pixels, each of the multiple pixels including a value for each of multiple bands; deriving, by the computing device, at least one index value for each of the multiple pixels of the image; generating, by the computing device, a map of cover crops for the one or more fields, using a trained model and the at least one index value for each of the multiple pixels of the image; storing, by the computing device, the map of cover crops for the one or more fields in a memory; and causing display of the map of cover crops for the one or more fields at an output devise.

Description

Attorney Docket No. 5089-000208-WO-POAMETHODS AND SYSTEMS FOR USE IN MAPPING COVER CROPS BASED ON REMOTE DATACROSS-REFERENCE TO RELATED APPLICATION

[0001] This application claims the benefit of, and priority to, U.S. Provisional Application No. 63 / 734,538, filed on December 16. 2024. The entire disclosure of the abovereferenced application is incorporated herein by reference.FIELD

[0002] The present disclosure generally relates to methods and systems for use in mapping the presence or absence of cover crops in fields, based on remote image data.BACKGROUND

[0003] This section provides background information related to the present disclosure which is not necessarily prior art.

[0004] The planting of cover crops is an important carbon-smart farming practice that can be implemented to reduce greenhouse gas emissions and sequester carbon in the soil. Cover crops provide valuable biomass to the soil when left on the field and provide soil health benefits for crops in the next planting. Cover crops are typically planted in the fall season after the cash crop has been harvested. The emergence of cover crops occurs later that year or the following spring. This timing helps differentiate cover crops from cash crops like corn and soybeans, which are harvested in the fall and planted in the spring. In the western U.S. com belt, cover crops can be directly planted into corn or soybean stubble remaining after harvest, as part of no-till management of the soil.

[0005] Carbon programs and registries have become increasingly prevalent. For example, a grower enrolled in a carbon program may report to the administrator of the program the amount of cover crops planted in his or her field, and in exchange may receive money, credits, or some other token of value. Conventionally, such programs rely on the grower to self-report the amount of cover crops planted, but such self-reporting is susceptible to errors and fraud. Therefore, there exists a need in the art for an automated method of detecting cover crops at scale.Attorney Docket No. 5089-000208-WO-POASUMMARY

[0006] This section provides a general summary of the disclosure, and is not a comprehensive disclosure of its full scope or all of its features.

[0007] Example embodiments of the present disclosure generally relate to computer-implemented methods for use in processing image data associated with fields. In one example embodiment, such a method generally includes accessing, by a computing device, an image of one or more fields, the image including multiple pixels, each of the multiple pixels including a value for each of multiple bands; deriving, by the computing device, at least one index value for each of the multiple pixels of the image; generating, by the computing device, a map of cover crops for the one or more fields, using a trained model and the at least one index value for each of the multiple pixels of the image, the map of cover crops indicating presence or absence of cover crops in the multiple pixels of the one or more fields: storing, by the computing device, the map of cover crops for the one or more fields in a memory; and causing display of the map of cover crops for the one or more fields at an output device.

[0008] Example embodiments of the present disclosure also generally relate to systems for use in processing image data associated with fields. In one example embodiment, such a system generally includes a computing device configured to perform one or more operations of the methods described herein. Example embodiments of the present disclosure also generally relate to computer-readable storage media including executable instructions for processing image data associated with fields. In one example embodiment, a computer- readable storage medium includes executable instructions, which when executed by at least one processor, cause the at least one processor to perform one or more operations described herein.

[0009] Further areas of applicability will become apparent from the description provided herein. The description and specific examples in this summary are intended for purposes of illustration only and are not intended to limit the scope of the present disclosure.Attorney Docket No. 5089-000208-WO-POABRIEF DESCRIPTIONS OF THE DRAWINGS

[0010] The drawings described herein are for illustrative purposes only of selected embodiments, are not all possible implementations, and are not intended to limit the scope of the present disclosure.

[0011] FIG. 1 illustrates an example system of the present disclosure configured for detecting cover crops in multiple fields, based on image data associated with the multiple fields;

[0012] FIG. 2 is a block diagram of an example computing device that may be used in the system of FIG. 1 ;

[0013] FIG. 3 illustrates a flow diagram of an example method, suitable for use with the system of FIG. 1, for preprocessing cover crop data used to train a model as disclosed herein;

[0014] FIG. 4 illustrates a flow diagram of an example method, suitable for use with the system of FIG. 1, for generating optical data features used to train a model as disclosed herein;

[0015] FIG. 5 illustrates a flow diagram of an example method, suitable for use with the system of FIG. 1, for generating Synthetic Aperture Radar (SAR) data features used to train a model as disclosed herein;

[0016] FIGS. 6A-6D illustrate example spectral signature plots (or graphs) of cover crop and non-cover crop fields of com and soy using minimum, maximum, mean, and median composites of optical bands, respectively;

[0017] FIG. 7 illustrates a flow diagram of an example method, suitable for use with the system of FIG. 1, for training and using a model as disclosed herein; and

[0018] FIG. 8 illustrates an example map (broadly, output) that may be generated in the system of FIG. 1 and / or the method of FIG. 7 illustrating presence of cover crops in different fields (e.g., on a pixel level, etc.) (e.g., as an interface for display at a computing device, etc.).

[0019] Corresponding reference numerals indicate corresponding parts throughout the several views of the drawings.Attorney Docket No. 5089-000208-WO-POADETAILED DESCRIPTION

[0020] Example embodiments will now be described more fully with reference to the accompanying drawings. The description and specific examples included herein are intended for purposes of illustration only and are not intended to limit the scope of the present disclosure.

[0021] Remote sensing, such as satellite imagery, etc., provides a tool for rapid detection of environmental conditions, such as, for example, cover crops. Various features embedded in remote sensing data obtained from such remote sensing may be used to detect cover crops, including optical and SAR (Synthetic Aperture Radar) features. Optical features use reflectance to capture crop biomass-proxies, including, for example, one or more vegetative indices (Vis). SAR includes microwave pulses at an oblique angle and measures the backscattered (in the direction of the sensor) portion of this signal(s) to analyze features on the surface. The intensity of each pixel represents a proportion of microwave backscattered from that area on the ground which depends on a variety of factors, such as, without limitation, types, sizes, shapes and orientations of the scatterers in the target area; moisture content of the target area; wavelength and polarization of the radar pulses; as well as the incident angle (source). How a SAR signal interacts with the surface is largely dependent on the wavelength of the microwave signal. Sentinel- 1, for example, operates in C-band (5.6 cm) wavelength and is mainly interacting with an upper part of the crop canopy and delivers information on the physical structure of the vegetation canopy, the surface roughness, and / or moisture status, etc. SAR is also characterized in being cloud-penetrating and thus is not impacted by cloud-cover, as opposed to optical remote sensing. Optical Vis also tend to saturate, limiting their ability to measure high-biomass cover crops, and therefore a combined use of optical and SAR features to estimate biophysical characteristics of cover crop may be effective.

[0022] Uniquely, the systems and methods herein rely on both optical and SAR features for detection of cover crops, thereby relying on SAR and optical features as complements to one another. While the energy captured by optical sensors relates to the amount of chlorophyll (or “greenness”) of vegetation, the amount of microwave energy scattered relates to the biophysical structure of the canopy. As such, the systems and methods herein provide a technical solution to a technical problem, through leveraging the uniqueAttorney Docket No. 5089-000208-WO-POA combination of SAR and optical features to provide enhanced performance in detecting cover crops.

[0023] FIG. 1 illustrates an example system 100 in which one or more aspects of the present disclosure may be implemented. Although the system 100 is presented in one arrangement, other embodiments may include the parts of the system 100 (or additional parts) arranged otherwise depending on, for example, types of images available, manner in which the images are obtained (e.g., via satellites, aerial vehicles, etc.), types of fields, size and / or number of fields, crops present in the fields, crop or management practices (e.g., cover crop planting, etc.) in the fields, etc.

[0024] As shown, the system 100 generally includes a computing device 102, and a database 104 coupled to (and in communication with) the computing device 102, as indicated by the arrowed line. The computing device 102 and database 104 are illustrated as separate in the embodiment of FIG. 1, but it should be appreciated that the database 104 may be included, in whole or in part, in the computing device 102 in other system embodiments. The computing device 102 is also coupled to (and in communication with) network 112. The network 112 may include, without limitation, a wired and / or wireless network, a local area network (LAN), a wide area network (WAN) (e.g., the Internet, etc.), a mobile network, and / or another suitable public and / or private network capable of supporting communication among two or more of the illustrated parts of the system 100, or any combination thereof.

[0025] In general, the computing device 102 is configured to initially access a data set (or multiple data sets) including images of one or more fields from the database 104 (e.g., where the images are collected as generally described herein, for example, from satellites, from other aerial vehicles, or from other imaging devices capable of capturing images of fields, etc.) along with cover crop data (e.g. whether cover crops are present at a particular location, etc.) for the field(s). The computing device 102 is then configured to train a model using the accessed data for identifying presence of cover crops in the field(s). And, once the model is trained, the computing device 102 is configured to access a data set including images of a particular field (or fields) and to then use the trained model to identify the presence of cover crops in the particular field(s) in the images. The computing device 102 is configured to then generate a map of the cover crops in the field(s) (see. e.g., FIG. 8, etc.) (e.g., an interface including the map for display at a computing device, etc.).Attorney Docket No. 5089-000208-WO-POA

[0026] In connection with the above, the system 100 includes various fields, which are represented herein by field 106. The fields, in general, are agricultural fields provided for planting, growing and harvesting crops, etc., in connection with farming or growing operations, for example. While only one field 106 is shown in FIG. 1, it should be appreciated that the field 106 may be representative of dozens, hundreds or thousands of fields associated with one or more growers. The fields may each cover less than an acre, an acre, or multiple or several acres (e.g., at least about two acres, about ten or more acres, about fifty or more acres, about one hundred or more acres, about two hundred or more acres, etc.). It should also be understood that the various fields may be understood to include (or to more generally refer to) growing spaces for crops, and which are exposed for satellite and aerial imaging regardless of size, etc. Further, it should be appreciated that the fields may be viewed as including one or multiple segments, which are different from one another in images of the fields, whereby the segment(s) may be one or more meters by one or more meters, or larger or smaller, etc.

[0027] In this example embodiment, each of the fields (including field 106) is subject to planting, growing and harvesting of crops in various different seasons. In connection therewith, the fields may be exposed to different machinery, management practices (e.g., treatments, harvesting practices, etc.), etc. One management practice, in particular, is the planting of cover crops in the field. For example, as mentioned above, in one field, corn may be harvested in the fall season, and subsequent to the harvesting, cover crops may be planted for emergence in the following spring season.

[0028] Further, the system 100 includes multiple data capture devices, including, in this example embodiment, a satellite 108 and an unmanned aerial vehicle (UAV) 110. In connection therewith, image data captured by (or from) the satellite 108 may be referred to as sat_data. And. image data captured by (or from) the UAV 110 may be referred to as UAV_data. The image data may included, for example, optical data, Synthetic Aperture Radar (SAR) data, etc., depending on, for example, the particular configuration of the capture device, etc. While only one satellite 108 and one UAV 110 are illustrated in FIG. 1, for purposes of simplicity, it should be appreciated that system 100 may include multiple satellites and / or multiple UAVs (or may include access to such satellite(s) and / or such UAV(s)).Attorney Docket No. 5089-000208-WQ-POA

[0029] Furthermore, the same and / or alternate data capture devices (e.g., including a manned aerial vehicle (MAV), etc.) may be included in other system embodiments.

[0030] With respect to FIG. 1, in particular, the satellite 108 is disposed in orbit about the Earth (which includes the field 106) and is configured to capture images of the field 106. As indicated above, the satellite 108 may be part of a collection of satellites (including multiple companion satellites) that orbits the Earth and captures images of different fields, including the field 106. Examples of satellite images may include, for instance, Copernicus Sentinel-2 images (e.g., Level-2A, etc.), Sentinel SI Synthetic Aperture Radar (SAR) imagery (e.g., in vertical transmit, vertical receive (VV mode), vertical transmit, horizontal receive (VH) mode, etc.), Landsat images, MODIS (Moderate Resolution Imaging Spectroradiometer) images, etc. In this example embodiment, the satellites (including the satellite 108) form a network of satellites, which, individually and together, may be configured to capture data (e.g., optical data, SAR data, etc.), at an interval of once per A days, where N may include one day. two days, five days, weekly, ten days, 15 days, 30 days, or another number of days, or on specific dates (e.g., relative to planting, harvest, etc.), etc. In addition, the satellite 108 is configured to capture data having a spatial resolution of about one meter or more by one meter or more per pixel, or other resolution, etc.

[0031] The UAV 110 may be configured to capture data (e.g., optical data, SAR data, etc.) at the same, similar or different intervals to that described for the satellite 108 (e.g., once per N days, where N may include one day, two days, five days, weekly, ten days, 15 days, 30 days, or another number of days, etc.) or on (or for) specific dates (e.g., relative to planting, harvest, etc.). The UAV 110, though, generally captures data of the area of interest at a higher spatial resolution than the satellite 108. For example, the UAV 110 may capture data having a spatial resolution of about five inches or less by about five inches or less per pixel, or other resolution, etc.

[0032] It should be appreciated that the captured satellite images and the UAV images may be upscaled or downscaled, from a spatial resolution perspective, as appropriate for use as described herein. It should also be appreciated that the satellite 108 and the UAV 110 may be configured to transmit, directly or indirectly, the captured satellite images and the captured UAV images, respectively, to the computing device 102 and / or the database 104 (e.g., via the network 112, etc.), whereby the images are stored in the database 104. TheAttorney Docket No. 5089-000208-WO-POA images may be organized, in the database 104, by location, date / time, and / or field, etc., as is suitable for use as described herein.

[0033] In this example embodiment, for certain one or more of the fields (e.g.. including the field 106, etc.), the database 104 further includes cover crop data for the field(s), if any (e.g., based on prior analysis consistent with the description herein, based on reporting by users associated with the field(s), based on prior scouting, etc.). The cover crop data is designated in a manner so that it is linked to one or more images of the fields, or vice versa. The cover crop data may indicate, for a specific field, or for one or more segments of the field, whether a cover crop is present (e.g. a binary indication of 1 for presence of a cover crop and 0 for absence thereof, a Boolean indication of TRUE for presence of a cover crop and FALSE for absence thereof, etc.). The cover crop data may further include or may be associated with other data related to the field 106, including, for example, the planting date for the cover crop in the field, soil condition(s), weather condition(s) (e.g., precipitation, clouds, sunshine, temperature, humidity, etc.), the type of the cover crop planted, etc.

[0034] The image data (regardless of whether captured as satellite images or UAV images) may include data indicative of various bands of wavelengths (e.g., within the electromagnetic spectrum, etc.). For example, optical images, and more specifically each pixel of the images, may include data (or wavelength band data or band data) related to the color red (R) (e.g., having wavelengths ranging between about 635 nm and about 700 nm, etc.), the color blue (B) (e.g., having wavelengths ranging between about 490 nm and about 550 nm, etc.), the color green (G) (e.g., having wavelengths ranging between about 520 nm and about 560 nm, etc.), and near infrared (NIR) (e.g., having wavelengths ranging between about 800 nm and about 2500 nm, etc.), etc.

[0035] In this example embodiment, the computing device 102 is configured to access (e.g., from the database 104, etc.) certain ones of the images for various fields (associated with the known cover crop data for the fields) and to train a model to classify the fields, or segments thereof.

[0036] FIG. 2 illustrates an example computing device 200 that may be used in the system 100 of FIG. 1. The computing device 200 may include, for example, one or more servers, workstations, personal computers, laptops, tablets, smartphones, virtual or cloudbased devices, etc. In addition, the computing device 200 may include a single computingAttorney Docket No. 5089-000208-WO-POA device, or it may include multiple computing devices located in close proximity or distributed over a geographic region, so long as the computing devices are specifically configured to operate as described herein. In the example embodiment of FIG. 1, the computing device 102 and the database 104 (and the satellite 108 and the UAV 110) may each include and / or be implemented in one or more computing devices consistent with (or at least partially consistent with) computing device 200. However, the system 100 should not be considered to be limited to the computing device 200, as described below, as different computing devices and / or arrangements of computing devices may be used. In addition, different components and / or arrangements of components may be used in other computing devices.

[0037] As shown in FIG. 2, the example computing device 200 includes a processor 202 and a memory 204 coupled to (and in communication with) the processor 202. The processor 202 may include one or more processing units (e.g., in a multi-core configuration, etc.). For example, the processor 202 may include, without limitation, a central processing unit (CPU), a microcontroller, a reduced instruction set computer (RISC) processor, a graphics processing unit (GPU), an application specific integrated circuit (ASIC), a programmable logic device (PLD), a gate array, and / or any other circuit or processor capable of the functions described herein.

[0038] The memory 204, as described herein, is one or more devices that permit data, instructions, etc., to be stored therein and retrieved therefrom. In connection therewith, the memory 204 may include one or more computer-readable storage media, such as, without limitation, dynamic random access memory (DRAM), static random access memory (SRAM), read only memory (ROM), erasable programmable read only memory (EPROM), solid state devices, flash drives, CD-ROMs, thumb drives, floppy disks, tapes, hard disks, and / or any other type of volatile or nonvolatile physical or tangible computer-readable media for storing such data, instructions, etc. In particular herein, the memory 204 is configured to store data including and / or relating to, without limitation, images, models, fields, plots, trials, and / or other types of data (and / or data structures) suitable for use as described herein.

[0039] Furthermore, in various embodiments, computer-executable instructions may be stored in the memory 204 for execution by the processor 202 to cause the processor 202 to perform one or more of the operations described herein (e.g.. one or more of the operations of method 300, method 400, method 500, etc.) in connection with the variousAttorney Docket No. 5089-000208-WO-POA different parts of the system 100, such that the memory 204 is a physical, tangible, and non- transitory computer readable storage media. Such instructions often improve the efficiencies and / or performance of the processor 202 that is performing one or more of the various operations herein, whereby such performance may transform the computing device 200 into a special-purpose computing device. It should be appreciated that the memory 204 may include a variety of different memories, each implemented in connection with one or more of the functions or processes described herein.

[0040] In the example embodiment, the computing device 200 also includes an output device 206 that is coupled to (and is in communication with) the processor 202. The output device 206 may output information (e.g., maps, etc.), visually or otherwise, to a user of the computing device 200, such as a researcher, a grower, etc. It should be further appreciated that various interfaces (e.g., as defined by the FIELD VIEW service, commercially available from Climate LLC, Saint Louis, Missouri; etc.) may be displayed at computing device 200, and in particular at output device 206, to display certain information to the user. The output device 206 may include, without limitation, a liquid crystal display (LCD), a light-emitting diode (LED) display, an organic LED (OLED) display, an “electronic ink” display, speakers, etc. In some embodiments, output device 206 may include multiple devices. Additionally, or alternatively, the output device 206 may include printing capability, enabling the computing device 200 to print text, images, and the like on paper and / or other similar media.

[0041] In addition, the computing device 200 includes an input device 208 that receives inputs from the user (i.e., user inputs) such as, for example, selections of fields, desired characteristics, etc. The input device 208 may include a single input device or multiple input devices. The input device 208 is coupled to (and is in communication with) the processor 202 and may include, for example, one or more of a keyboard, a pointing device, a touch sensitive panel, or other suitable user input devices. It should be appreciated that in at least one embodiment an input device 208 may be integrated and / or included with an output device 206 (e.g., a touchscreen display, etc.).

[0042] Further, the illustrated computing device 200 also includes a network interface 210 coupled to (and in communication with) the processor 202 and the memory 204. The network interface 210 may include, without limitation, a wired network adapter, a wireless network adapter, a mobile network adapter, or other device capable ofAttorney Docket No. 5089-000208-WO-POA communicating to one or more different networks (e.g., one or more of a local area network (LAN), a wide area network (WAN) (e.g., the Internet, etc.), a mobile network, a virtual network, and / or another suitable public and / or private network capable of supporting wired and / or wireless communication among two or more of the parts illustrated in FIG. 1, etc.) (e.g., network 112, etc.), including with other computing devices used as described herein.

[0043] In connection therewith, the computing device 102 is configured to access images (broadly, image data) associated with a time window associated with a planting date of cover crops, for example, a time interval in the winter months after harvesting. The interval(s) may be about one month, about two months, about three months, about four months, about five months, about six months, or more or less, etc. In one specific example, satellite images from about November to April are accessed. In addition to images, ground measurement data (e.g. ground truth data regarding the presence or absence of cover crops, etc.) may also be accessed. Detailed descriptions of the ground measurement data and preprocessing thereof are provided below in connection with FIG. 3.

[0044] After accessing the desired images, the computing device 102 may be configured to generate optical and SAR remote sensing data features from the images. In connection therewith, the computing device 102 may be configured to transform the image data, as necessary, into a number of optical bands, which may include the above R, G, B, and NIR bands, and then add the bands as desired combinations thereof. In one example, bands and indices of the images, per pixel, may include (without limitation): blue; green; red; shortwave infrared 1 or swirl; short-wave infrared 2 or swir2; red-edge-4; NIR; ND VI (Normalized Difference Vegetation Index) (i.e., (NIR-red) / (NIR+red)); CAI (Cellulose Absorption Index) (i.e., (0.5(blue+swir2)-red); LCAI (Lignin Cellulose Absorption Index) (i.e., 100((blue-red)+(blue-swir2))); NDRI (Normalized Difference Residue Index) (i.e., ((swir2-red) / (swirl+swir2)); NDTI (Normalized Difference Tillage Index) (i.e., (swirl- swir2) / (swirl+swir2)); GCVI (Green Chlorophyll Vegetation Index) (i.e., (NIRZgreen)-l); STI (simple tillage index) (i.e., swirl / swir2); ND 15 (Normalized Difference Index 5) (i.e., (NIR- swirl) / (NIR+swirl)); NDI7 (i.e., (NIR-swir2) / (NIR+swir2)); and CRC (Crop Residue Cover) (i.e., (swirl-green) / (swirl+green)); etc. Swirl and swir2 are at different wavelengths - 1.57- 1.65pm versus 2.08-2.35pm, respectively. The pixels are then expressed in optical bands andAttorney Docket No. 5089-000208-WO-PQA derived indices. At least some of these data features are also referred to as vegetative indices (VI) herein.

[0045] For the SAR data features, VV (backscatter after terrain correction) and VH (backscatter after terrain correction) data may be used. As it relates to the VV and VH, the image data is subject to one or more of terrain correction, radiometric correction, and / or backscatter normalization to provide the data in a form to be input herein. It should be appreciated that such image data may be included in and / or provided as part of satellite images, and in particular, the Sentinel SI Synthetic Aperture Radar (SAR) imagery (e.g., in VV mode, VH mode, etc.).

[0046] The computing device 102 is then configured to compile a composite data set of the optical data and SAR data features above (e.g., indices, correction data, etc.). For example, the computing device 102 may be configured to calculate seasonal composites of minimum, maximum, mean, and median of the various data features disclosed above for winter (z.e.. November, December, January) and spring (z.e., February, March, April) months. The computing device 102 is next configured to append the composite data set to the imagery data. It should be appreciated that the image pixels may be expressed otherwise as optical bands or derived indices (e.g., through other combinations of band data, etc.) in other system embodiments. Detailed descriptions of the generation of data features and seasonal composites thereof are provided below in connection with FIGS. 4 and 5.

[0047] In addition to optical and SAR data features, elevation-derived features (e.g., slope information, etc.) may also be used, as fields with higher slope are more likely to implement conservation practices, due to greater benefits for erosion prevention relative to flatter fields. The presence of cover crops in the off-season months may be detected by higher values of Vis consisting of NIR and red-edge-4 bands, and higher values of backscatter for SAR bands. That is, higher field slope values could be an important indicator in detection of cover crops.

[0048] Next in the system 100, the computing device 102 is configured to split the data set into a training subset and a validation subset. Pixels within each field of the image data may be highly correlated in one or more implementations, whereby the pixels should be retained in either training subsets or the validation subsets, to prevent, for example, dataAttorney Docket No. 5089-000208-WO-PQA leakage and / or overfitting. The computing device 102 may employ gridding over pl ots / fields as a spatial stratification.

[0049] The computing device 102 is then configured to train the model, which may include, for example, a logistic regression model, a Residual Neural Network or RESNET model (e.g., RESNET ID, etc.), or another suitable model, etc. The model is trained to produce a binary output based on the inputs, i.e., 1 for presence of cover crop and 0 for absence. And, in turn, the computing device 102 may be configured to validate the trained model, based on the validation subset of the data set, which, again, includes the same type of input data. The model is validated when sufficient performance of the model is achieved.

[0050] After training, the computing device 102 is configured to access an image (or images) of a particular field, such as. for example, the field 106. The computing device is then configured to process the data for the image in the same manner as above (e.g., derive one or more indices, etc.), and then to employ the trained model to identify a cover crop (or cover crops) in the field 106, as a whole or by segments included therein. Then, finally in the system 100, in this example, the computing device 102 is configured to generate a map of the field showing the presence or absence of cover crop(s). The computing device 102 is configured to then display the map to one or more users (e.g., via the FIELDVIEW service from Climate LLC, Saint Louis, Missouri; etc.). As described, the map may then be used and / or leveraged to inform one or more crop management decisions with regard to the field 106 (e.g., direct operation of a farm implement to apply desired treatments to the fields such as pesticides, a herbicides, and / or a fertilizers based on the cover crop(s); etc.).

[0051] For example, from the above, based on the identified presence of a cover crop(s) in the field 106 (e.g., and the mapping of location(s) thereof in the field 106, etc.), the computing device 102 may be configured to generate one or more instructions (e.g., scripts, plans, etc.) for treating the field 106 (e.g., the crop in the field 106, etc.). The computing device 102 may then transmit the instructions to an agricultural machine, etc., whereby upon receipt, the agricultural machine, etc. automatically operate(s), in response to the instructions, to treat the crop in the field 106 (e.g., the instructions are used to control an operating parameter of the agricultural machine, etc.). Such treatment, processing, etc. of the crop, as defined by the instructions, may include directing the agricultural machine (e.g., causing operation of the machine, etc.) to apply one or more fertilizers, herbicides, pesticides, etc.Attorney Docket No. 5089-000208-WO-POA(e.g., as part of a treatment plan, etc.); etc. In this way, the agricultural machine, etc. operates in an automated manner, in response to the identified cover crop presence in the field 106, to perform one or more subsequent agricultural tasks.

[0052] FIG. 3 illustrates a flow diagram of an example method 300, suitable for use with the system of FIG. 1, for preprocessing cover crop data used to train a model as disclosed herein. The method 300 is described herein in connection with the system 100, and may be implemented, in whole or in part, in the computing device 102 of the system 100, and also the computing device 200. However, it should be appreciated that the method 300, or other methods described herein, are not limited to the system 100 or the computing device 200. And, conversely, the systems, data structures, and the computing devices described herein are not limited to the example method 300.

[0053] At the outset in the method 300, the computing device 102 accesses (or requests) the desired data at step 302. In this particular example, for instance, the computing device 102 may access and store ground measurement or ground truth data regarding the presence or absence of cover crops in various plots or fields in, for example, the United States. In still a more particular example, the ground measurement data may be stored as an Excel file or a similar spreadsheet format, and may contain the following data fields: FIELD ID, FIELD NAME, LOCATION (e.g., longitudinal and latitudinal information), AREA OF FIELD (in, e.g.. acres, hectares, square feet, square meters, etc.), COUNTRY CODE (e.g., US, AR (Argentina), BR (Brazil), etc.), CROP (e.g., soy, corn, cotton, etc.), PLANTING DATE, YEAR, and COVER CROP label (e.g., TRUE for presence of cover crop, FALSE for absence). Each row of the spreadsheet may correspond to a particular field of a particular grower or farmer. In addition, the cover crop label may be implemented as three separate labels: a PRE-SEASON COVER CROP label, an IN-SEASON COVER CROP label, and a POST-SEASON COVER CROP label.

[0054] At step 304, the computing device 102 performs a de-duplication operation on the data file such that duplicative rows are removed. At step 306, the computing device 102 converts the cover crop labels in the data file to binary. For example, TRUE is converted to 1 and FALSE is converted to 0. At step 308, the computing device 102 performs another de-duplication operation, where overlapping geometries for the same year are removed. For example, using the location information (e.g., longitudinal and latitudinalAttorney Docket No. 5089-000208-WO-POA information, etc.) in two rows of the data file, it may be determined that the two fields corresponding to these two rows have greater than 90% overlap. And using the year information in the two rows, it may be determined that the two rows correspond to the same year (e.g., 2022). In this case, one of the rows is removed from the data file.

[0055] At step 310, the computing device 102 combines multi-year data. In one example, yearly data may be stored in separate data files or spreadsheets, i.e., there are separate data files for the years 2020, 2021, and 2022. At step 310, the computing device 102 combines these files into a single file. At step 312, the computing device 102 identifies overlapping geometries across different years. The operation at step 312 is similar to the operation in step 308, except that the determination is not limited to same-year data. For example, using the location information, two rows may be determined to be overlapping if they overlap by greater than 90%, regardless of the year information in the rows.

[0056] Next at step 314, the computing device 102 removes overlapping rows with conflicting cover crop labels. For example, a first row in the combined data file may correspond to a Field A in the year 2020, and the post-season cover crop label in the first row may be set as TRUE. A second row in the combined data file may have location information that overlaps with the location information of the first row, indicating that the first and second rows correspond to the same Field A. But in the second row, the year information may be 2021 and the pre-season cover crop label for 2021 may be set as FALSE. This is indicative of a data error. If a field has post-season TRUE cover crop label in 2020, then pre-season cover crop label for 2021 should also be TRUE. Therefore, farm fields where these labels were contradictory for consecutive years are removed.

[0057] At step 316, the computing device 102 may query the multi-year combined data file for only the corn and soy rows. For example, the multi-year combined data file may be filtered according to the CROP field mentioned above, and only the rows indicating soy or corn in the CROP field may be obtained. It should be noted that filtering the data according to the planting of com or soy is only an example, and the data may be filtered or queried based on other crops (e.g., cotton, wheat, etc.) or not be filtered at all.

[0058] Finally at step 318, the computing device 102 modifies the filtered data file to add a column or field indicating U.S. county information (or other geographicAttorney Docket No. 5089-000208-WO-POA information). In one embodiment, the pre-processed data file obtained at the end of step 318 is used to train the model discussed in detail below.

[0059] FIG. 4 illustrates a flow diagram of an example method 400, suitable for use with the system of FIG. 1, for generating optical data features used to train a model disclosed herein. The method 400 is described herein in connection with the system 100, and may be implemented, in whole or in part, in the computing device 102 of the system 100, and also the computing device 200. However, it should be appreciated that the method 400, or other methods described herein, are not limited to the system 100 or the computing device 200. And, conversely, the systems, data structures, and the computing devices described herein are not limited to the example method 400.

[0060] The method 400 starts with step 402, in which the computing device 102 queries Sentinel-2 Level 2A (L2A) satellite data, and relevant data is retrieved from the database 104. Two parameters are used to query the L2A satellite data. The first parameter is location, for example, field boundary information and / or longitudinal and latitudinal information of fields, which can be extracted from the combined data file described in connection with FIG. 3 above. The second parameter is time, hi one embodiment, satellite data from November 1, [YEAR-1] to April 30, [YEAR] may be queried. These months were chosen, in this example, because they are the typical off-season for com and soybean, thereby allowing these months to be the growing period of cover crops. In major com and soybean regions, soybeans are planted in late May and early June, while com tends to be planted earlier, in late April and early May. Harvesting occurs in October-November months depending on the planting date. Accordingly, in one example, for the fields identified in the combined data file generated as a result of the method 300, satellite data from November 1, 2020-April 30, 2021, November 1, 2021-April 30, 2022, and November 1, 2022-April 30, 2023 may be obtained in step 402.

[0061] Next, in step 404, the computing device 102 applies a cloud mask, for example, a cloud mask extracted from the Scene Classification Layer (SCL) band of the L2A data, to the satellite data to eliminate data points (z.e., pixels) influenced by clouds, cloud shadows, and snow. In step 406, for each day, the computing device 102 calculates a mean composite when, for example, multiple images are taken on the same day. In step 408, the computing device 102 calculates various vegetative indices for each day using its meanAttorney Docket No. 5089-000208-WO-PQA composite at the pixel level. As disclosed above, in one embodiment, the relevant optical bands and Vis are blue, green, red, swirl, swir2, red-edge-4, NIR, ND VI, CAI, LCAI, NDRI, NDTI, GCVI, STI. NDI5. NDI7, and CRC.

[0062] Next, in step 410, the computing device 102 masks, i.e., removes, all pixels where NDVI<0. Negative ND VI values commonly represent cloud, water, or snow pixels. Therefore. NDVI<0 pixels likely correspond to pixels affected by cloud, rain, or snow that were missed by the SCL cloud mask in step 404. In step 412, the computing device 102 calculates seasonal composites of the bands and Vis. As shown below in connection with FIGS. 6A-6D, the seasonal composites are the minimum, maximum, mean, and median for winter and spring seasons. For the given time-window (winter or spring), the fields with cover crops should have higher minimum reflectance in the NIR region which would be higher than a field with no cover-crop (weeds might be contributing to the reflectance). In addition, minimum and maximum composites will help identify cover crop fields from bare ground or fields with weeds based on the observation that reflectance in the NIR and red- edge-4 region of the EM spectrum would be higher for cover crops.

[0063] Finally, in step 414, the computing device 102 performs spatial aggregation (e.g., calculating the mean over a particular field, etc.) for each band and VI for each image, for each defined field boundary. In connection therewith, temporal averaging, specifically within the predetermined winter or spring timeframe, may also be done. Temporal averaging may facilitate an in-depth assessment of the seasonal vegetation dynamics within the specified field boundary.

[0064] FIG. 5 illustrates a flow diagram of an example method 500, suitable for use with the system of FIG. 1, for generating SAR data features used to train a model disclosed herein. The method 500 is described herein in connection with the system 100, and may be implemented, in whole or in part, in the computing device 102 of the system 100, and also the computing device 200. However, it should be appreciated that the method 500, or other methods described herein, are not limited to the system 100 or the computing device 200. And, conversely, the systems, data structures, and the computing devices described herein are not limited to the example method 500. The method 500 serves a similar purpose as the method 400, but whereas the method 400 is used to generate optical (e.g., band and VI, etc.) data features, the method 500 is used to generate SAR data features.Attorney Docket No. 5089-000208-WQ-POA

[0065] At the onset in step 502, the computing device 102 queries SAR data (sigmaO, vv, vh) and Radiometric Terrain Corrected (RTC) data (factor) from the database 104. SAR data is used here based on the observation that SAR backscatter from a field with cover crop would be higher compared to bare ground or snow. As in the method 400, the SAR and RTC data are queried using location and time parameters. Next, in step 504, the computing device 102 applies the RTC factor to sigmaO.

[0066] At step 506, for each day, the computing device 102 calculates a mean composite of the SAR data. This step generally corresponds to step 406 in the method 400. Next, in step 508, the computing device 102 masks pixels with backscatter >= 0. This filters out invalid or noisy pixels of SAR. In step 510, the computing device 102 calculates seasonal composites of VV and VH. This step generally corresponds to step 412 in the method 400, and therefore min, max, mean, and median are calculated for the winter and spring months. Finally, in step 512, the computing device 102 performs spatial aggregation (e.g., calculating the mean over a particular field, etc.) for VV and VH, generally corresponding to step 414 in the method 400.

[0067] In addition to optical and SAR data features, elevation features may also be used in one or more embodiments disclosed herein. For example, elevation features such as “height,” “slope,” “aspect,” “prof_curvature,” “plan_curvature,” “curvature,” “flow_accum,” and “relative_altitude” may be obtained by querying a 10-meter Digital Elevation Model (DEM). The querying may be based on the location information (e.g. field boundary) stored in the combined data file of FIG. 3. Field-level min, max, mean, and median of the queried features can also be calculated.

[0068] FIGS. 6A-6D illustrate spectral signature plots of cover crop and non-cover crop fields of com and soy using minimum, maximum, mean, and median composites of the optical bands, respectively. The numbered lines, that is, 602, 604, 606. and 608 in FIG. 6A, 612, 614, 616, and 618 in FIG. 6B, 622, 624, 626, and 628 in FIG. 6C, and 632, 634, 636, and 638 in FIG. 6D, represent the optical indices of fields with cover crop (e.g., represent a TRUE label, etc.), while the other, unnumbered lines in each of the corresponding plots represent fields without cover crop (e.g., represent a FALSE label, etc.). These plots demonstrate the differences between cover crop and non-cover crop fields in the near-infrared (NIR) band using the minimum composites, while differences in short-wave infrared (SWIR) areAttorney Docket No. 5089-000208-WO-PQA noticeable using the maximum composites. The near-infrared (NIR) band is crucial for vegetation analysis as healthy vegetation reflects more NIR and green light compared to other wavelengths. This is because chlorophyll, a key component of photosynthesis, strongly absorbs red light and reflects NIR light. Higher reflectance of NIR and red-edge-4 band (band after NIR band), for the cover crop fields can thus be observed in FIGS. 6A-6D because of presence of vegetation. The short-wave infrared (SWIR) bands, which have longer wavelengths than the NIR and red-edge bands, are sensitive to vegetation water content changes. SWIR bands are useful for discriminating between different vegetation types, as well as for monitoring soil moisture and heat stress in crops.

[0069] FIG. 7 illustrates a flow diagram of an example method 700, suitable for use with the system of FIG. 1, for training and using a model as disclosed herein (e.g.. the model referenced in the system 100, in the method 300, in the method 400, in the method 500, etc.). The method 700 is described herein in connection with the system 100, and may be implemented, in whole or in part, in the computing device 102 of the system 100, and also the computing device 200. However, it should be appreciated that the method 700, or other methods described herein, are not limited to the system 100 or the computing device 200. And, conversely, the systems, data structures, and the computing devices described herein are not limited to the example method 700.

[0070] At step 702, the computing device 102 obtains, and pre-processes, ground measurement data of whether cover crop exists at various fields at various time intervals (e.g., winter and fall months of the years 2020-2022). A specific embodiment of the step 702 is described in detail above in connection with FIG. 3. Next, at step 704, the computing device 102 obtains, and processes, satellite imagery data corresponding to the ground measurement data (e.g., various optical and SAR indices and composites thereof are calculated and stored, etc.). A specific embodiment of the step 704 is described in detail above in connection with FIGS. 4 and 5.

[0071] Next at step 706, the computing device 102 trains a model using the obtained ground measurement and satellite image data. In one embodiment, the model used is a logistic regression model. Logistic regression is an algorithm that is commonly used in the art for binary classification tasks. Regularization techniques, such as LI (Lasso) and L2 (Ridge), are instrumental in preventing overfitting and improving model generalization. ThisAttorney Docket No. 5089-000208-WO-POA work uses a two-step approach to refine logistic regression models by initially applying LI regularization for feature selection, followed by L2 regularization for enhanced predictive capabilities. As described above, prior to model development, the dataset is pre-processed. In one embodiment, in the pre-processing, feature generation of optical, SAR, and topographic features generate approximately two hundred features that may be considered for modeling. LI regularization is introduced to the logistic regression model. Feature coefficients obtained from the LI -regularized model are utilized to select, for example, the top twenty features with the highest absolute coefficients. In one embodiment, these features include “green_median_winter,” “blue_median_winter,” “ndi5_min_spring,” “red_median_winter,” “slope_max,” “ndi7_min_spring,” “height_min,” “ndvi_max_spring,” “nir_mean_winter,” “ndi5_max_spring,” “relative_altitude_max,” “vv_mean_winter,” “green_mean_winter,” “vv_max_spring,” “nir_median_spring,” “ndi7_max_spring,” “vh_mean_winter,” “ndi7_max_winter,” “ndi5_max_winter,” and “swir2_max_spring.”

[0072] This feature selection process enhances model interpretability. Following feature selection, the logistic regression model is retrained, this time with L2 regularization. The model is trained exclusively on the twenty selected features, promoting a robust and well- generalized classifier. Balanced class weights are used to address class imbalance. Binary Cross-Entropy may be used as the loss function. For validation, in an example embodiment, the model can be evaluated on the data set using a train-test split of 80:20, and the data set may be stratified by U.S. state.

[0073] In another embodiment, the model used may be ResNetlD. ResNetlD may be chosen for its ability to handle long sequences efficiently and capture localized patterns through hierarchical layers, aligning well with the high-frequency and temporal regularity of the data. To train the ResNetlD model, fields are randomly split into training and testing sets. Balanced class weights are used as well to address class imbalance. In this example, the RESNET model network includes an architecture illustrated in FIG. 7, where a number n of residual blocks is included, along with a final SoftMax classifier. In this example embodiment, the number n is three to indicate three residual blocks. In addition, the architecture includes fully connected layers (FC) and a Global Averaging Pooling (GAP) layer (not shown) following the residual blocks. Also, the final SoftMax classifier includes a number of neurons equal to the number of classes in a dataset. Further, the example RESNETAttorney Docket No. 5089-000208-WO-POA model includes a shortcut residual connection between consecutive convolutional layers, and a linear shortcut is added to link the output of a residual block to its input thus enabling the flow of the gradient directly through these connections. In this manner, the training is simplified by reducing the vanishing gradient effect. The trained model is then validated (and / or evaluated) through the validation subset of the composite data set.

[0074] After the model is trained, the computing device 102, for example, requests particular field data, at step 708, by identifying a specific field (e.g., field 106, etc.) for which cover crop is to be evaluated (e.g., automatically, in response to an input from a grower or user, etc.). In connection therewith, the computing device 102 accesses one or more images for the field during winter or spring seasons.

[0075] The identified images of the field are then inputted into the trained model, whereby each pixel of the accessed / received field images is assigned binary digit indicating presence or absence of cover crops. The computing device 102 may then aggregate the predicted cover crop indications into a map, at step 710, so that a cover crop map for the field (e.g., for field 106, etc.) is compiled. As described, the map may then be used and / or leveraged to inform one or more crop management decisions with regard to the field 106 (e.g., practice verification / monitoring, application, etc., of desired treatments to the fields such as pesticides, a herbicides, and / or a fertilizers; etc.). In connection therewith, FIG. 8 illustrates an example map 800 that may be generated by the computing device 102 to illustrate presence of cover crops in fields (at a pixel level). As shown, the map 800 includes multiple fields (or field regions / portions), identified at 802-810, and an indication of which of the fields (or field regions / portions) 802-810 includes a cover crop. In particular in this example, the fields (or field regions / portions) 804, 806, and 808 are identified (via a cross-hatching pattern in this example) as including a cover crop. While the fields (or field regions / portions) 802 and 810 are identified (via the lack of a cross-hatching pattern in this example) as not including a cover crop. It should be appreciated that other indicators may be used in other example embodiments to identify presence, or lack of presence, of a cover crop in a field or portion of a field.

[0076] In view of the above, the systems and methods herein provide for mapping the presence or absence of cover crops in regions (e.g., in fields in the regions, etc.), based on images of the regions, through a trained classifier model. In this manner, an objective (andAttorney Docket No. 5089-000208-WO-POA generally automated) identification of cover crops in the regions, based on image data (including both optical data and SAR data), is provided, which avoids manual intervention and data compilation by individual growers, etc. In turn, from the mapping, one or more crop management decisions may be implemented with regard to the regions and, more particularly, the fields in the regions (e.g., practice verification / monitoring, application, etc. of desired treatments to the fields such as pesticides, a herbicides, and / or a fertilizers: etc.).

[0077] It should be appreciated that the functions described herein, in some embodiments, may be described in computer executable instructions stored on a computer readable media, and executable by one or more processors. The computer readable media is a non-transitory computer readable media. By way of example, and not limitation, such computer readable media can include RAM. ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. Combinations of the above should also be included within the scope of computer-readable media.

[0078] It should also be appreciated that one or more aspects, features, operations, etc. of the present disclosure may transform a general-purpose computing device into a special-purpose computing device when configured to perform the functions, methods, and / or processes described herein.

[0079] As will be appreciated based on the foregoing specification, the abovedescribed embodiments of the disclosure may be implemented using computer programming or engineering techniques, including computer software, firmware, hardware or any combination or subset thereof, wherein the technical effect may be achieved by performing at least one of the following operations: (a) accessing, by a computing device, an image of one or more fields, the image including multiple pixels, each of the multiple pixels including a value for each of multiple bands; (b) deriving, by the computing device, at least one index value for each of the multiple pixels of the image: (c) generating, by the computing device, a map of cover crops for the one or more fields, using a trained model and the at least one index value for each of the multiple pixels of the image, the map of cover crops indicating presence or absence of cover crops in the multiple pixels of the one or more fields: (d) storing, by theAttorney Docket No. 5089-000208-WQ-POA computing device, the map of cover crops for the one or more fields in a memory; and / or (e) causing display of the map of cover crops for the one or more fields at an output device.

[0080] Examples and embodiments are provided so that this disclosure will be thorough, and will fully convey the scope to those who are skilled in the art. Numerous specific details are set forth such as examples of specific components, devices, and methods, to provide a thorough understanding of embodiments of the present disclosure. It will be apparent to those skilled in the art that specific details need not be employed, that example embodiments may be embodied in many different forms and that neither should be construed to limit the scope of the disclosure. In some example embodiments, well-known processes, well-known device structures, and well-known technologies are not described in detail, hi addition, advantages and improvements that may be achieved with one or more example embodiments disclosed herein may provide all or none of the above-mentioned advantages and improvements and still fall within the scope of the present disclosure.

[0081] Specific values disclosed herein are example in nature and do not limit the scope of the present disclosure. The disclosure herein of particular values and particular ranges of values for given parameters are not exclusive of other values and ranges of values that may be useful in one or more of the examples disclosed herein. Moreover, it is envisioned that any two particular values for a specific parameter stated herein may define the endpoints of a range of values that may also be suitable for the given parameter (z.e., the disclosure of a first value and a second value for a given parameter can be interpreted as disclosing that any value between the first and second values could also be employed for the given parameter). For example, if Parameter X is exemplified herein to have value A and also exemplified to have value Z, it is envisioned that parameter X may have a range of values from about A to about Z. Similarly, it is envisioned that disclosure of two or more ranges of values for a parameter (whether such ranges are nested, overlapping or distinct) subsume all possible combination of ranges for the value that might be claimed using endpoints of the disclosed ranges. For example, if parameter X is exemplified herein to have values in the range of 1 - 10, or 2 - 9, or 3 - 8, it is also envisioned that Parameter X may have other ranges of values including 1 — 9, 1 — 8, 1 — 3, 1 - 2, 2 — 10, 2 — 8, 2 — 3, 3 — 10, and 3 - 9.

[0082] The terminology used herein is for the purpose of describing particular example embodiments only and is not intended to be limiting. As used herein, the singularAttorney Docket No. 5089-000208-WO-POA forms "a,” “an,” and “the” may be intended to include the plural forms as well, unless the context clearly indicates otherwise. The terms “comprises,” “comprising,” “including,” and “having,” are inclusive and therefore specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. The method steps, processes, and operations described herein are not to be construed as necessarily requiring their performance in the particular order discussed or illustrated, unless specifically identified as an order of performance. It is also to be understood that additional or alternative steps may be employed.

[0083] When a feature is referred to as being “on,” “engaged to,” “connected to,” “coupled to,” “associated with,” “in communication with,” or “included with” another element or layer, it may be directly on, engaged, connected or coupled to, or associated or in communication or included with the other feature, or intervening features may be present. As used herein, the term “and / or” and the phrase “at least one of” includes any and all combinations of one or more of the associated listed items.

[0084] Although the terms first, second, third, etc. may be used herein to describe various features, these features should not be limited by these terms. These terms may be only used to distinguish one feature from another. Terms such as “first,” “second,” and other numerical terms when used herein do not imply a sequence or order unless clearly indicated by the context. Thus, a first feature discussed herein could be termed a second feature without departing from the teachings of the example embodiments.

[0085] The foregoing description of the embodiments has been provided for purposes of illustration and description. It is not intended to be exhaustive or to limit the disclosure. Individual elements or features of a particular embodiment are generally not limited to that particular embodiment, but. where applicable, are interchangeable and can be used in a selected embodiment, even if not specifically shown or described. The same may also be varied in many ways. Such variations are not to be regarded as a departure from the disclosure, and all such modifications are intended to be included within the scope of the disclosure.

Claims

Attorney Docket No. 5089-000208-WO-POACLAIMSWhat is claimed is:

1. A computer-implemented method for use in processing image data associated with fields, the method comprising: accessing, by a computing device, an image of one or more fields, the image including multiple pixels, each of the multiple pixels including a value for each of multiple bands; deriving, by the computing device, at least one index value for each of the multiple pixels of the image; generating, by the computing device, a map of cover crops for the one or more fields, using a trained model and the at least one index value for each of the multiple pixels of the image, the map of cover crops indicating presence or absence of cover crops in the multiple pixels of the one or more fields; storing, by the computing device, the map of cover crops for the one or more fields in a memory; and causing display of the map of cover crops for the one or more fields at an output device.

2. The computer-implemented method of claim 1, further comprising: accessing, by the computing device, ground measurement data of the cover crops for the one or more fields; and pre-processing the ground measurement data; and wherein the trained model is trained on the pre-processed ground measurement data and the at least one index value for each of the multiple pixels of the image.

3. The computer-implemented method of claim 2, wherein the pre-processing of the ground measurement data further comprises: de-duplicating overlapping fields in same-year data; combining same-year data of a plurality of years into multi-year data; de-duplicating overlapping fields in the multi-year data; and removing data with conflicting cover crop information.Attorney Docket No. 5089-000208-WO-POA4. The computer-implemented method of claim 1, wherein the multiple bands include red, blue, green, and near infrared.

5. The computer-implemented method of claim 1, wherein the deriving of the at least one index value for each of the multiple pixels of the image is based on at least one of:ND VI - (NIR-red) / (NIR+red);CAI = 0.5(blue+swir2)-red);LCAI = 100((blue-red)+(blue-swir2));NDRI = (swir2-red) / (swirl+swir2);NDTI = (swirl-swir2) / (swirl+swir2);GCVI = NIR / green;STI = swirl / swir2;NDI5 = (NIR-swirl) / (NIR+swirl);NDI7 = (NIR-swir2) / (NIR+swir2); and / orCRC = (swirl-green) / (swirl+green); and wherein NIR is a near infrared band value, red is a red band value, green is a green band value, swirl is a short-wave infrared 1 value, and swir2 is a short-wave infrared 2 value.

6. The computer-implemented method of claim 1, further comprising: accessing, by the computing device, Synthetic Aperture Radar (SAR) backscatter data and / or elevation data features for the one or more fields; and wherein the trained model is further trained on the SAR backscatter data and / or the elevation data features.

7. The computer-implemented method of claim 6, wherein the SAR backscatter data comprises data in vertical transmit, vertical receive (VV) mode and vertical transmit, horizontal receive (VH) mode; and wherein the elevation data features comprise at least one of height, slope, aspect, prof_curvature, plan_curvature, curvature. flow_accum, and relative_altitude data queried from a 10-meter Digital Elevation Model (DEM).Attorney Docket No. 5089-000208-WO-POA8. The computer-implemented method of claim 6, wherein the deriving of the at least one index value for each of the multiple pixels of the image is based on the following:ND VI = (NIR-red) / (NIR+red); and wherein NIR is a near infrared band value and red is a red band value, and wherein the method further comprises: removing pixels corresponding to ND VI < 0; removing pixels corresponding to backscatter >= 0; calculating first seasonal composites of the at least one index value; calculating second seasonal composites of the SAR backscatter data; calculating first field level aggregates of the at least one index value; and calculating second field level aggregates of the SAR backscatter data.

9. The computer-implemented method of claim 8. wherein the first seasonal composites and the second seasonal composites are calculated for winter and spring seasons; and wherein the first seasonal composites and the second seasonal composites comprise at least one of minimum, maximum, mean, and median of the at least one index value and the SAR backscatter data.

10. The computer-implemented method of claim 1, wherein the trained model includes at least one of a logistic regression model and a Residual Network (RESNET) model.

11. A system for use in processing image data associated with fields, the system comprising a computing device configured to: access an image of one or more fields, the image including multiple pixels, each of the multiple pixels including a value for each of multiple bands; derive at least one index value for each of the multiple pixels of the image; generate a map of cover crops for the one or more fields, using a trained model and the at least one index value for each of the multiple pixels of the image, the map of cover crops indicating presence or absence of cover crops in the multiple pixels of the one or more fields;Attorney Docket No. 5089-000208-WO-POA store the map of cover crops for the one or more fields in a memory; and cause display of the map of cover crops for the one or more fields at an output device.

12. The system of claim 11, wherein the computing device is further configured to: access ground measurement data of the cover crops for the one or more fields; and pre-process the ground measurement data, wherein the trained model is trained on the pre-processed ground measurement data and the at least one index value for each of the multiple pixels of the image.

13. The system of claim 12, wherein to pre-process the ground measurement data, the computing device is further configured to: de-duplicate overlapping fields in same-year data; combine same-year data of a plurality of years into multi-year data; de-duplicate overlapping fields in the multi-year data; and remove data with conflicting cover crop information.

14. The system of claim 11, wherein the multiple bands include red, blue, green, and near infrared.

15. The system of claim 11, wherein the at least one index value for each of the multiple pixels of the image is based on at least one of:ND VI = (NIR-red) / (NIR+red):CAI = 0.5(blue+swir2)-red);LCAI = 100((blue-red)+(blue-swir2));NDRI = (swir2-red) / (swirl+swir2);NDTI = (swirl-swir2) / (swirl+swir2);GCVI = NIR / green;STI = swirl / swir2;NDI5 = (NIR-swirl) / (NIR+swirl);NDI7 = (NIR-swir2) / (NIR+swir2); and / orCRC = (swirl-green) / (swirl+green); andAttorney Docket No. 5089-000208-WO-POA wherein NIR is a near infrared band value, red is a red band value, green is a green band value, swirl is a short-wave infrared 1 value, and swir2 is a short-wave infrared 2 value.

16. The system of claim 11, wherein the computing device is further configured to: access Synthetic Aperture Radar (SAR) backscatter data and / or elevation data features for the one or more fields: wherein the trained model is further trained on the SAR backscatter data and / or the elevation data features.

17. The system of claim 16, wherein the SAR backscatter data comprises data in vertical transmit, vertical receive (VV) mode and vertical transmit, horizontal receive (VH) mode; and wherein the elevation data features comprise at least one of height, slope, aspect. prof_curvature, plan_curvature. curvature. flow_accum, and relative_altitude data queried from a 10-meter Digital Elevation Model (DEM).

18. The system of claim 16, wherein the at least one index value for each of the multiple pixels of the image is based on the following:NDVI = (NIR-red) / (NIR+red)', and wherein NIR is a near infrared band value and red is a red band value, and wherein the computing device is further configured to: remove pixels corresponding to NDVI < 0; remove pixels corresponding to backscatter >= 0; calculate first seasonal composites of the at least one index value; calculate second seasonal composites of the SAR backscatter data; calculate first field level aggregates of the at least one index value; and calculate second field level aggregates of the SAR backscatter data.

19. The system of claim 18, wherein the first seasonal composites and the second seasonal composites are calculated for winter and spring seasons; andAttorney Docket No. 5089-000208-WO-POA wherein the first seasonal composites and the second seasonal composites comprise at least one of minimum, maximum, mean, and median of the at least one index value and the SAR backscatter data.

20. The system of claim 11, wherein the trained model includes at least one of a logistic regression model and a Residual Network (RESNET) model.

Images