Beijing LICA United Technology Limited

Novel Instruments Provide New Opportunities

Hotline: 010-51292601
Technical Technical
News Technical

ASD | Applicability of PROSPECT model

Date: 2021-12-03
浏览次数: 18

PROSDM: Applicability of PROSPECT model coupled with spectral derivatives and similarity metrics to retrieve leaf biochemical traits from bidirectional reflectance

ASD |

Leaf biochemical traits provide valuable information on understanding plant photosynthetic function, dynamic growth, nutrient cycling, and primary production. Leaf chlorophyll content (Cab), carotenoid content (Cxc), water content (Cw), and dry matter content (Cm) are considered as four main important leaf biochemical traits, which are connected to plant healthy and growth status, such as photosynthesis, nitrogen, stresses, and senescence. Approaches that enable high-throughput measurements of these leaf biochemical traits are critical to characterize plant physiological status and critical functional processes. The PROSPECT model, one of the most popular leaf radiative transfer models, has long been used for retrieving leaf biochemical traits from leaf directional-hemispherical reflectance factor (DHRF) spectra, while remains underexplored for applications to the leaf bidirectional reflectance factor (BRF) spectra to retrieve leaf biochemical traits. Existence of leaf surface reflectance and anisotropic property could be the main issues that constrain the applicability of the PROSPECT inversion to assess leaf biochemical traits from leaf BRF spectra.

Based on this, in this article, a group of Chinese scientists present an inversion method by integrating the PROSPECT model with spectral derivatives and similarity metrics (SDM), called PROSDM, to remove the difference between leaf BRF and DHRF spectra and retrieve leaf biochemical traits from leaf BRF spectra. The specific objectives are: (1) to investigate how the difference between leaf BRF and DHRF spectra varies with the wavelength and influences the retrievals of Cab, Cxc, Cw, and Cm by the PROSPECT inversion, (2) to explore the PROSDM to eliminate the difference between BRF and DHRF spectra, and retrieve Cab, Cxc, Cw, and Cm from leaf BRF spectra with the comparisons with the PROSPECT and PROCOSINE inversions as well as PROCWT, and (3) to assess the effects of the PROSPECT versions, spectral subdomains, spectral noises, and the range of model parameters on the performance of the PROSDM.

In order to obtain a wide range of leaf biochemical properties and reflectance, ten datasets, including one measured and nine publicly available, were collected to cover various plant species with varied growth stages, nutrition status, and planting regions. The Dataset #1 was developed from 2279 plant leaves that were randomly sampled from oilseed rape (Brassica napus L.), rice (Oryza sativa L.), and citrus (Citrus aurantium L.) by ASD FieldSpec 4. Nine publicly available datasets from the EcoSIS Spectral Library (online https://ecosis.org/) with widely variable leaf spectral and biochemical properties were used in this study. Among them, leaf BRF spectra in 7 datasets were measured by an ASD spectroradiometer (Analytical Spectral Devices, Inc., Boulder, CO, USA) equipped with the ASD leaf clip.

[Results]

ASD |

The difference between mean BRF and DHRF spectra (a) as well as the contribution of the difference to the mean BRF spectra (b). Oilseed rape (red line) was obtained in Dataset #1, and other plant species were obtained in Dataset #5.


ASD |

Differences between leaf BRF (green line) and DHRF (orange line) spectra using the first- (a–c) and second order (d–f) derivatives by considering two cases of wavelength-independent f (a, d) and wavelength-dependent f (b, c, e, f). As for wavelength-independent f, leaf DHRF spectra (DHRF = BRF – f) were calculated from the same leaf BRF spectrum with different f (a, d). As for wavelength-dependent f, leaf BRF and DHRF spectra at 400–800 nm (b, e) and 800–2500 nm (c, f) was obtained from the leaf sample in Dataset #1, and other leaf samples with the similar tendency were not presented. Comparisons among the simulated DHRF spectra from the PROSPECT inversion and PROSDM-FMD and the measured BRF and DHRF spectra from the leaf sample in Dataset #1 (g–i).

 

ASD |

Measured and estimated Cab (a, e, i, m), Cxc (b, f, j, n), Cw (c, g, k, o), and Cm (d, h, l, p) in all datasets (Dataset #1–#10) from leaf BRF spectra using the PROSPECT inversion (a–d), PROCOSINE inversion (e–h), PROCWT-S4 (i–l), and PROSDM (m–p) based on the PROSPECT-PRO with the full spectral domain. For the PROSDM, the PROSDM-SED was used to retrieve Cab and Cxc, and the PROSDM-FMD was used to retrieve Cw and Cm.

[Conclusions] In this study, a novel inversion method, called PROSDM, was proposed to retrieve leaf biochemical properties from leaf BRF spectra. It was found that spectral derivatives were able to eliminate the wavelength-independent difference between leaf BRF and DHRF spectra. When the difference between BRF and DHRF spectra varied with the wavelength, spectral derivatives only removed part of the difference, while implementation of MD compensated the limitation of the spectral derivative to further reduce the difference. As a result, the PROSDM accurately retrieved Cab, Cxc, Cw, and Cm from leaf BRF spectra across different plant species. Unlike that the standard PROSPECT inversion requires the measurements of leaf DHRF spectra from a spectroradiometer with an integrating sphere, the PROSDM extended the application of the PROSPECT inversion to leaf BRF spectra for the retrievals of leaf biochemical traits. Potential, it can be applied to various handheld spectroradiometer coupled with the leaf clip for the in situ retrievals of leaf biochemical traits.

With the full spectral domain, the PROSDM-SED achieved the optimal retrievals of Cab and Cxc with the RMSEs of 7.64 μg/cm2 and 2.77 μg/cm2, respectively, and the best estimations of Cw (RMSE = 0.0041 g/cm2) and Cm (RMSE = 0.0024 g/cm2) was produced by the PROSDM-FMD. The RMSEs of Cab, Cxc, Cw, and Cm retrieved from the PROSDM were reduced by 20.33%, 29.34%, 25.45%, and 44.19%, respectively, compared with the PROSPECT inversion. Results demonstrated that the retrieval accuracies of Cab, Cxc, Cw, and Cm from the PROSPECT and PROCOSINE inversions as well as PROCWT were greatly affected by the spectral saturation, PROSPECT versions, spectral subdomains, and the range of model parameters. The retrieval results by different inversion methods may be improved with the suitable spectral subdomains and ranges of model parameters. This requires the knowledge of a priori information of leaf biochemical and structural traits from field measurements and reported studies. Compare with these inversion methods, the proposed PROSDM showed the great potential to alleviate these negative effects on the retrievals of Cab, Cxc, Cw, and Cm. For different PROSPECT versions, the PROSPECT-PRO should be recommended to retrieve leaf biochemical traits from leaf BRF spectra.

Further research is needed to measure the spectral and directional variations of the leaf BRF spectra based on the leaf BRDF model, and coupling the BRDF model with the proposed PROSDM could improve the characterization of variations between leaf BRF and DHRF spectra. In addition, the PROSDM did not obtain the consistent retrieval results in different datasets due to the variations in the difference between BRF and DHRF spectra across plant species. More work is expected to focus on challenges of understanding variability in leaf optical properties under different viewing and illumination angles across plant species. The PROSDM framework is expected to be performed on different scales to advance the applicability of mechanistic approaches in remote sensing, ecology, and environmental studies.


ASD | 6377414053493222255901487.pdf



News / Related News More
2022 - 11 - 30
The pine wood nematode (Bursaphelenchus xylophilus) is a dangerous invasive species that causes pine wilt disease (PWD), which has already destroyed enormous areas of pine forest in China. Therefore, early-stage PWD detection is crucial for a prompt management response. Hyperspectral RS uses data from hundreds of bands and continuous wavelengths, and these bands can capture physiological changes i...
2022 - 11 - 24
Microplastics (particles3. A considerable part of manufactured plastics is heavier than water when they enter water environment, and they can enter sediments system. Some studies have showed that the presence of microfibers in sediment could change the composition of marine sediment microbial communities and significantly affect the nitrogen cycling, and affect the biogeochemical process of marine...
2022 - 11 - 14
Identification and characteristic analysis of urban vegetation spectra under different dust depositionIn recent years, the ecological environment of cities has suffered a lot from damages due to the acceleration of urbanization. The transportation network, traffic flow, industrial activities, and the use of fossil fuels are sources of serious particulate pollutants in the urban greening environmen...
2022 - 11 - 04
Application of Resonon Pika L Hyperspectral Imaging on Individual tree segmentation and tree species classification in subtropical broadleaf forestsAccurate information on tree species is essential in sustainable forest management, ecosystem service assessment, biodiversity monitoring, and ecological environment protection. Therefore, efficient, and cost-effective ways for classifying individual t...
Close window】【Print
Copyright ©2018-2023 LICA United Technology Limited
犀牛云提供企业云服务

LICA United Technology Limited

Address:The 5th.Building,No.18,Anningzhuang East Road,Haidian District, 100085, Beijing, China.

Tel:010-51292601
Fax:010-82899770-8014
E-mail:info@li-ca.com

 


 

 


 
  • Name:
  • *
  • 公司名称:
  • *
  • 地址:
  • *
  • 电话:
  • *
  • 传真:
  • *
  • E-mail:
  • *
  • 邮政编码:
  • *
  • 留言主题:
  • *
  • Details:
  • *
Feedback
Follow us
  • Wechat
  • Mobile Website
友情链接: