Spatial Dependence of Growth and Yield of Winter Wheat and Weeds Using Geostatistical and Remote Sensing Methods

Understanding the spatial relationship of weed populations and crop growth and yield within fields is increasingly important as methods are being developed for the site-specific management of weeds. Knowledge of spatial variability helps to improve weeds and crop management efficiency. It is well known that weed populations have a patchy distribution and a weed patch is considered stable if is consistent with density and location over time. Most of the spatial variability studies about weed populations have provided very precise information for site-specific recommendations and they are referred to as density data. The knowledge of spatial patterns within a field is critical not only to farmers for potential variable rate applications, but also to select homogenous zones within the field to run crop models with site-specific input to better understand and predict the impact of weeds, soil and landscape characteristics on spatial and temporal patterns of crop yields to enhance resource use efficiency at field level. Remote sensing and geostatistic methods have been widely explored as a tool for detection and mapping of weeds distribution in agricultural crops. Remote sensing is the science and art of obtaining information about an object, area, or phenomenon through the analysis of data acquired by a device that is not in contact with the object, area, or phenomenon under investigation. Thus, on the basis of different types of platforms carrying specific sensors, agricultural remote sensing could generally be categorized into satellite, aerial, and handheld or tractor mounted remote sensing. The cross-semivariogram analysis is one of geostatistical methods to describe the joint spatial dependence, or continuity between two variables such as crop yield and density of weeds and other pairs of variables that are influenced by the same local conditions or spatial processes. The objective of this study was to describe the spatial relationship between winter wheat and weeds using geostatistic and remote sensing.

Geostatistic and remote sensing techniques were used to describe spatial relationship between weeds distribution and wheat biological and grain yield on a winter wheat fields located in Faculty of Agriculture, Shahrood University of Technology in growing season 2017-2018. Wheat traits such as biological and grain yield, weed density, and biomass were measured at 110 points of field based on an 8×8 m grid. Also, in order to investigation of biological and grain yield of wheat using remote sensing, an aerial image was taken by a quadcopter from the height of 120 m and then numerical values of aerial image pixels extracted and analyzed using geostatistics methods. Before subjecting data to spatial statistical analysis, they were log-transformed and detrended by a median polishing procedure as described previously. To describe the co-continuity of two variables (wheat yield and weeds populations) cross-semivariance was used. Semivariograms and cross-semivariograms were calculated both using GS+ and Variowin (Software for spatial data analysis in 2D, Spring Verlag, New York, USA) softwares, respectively.

Semivariogram analysis showed that total weed density and biomass had patchy patterns on the field. Cross-semivariograms showed weed density and biomass had negative spatial correlation of 75.89 to 62.50% with wheat biological yield and negative spatial correlation of 91.9 and 56.25 with grain yield of wheat, respectively. Remote sensing data showed a strong correlation of 88.23 and 78.12% with biological and grain yield of wheat, respectively. High spatial dependence was observed between the maps of biological and grain yield of wheat. Also, weed density and biomass patterns on the maps of weeds were similar to patterns of wheat biological and grain yield on the maps. Competition for resources among plants has long been considered to generate stress for plants and to be important for determining the distribution of species, as well as their evolution. Weeds reduce crop yields primarily by competing with the crop for limited resources (light, water, nutrients, etc.). Knowledge of the distribution of weeds across the field creates an opportunity for site-specific weed management. One potential way to utilize this information is through variable-rate herbicide application (VRA). While mapping weeds via field scouting or remote sensing for precisely varying application can be expensive, photographs taken from an airplane at times when weed-crop discrimination is good may be a more affordable method of roughly mapping weed patches to allow for some reduction in herbicide use.

Base on this experiment results, weed density and biomass had a patchy distribution and so they can, depending on their impact degree, affect the growth and yield patterns of wheat in a non-uniform manner. Also geostatistical and remote sensing data has great potential for study of plant distribution and their relationships on the fields.