JOURNAL OF NATURAL RESOURCES ›› 2019, Vol. 34 ›› Issue (2): 400-411.doi: 10.31497/zrzyxb.20190215

• Resource Evaluation • Previous Articles     Next Articles

Intercomparison and evaluation of influencing factors among different LAI products over mountainous areas

JING Jin-cheng1,2,3(), JIN Hua-an1, TANG Bin2, LI Ai-nong1()   

  1. 1. Research Center for Digital Mountain and Remote Sensing Application, Institute of Mountain Hazards and Environment, CAS, Chengdu 610041, China
    2. College of Earth Science, Chengdu University of Technology, Chengdu 610059, China
    3. Sichuan Provincial Coal Design & Research Institute, Chengdu 610031, China;
  • Received:2018-08-03 Revised:2018-12-15 Online:2019-02-28 Published:2019-02-28
  • Contact: LI Ai-nong;


The validation of LAI products not only provides quantitative quality description, but also plays an important part in the improvement of retrieval algorithm and understanding of uncertainties regarding to land surface process models over rugged surfaces. This study evaluated the performance of MODIS (C5, C6), Geoaldn2 (GEOV1, GEOV2) LAI products using intercomparison methods over Southwestern China. The spatiotemporal distribution of integrity and consistencies, such as the percentage of main algorithm, smoothness index, average value during growing season, yearly mean LAI bias and root mean square error (RMSE), respectively, were investigated during the period 2001-2016. Meanwhile, different versions of the same data source LAI products were compared so as to get a clear understanding of improvement about the new one over heterogeneously hilly regions. Lastly, four factors, such as topography (altitude and relief amplitude), vegetation types, and climate regionalization were selected to assess the influence of different underlying surfaces on LAI products using the Geodetector. The results show that spatial and temporal consistency of these LAI products is good over most areas. All LAI products exhibit a higher percentage of good quality data (i. e. successful retrieval) in mountainous areas, and GEOV2 LAI is higher than others. The percentage in altitude and higher relief amplitude area seems to be low. GEOV2 LAI shows smoother temporal profiles than others, and Geoland2 is smoother than MODIS LAI. It is clear that MODIS C5 is smoother than MODIS C6, and GEOV2 is superior to GEOV1. RMSE and yearly mean LAI bias between one and another LAI product are vulnerable to topographic indices, especially to altitude. Q-statistic in Geodetector is smallest related to relief amplitude, biggest to vegetation for MODIS LAI product, and biggest to climate for Geoland2 LAI. Altitude, vegetation, and climate play a dominant role in spatial distribution of LAI. The validation experience demonstrates the importance of topography, vegetation and climate for LAI estimation over mountainous areas. Considerable attention will be paid to the production of higher quality LAI products in topographically complex terrain.

Key words: leaf area index, Southwestern China, intercomparison, Geodetector, spatial heterogeneity