Comparing a random forest based prediction of winter wheat yield to historical production potential

<p>Predicting wheat yield is crucial due to the importance of wheat across the world. When modeling yield, the difference between potential and actual yield consistently changes because of technology. Considering historical yield potential would help determine spatiotemporal trends in agricultural development. Comparing current and historical production in Denmark is possible because production has been documented throughout history. However, the current winter wheat yield model is solely based on soil. The aim of this study was to generate a new Danish winter wheat yield map and compare the results to historical production potential. Utilizing random forest with soil, climate, and topography variables, a winter wheat yield map was generated from 876 field trials carried out from 1992 to 2018. The random forest model performed better than the model based only on soil. The updated national yield map was then compared to production potential maps from 1688 and 1844. While historical time periods are characterized by numerous low production potential areas and few highly productive areas, present-day production is evenly distributed between low and high production. Advances in technology and farm practices have exceeded historical yield predictions. Thus, modeling current yield could be unreliable in future years as technology progresses.</p>

Comparing a Random Forest Based Prediction of Winter Wheat Yield to Historical Yield Potential

Predicting wheat yield is crucial due to the importance of wheat across the world. When modeling yield, the difference between potential and actual yield consistently changes because of advances in technology. Considering historical yield potential would help determine spatiotemporal trends in agricultural development. Comparing current and historical yields in Denmark is possible because yield potential has been documented throughout history. However, the current national winter wheat yield map solely uses soil properties within the model. The aim of this study was to generate a new Danish winter wheat yield map and compare the results to historical yield potential. Utilizing random forest with soil, climate, and topography variables, a winter wheat yield map was generated from 876 field trials carried out from 1992 to 2018. The random forest model performed better than the model based only on soil. The updated national yield map was then compared to yield potential maps from 1688 and 1844. While historical time periods are characterized by numerous low yield potential areas and few highly productive areas, current yield is evenly distributed between low and high yields. Advances in technology and farm practices have exceeded historical yield predictions, mainly due to the use of fertilizer, irrigation, and drainage. Thus, modeling yield projections could be unreliable in the future as technology progresses.

Sensors for crop productivity, yield maps

For gathering the final results of a season-round crop work, the georeferenced yield is a key piece of information. Harvesting equipment can be equipped with sensors to gather such information. Systems based on different technologies (impact, volume, optics, density, gravity…) will be explained for recording the yield flow inside the machinery, during the harvesting. Adaptations of yield sensors depending on the commodity, along with new sensing systems will be discussed. Sensor for quality quantification will also be explained, as they are important for certain crops. Basic procedures for the calibration of the sensing system and the proper registration of yield data to generate a successful yield map are presented.

Integrated Method for Rice Cultivation Monitoring Using Sentinel-2 Data and Leaf Area Index

Rice is an essential crop for national food security in Egypt. Increasing the population calls for regular increases in rice production. At the same time, cultivated rice crop areas should be decreased because of the gradual scarcity of irrigation water. This means more rice production should be gained from less rice area. This situation calls for the annual accurate system for rice monitoring and yield estimation. Therefore, it is necessary to apply a remotely sensed based system for rice cultivation assessment using satellite imagery parallel with field measurements of some biophysical parameters. Multi-temporal normalized difference vegetation index (NDVI) extracted from twelve sentinel-2 imagery cover the whole summer season with variance and maximum value assessed by ground control points (GCPs), were used to isolate uncultivated areas, then to isolate rice areas and other vegetation covers. object-based classification methods with kappa co-efficient 0.9261 and overall accuracy 94.92% was generated to discriminate rice crop area and other summer crops on the study area. Leaf area index (LAI) for the experiment the l site was calculated using the surface energy balance algorithm for Land (SEBAL) model and then tested versus measured (LAI). NDVI and LAI were used to generate an empirical ran rice yield prediction model. Then, this model was used to produce rice to yield a map. The study was carried out in an experimental site in Kafr Elsheikh governorate with a total area of 5040 Hectare. Produced cultivated land use map showed 95% overall accuracy. High similarity was observed between measured and calculated (LAI) with high accuracy of R2 = 0.94. of Rice, yield map showed expected to yield more to than a month before harvest. The generated yield map was tested using a correlation coefficient between actual yield and estimated yield with high accuracy R2 = 0.9. This method is applicable to estimate the acreage and productivity of rice in the northern Nile delta in adequate time before harvest.

Integrating SEBAL with in-Field Crop Water Status Measurement for Precision Irrigation Applications—A Case Study

The surface energy balance algorithm for land (SEBAL) has been demonstrated to provide accurate estimates of crop evapotranspiration (ET) and yield at different spatial scales even under highly heterogeneous conditions. However, validation of the SEBAL using in-field direct and indirect measurements of plant water status is a necessary step before deploying the algorithm as an irrigation scheduling tool. To this end, a study was conducted in a maize field located near the Venice Lagoon area in Italy. The experimental area was irrigated using a 274 m long variable rate irrigation (VRI) system with 25-m sections. Three irrigation management zones (IMZs; high, medium and low irrigation requirement zones) were defined combining soil texture and normalized difference vegetation index (NDVI) data. Soil moisture sensors were installed in the different IMZs and used to schedule irrigation. In addition, SEBAL-based actual evapotranspiration (ETr) and biomass estimates were calculated throughout the season. VRI management allowed crop water demand to be matched, saving up to 42 mm (−16%) of water when compared to uniform irrigation rates. The high irrigation amounts applied during the growing season to avoid water stress resulted in no significant differences among the IMZs. SEBAL-based biomass estimates agreed with in-season measurements at 72, 105 and 112 days after planting (DAP; r2 = 0.87). Seasonal ET matched the spatial variability observed in the measured yield map at harvest. Moreover, the SEBAL-derived yield map largely agreed with the measured yield map with relative errors of 0.3% among the IMZs and of 1% (0.21 t ha-1) for the whole field. While the FAO method-based stress coefficient (Ks) never dropped below the optimum condition (Ks = 1) for all the IMZs and the uniform zone, SEBAL Ks was sensitive to changes in water status and remained below 1 during most of the growing season. Using SEBAL to capture the daily spatial variation in crop water needs and growth would enable the definition of transient, dynamic IMZs. This allows farmers to apply proper irrigation amounts increasing water use efficiency.

Rapeseed crop damage by wildlife assessed from Landsat images

Rapeseed is the fourth most important crop in Hungary regarding its cultivation area. Crop damage by deer and boar has been becoming strongly debated issue in the last few years. More exact clarification of damage was aimed at in this study with help of Landast images. Six rapeseed fields were investigated both in 2012 and 2013 in the administrative area of Várfölde, Bánokszentgyörgy, Bázakerettye and Borsfa (Zala County, Hungary). The total area in 2013 was 43 hectares. 100 % wildlife damage affected 3.9 hectares and 10 hectares were free from any damage. The total area in 2012 was 40 hectares in which 3.3 hectares were free from damage but neither fields suffered total damage. Yield map from 2017 of a 26 hectares field near to Gutorfölde and Szentkozmadombja was used to validate the space image based assessment method with real yield data. Landsat 7 images with acquisition dates of 2013.04.16., 2013.05.18. and 2013.06.03. were downloaded from the website of US Geological Service. All bands and NDVI index were investigated for all dates to establish best estimator for differences between damaged and not damaged fields. Band 5 (SWIR: 1.55-1.75 μm) in 16th of April proved to be the best. It was concluded with help of the reflectance data (zero damage: 96.4, complete damage:164.5, partial damage:124.7 on the average) that yield reduction was 41.5 % on areas with partial damage. There was no complete damage in 2012. However, reflectance data of not damaged fields were very close to each other in the two years (96.4 in 2013 and 89.9 in 2012 on the average) thus, it could be assumed that the same is true for reflectance data of completely damaged fields, so data from 2013 can be used for the comparison. Based on the data (zero damage: 89.9/2012, complete damage:164.5/2013, partial damage:118.4/2012 on the average) it was calculated, that one field suffered 38 % yield reduction, one hectare portion of another field suffered 23 % yield reduction, and other fields were not damaged significantly. Yield map from 2017 and Landsat 8 SWIR reflectance (Band 6: 1.566 – 1.651 μm) in 3rd of April have shown strong correlation (R2=0,634), which was a direct evidence that both yield and wildlife damage of rapeseed can be reliably assessed from Landsat SWIR reflectance data acquired in April.

Vegetation indexes and delineation of management zones for soybean1

ABSTRACT The delimitation of site-specific management zones may be an operational and economically feasible approach in precision agriculture. This study aimed at investigating the spatial correlations between spectral indexes sampled during different growth stages of soybean and crop yield. Soil attributes stratified in each zone and the influence of altitude were also assessed. The simple ratio index, normalized difference vegetation index and soil-adjusted vegetation index were calculated for soybean at the V6, R5 and R5.5 stages. Spatial dependence analysis via semivariogram was performed for the vegetation indexes, soybean yield and terrain elevation. The crop yield map was taken as a reference to assess the spatial agreement with the different maps generated from the spectral indexes. The average values for chemical and granulometric soil attributes were calculated and analyzed by their means among the zones delineated. The field division into two management zones, due to the combination of altitude, simple ratio index of the V6 stage and soil-adjusted vegetation index of the R5.5 stage, showed the highest agreement with the soybean yield map. Differences between the delineated zones were identified for the phosphorus, clay and silt contents.