1. College of Geographical Science, Inner Mongolia Normal University, Huhhot 010022, China
2. College of Earth and Environmental Sciences, Lanzhou University, Lanzhou 730000, China
3. Key Laboratory of Remote Sensing & Geography Information System, Inner Mongolia Normal University, Huhhot 010022, China
Abstract：In recent years, the Mongolian Jas’s Larch Inchworm disaster have been increasing and gradually approaching the Daxinganling area, which will threaten the safety of the forest ecosystem in the north of China. It is necessary to replace the Mongolian traditional detection method with the modern remote sensing monitoring method, and it is important to control the rule of occurrence and development of the insect at the first opportunity. A detection model was developed for rapid and wide range remote sensing monitoring of Jas’s Larch Inchworm damage based on spectral analysis technology. Spectra in situ measurements were conducted for healthy larch and mild, moderate, severe damaged larch to compute and compare the raw spectra and continuum removal spectra. On this basis, spectral sensitive bands and sensitivities of continuum removal spectra were uncovered. Then some spectral characteristic parameters were obtained by the first derivative transformation of continuum removal spectrum, and its change features vs damaged level of larch were analyzed to construct the detection model based on CART (classification and regression tree) algorithm. The results showed that the sensibility of continuum removal spectra was more significant than that of raw spectra, especially in 480～520 nm (blue edge), 640～720 nm (red edge) and 1 416～1 500 nm (short wave infrared valley). In these bands, the value for the sensitivity of continuum removal spectra was between 0.1 and 0.2. At the same time, it appeared sensitive peaks in the sensitive bands. With the increase of damaged level, the sensibility of continuum removal spectra had an enhanced trend which was more evident compared with raw spectra. In addition, the sensitive peak position of the continuum removal spectra in the blue band was shifted to the short wave direction, that is, 502 nm→490 nm, and the spectral sensitive peak position in the red valley and red edge, short wave infrared valley moved to the long wave direction, that is, 664 nm→672 nm and 1 436 nm→1 448 nm. Furthermore, the normalized reflectance of red valley position and short wave infrared valley position as well as areas of red valley and short wave infrared valley were found a trend of rising. First derivative of continuum removal spectra had obvious in response for damaged level in blue edge and red edge. Moreover, the band peaks were arisen. The red edge position blue shift arose (718 nm→700 nm) as the damaged level increased, while the slopes and areas for blue edge and red edge decreased. Hence, the CART model that was established based on several continuum removal spectral parameters such as red edge slope, normalized reflectance of red valley position and short wave infrared valley position, areas of red valley and short wave infrared valley and both slope and area of blue edge had superior detection ability in the damaged level of larch. Compared to multi-linear regression model, the CART model performed better with the Kappa value of 0.875. These results will play important roles on remote sensing monitoring of the damage of Jas’s Larch Inchworm.
Key words：Jas’s larch inchworm; Spectral sensibility; Spectral characteristics of continuum removal; Larch damaged level; Classification and regression tree (CART)
[1] Wulder M A, Dymond C C, White J C, et al. Forest Ecology & Management, 2006, 221(1-3): 27.
[2] Niemann K O, Quinn G, Stephen R, et al. Canadian Journal of Remote Sensing, 2015, 41(3)：191.
[3] White J C, Coops N C, Hilker T, et al. International Journal of Remote Sensing, 2007, 28(10): 2111.
[4] Mutanga O, Ismail R. Southern Forests: a Journal of Forest Science, 2010, 72(1): 1.
[5] WU Nan, LIU Jun-ang, YAN Rui-kun, et al(伍 南, 刘君昂, 闫瑞坤, 等). Chinese Agricultural Science Bulletin(中国农学通报), 2012, 28(4): 51.
[6] Kobayashi T, Kanda E, Kitada K, et al. Phytopathology, 2001, 91(3): 316.
[7] Fassnacht F E, Latifi H, Ghosh A, et al. Remote Sensing of Environment, 2014, 140(1): 533.
[8] Ahern F J. International Journal of Remote Sensing, 1988, 9(9): 1451.
[9] Kokaly R F, Clark R N. Remote Sensing of Environment, 1999, 67(3): 267.
[10] Zhi H, Urner B J, Dury S J, et al. Remote Sensing of Environment, 2004, 93(1/2): 18.
[11] Mutanga O, Skidmore A K, Wieren S V. Journal of Photogrammetry and Remote Sensing, 2003, 57(4): 263.
[12] Cheng T, Rivard B, Sánchezazofeifa G A, et al. Remote Sensing of Environment, 2010, 114(4): 899.
[13] Lausch A, Heurich M, Gordalla D, et al. Forest Ecology & Management, 2013, 308(4): 76.
[14] Helmi Z M S, Mohamad A M S, Azadeh G. American Journal of Applied Sciences, 2006, 3(6): 1864.
