Rice Area Estimation by Using Multi-temporal Classification Method and AVHRR Data

Monitoring of agricultural production is very important for the management of the world΄s food supply for social security. As the population increases, and the economy and industry develop, the quantity and quality of cultivated lands are decreasing rapidly. In this regard, accurate measurements of the crop production rates are of prime importance in agricultural statistics. Generally, in order to estimate crop productions, information such as the area and the corresponding yield is significant. Remote sensing technology can provide timely and accurate information about crop distribution, area planted and potential production for decision makers at many levels. In this paper, we focus on a crop area estimation method by using remote sensing satellite images. Over the recent decades, different methods and algorithms have been proposed for classifying and estimating the crop areas from the remotely sensed data. The aim of this research is to estimate rice cultivated area in Guilan province (located in the north of Iran) by using AVHRR (a sensor installed on NOAA satellite) multitemporal data. For this purpose, we used a multi-temporal classification method based upon a maximum value-based composition technique to produce multi-temporal and cloud-free data sets of NDVI. The innovation presented in this research is related to applying a variety range of algebraic operations on the images for discrimination of the crops that have different maximum greenness during the growing season. In spite of the fact that NOAA/AVHRR is a low resolution data, the results of the area estimation are promising particularly when regional data of major agricultural crops and their status are required. In this regard, the accuracy of rice area estimation is 91.96 percent in comparison with the reference map.Because of the acceptable results and low costs of the approach, further investigations and evaluations are recommended.