Comparison of spectro-biophysical and yield parameters of cotton in irrigated lowlands of Amudaria River

Aybek Arifjanov,Shavkat Kenjabaev,Umida Voxidova,Christopher Conrad,Dilbar Abduraimova,Odilbek Eshchanov

doi:10.35595/2414-9179-2020-3-26-294-308

Abstract

This study aims defining the best predictors of biophysical parameters and yield with vegetation indices derived from Landsat 8 OLI surface reflectance data. The study was conducted in 2015 at five crop fields in Kulavat canal irrigation system in Khorezm province, Uzbekistan. The Environment for Visualizing Images (ENVI) ver. 4.5 and R programming software ver. 1.0.143 were used to process, calculate seven vegetation indices (VIs) and predict biophysical parameters and yield of cotton. The trend analysis show that in-situ measured biophysical parameters for the whole growth stage of cotton follows the 3rd order polinomial curve (R2 = 0.96-0.99). The NDVI, SAVI, TVI and RVI had linear interrelationship between each other with strong positive correlation of R2>0.9 (highly significant with p-value=0). The VIs showed a logarithmic function relationship with crop height (crht), power function relationship with green biomass (gbm) and leaf area index (LAI), and linear function relationship with the fraction of photosyntetically active radiation below the plant canopy (FPAR) during the entire growing period of cotton. Among seven VIs tested in this study, the NDVI/SAVI and GCI explained 88 and 91 % of variation in crht, respectively. These three indices also well explained gbm variation (R2=0.86). The TVI was slightly better explained FPAR than NDVI and SAVI (all R2>0.87). The NDVI, SAVI and TCG explained 82 % of variation in LAI. Among all VIs, GCI, NDGI and RVI were found to be the best predictor of cotton yield during August, explaining 76-79 % variability (p<0.001). Based on spectro-biophysical analysis, VIs derived from RS data on July and August (anthesis and peak growth stages of cotton) is more reliable to use for modeling cotton yield (seed and lint yields together). Therefore, field data collection is recommended to perform during these months taking into account in-field crop condition and remotely sensed data acquisition date. In addition, September 5-20 is the second important period (i.e., cotton pick-up) to conduct yield data collection for establishment of relationships between cotton yields with VIs (July-August).

Highlights

Population growth and climate variability necessitate using available land and water resources effectively to maintain sustainable agricultural production in the Central Asian countries, especially in lowlands of the Amudarya and Syrdarya Rivers
The height of cotton for the whole growth stage follows the 3rd order of polinomial curve (R2 = 0.96)
Yield related with vegetation indices (VIs) Among all VIs, green chlorophyll index (GCI), normalized difference greenness index (NDGI) and ratio vegetation index (RVI) were found to be the best predictor of cotton yield during August, explaining 76−79 % variability (p

Summary

Introduction

Population growth and climate variability necessitate using available land and water resources effectively to maintain sustainable agricultural production in the Central Asian countries, especially in lowlands of the Amudarya and Syrdarya Rivers. During the last 40 years, the expansion of cropland in Uzbekistan has slackened, but land is used much more intensively To understand land use dynamics and to be able to predict possible future developments, constant monitoring is needed. In this context, in-situ phenological observation of crop development and growth is very important. Based on proper and in-time phenology observations, more stable crop yields and quality can be recommended for land users, which can facilitate future improvement of crop, water and land management.

Objectives

Results

Conclusion