Development of regional wheat VI-LAI models using Resourcesat-1 AWiFS data

• Published in: Journal of Earth System Science 2011-12, Vol.120 (6), p.1113-1125
• Main Authors: CHAURASIA, SASMITA, NIGAM, R, BHATTACHARYA, B K, SRIDHAR, V N, MALLICK, K, VYAS, S P, PATEL, N K, MUKHERJEE, J, SHEKHAR, CHANDER, KUMAR, DHIRAJ, SINGH, K R P, BAIRAGI, G D, PUROHIT, N L, PARIHAR, J S
• Format: Article
• Language: English
• Subjects: Agroclimatic conditions; Biomass; Clear sky; Climatic conditions; Crop science; Crops; Earth and Environmental Science; Earth Sciences; Earth, ocean, space; Errors; Exact sciences and technology; Normalized difference vegetative index; Regional development; Remote sensing; Saturation; Simulation; Simulation models; Space Exploration and Astronautics; Space Sciences (including Extraterrestrial Physics; Spatial discrimination; Spatial resolution; Temporal resolution; Vegetation; Vegetation index; Wheat
• ISSN: 0253-4126; 2347-4327; 0973-774X
• DOI: 10.1007/s12040-011-0126-x

The time of forcing of spatial LAI to crop models at single or multiple stages is important to simulate crop biomass and yield in varying agro-climatic conditions and scales. The high temporal resolution (5-day) by Advanced Wide Field Sensor (AWiFS) on-board Resourcesat-1 Satellite IRS-P6 with 56 m spatial resolution and large swath (740 km) has substantially increased the availability of regional clear sky optical remote sensing data. The present study aimed at developing empirical vegetation index VI-LAI models for wheat using AWiFS optical data in four bands and in-situ measurements sampled over five different agro-climatic regions (ACRs) during 2005–2006 followed by validation during 2006–2007. While nonlinear relations exist for all the three normalized indices such as normalized difference vegetation index (NDVI), normalized difference water index (NDWI) and Green NDVI, linear relation was the best fit for ratio vegetation index (RVI). Both NDVI and RVI models generally showed better correlation ranges (0.65–0.84 for NDVI and 0.37–0.76 for RVI) than other indices. The common NDVI-LAI model was found to produce lower root mean square errors (RMSE) between 0.5 and 1.1 from pooled model than those between 0.5 and 1.32 from regional models. The rate of substantial increase in errors from NDVI-LAI model (RMSE of modeled LAI: 0.85 to 1.28) as compared to RVI-LAI model (RMSE of modeled LAI: 1.12 to 1.17) at LAI greater than 3, than below 3 revealed the early saturation of NDVI than RVI. It is therefore recommended that LAI estimates can be used to force crop simulation model upto early vegetative stage based on NDVI and maximum vegetative to reproductive stages based on RVI.