Application of Remote Sensing to Agricultural Field Trials

摘要: Remote sensing techniques enable quantitative information about a field trial to be obtained instantaneously and non-destructively. The aim of this study was identify method that can reduce inaccuracies in analysis, how remote support and/or replace conventional measurements trials. In the literature there is certain consensus best bands from which characteristic spectral vegetation extracted are those visible (green red) infrared regions electromagnetic spectrum. This confirmed present by an analysis multispectral scanner data ('Daedalus scanner') trials with cereals. optimal were thereby selected for explaining grain yield mostly contained channels 5 (550- 600 nm), 7 (650-700 9 (800-890 nm). Multispectral aerial photography found most appropriate recording extensive short period. study, recordings carried out single-engine aircraft, using two Hasselblad cameras obtaining vertical photographs on black white 70-mm films. way, costs stayed within acceptable limits. scale chosen, given dimensions at experimental farm Wageningen Agricultural University, where research out, 1:8 000. Photographs taken approximately fortnightly keep step sampling. film/filter combinations high resolution matching 5, Daedalus scanner, resulted following passbands: green : 555-580 nm; red :665-700 :840-900 nm, densities objects film measured means automated Macbeth TD-504 densitometer. An aperture diameter 0.25 mm densitometer, order obtain spatial 1:8000, applicable plots 3 metres wide. converted into exposure values, corrected light falloff, then linear function applied convert them reflectance factors. time, relative aperture, transmittance optical system, irradiance, path radiance atmospheric attenuation incorporated. Reference targets known characteristics set up during missions recorded same camera setting under conditions as trials, ascertain parameters function. Information crop suggested reflectances region spectrum or would suitable estimating soil cover, whereas might leaf area index (LAI). Other plant characteristics, such dry matter weight yield, may estimated indirectly reflectances. Field cereals analysed showed treatment effects shown tended opposite LAI. Treatment similar LAI, even large LAI (6-8). manifest more stable time than Coefficients variation residuals resulting analyses variance systematically smaller all experiments: particularly small. general, critical levels testing indicates power larger Soil moisture content not constant growing season differences greatly influence reflectance. Since multitemporal required, correction had made background when ascertaining relationship between characteristics. no model stood being agricultural Thus, monograph simplified presented cover vegetation. First all, redefined as: projection shadows included, seen sensor pointing vertically downwards, total (in definition depends position sun). Then, based expression composite plants soil: various passbands combination its complement, coefficients, respectively. By model, it should, theoretically, possible correct passbands. practice, however, procedures derived yielded poor results because difference so attention focussed For calculated subtracting contribution Theoretically, combining green, passbands, enables calculated, without knowing main assumption ratio bare different independent content: valid many types. type approximated Subsequently used according inverse special case Mitscherlich contains have ascertained empirically. Model simulations SAIL (introduced Verhoef, 1984) potential simplified, semi-empirical, Analogous derivations generative canopy (cereals) yellowing leaves. estimation good study. presence could coefficients one field. Regression curves differed significantly crop, stage. been caused systematic discrepancies sampling subjectivity separating yellow To date, approach regression each incorporating few additional plots, both measured.