Extraction Of Growth Parameters Of Winter Wheat Based On Terrestrial LiDAR Data

This study was a part of "Impacts of climate change on food systems in China and its adaptation". The goal of this study was to investigate how laser point data could be explored systematically to measure important growth parameters of wheat such as crop height, degree of coverage, and biomass density in crop yield monitoring. Based on the previous research, this study presented a hierarchical algorithm that using terrestrial data to extract growth parameters of wheat stands. The major contents and conclusions were summarized as follow:(1) Terrestrial LiDAR data processing and wheat points extraction:We proposed a hierarchical processing algorithm to improve the process flow according to the terrestrial LiDAR data characteristics and the research objectives. Firstly, we divided laser points into two classes as outsider and insider points depending on whether the point belonged to the wheat fields or not, and then we processed the two classes in separate way. We could obtain a reference value of ground elevation after ground points filtering and statistical analysis. At last, we could extract wheat points after systematic process of insider points, including removal of gross errors, data thinning and removal of interference feature using a variety of methods and experimental comparison. This hierarchical algorithm reduced the hardness of LiDAR points classification based solely on3D coordinate information, Improved the classification accuracy and reduced the workload. (2) Modeling of terrestrial LiDAR data:We concluded the difficulties of DSM generation, compared interpolation methods with summarizing the interpolation principle and experimental analysis. We chose two methods including1DW and TIN to build DSM and found out the optimal interpolation size. Validation of measurement value showed a high R2. Then the CHM which can accurately describe the conTINuous surface characteristics of the wheat canopy could be gained from the difference operation between the digital surface model (DSM) and the reference value of ground elevation.(3) Measurement of growth parameters:Degree of coverage was calculated directly from grid Canopy height model (CHM) with raster calculator; then we applied the watershed segmentation algorithm to divide the CHM into non-overlapping areas, and then detected peak location of each area to get local maxima. We could extract the wheat height and biomass density maps after this step. Accuracy assessment and reliability analysis based on field data showed an ideal result:measured wheat height was66cm and model-derived wheat height was68cm, and distribution of biomass density and wheat fields growing trend were consistent. It let us know that the method raised in this study possessed sufficient feasibility and relatively high utility value, which would effectively assist for the wheat growth parameter extraction and crop yield monitoring.