大田玉米作物系数无人机多光谱遥感估算方法

Estimating Method of Crop Coefficient of Maize Based on UAV Multispectral Remote Sensing

作物系数K_c快速获取是大田作物蒸散量(Evapotranspiration,ET)估算的关键,为研究无人机多光谱遥感估算玉米作物系数的可行性和适用性,以2017年内蒙古达拉特旗昭君镇实验站大田玉米、土壤、气象等数据为基础,采用经气象因子和作物覆盖度校正后的双作物系数法计算不同生长时期与不同水分胁迫玉米的作物系数,并使用自主研发的无人机多光谱系统航拍玉米的冠层多光谱(蓝、绿、红、红边、近红外,475~840 nm)影像,研究了不同生长时期(快速生长期、生长中期和生长后期)玉米的6种常用植被指数(Vegetation indices,VIs):归一化差值植被指数(NDVI)、土壤调节植被指数(SAVI)、增强型植被指数(EVI)、比值植被指数(SR)、绿度归一化植被指数(GNDVI)和抗大气指数(VARI),与作物系数K_c的关系模型及水分胁迫对其的影响。结果表明:玉米生长时期和水分胁迫是影响玉米VIs-K_c模型相关性的两个重要因素。不同生长时期玉米植被指数和K_c相关性不同:充分灌溉情况下,快速生长期玉米VIs-K_c模型的相关性(R2为0.731 2~0.940 1,p<0.05,n=25)与生长中期至生长后期VIs-K_c模型的相关性(R2为0.276 5~0.373 2,p<0.05,n=40)不同;水分胁迫情况下,快速生长期玉米VIs-K_c模型的相关性(R2为0.0002~0.0830,p<0.05,n=25)与生长中期至生长后期VIs-K_c模型的相关性(R2为0.366 2~0.848 7,p<0.05,n=40)不同。水分胁迫对VIs-K_c模型的相关性影响较大:快速生长期,充分灌溉玉米VIs-K_c模型的相关性(R2最大为0.940 1)比水分胁迫玉米VIs-K_c模型的相关性(R2最大为0.083 0)强;生长中期至生长后期,充分灌溉玉米VIsK_c模型的相关性(R2最大为0.373 2)比水分胁迫玉米VIs-K_c模型的相关性(R2最大为0.848 7)弱。部分植被指数和作物系数相关性较强;快速生长期充分灌溉玉米的VIs-K_c模型的相关性由大到小依次为:SR、EVI、VARI、GNDVI、SAVI、NDVI;生长中期至生长后期水分胁迫玉米的VIs-K_c模型的相关性由大到小依次为:SR、GNDVI、VARI、NDVI、SAVI、EVI;其中比值植被指数SR与作物系数K_c的相关性最好。结果表明采用无人机多光谱技术估算K_c具有一定的可行性。

Rapid acquisition of crop coefficient Kc is the key to estimation of field evapotranspiration （ET）, in order to study the feasibility and applicability of unmanned aerial vehicle （UAV） multispectral remote sensing in estimation of maize crop coefficient, based on the data of field maize in experimental station, soil and meteorology in Zhaojun Town, Dalate Qi, Inner Mongolia in 2017, by using meteorological factors and crop canopy cover to correct dual crop coefficient method at different growth stages and different water stresses. The multi spectral （blue, green, red, red edge, near IR, 475-840nm） images from UAV were used to calculate vegetation indices （normalized difference vegetation index （NDVI）, soil adjusted vegetation index （SAVI）, enhanced vegetation index （EVI）, simple ratio （SR） and green normalized difference vegetation index （GNDVI）, visible atmospherically resistant index （VARI）） of maize in different growth stages （rapid growth stage, mid-growth stage and late growth stage）. Thus the model relation of VIs and crop coefficient Kc could be established, and the effect of water stress on it was studied. Results demonstrated that maize growth period and water stress were two important factors influencing the VIs-Kc model. The correlation between VIs and Kc in different growth stages was different： under full irrigation condition, the correlation of VIs-Kc model in the rapid growth stage （R^2 was 0.7312-0.9401, p〈0.05, n=25） was different with the correlation of VIs-Kc model from mid to late growth stage （R^2 was 0.2765-0.3732,p〈0.05,n=40）;under water stress condition, the correlation of VIs-Kc model in the rapid growth stage （R^2 was 0.0002-0.0830, p〈0.05, n=25） was different with the correlation of VIs-Kc model from mid to late growth stage （R^2 was 0.3362-0.8487,p〈0.05,n=40）. Water stress had a significant effect on the correlation of VIs-Kc model： in the rapid growth stage, the correlation of VIs-Kc model for full irrigation maize （the maximum value of R^2 was 0.9401） was better than the correlation for water stress maize （the maximum value of R^2 was 0.0830）;from mid to late growth stage, the correlation of VIs-Kc model for full irrigation maize （the maximum value of R^2 was 0.3732） was worse than the correlation for water stress maize （the maximum value of R^2 was 0.8487）. The correlation of part of VIs and crop coefficient Kc was good;the descending order of correlation of VIs-Kc model for full irrigated maize in the rapid growth stage was SR, EVI, VARI, GNDVI and SAVI;the descending order of correlation of the VIs-Kc model for water stress maize from mid to late growth stage was SR, GNDVI, VARI, NDVI, SAVI and EVI;the correlation of SR and crop coefficient Kc was the best. Estimation of Kc based on UAV multispectral technology was feasible.