大豆残茬光谱响应特征研究

利用HR-768型便携式光谱仪,测定了不同大豆残茬覆盖度下的地面光谱,利用照相法获取对应的大豆残茬覆盖度。采用线性回归方法分析了单波段反射率、反射率一阶导数、归一化指数、比值指数与大豆残茬覆盖度的相关性,获取了不同覆盖度水平下大豆残茬的光谱响应特征,并结合MODIS、TM、HJ-1B星的波段响应函数建立了大豆残茬覆盖度最优估算模型。结果表明,在2050—2150nm和2250—2350nm两个波段范围内,大豆残茬与裸土的光谱差异最显著,可用于二者的区分;大豆残茬的光谱特征与玉米、小麦残茬的光谱特征相似,仅在920—967nm范围内存在特殊的吸收峰;以高光谱数据为数据源,941.6nm处的反射率、2151.8nm处反射率一阶导数、1461.3nm和2404.4nm反射率构建的归一化指数以及2247nm和608.6nm反射率构建的比值指数适宜用于作物残茬覆盖度估算,以宽波段数据为数据源,短波红外与红波段反射率构建的比值指数适宜用于大豆残茬覆盖度估算。

Spectral responses analysis of soybean residues

Ground refl ectance with different soybean residue covers (SRC) were measured with HR-768 portable spectroradiometer. Coincident with spectrum measurements, soybean residue cover were measured using photographic method. Linear regression analysis was used to evaluate the correlations between soybean residue cover and spectral refl ectance, as well as its f irst order derivative and derived indices (namely ratio index and normalized difference index). Characteristics of spectral responses of soybean residue were analyzed. With relative spectral response functions, bands refl ectance of MODIS, TM, HJ-1B CCD1 and IRS were simulated respectively to establish soybean residue cover estimation model. The results indicate that soybean residue can be distinguished from soil using its unique absorption feature in the spectral range of 2050 nm to 2150 nm and 2250 nm to 2350 nm. The spectral characteristics of soybean residue are similar to that of corn and wheat residue except for an absorption feature in the spectral range of 920 nm to 967 nm. Parameters that are suitable for soybean residue cover estimation were identif ied using hyperspectral data, which includes refl ectance at wavelength of 941.6 nm, f irst derivation of refl ectance at wavelength of 2151.8 nm, normalized difference index between refl ectance at wavelengths of 1461.3 nm and 2404.4 nm, and ratio index between refl ectance at wavelengths of 2247 nm and 608.6 nm. Taking broad band data as data source, ratio index between shortwave infrared refl ectance and red refl ectance performs the best in soybean residue cover estimation.