نوع مقاله : علمی- پژوهشی
نویسندگان
1 دانشجوی دکتری، گروه جنگلداری و اقتصاد جنگل، دانشکده منابع طبیعی، دانشگاه تهران، کرج، ایران
2 دانشیار، گروه علوم و صنایع چوب و کاغذ، واحد کرج، دانشگاه آزاد اسلامی، کرج، ایران
3 نویسنده مسئول، استاد، گروه جنگلداری و اقتصاد جنگل، دانشکده منابع طبیعی، دانشکدگان کشاورزی و منابع طبیعی، دانشگاه تهران، کرج، ایران
4 استاد، گروه جنگلداری و اقتصاد جنگل، دانشکده منابع طبیعی، دانشکدگان کشاورزی و منابع طبیعی، دانشگاه تهران، کرج، ایران
5 استادیار، گروه جنگلداری، دانشکده کشاورزی و منابع طبیعی اهر، دانشگاه تبریز، ایران
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
Background and Objectives: Indirect methods for measuring the leaf area index (LAI) such as hemispherical photography using a digital camera equipped with a fisheye lens, offer high accuracy; however, they are cost-prohibitive and logistically challenging for field applying. Recent technological advancements in smartphones and the availability of clip-on fisheye lenses have enabled the potential use of these devices for LAI estimation. This study aimed to compare the accuracy of hemispherical images from smartphones compared to digital cameras in estimating LAI in species of Robinia pseudoacacia L., Pinus eldarica Medw., Celtis australis L., and Fraxinus rotundifolia Mill. in Chitgar Forest Park in Tehran County, Iran.
Methodology: A total of 169 sampling points were selected across four stands, including 47 points in R. pseudoacacia, 36 in P. eldarica, 47 in F. rotundifolia, and 39 in C. australis. For sampling, the first point was randomly selected at each location, and subsequent photographs were taken at 10-meter intervals beneath the canopy. At each point, two hemispherical images were taken simultaneously using both digital camera and a smartphone. All photographs were taken at a height of 0.5 meters under pre-sunrise conditions to ensure uniform lighting. The leaf area index was calculated from images using GLA software, data normality was assessed in SPSS, followed by paired t-tests. For linear regression modeling, 70% of the data was randomly selected for model calibration, while the remaining 30% was used for validation. This analysis was carried out in Microsoft Excel. The F-test was applied to evaluate the overall significance of regression coefficients. Additionally, the root mean square error (RMSE) and relative root mean square error (RRMSE) were calculated to quantify the regression model’s error, and bias was used to determine the model's underestimation or overestimation.
Results: The mean LAI measured by the digital camera was 0.8 (±0.27 SD) for R. pseudoacacia, 0.89 (±0.15 SD) for P. eldarica, 1.24 (±0.16 SD) for C. australis, and 0.77 (±0.27 SD) for F. rotundifolia. These values for the smartphone were 0.66 (±0.31 SD), 0.83 (±0.19 SD), 1.07 (±0.14 SD), and 0.75 (±0.3 SD), respectively. The paired t-test revealed statistically significant differences in mean values (p < 0.05) for R. pseudoacacia, P. eldarica, and C. australis, while F. rotundifolia showed no significant difference (p > 0.05). The mean difference values were positive for all species, with the highest value observed in C. australis (0.2) and the lowest in F. rotundifolia (-0.028). All simple linear regression models were statistically significant based on F-tests. Also, based on the t-test, intercept was not significant in all linear regression models. The coefficients of determination (R²) values were determined 0.53 for R. pseudoacacia, 0.59 for P. eldarica, 0.73 for C. australis, and 0.71 for F. rotundifolia. Evaluation using the 30% validation dataset yielded RMSE and RRMSE values of 0.07 and 9% for P. eldarica, 0.1 and 10% for C. australis, 0.18 and 25% for F. rotundifolia and 0.18 and 33% for R. pseudoacacia. Smartphone measurements without modeling consistently underestimated values across all species, most markedly in C. australis with a 13.7% underestimation compared to reference values.
Conclusion: In the present study, digital camera data were used solely as a reference for model calibration and validation. After model development, the smartphone can independently estimate the leaf area index without requiring camera data. However, the model has only been validated under the conditions and species examined, and its application to other regions or species would require further calibration and validation. Moreover, a mid-range smartphone and a clip-on fisheye lens to achieve a 180-degree field of view. Although their accuracy and image quality are lower than digital cameras, they can serve as an effective and practical alternative for remote areas or large-scale field monitoring, particularly in species with homogeneous canopy structures.
Keywords: Fisheye lens, hemispherical photography, linear regression, model validation.
کلیدواژهها [English]
)