基于实测光谱的潮滩土壤含水量遥感反演模型研究

Remote Sensing Retrieval Model on Soil Moisture Content of Tidal Flat based on Measured Spectra

以王港潮滩为研究区,采集土壤样品,测量土样的光谱,分析不同含水量土壤的光谱特征;利用统计相关法,建立TM遥感影像各波段对应波长处的光谱反射率与土壤含水量之间的关系模型。研究结果表明,在干燥条件下,土壤颗粒越细小,土壤表面越光滑平整,土壤光谱反射率越高,随着土壤含水量的增加,土壤反射率呈下降趋势;波长越长,土壤光谱反射率与含水量的相关性越高,其中2220nm处的土壤光谱反射率与含水量具有最高的相关性;通过不同的波段组合,建立新的土壤含水量反演因子,其中由490nm与2220nm处光谱反射率的差值建立的因子与土壤含水量具有较高的相关性;由单波段与波段差值因子建立的土壤含水量反演模型的预测误差分别为4.590和5.147,反演精度较高。因此所建模型可以用于获取潮滩土壤含水量的空间分布,能够为研究潮滩土壤含水量与潮滩地形地貌之间的关系提供数据支持。

Taking one piece of tidal flat as study area, which is located at the south of Dafeng port in Jiangsu province, soil samples were collected from the surface soil of 5 cm depth along 16 transects in 2 tidal mudflats at low tides in June of 2008. These sampling points were distributed between the bare surface near shore and the low tidal line offshore. In total, 162 soil samples were collected. The positions of these samples were determined using a differential GPS unit. Twenty five samples were selected from the overall samples as there is little difference in the spectra for the same type of soil. This selection was based on the consideration of soil type and its spatial distribution. This sub-group samples can representative the major characteristics of the whole samples both geographically and in soil substances. Spectral measurements were taken in an outdoor setting from the nadir at 10 cm high using a full-range （350-2 500 nm） spectrometer with a 1° view angle. Samples were placed under natural light. The measured reflectance was calibrated using a white spectral on panel. The 25 soil samples were spread out evenly in 9 cm round culture dishes to a depth of 0.5 cm. These dishes had been previously weighed and painted pitch black to avoid contamination. Distilled water was added to these culture dishes until soil moisture content reached full saturation. Under natural conditions moisture is gradually lost from the soil, so reflectance was measured at a short time interval while samples were weighted simultaneously to an accuracy of 0.001 g using an analytical balance. Instantaneous moisture content was calculated as a ratio of the lost （evaporated） water weight to oven dried sample mass minus the weight of the culture dish. The results showed that under dry condition, the much smaller particle size of intertidal soil is, the much smoother of soil surface becomes. In this case the reflectance of soil was relatively higher. However, when water come out in the soil, above phenomenal will be disappeared as electromagnetic wave is absorbed by water severely and has much further influence than that only from particle size. With the increase of water content, the unit area of soil surface would hold more water which leads to stronger absorption to electromagnetic wave, correspondingly, the reflectance of soil show a decreasing trend. It is easily found at VNIR （visible and near-infrared radiometer） wavebands and absorption, especially at those positions with longer wavelength. Compared with other wavebands of TM sensor, TM7 has the longest wavelength, which is correspond to 2 220 nm approximately, so it is more sensitive to the variation of soil moisture, meanwhile statistical analysis provided that the reflectance of TM7 enjoys high correlation with moisture content and the prediction model constructed via this sensitive band possesses potential to retrieve moisture content. One reflectance-moisture model was constructed according to the difference between the 490 nm and 2 220 nm spectral reflectance also had good predictive ability for moisture detection and it could be also used to eliminate the influence of background.