Teilprojekt (TP) 13

Project partner: Dr. Heike Bach, VISTA Geowissenschaftliche Fernerkundung GmbH

TP-Title: Canopy status assesment using remote sensing for model based yield estimation

TP short title: Model based analyses with remote sensing

TP-aim: Application of hyperspectral remote sensing data, respectively their coupling with model simulations for the description of canopy status and yield estimation on heterogeneous surfaces. The results shall serve as basis for the decision making process in precision farming.

Status of the TP combination: Thematical cooperation with TP 4, 9, 10, 12, 14, 15, 22 and 23. Provision of canopy information of the test fields for the project partners.

Short description

The aim of subproject 13 will be the model based analysis of remote sensing data of an imaging spectrometer to retrieve canopy status on the pilot farms of the cooperative project. The imaging spectrometer AVIS (Airborne Visible Infrared Imaging Spectrometer) will acquire multitemporal datasets with a high spatial resolution (2-4m) to monitor the reflectance of the canopies (approx. 4 acquisitions per vegetation period). Furthermore the ESA will provide remote sensing data of the hyperspectral directional imaging spectrometer CHRIS.

Canopy parameters will be derived on the basis of spectral measurements alone. In addition, the radiation transfer model SLC, an enhancement of GeoSAIL (Verhoef & Bach, 2003), will be applied for simulation of the canopy reflectance. This model aided analysis allows for quantitative analyses and regional transferability of the results. Comparison between the retrieved canopy parameters and other methods of gathering the surface information (e.g. Yara- and Pendel sensor and the results of on-field optical sensors) will be drawn. Accuracy assessment on the basis of the field measurements on the pilot farms will be conducted for validation purposes.

In a second step of model supported analyses, an assessment of the information acquisition through coupling of remote sensing information with meteorological data, soil information and a plant growth model will be developed. Through gradual adaptation of the canopy parameters (e.g. leaf area and plant density) the modelled reflectance values for each pixel are fitted to the measured values. Thus a calibrated spatial distribution of plant parameters as an improved input for the model is determined.

The results will be used for yield estimation and as the basis for the decision making process in precision farming. The first two years of investigation are reserved for method development. A test application and a test of the transferability to larger areas will follow in year 3.

Status Quo

Three aerial flights will be conducted with the imaging spectrometer AVIS in 2005. Acquisition planning is in progress. The acquisitions will be adapted to the 2005 field selection and the desired spatial resolution in each of the two investigation areas.

Wimex has already been established as a CHRIS test site of ESA last year, Groß Twülpstedt is added as a test-site in 2005. Some CHRIS scenes of 2004 are existing for Wimex. A clear cloud scene of Sep 9 2004 was corrected for radiometric and atmospheric errors. As result spectral reflectance values were derived, which describe the absorption and reflectance of the land surface. As CHRIS is a multiangular imaging spectrometer, scenes are available at different observation angles. A geometrical correction was performed on some of the angular scenes. The following illustration shows the Sep 9th nadir scene overlaid with the test fields of 2005. The analyses of the data will take place in the coming investigation period.

Also the development of the canopy radiation model SLC, which provides a method for extraction of canopy parameters from hyperspectral data, will continue.

Substantial questions:

· Is the application of hyperspectral remote sensing data using AVIS suitable for decision making in precision farming?

· How accurate are model simulations based on remote sensing for the description of current canopy status and yield estimation on heterogeneous surfaces?

· Are the results of Yara- and Pendelsensor and of on-field optical sensors comparable to the results attained on the basis of hyperspectral data?

· Which possibilities offers the model aided analysis of remote sensing data in regard to plant protection?

· Are the retrieved yield data of the subarea transferable to a larger area, so that a regionally transferable modelling method is obtained?

Date of issue: 08.04.2005