ارزيابي شاخص هاي طيفي استخراج شده ازتصاوير ALOS-AVNIR2 به منظورتخمين ميزان بايومس محصول برنج

Evaluation of ALOS-AVNIR2 Spectral Indices for Prediction of Rice Biomass

داده‌هاي ماهواره‌اي به منظور گسترش و مديريت منابع كشاورزي همواره قادر هستند درتامين اطلاعات لازم در جنبه‌هاي مختلف جوامع گياهي من جمله بايومس نقش مهمي را ايفا نمايند. در تحقيق حاضر به بررسي توانايي شاخص‌هاي طيفي در تخمين بايومس محصول برنج در شهرستان آمل به عنوان يكي از قطب هاي مهم توليد برنج در كشور پرداخته شده است. محصول برنج به دليل اهميتي كه تامين نيازهاي غذايي و كالري بخش عظيمي از جامعه برعهده دارد براي اين تحقيق انتخاب شد. عمليات ميداني اندازه‌گيري بايومس در زماني كه برنج در منطقه مطالعاتي در حداكثر رشد رويشي خود قرار داشت( خرداد 1389)، انجام شد. دو تصوير ماهواره ALOS-AVNIR-2 كه همزمان با اندازه‌گيري‌هاي ميداني اخذ گرديده بود جهت استخراج و تعيين شاخص هاي گياهي مورد استفاده قرار گرفتند. سپس همبستگي بين داده‌هاي زميني و شاخص‌هاي گياهي حاصل از تركيب باندهاي مختلف، ارزيابي و پس‌از آن، شاخص‌هاي گياهي مناسب تشخيص داده شدند. در نهايت محاسبات و بررسي‌هاي آماري براي معرفي مدل مناسب ارايه گرديد. پس از آزمايش مدل‌ها نتايج به دست آمده نشان داد كه شاخص DVI با ضريب تعيين 72 درصد نسبت به ديگر شاخص هاي استفاده شده از دقت بالاتري در تخمين بايومس برخوردار است. بر اساس اطلاعات به دست آمده از اين تحقيق مي‌توان بيان نمود كه با استفاده از تصاوير ماهواره ALOS امكان تخمين بايومس با دقت قابل قبولي وجود دارد

Extended Abstract Introduction Quantification and monitoring of biophysical and biochemical parameters of vegetation play a vital role in the terrestrial ecosystems. Biophysical parameters such as biomass and leaf area index (LAI) are the indicators of the productivity and function of crop. Rice is one of the major staples and widely planted crop in northern part of Iran. Rapid population growth, demands more rice production, while agricultural lands are gradually reducing with urban expansion. Monitoring rice production by remote sensing has been widely carried out in many countries where rice is the main food. The advantages of remotely sensed data, such as repetition of data collection, a synoptic view, a digital format that allows fast processing of large quantities of data, and the high correlation between spectral bands and vegetation parameters, make it the primary source for large area biomass estimation, especially in areas of difficult access. Therefore, remote sensing-based biomass estimation has increasingly attracted scientific interest. The possibility of estimation biomass by satellite remote sensing has been investigated in several studies at various spatial scales and environments. However this study aims to estimate the amount of fresh biomass of rice using Vegetation indices from ALOS satellite image which is rather new satellite. Research Methodology During June 29 - July 14, 2010 an extensive field campaign was conducted in Amol, Northern part of Iran. Most of the agricultural activities in this area are characterized by rice cropping. Most of the rice crops were in grain-filling stage. Sixty three plots of 30m by 30m were chosen by adopting stratified random sampling, in each plot three to five subplot of 1m by 1m were randomly selected (depending on the homogeneity of sample plot). For measuring the biomass, the rice shrubs were collected in each sub plots. The fresh biomass was determined by dividing the weight of the clipped shrubs by the surface area of the subplots. The results of the calculation were expressed in g m2.This method for fresh biomass calculation is reported in. The plot biomass was calculated by averaging subplots biomass. Two images of AVNIR-2 were processed. Since JAXA does not deliver AVNIR-2 images atmospherically corrected, the images were corrected by FLAASH module installed with ENVI 4.7 software. Geometric corrections were done with second-degree polynomials, using both the 1:25000 topographic map and handheld GPS-derived control points. Most studies have focused on developing empirical relationship between ground-measured vegetation parameters and spectral indices commonly known as vegetation indices. these Vegetation indices can combine reflectance measurements from different portion of the electromagnetic spectrum to provide information about crop biomass. Some vegetation indices have been developed to remove variability caused by canopy geometry, soil background, sun view angels, and atmospheric conditions when measuring biophysical properties. In this study for fresh rice biomass estimation, in addition to ALOS spectral bands, twenty vegetation indices have been proposed and used. Results The calculated vegetation indices and the rice biomass measurements were used to carry out the correlation analysis. The purpose of these analyses was to examine how close the relationships between vegetation indices and the measured biomass are. Table 1 shows these results. R2 values indicate a considerable variation in the performance of different vegetation indices for estimating fresh biomass. A high correlation exists between some vegetation indices and fresh biomass. As can be observed from the table, the best vegetation indices for biomass estimation were DVI, and WDVI, with R2 values of approximately 0.72. As it is demonstrated by Roujean and Breon (1995) in comparison to NDVI the DVI index is less affected by background soil. The DVI uses the deduction of NIR and RED in its calculation. As an increase in biomass will cause the increase of NIR and decrease in RED reflectance, the derived result is justified. Conclusion In this study, several vegetation indices were compared for their abilities to estimate rice fresh biomass. Empirical equations were obtained to estimate fresh biomass using different vegetation indices. Results show that using vegetation indices such as DVI and WDVI the rice fresh biomass can be estimated with an acceptable accuracy.