Evaluation of the Relationships Between Wheat Yield and Biomass with Remote Sensing Vegetation Indices on the basis of Agricultural Climate Regions

Vegetation indices obtained from satellite images are widely used in the monitoring of agriculturalcrops and the prediction of yields. The changes in meteorological factors and especially the irregularrainfall regime have positive and negative effects on different ecological regions. The aim of thisstudy is to investigate the relationships between spectral vegetation indices, yield and biomass onthe basis of agricultural climatic regions (ACZ) and to determine effective vegetation indices ona temporal basis. In this context, 5 different regions were determined by using aridity index andgrowing degree days (GDD) parameters in establishing ACZ. In the field studies, the plant samples were collected from 108 plots of wheat cultivated in Ankara and Konya provinces and yield and biomass values were obtained.During the crop growing period, Normalized Vegetation Index (NDVI) and Enhanced VegetationIndex (EVI) of 15 different periods from MODIS satellite image were used. The relationship betweenvegetation indices and biomass and yield was examined by Pearson correlation coefficient. Thehighest correlation between vegetation indices and biomass was observed in May in 12th period(75%) and 13th period (72%) for NDVI, 12th period (71%) and 13th period (72%) for EVI. Thehighest correlation rate between vegetation indices and yield was determined as NDVI (72%) andEVI (70%) in the 12th period. During the vegetation period, the 12th and 13th periods correspondto the heading / anthesis period.Linear regression models were created according to the significance levels (p <0.05) of theparameters. The R2 values for the vegetation indexes determined for ACZ-based models rangedfrom 0.37 to 0.87 for biomass-NDVI and 0.56 for the whole area, 0.24 to 0.68 for the biomass-EVIand 0.52 for the whole area. These ratios vary between 0.28 and 0.72 between yield and NDVI,whereas 0.53 in whole area. The yield-EVI ranges from 0.32 to 0.54, while it is 0.48 for the wholearea. In ACZ based estimation models, a higher level of relationship was obtained than the modelmade for the whole area.