Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → Turquie → Crop Yield Estimation Model Design For Cotton And Maize Crops Using Satellite Imagery, Meteorological Data And Camera Photographs : Sanliurfa Case Study

İstanbul Teknik Üniversitesi (2015)

Crop Yield Estimation Model Design For Cotton And Maize Crops Using Satellite Imagery, Meteorological Data And Camera Photographs : Sanliurfa Case Study

ALGANCI Uğur

Titre : Crop Yield Estimation Model Design For Cotton And Maize Crops Using Satellite Imagery, Meteorological Data And Camera Photographs : Sanliurfa Case Study

Uydu Görüntüleri, Meteorolojik Veriler Ve Kamera Fotoğrafları İle Pamuk Ve Mısır Bitkileri İçin Rekolte Tahmin Modeli Tasarımı : Şanlıurfa Örneği

Auteur : ALGANCI Uğur

Université de soutenance : İstanbul Teknik Üniversitesi

Grade : Doctor of Philosophy (PhD) 2015

Résumé partiel
Accurate identification of crop types and estimates of their area at agricultural parcel level from satellite sensor data can provide important information to support agricultural policies, verify claims by farmers who apply for public subsidies and assist in the practice of precision agriculture. Since agricultural practices contribute to greenhouse gas emissions, identification of crop types and their spatial distribution assumes importance from a climate perspective as well. Several countries in the European Union have developed their Land Parcel Identification Systems (LPIS) based on satellite sensor or aerial images to support area-based subsidies and agricultural policies. In addition to identification of crop types and their spatial distribution, timely and accurate estimation of crop yield at local to regional scales is of paramount importance for societal, economic, agricultural, and policy considerations. For example, regional-scale yield estimation enables estimation of potential agricultural production capacity so that food pricing and stock management can be performed for the forthcoming year. Yield estimates also add valuable data to modeling of agriculture – ecosystem relationships in terms of carbon cycling and climate change. Using allometric relationships, crop yields can be converted to terrestrial net primary productivity (NPP) to be used for carbon budget determination. Determination of the crop cultivations and their yield information is an important task from this perspective and should be considered as a spatial problem. Gathering this information in regional or country scale and in parcel basis seems nearly impoossible with classical terrestrial meyhods. At this point, remote sensing systems with their synoptic viewing capability and variety of temporal and spatial resolution are important data source to derive agricultural information. Remote sensing methods are superior to conventional methods since fast and economic data acquisition and fast processing of data using computer based analyses are possible in remote sensing. Characteristics of agricultural field, difference of spectral reflectance of different crop types and differences in feature characteristics such as shape and texture are important parameters that should be considered while working agricultural areas with remote sensing. Remotely sensed data provide identifiable signatures for crop type, crop density, crop geometry etc, in order to perform agricultural survey and analysis. This study aimed to perform a two step analysis of cultivated areas of Harran Plain at Sanliurfa Province in Southeastern region of Turkey in order to develop an adaptive yield model for cotton and maize crops at parcel level and regional scale. An agro-meteorological model with multi sensor and multi temporal satellite images, digital photos and meteorological measurements were combined with cultivated area detection analysis in order to achieve the model construction. First part of the study investigates the accuracy of pixel- and object-based classification techniques across varying spatial resolutions to identify crop types at parcel level and estimate the area of six test sites to find the optimum data source for the identification of crop parcels. Multi sensor data with spatial resolutions of 2.5m, 5m and 10m from SPOT-5 and 30 m from Landsat 5 TM were used. Maximum Likelihood (ML), Spectral Angle Mapper (SAM) and Support Vector Machines (SVM) were used as pixel-based methods in addition to object-based image classification (OBC). Post-classification methods were applied to the output of pixel-based classification to minimize the noise effects and heterogeneity within the agricultural parcels. OBC results provided comparatively the best performance for both parcel identification and area estimation at 10m and finer spatial resolution levels. SVM followed OBC at 2.5m and 5m resolutions but accuracies decreased dramatically with coarser resolutions. ML and SAM results were worse up to 30m resolution for both crop type identification and area estimation. In general, parcel identification efficiency was strongly correlated with spatial resolution while the classification algorithm was a more effective factor than spatial resolution for area estimation accuracy. Results also provided an opportunity to discuss the effects of image resolution and the classification algorithm independent factors such as parcel size, spatial distribution of crop types and crop patterns. According to the regional scale analysis performed in Harran, Sanliurfa and Hilvan districts

Présentation

Version intégrale (2,75 Mb)

Page publiée le 17 décembre 2020