Mapping water status based on aerial thermal imagery: comparison of methodologies for upscaling from a single leaf to commercial fields

Aerial thermal remote sensing can provide a means for collecting spatial plant water status data. Many studies have shown their potential in irrigation management but the adaptation of this technology is not straight forward. In this paper, knowledge accumulated in recent years on thermal imagery analysis methodology for water status mapping is summarized aiming at indicating alternatives to calculate the Crop Water Stress Index (CWSI) for commercial scale water status mapping. Based on literature overview, four forms of wet-baselines to calculate CWSI were selected, namely: artificial wet reference surface, two theoretical calculations and a statistical bio-indicator. These baselines were used to calculate CWSI based on multi-temporal aerial thermal images of cotton fields. CWSI based on a statistical bio-indicator and one of the theoretical wet-baselines provided the best correlations. It is argued though that the statistical one is preferable since it includes the plant characteristics and it is farmer-friendly. Based on bio-indicators, leaf water potential maps of three commercial fields were produced on several dates through the season. Water status spatial patterns were not static and the effect of static factors like sandy soil patches also changed through the season. The maps show the importance of in-season variability mapping for rational irrigation management. To improve current variable-rate irrigation, the concept of in-season irrigation management zones (IMZ) based on thermal-images should be considered and integrated with the delineation of static irrigation IMZ.     

Estimation of yield zones using aerial images and yield data from a few tracks of a combine harvester

Yield maps derived from yield mapping systems are often erroneous not only due to limitations in measuring the yield precisely but due to insufficient consideration of the requirements of yield mapping systems in practice as well. Aerial images of cultivated crop fields at an advanced growth stage frequently provide a spatial pattern similar to that of yield maps. Therefore, the possibility of generating a yield map using aerial images and measured yield data of a few tracks was examined for a period of 2 years in two fields grown with cereals. Yield zones based on Visible Atmospherically Resistant Index (VARI) values were compared with yield zones based on measured yield data of the whole field. About half of the grid cells of a field were allocated to the same yield zones irrespective of the mode of yield determination. Using the Kruskal–Wallis test, the data sub-sets of measured yield within the yield zones based on the VARI values differed significantly for all tested yield zones. As a result, the approach was successful in the case of these experimental sites.     

Underlying causes of yield spatial variability and potential for precision management in rice systems

Our current understanding of the mechanisms driving spatiotemporal yield variability in rice systems is insufficient for effective management at the sub-field scale. The overall objective of this study was to evaluate the potential of precision management for rice production. The spatiotemporal properties of multiyear yield monitor data from four rice fields, representing varying soil types and locations within the primary rice growing region in California, were quantified and characterized. The role of water management, land-leveling, and the spatial distribution of soil properties in driving yield heterogeneity was explored. Mean yield and coefficient of variation at the sampling points within each field ranged from 9.2 to 12.1 Mg ha^?1 and from 7.1 to 14.5 %, respectively. Using a k-means clustering and randomization method, temporally stable yield patterns were identified in three of the four fields. Redistribution of dissolved organic carbon, nitrogen, potassium and salts by lateral flood water movement was observed across all fields, but was only related to yield variability via exacerbating areas with high soil salinity. The effects of cold water temperature and land-leveling on yield variability were not observed. Soil electrical conductivity and/or plant available phosphorus were identified as the underlying causes of the within-field yield patterns using classification and regression trees. Our results demonstrate that while the high temporal yield variability in some rice fields does not permit precision management, in other fields exhibiting stable yield patterns with identifiable causes, precision management and modified water management may improve the profitability and resource-use efficiency of rice production systems.     

Frequency Analysis of Yield for Delineating Yield Response Zones

The yield in any given field or management zone is a product of interaction between many soil properties and production inputs. Therefore, multi-year yield maps may give better insight into determining potential management zones. This research was conducted to develop a methodology to delineate yield response zones by using two-state frequency analysis conducted on yield maps for 3 years on two commercial corn fields near Wiggins, Colorado. A zone was identified by the number of years that yield was equal and greater than the average yield in a given year. Classes producing statistically similar yield were combined resulting in three potential yield zones. Results indicated that the variability of yield over time and space could successfully be assessed at the same time without the drawbacks of averaging data from different years. Frequency analysis of multi-year yield data could be an effective way to establish yield response zones. Seventeen percent of the field #1 consistently produced lower yield than the mean while 43 of the field produced yield over the mean. Corresponding values for field #2 were 6% and 42%.The remainder of the fields produced fluctuating yields between years. These spatially and temporally sound yield response maps could be used to identify the yield-limiting factors in zones where yield is either low or fluctuating. Yield response maps could also be helpful to delineate potential management zones with the help of resource zones such as electrical conductivity and soil maps, along with the directed soil sampling results.     

Identifying and characterizing yield limiting factors in paddy rice using remote sensing yield maps

Identification and characterization of yield limiting factors based on multi-year yield maps is important for delineating field management zones. Multi-year yield maps were derived from satellite images of a paddy-rice (Oryza sativa L.) study site with a conventional two-cropping system in central Taiwan. Spatiotemporal yield-trend maps with consistently high, average and low yields, and inconsistent yield areas were delineated based on temporal variation and the means of the normalized yields on a per pixel basis. Soil and plant samples were collected and grouped for statistical analysis based on the derived yield-trend maps. Comparison of soil properties and rice yield components among yield classes indicated that differences in leaching loss of basal and top-dressed N fertilizers were the likely limiting factor affecting the spatial variation of yield within the study site.     

Spatial and temporal stability of soil phosphate concentration and pasture dry matter yield

Several methods are described that could be used by a farm manager to define the spatial and temporal stability within a field from a series of yield maps. A quantitative analysis of soil phosphate concentration and pasture dry matter yield data over 4 years (2004–2007) were investigated to identify the spatial and temporal stability in a 6 ha pasture field. The data were combined into two maps that characterize the spatial and temporal variation recorded over the 4 years. The two maps were then combined to create a single map with five management classes, each with different characteristics that can have an impact on the way the field is managed. These categories are: high yielding and stable, high yielding and moderately stable, low yielding and stable, low yielding and moderately stable and unstable. The unstable class represents 83 and 93% of the total area with regard to soil phosphate concentration and pasture dry matter yield, respectively. Results from this study show that the significant temporal stability found cancels out over time, leaving a relatively homogenous map of spatial variation. The implication of the findings is that each pasture field should be managed according to the current year’s conditions. These results also justify a further study that evaluates the soil phosphorous dynamics under Mediterranean conditions.     

产量分布图生成系统的研究

使用Visual Basic6．0编程语言开发了YMapper产量分布图生成软件系统。研究了一种具有粗大误差数据过滤功能的局部平均插值方法，分析了产量分布图生成过程中涉及到的坐标系定义、产量数据分类与统计分析、图形配色与绘制等问题。对谷物和棉花等作物产量数据的处理结果表明，该系统能够对联合收割机测产系统记录的产量数据文件进行处理，通过插值运算将离散分布的产量数据点生成连续的产量分布图；其误差数据过滤功能能够防止产量过低和过高的粗大误差数据点参与插值运算，使产量分布图的精度得到了保证。系统能够按照用户设置的分类和着色方式，将作物产量的空间分布情况以产量数据点图或产量分布图的形式显示。并且能够对作物的产量水平进行统计分析。     

Comparison of QuickBird Satellite Imagery and Airborne Imagery for Mapping Grain Sorghum Yield Patterns

Timely and accurate information on crop conditions obtained during the growing season is of vital importance for crop management. High spatial resolution satellite imagery has the potential for mapping crop growth variability and identifying problem areas within fields. The objectives of this study were to use QuickBird satellite imagery for mapping plant growth and yield patterns within grain sorghum fields as compared with airborne multispectral image data. A QuickBird 2.8-m four-band image covering a cropping area in south Texas, USA was acquired in the 2003 growing season. Airborne three-band imagery with submeter resolution was also collected from two grain sorghum fields within the satellite scene. Yield monitor data collected from the two fields were resampled to match the resolutions of the airborne imagery and the satellite imagery. The airborne imagery was related to yield at original submeter, 2.8 and 8.4 m resolutions and the QuickBird imagery was related to yield at 2.8 and 8.4 m resolutions. The extracted QuickBird images for the two fields were then classified into multiple zones using unsupervised classification and mean yields among the zones were compared. Results showed that grain yield was significantly related to both types of image data and that the QuickBird imagery had similar correlations with grain yield as compared with the airborne imagery at the 2.8 and 8.4 m resolutions. Moreover, the unsupervised classification maps effectively differentiated grain production levels among the zones. These results indicate that high spatial resolution satellite imagery can be a useful data source for determining plant growth and yield patterns for within-field crop management.     

Field methods for making productivity classes for site-specific management of wheat

Reducing the decision-making unit to classes within fields can improve yields, efficiency in the use of nutrients and profitability of crops. The objectives were to compare methods for class delimitation in wheat (Triticum aestivum L.) crops based on apparent productivity levels and establish similarities among them in terms of spatial overlapping, productive attributes and the use of nitrogen. In three wheat fields, high and low apparent productivity classes (APC) were defined based on eight methodologies: yield maps, soil maps, gramineae vegetation index, rotation crop index, interpretation of satellite images, management records, elevation and integrated soil and yield maps. In each APC, soil and crop yield components were determined under five nitrogen fertilization levels. Among delimitation methodologies, the degree of coincidence varied from 1.4 to 81.7%. The differences in soil properties, nitrogen use efficiency and grain yields were greater among fields than among APC within each field. In each field, the delimitation methodologies identified different single factors that discriminated among the potential management classes and were partially associated with the crop grain yields. The wheat crops at the low APC yielded 39% less and 12% less than at the high APC, respectively. The nitrogen fertilization, at the rate for maximum productivity for each ACP, reduced the yield differences between contrasting APC. Nitrogen fertilization also modified clustering of classes based on expected yields. Making management classes for wheat based on expected productivity is more accurate when based on previous crop production information under similar nitrogen fertilization conditions than the targeted crop.

     

Strategies for monitoring within-field soybean yield using Sentinel-2 Vis-NIR-SWIR spectral bands and machine learning regression methods

Soybean crop plays an important role in world food production and food security, and agricultural production should be increased accordingly to meet the global food demand. Satellite remote sensing data is considered a promising proxy for monitoring and predicting yield. This research aimed to evaluate strategies for monitoring within-field soybean yield using Sentinel-2 visible, near-infrared and shortwave infrared (Vis/NIR/SWIR) spectral bands and partial least squares regression (PLSR) and support vector regression (SVR) methods. Soybean yield maps (over 500 ha) were recorded by a combine harvester with a yield monitor in 15 fields (3 farms) in Paraná State, southern Brazil. Sentinel-2 images (spectral bands and 8 vegetation indices) across a cropping season were correlated to soybean yield. Information pooled across the cropping season presented better results compared to single images, with best performance of Vis/NIR/SWIR spectral bands under PLSR and SVR. At the grain filling stage, field-, farm- and global-based models were evaluated and presented similar trends compared to leaf-based hyperspectral reflectance collected at the Brazilian National Soybean Research Center. SVR outperformed PLSR, with a strong correlation between observed and predicted yield. For within-field soybean yield mapping, field-based SVR models (developed individually for each field) presented the highest accuracies. The results obtained demonstrate the possibility of developing within-field yield prediction models using Sentinel-2 Vis/NIR/SWIR bands through machine learning methods.

     

An approach for assessing the effects of site-specific fertilization on crop growth and yield of durum wheat in organic agriculture

Precision agriculture (PA) technologies allow us to assess field variability and support site-specific (SSP) application of inputs. The joint application of PA and organic farming practices might be synergetic. The objective of this 3-year study was to propose a multivariate statistical and geostatistical approach, to evaluate the effects of SSP nitrogen (N) fertilization on durum wheat in transition to organic farming. Soil parameters were measured to assess soil fertility level before the SSP fertilization on wheat, which was carried out by management zones in the third year. Radiometric measurements were performed with a hyperspectral spectroradiometer and N-uptake at anthesis and grain yield were determined. The expected values and 95 % confidence intervals of the soil parameters, N-uptake and yield data were estimated with polygon kriging for each management zone. Reflectance data were reduced through principal component analysis and the retained principal components were submitted to factorial co-kriging analysis to estimate orthogonal scale-dependent factors. Comparisons between N-uptake and yield and between the retained regionalized factors (F1) and yield were performed. The spatial pattern of F1 at shorter scales was mostly reproduced in the N-uptake map, suggesting the predictive capacity of hyperspectral data for crop N-status. Within-cluster variance for yield was reduced, quite probably as a combined effect of meteorological pattern and management. The preliminary results seem to be promising in the perspective of PA. Moreover, an inverse relationship between grain yield and crop N-status was observed.     

Interdisciplinary Irrigated Precision Farming Research

The USDA-Agricultural Research Service and Colorado State University are conducting an interdisciplinary study that focuses on developing a clearer scientific understanding of the causes of yield variability. Two years of data have been collected from two commercial center pivot irrigated fields (72 and 52 ha). Cooperating farmers manage all farming operations for crop production and provide yield maps of the maize grown on the fields. The farmers apply sufficient inputs to minimize risk of yield loss. The important variables for crop production have been sampled at a grid spacing of 76 m for two seasons. A spatial auto-regressive model was fitted to the data to determine the critical factors affecting yield variability. Thirty one layers of data were included in the analysis, and a total of over 140,000 models were examined. Up to five predictors were used in each model. Variability in water application, nitrate nitrogen, organic matter, phosphorus, topology, percent silt and soil electrical conductivity were significant in explaining the yield variability for Field 1. Variability in water application, ammonium, nematodes, percent clay, insects, potassium, soil electrical conductivity, and topology were significant in explaining the yield variability for Field 2. The tentative conclusion is that the potential economic benefit of site specific management is small where the farmer's management tolerance for risk is low. The potential of site specific management is in reducing the cost of inputs and environmental impact, but could increase risk.     

From yield history to productivity zone identification with hidden Markov random fields

Modern precision agriculture equipment enables site-specific management by allowing different treatments for different parts of a field. This ability to subdivide the field calls for identifying management zones. A compromise between treating a field uniformly and treating every plant individually is needed, as the former does not maximize yields and the latter is often impractical. This work presents an algorithm for inferring the yield productivity zones (YPZ) for a field based on yield data from multiple years. The algorithm uses a hidden Markov random field model (HMRF) to find regions of the field which likely correspond to the same underlying yield distribution (i.e., productivity zones). These regions are modeled to be the same every year, but their distributions (i.e., yield characteristics) are allowed to vary with time to account for year-to-year variability (from e.g., weather effects, differing crops or crop varieties). The zone assignments and distributions are estimated using stochastic expectation maximization (SEM) and the maximizer of the posterior marginals (MPM). The underlying assumption of the model and algorithm is that the yields corresponding to a given YPZ will behave similarly and therefore derive from the same probability distribution. YPZs are useful inputs for determining management zones. An advantage of this method is that it is able to run with only the yield data which are automatically collected during harvest. Also, this method requires no crop specific calibration or configuration or normalization of the data by year.

     

Evaluation of an analytical method to identify determinants of rice yield components and protein content

Modern information technologies have facilitated the collection of data to assess various aspects of rice production such as yield, quality, soil properties and growth conditions. Currently, farmers can identify any variation of these indicators within a field, between fields or with other farmers. However, a comprehensive analytical method to identify the determinants of variability has not been developed, and the data collected are not efficiently utilized to diagnose and improve the production skills of farmers. Our study focused on the development of an analytical method that can identify the determinants of rice yield and quality. The analytical method used applied cluster analysis (Ward method) to assess the data from 82 paddy fields where rice is produced in various environments and with various management styles. Initially, the 82 paddy fields were classified into 11 clusters based on five indicators of yield components and rice quality; number of panicles, number of spikelets, percentage of ripened grains, 1000-grain weight (GW) and protein content of brown rice. Then, 9 of 11 clusters (two clusters were excluded due to insufficient data to form a cluster) were divided into four groups based on yield capacity. As a result, common characteristics of fertilizer application, meteorological environment and growth conditions were extracted from each cluster. Furthermore, determinants of yield components and protein content were efficiently identified based on the common characteristics extracted.     

Geostatistical analysis of fruit yield and detachment force in coffee

The aim of this study was to use geostatistical analysis to evaluate the spatial variation in the detachment force of coffee fruit and coffee yield by variograms and kriging for precision agriculture. This study was conducted at Brej?o farm, Três Pontas, Minas Gerais, Brazil. The detachment force of green and mature coffee fruit was measured with a prototype dynamometer and georeferenced. The yield data were obtained from manual harvesting and were georeferenced. The data were evaluated by variograms estimated by residual maximum likelihood (REML), which provided a satisfactory approach for modeling all the variables with a small sample size. Spherical and exponential models were fitted, the first provided the better fit to mature fruit detachment force and the latter provided the better fit to coffee yield and green fruit detachment force. They were used to describe the structure and magnitude of spatial variation in the variables studied. Kriged estimates were obtained with the best fitting variogram models and mapped. The statistical and geostatistical analyses enabled us to characterize the spatial variation of the detachment force of green and mature coffee fruit and coffee yield and to visualize the spatial relations among these variables. The precision agriculture techniques used in this paper to collect, map and analyze the variables studied will help coffee farmers to manage their fields. Maps of coffee yield will enable farmers to apply nutrients site-specifically and manage harvesting either manually or mechanically. In addition, maps of detachment force of coffee fruit can enable farmers to harvest coffee selectively by choosing the appropriate places and the right time to start. This will improve the quality of the final product and also increase profits.     

Predicting Cotton Lint Yield Maps from Aerial Photographs

It is generally accepted that aerial images of growing crops provide spatial and temporal information about crop growth conditions and may even be indicative of crop yield. The focus of this study was to develop a straightforward technique for creating predictive cotton yield maps from aerial images. A total of ten fields in southern Georgia, USA, were studied during three growing seasons. Conventional (true color) aerial photographs of the fields were acquired during the growing season in two to four week intervals. The aerial photos were then digitized and analyzed using an unsupervised classification function of image analysis software. During harvest, conventional yield maps were created for each of the fields using a cotton picker mounted yield monitor. Classified images and yield maps were compared quantitatively and qualitatively. A pixel by pixel comparison of the classified images and yield maps showed that spatial agreement between the two gradually increased in the weeks after planting, maintained spatial agreement of between 40% and 60% during weeks eight to fourteen, and then gradually declined again. The highest spatial agreement between a classified image and a yield map was 78%. The highest average agreement was 52% and occurred 9.9 weeks after planting. The visual similarity between the classified images and the yield maps were striking. In all cases, the dates with the best visual agreement occurred between eight and ten weeks after planting, and generally, during July for southern Georgia. This method offers great potential for offering cotton farmers early-season maps that predict the spatial distribution of yield. Although these maps can not provide magnitudes, they clearly show the resulting yield patterns. With inherent knowledge of past performance, farmers can use this information to allocate resources, address crop growth problems, and, perhaps, improve the profitability of their farm operation. These maps are well suited to be offered to farmers as a service by a crop consultant or a cooperative.     

Subfield crop yields and temporal stability in thousands of US Midwest fields

Understanding subfield crop yields and temporal stability is critical to better manage crops. Several algorithms have proposed to study within-field temporal variability but they were mostly limited to few fields. In this study, a large dataset composed of 5520 yield maps from 768 fields provided by farmers was used to investigate the influence of subfield yield distribution skewness on temporal variability. The data are used to test two intuitive algorithms for mapping stability: one based on standard deviation and the second based on pixel ranking and percentiles. The analysis of yield monitor data indicates that yield distribution is asymmetric, and it tends to be negatively skewed (p?<?0.05) for all of the four crops analyzed, meaning that low yielding areas are lower in frequency but cover a larger range of low values. The mean yield difference between the pixels classified as high-and-stable and the pixels classified as low-and-stable was 1.04 Mg ha^?1 for maize, 0.39 Mg ha^?1 for cotton, 0.34 Mg ha^?1 for soybean, and 0.59 Mg ha^?1 for wheat. The yield of the unstable zones was similar to the pixels classified as low-and-stable by the standard deviation algorithm, whereas the two-way outlier algorithm did not exhibit this bias. Furthermore, the increase in the number years of yield maps available induced a modest but significant increase in the certainty of stability classifications, and the proportion of unstable pixels increased with the precipitation heterogeneity between the years comprising the yield maps.

     

Yield prediction in apple orchards based on image processing

It has been suggested that apple ( Malus * domestica Borkh) flowering distribution maps can be used for site-specific management decisions. The objectives of this study were (i) to study the flower density variability in an apple orchard using image analysis and (ii) to model the correlation between flower density as determined from image analysis and fruit yield. The research was carried out in a commercial apple orchard in Central Greece. In April 2007, when the trees were at full bloom, photos of the trees were taken following a systematic uniform random sampling procedure. In September 2007, yield mapping was carried out measuring yield per ten trees and recording the position of the centre of the ten trees. Using this data (the measured yield of the trees and the pictures samples, representing the flower distribution), an image processing-based algorithm was developed that predicts tree yield by analyzing the picture of the tree at full bloom. For the evaluation of the algorithm, a case study scenario is presented where the error of the predicted yield was set at 18%. These results indicated that potential yield could be predicted early in the season from flowering distribution maps and could be used for orchard management during the growing season.     

Changes in yield classification in a soybean-rice rotation

For yield based site-specific management to be successful in fields with crop rotations, changes in management zones between crops must be determined. The study objectives were to determine if yield classes change between crops within a rotation and whether soil properties can predict the yield classes or the year-to-year changes. A percentile classification method was used to categorize yearly soybean (Glycine max) and rice (Oryza sativa) yield in two fields with soybean-rice-soybean rotations into low, medium and high yield classes. There was little agreement in yield classifications between years. Yield class based on soil properties was predicted accurately by linear discriminant analysis in Field 1 20–67% of the time and in Field 2 13–83% of the time. Predictions in Field 1 were based on soil available Mg and P, elevation and the deep soil apparent electrical conductivity (EC_a). Predictions in Field 2 were based on soil texture, soil available P, K and Mg, and pH. The linear discriminant analysis was also able to predict year-to-year changes in yield class. Changes in class in Field 1 could be predicted by total soil C and N, silt, and soil available Mg and P depending on the year. Soil texture, soil available P, K and Mg, total soil C and pH, elevation and deep soil EC_a predicted yield changes in Field 2 depending on the year. The results of this study indicate only limited success at management zone definition in a soybean-rice rotation. Further investigation is needed with other crop rotation sequences to verify the findings of this study.