Application Of Remotely Sensed Images On Alfalfa Grassland Discrimination And Yield Estimation

Alfalfa (Medicago sativa) is a kind of forage which is widely distributed in the world.It has good functions of soil amendment, soil and water conservation.In this study, research was conducted in Alukeerqin. Combined with the dynamic change characters of the growth and the production of alfalfa in a year, we conducted field investigations, hyperspectral measurements and acquired multisource remote sensing images. In order to provide technical support for the macro-management of alfalfa industry in our country, we took up researching extraction artificial alfalfa grassland information and yield estimation.The main results were as follows:1.Both summation of NDVI approaches and unsupervised classification method can be used to extract Artificial alfalfa grassland information.The extaction accuracy of summation of NDVI was high and it was easy to use and master. Although extaction accuracy of unsupervised classification method was 100%, classification process was complicated and it need experienced professionals to implement.2. Artificial alfalfa Grassland was mainly distributed in the south of the research region. The total area of Artificial alfalfa Grassland was 99.49km2 in 2013 (342 Irrigation circles),158.98km2 in 2014(543 Irrigation circles) and 238.85km2 in 2015 (689 Irrigation circles). Its area has increased from the year of 2013 to 2015.3.The result of alfalfa yield estimation using HJ1A/B remote sensing datas was that coefficient determination of exponential model and power function model between NDVI、DVI、GNDVI、TVI、ARVI、MSAVI、PVI and fresh yield were more than 0.55.All of them could be used to estimate fresh yeild of alfalfa.4. In term of hyperspectral remote sensing estimation, both coefficient determination of power function model between (SDr-SDb)/(SDr+SDb) and fresh yield and coefficient determination of exponential model between (SDr-SDb)/(SDr+SDb) and hay yield were more than 0.7. The coefficient determination of exponential model between SDr/SDb and fresh yield and dry yield were more than 0.69.The coefficient determination of exponential model between Rg and fresh yield and hay yield were more than 0.63.Three Hyperspectral characteristic variables could be used to estimate yeild of alfalfa.5. The annual total hay yield of alukeerqin was 1.62×105t and the annual average hay yield was 6753.77 kg/hm2. The total hay yield of first cutting was 4.78×104t and average hay yield was 1994.55 kg/hm2. The total hay yield of second cutting was 5.08×104t and average hay yield was 2116.88 kg/hm2. The total hay yield of third cutting was 6.34x 104t and average hay yield was 2642.34 kg/hm2.