Hyperspectral Remote Sensing-based Drought Monitoring Models Of Summer Maize

Crop drought is the main factor affecting crop growth and yield formation.Real-time identification and monitoring of the occurrence and development of crop drought is the key to scientifically prevent and mitigate the effects of crop drought.In this study,the characteristics of three common drought sensitive indicators(the land surface-air temperature difference,the canopy water content and canopy fraction of absorbed photosynthetically active radiation(fAPAR))and their remote sensing-based monitoring models for summer maize were studied in terms of the hyperspectral data and their responsble biological and environmental data fromsummer maize field manipulation experiments with different irrigation water amounts at different development stages during the years of 2014-2015.The results may provide scientific support for the agricultural drought remote sensing application.The main conclusions are as follows:(1)The land surface-air temperature difference fluctuates with the development period of summer maize.In the same observation day,the land surface-air temperature difference was larger in the treatment with less irrigation and smaller in the treatment with more irrigation.With the advancement of the growth process of summer maize,the land surface-air temperature difference of summer maize farmland was significantly affected by the change of soil moisture.The soil moisture and the land surface-air temperature difference showed an anticorrelation.The more serious the soil moisture deficit was,the higher the land surface-air temperature difference was.During the whole water treatment,the normalized difference vegetation index was main impact factor for the land surface-air temperature difference and two of them show a significant linear relationship.But the land surface-air temperature difference also was influenced by other factors at different stages: after 3 leaf stage,the land surface-air temperature difference was affected by canopy fraction of absorbed photosynthetically active radiation and they had significant linear relationship;from 3 leaf stage to jointing stage,it was influenced by soil relative humidity and air relative humidity,and they had significant linear relationship.The remote sensing inversion model of the land surface-air temperature difference was established based on soil,biological and meteorological factors.The model for the whole growth period of summer maize can explain 63% of the variation of the land surface-air temperature difference in 2015.However,the models for vegetative and reproductive stages can explain 79% of the variation of the land surface-air temperature difference in 2015.(2)The canopy water content(EWTc)first increased and then decreased with the growth period.None of the six spectral indexes(WI?MSI?GVMI?The area of the red-edged reflectivity curve?WI/NDVI?WI/CIgreen)of EWTc model in 3 leaf stage of summer maize have passed the significance test,and all the indexes of EWTc model after 3 leaf stage have passed the significance test at 0.001 level,and the model accuracy ranged as: tasseling stage > knotting stage > filling stage> milk-ripening stage > seven-leaf stage.Fuel moisture content(FMC)at 7 leaf and jointing stage given by six indicators passed the significance test at the 0.001 level.FMC at the three-leaf stage given by WI/NDVI index passed the significance test.FMC at the summer maize after jointing stage could not be inverted by the six spectral indicators.The results showed that the precision of leaf and canopy water content of summer maize with the same spectral index varied greatly in different growth stages.Both canopy and leaf water content indexes could reflect the growth status of summer maize.The precision of spectral index inversion of summer maize leaf and canopy water content index was closely related to the growth period of summer maize,and then the inversion model of summer maize water content in different growth periods was proposed.(3)Under mild drought conditions,fAPAR value increased with the advancement of growth period,and the range of fAPAR value changed greatly.The more severe the drought was,the worse the growth condition of summer maize was,and the smaller the fAPAR value was.With the increase of drought degree,the reflectance of visible light region and short-wave infrared region increased,the reflectance of near-infrared region showed a decreasing trend,and the response of water absorption band in near-infrared region to water stress was weak.The hyperspectral reflectivity and its transformation data could effectively estimate canopy fAPAR,which was significantly negatively correlated with the reflectivity of visible light region and short-wave infrared region,and positively correlated with the reflectivity of near-infrared band.fAPAR had the best correlation with the reflectivity of the visible light region,followed by the short-wave infrared region,and the near-infrared region was relatively poor.The first derivative spectral reflectance was negatively correlated with fAPAR at the yellow edge(550-640 nm)and had the strongest correlation,and positively correlated with fAPAR at the red edge(680-760 nm),and positively correlated with fAPAR at the short-wave infrared band(1500-1641nm)and was relatively stable.The correlation between reflectivity and fAPAR was stronger than that between reflectivity and fAPAR.fAPAR estimation models based on EVI,RDVI,SAVI and MSAVI was the best,followed by fAPAR estimation model based on first-order derivative spectral reflectance,and fAPAR estimation model based on reflectance was the worst.