| Drought is the most important limiting factor for crop production and it becoms more and more severe problem in many regions of the world, especially in arid and semi-arid regions. The early detection of water stress with non-destructive methods is crucial because it can help to identify water stress status at larger temporal and spatial scales before any damage is clearly visible. With the development of precision agriculture, hyprespectral remote sensing technologies for crop management have the potential to provide more information for making decisions on a canopy scale in real time. The spring wheat is a more important crop in the study area. The field experiments of spring wheat have been conducted for four years. We designed four treatments according to soil water gradient (i.e., different water stress). First, the hyperspectral reflectance and the physiological parameters of spring wheat were measured in different water stress. Then, we have analyzed the relationship between hyperspectral reflectance and the physiological parameters in different water stress. Finally, models for monitoring the physiological parameter in typical growth stages have been built. We hope that the research results could be used to dynamically monitor the change of spring wheat growth under drought stress, and lay the key technological and theoretical foundation for strengthening agriculture management and improving the crop water use effections. From the study, we obtained the main results as follows:(1) The physiological parameters of spring wheat such as the canopy water content, leaf water potential, soil water potential, soil relative moisture, chlorophyll content, LAI and plant height were measured and analyzed in different water stress level. The hyperstral reflectance change were obtained in whole spectral region in 350-2500nm corresponding to different water stress level. We found the response of hyperstral reflectance to water stress level was very sensitive. But the response of different spectral range to water stress is different, there is negative relationship between the visible and shortwave-infrared regions and water stress level, positive relationship between near infrared band and water stress level, especially, there is very close relationship between short-infrared region and water stress level, which is the optimal selection for build water index for monitoring the drought.(2) The relationship between canopy water content and soil water content in different depth is different during the growth stage of spring wheat. In early growth stage, canopy water content is obvious relative to soil water content in the depth of 10cm?20cm. In the middle stage, it is close relative to soil water content in the depth of 30cm?50cm. canopy water content and leaf water potential are very sensitive to water stress level, which are important indices to monitor water stress.(3) Based on the relationship between the canopy hyperspectral reflectance and soil water parameters, two sensitive spectral bands were identified, that is 780nm and 1750nm. We used the two bands to build two new indices, named semi-arid water index-1 (R780/R1750) and semi-arid water index-2 ((R780-R1750)/(R780+R1750)). They were very stable for estimation water stress levels and canopy water content.(4) We analyzed the correlation of different spectral index to leaf chlorophyll content (LCC) and canopy chlorophyll content (CCC) during all growth stage and different growth stage of spring wheat. The results show that SACI-2 (the new index we built in this study named semi-arid chlorophyll index-2) and MTCI (MERIS total chlorophyll index) are good indices to monitor chlorophyll content at canopy and leaf scales.(5) Regression model of leaf area index (LAI) and spectral indices (Wl, SACI) was built in different water stress. By testing, we found the model performed well to estimate LAI in growth stages.(6) We tried to estimate the height of spring wheat in different water stress level by many spectral indices. WI (R900/R970) and GLI (green leaf index) was found to be good ones. The regression relationships between plant height and WI and GLI were built in different growth stage of spring wheat.(7) We developed the spectral indices to estimate the yield of spring wheat by the combination of many kinds spectral bands from only-one-time and multiple-time models. By verifying, the spectral indices from the multiple-time model could predict the yield more accuracy.In the semi-arid Loess plateau, northwest China, rainfall is scare, and soil erosion and water loss are severity, both decreaseing the soil moisture that can be effectively used to improve productivity. Therefore, the area was constantly hit by drought. From the study, we can draw the conclusion that canopy reflectance can be used to develop a certain index for monitoring agriculture drought in future. We hope that the results can provide some application for agricultural activities in the entire Chinese Loess Plateau by using remote sensing data. |
PDF Full Text Request