基于头季关键生育期的NDVI值诊断再生稻的生长状况

明确再生稻头季光谱数据与再生季生长的相关性,是一种预测和诊断再生稻生长状况的便捷方法。研究采取主动式的Greenseeker 光谱仪,测定了不同再生稻品种(组合)关键生育期的NDVI,结合再生稻的生理指标分析计算了头季抽穗前后的NDVI和再生季抽穗后的干物质、LAI和叶片氮含量(累积)的相关性。结果表明头季抽穗后的NDVI与再生季抽穗后的绿叶生物量、LAI、营养器官和绿叶的氮累积具有显著的指数正相关,且不受再生稻品种(组合)类型的影响。利用不同密度试验的数据对方程进行了较充分的验证和检测,表明模拟值与实测值之间符合度较高,估算精度为0.9765,估计的RMSE为1.464,平均相对误差为-0....

Full description

Saved in:
Bibliographic Details
Published in江西农业学报 Vol. 27; no. 2; pp. 7 - 11
Main Author 柳开楼 刘璇 胡惠文 段红霞 夏桂龙 李亚贞
Format Journal Article
LanguageChinese
Published 江西省红壤研究所/国家红壤改良工程技术研究中心/农业部江西耕地保育科学观测实验站,江西 南昌,330046%江西省余江县农业局,江西 余江,335200%江西省邓家埠水稻原种场,江西 余江,335200 2015
Subjects
Online AccessGet full text
ISSN1001-8581

Cover

More Information
Summary:明确再生稻头季光谱数据与再生季生长的相关性,是一种预测和诊断再生稻生长状况的便捷方法。研究采取主动式的Greenseeker 光谱仪,测定了不同再生稻品种(组合)关键生育期的NDVI,结合再生稻的生理指标分析计算了头季抽穗前后的NDVI和再生季抽穗后的干物质、LAI和叶片氮含量(累积)的相关性。结果表明头季抽穗后的NDVI与再生季抽穗后的绿叶生物量、LAI、营养器官和绿叶的氮累积具有显著的指数正相关,且不受再生稻品种(组合)类型的影响。利用不同密度试验的数据对方程进行了较充分的验证和检测,表明模拟值与实测值之间符合度较高,估算精度为0.9765,估计的RMSE为1.464,平均相对误差为-0.1191,RRMSE为0.1814。
Bibliography:36-1124/S
To make clear the relationship between the growth of regenerated rice and the canopy NDVI value of main rice is a convenient method for forecasting and diagnosing the growth status of regenerated rice .This study measured the NDVI values of dif-ferent regenerated rice varieties ( combinations ) at key growth stages by using active spectrometer Greenseeker .Combining with the physiological indexes of regenerated rice , we analyzed and calculated the correlations of the NDVI of main rice before and after head-ing with the dry matter , LAI, and leaf nitrogen content ( accumulation ) of regenerated rice after heading .The results showed that there were significant positive correlations between the NDVI of main rice after heading and the green leaf biomass , LAI, and nitro-gen accumulation in vegetative organ and green leaf of regenerated rice after heading , and these correlations were not affected by the type of regenerated rice varieties ( combinations ) .These equations were adequately verified and de
ISSN:1001-8581