|Title:||衛星及地面遙測資料應用於水稻生長及產量監測初步研究||Other Titles:||Preliminary Study on Using Spaceborne and Ground-based Remotely Sensed Data for Rice Growth and Yield Monitoring||Authors:||吳啟南
|Keywords:||水稻產量;回歸分析;生長監測;Rice yield;Regression analysis;Growth monitoring||Issue Date:||Jun-2002||Publisher:||農業試驗所||Related Publication(s):||農業試驗所特刊第101號||Start page/Pages:||19-38||Source:||應用於水稻精準農業體系之知識與技術||Conference:||水稻精準農業體系之研究計畫階段性成果研討會||Abstract:||
本研究利用衛星影像及地面遙測資料監測水稻生長及預估產量。其中水稻生長監測以影像攝像日對照生長天數後，繪出全生育期水稻對應SPOT影像的NDVI指數分佈；另亦利用NDVI植被指數分析與比對正常與罹患稻熱病水稻的差異。水稻產量預估則利用四個時段SPOT影像的NDVI、RVI 及TVI指數，配合農試所89 年精準農業試驗田的坪割產量資料，以向後消去法逐步回歸(stepwise regression by backward e1inilnation)估算水稻產量，並建立預估水稻產量的回歸方程式。地面遙測以具有綠、紅及近紅外光等三個波段的Duncan MS3100/CIR 近地面紅外成像儀資料監測水稻生長，並建立水稻正常生長及不同氮肥力區的NDVI指數對應生長天數分佈曲線。衛星影像監測正常水稻生育，在綠色植生反應最敏感的近紅外光段光譜反射以幼穗形成期～孕穗期值最高，NDVI指數分佈亦相同；而逆境生長的稻熱病水稻區，在近紅外光段光譜反射及NDVI指數均減少。另以SPOT影像估測水稻產量；單一時期或多時段SPOT影像與一期稻產量資料回歸分析後，相關係數最高為066；相同的方法應用於二期稻的產量預估，相關係數有明顯提昇，最高分蘗期SPOT影像預估產量的相關係數為0.71，加上幼穗形成期或乳熟期影像的相關係數為0.84，再結合黃熟期前的影像，回歸相關係數可達0.90，其預估產量的標準差為0.39公噸／公頃。另近地面紅外成像儀監測以農試所試驗田90 年兩期稻作為攝取對象，攝像資料經增益調整後，正常生長水稻的NDVI指數極值在幼穗形成期～孕穗期問；另不同氮肥力區與正常生育水稻NDVI指數的比較發現，氮變異區水稻在孕穗期以前的NDVI指數均較低，抽穗期以後的NDVI指數反而較高，且反應在黃熟期的水稻生育有較晚成熟的現象。
SPOT and Duncan M53100/CIR imagery were adopted for the study of rice growth status and yield monitoring. The full-growth-period spectral responses of rice were characterized by NDVI distribution derived by using SPOT bands, and the NTYVI responses of normal and leaf blast infected rice plants were compared. The NDVI, RVI and TVI indices derived from SPOT imagery during two cropping seasons in 2000 were used to correlate with the yields of the parcels by means of stepwise regression technique. Meanwhile, data from ground-based Duncan M53100 imagery (G, R, NIR) were taken from the normal and varied nitrogen fertilized parcels to study their spectral response in the two cropping seasons of 2001. The results show that both MR and NDVI indices rearched the maximum in panicle formation to grain filling stage, while their values were lower in the leat blast infected rice compared to the normal plants. When single or multiple SPOT images were employed to correlate with the yield of the first crop, the maximum regression coefficient (MRC) was 0.66. The MRC was increased to 0.71, 0.84 and 0.90 when the images from the maximum tillering stage, panicle formation stages, and grain-filling stage were incorporated into the analysis of the second crop, respectively. For the latter case, the standard deviation of the yield was 0.39 ton/ha. On the other hand, the images acquired by Duncan were gain adjusted and the NDVIs caculated from the data of the two cropping seasons in 2001 were analyzed. It was found that the maximum of NDVI was also located at the panicle formation to grain filling stage, and the NDVIs of varied nitrogen fertilized paddies were lower before grain filling stage, but higher than the normal paddy afier heading stage. The preliminary results suggest that the ground-based imagery can be used to distinguish the growth variation by the different rates of nitrogen fertilization.
|Appears in Collections:||作物組|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.