This study was conducted to focus on the pattern of the change of multi-temporal NDVI data of irrigated and non-irrigated crops at local scale and this page shows the recent result of this study using vegetation instrument data. Suphanburi province in Thailand was selected as the pilot study area  because the dominant of agriculture activities in this area with and without irrigation system. Rice the major crop in this area has been selected for the analysis. The NDVI time series analysis of irrigated and non-irrigated rice cultivation was carried out in Suphanburi province. Once the model or pattern of NDVI and other metrics for identifying irrigated and non-irrigated rice areas are extracted, this model will be tested in other and/or larger rice areas. The agriculture pattern of other crops will also be observed. Finally the reliable model, which relies on more general parameters, is expected to extract for irrigated area mapping in global scale.

Irrigated area

Rainfed area

Data used

The SPOT 4, currently in operation, is orbiting the earth at an altitude of 832 km, in a 26-day cycle. The SPOT 4 loads two kinds of sensors: Haute Resolution Visible Infra-Red (HRVIR) instrument and Vegetation instrument, allowing the daily observation of 2,250 km in swath width with a resolution of 1.15 km for almost all the area, excluding some parts around the equator. While the Vegetation instrument covers almost all of the earth everyday, therefore it’s possible to receive data on cloud free day. In this study 10-day composite of NDVI data observed from January 1999 to December 2001 were analyzed. Local Maximum Fitting (LMF) was executed for the correction of 10-day composite data in order to pursuit cloudless data.

In this study, the areas of paddy were defined based on Landsat TM (07 Dec 2001) with different textures/patterns (such as homogenous and heterogeneous areas). Table 1-4 shown false color composite and NDVI of Landsat TM (resolution 30x30m), which was used as the reference image for defining irrigated and non-irrigated areas of paddy. Then the defined areas were transferred to SPOT VI to calculate mean of NDVI. For more accuracy and better understanding, field survey picture was compared together with satellite image.The area observed here is located in Doembang Nangbuat district, northern Suphanburi. The representative picture above shown rather homogeneous or large continuous field under irrigated. The stage of rice in this picture is first or growing from wet to growing stage while in Landsat image shown growth stage.

Landsat TM image showing homogenous irrigated paddy

Training Sample of homogenous irrigated paddy

The NDVI time series show clear differences between irrigated and non-irrigated areas. In non-irrigated areas, the NDVI pattern is similar throughout the study area. The pattern agrees with actual condition on the ground, such as one high peak during cultivation followed by a low peak that indicates grass and other vegetation growth after harvesting. The NDVI fluctuation pattern in an irrigated area varies from place to place as a function of the cultivated land unit size (large contiguous or small disaggregated field plots) and water distribution from irrigation system. Homogeneously cultivated areas show a two large peaks per year. Fluctuation pattern in continuous rice cultivation shows three low peaks which reflect a mix of cultivated and bare land and multiple growth stages that characterize continuous cultivation. The next steps of this study are the analysis on more detail of the NDVI fluctuation pattern such as the number of peaks, distance from peak-to-peak or trough-to-trough per year, together with some other metrics to categorize and classify the pattern in order to separate irrigated and non-irrigated rice and allow classification of low cost images.

NDVI fluctuations of Irrigated and Non-irrigated paddy in Suphanburi, Year 1999-2001