. .  , 2003, . 24, . 14, 2945–2957 Determination of scaling characteristics of AVHRR data with wavelets: application to SGP97 N. A. BRUNSELL* and R. R. GILLIES*† *Department of Plants, Soils, and Biometeorology, Utah State University, Logan, UT 84322-4820, USA †Department of Aquatic, Watershed and Earth Resources, Utah State University, Logan, UT 84322-5210, USA (Received 2 August 2001; in final form 4 April 2002 ) Abstract. A wavelet multi-resolution analysis was conducted using vegetation and radiometric temperature data derived from the Advanced Very High Resolution Radiometer (AVHRR) data from the Southern Great Plains 1997 (SGP97) Hydrology Experiment. The wavelet coefficients were used to investigate whether the Normalized Difference Vegetation Index (NDVI) and radiometric temperature fields exhibit self-similar scaling behaviour. Using the first six moments of the wavelet coefficients through eight levels of dyadic decomposition, the NDVI data are shown to be statistically self-similar with a slope of approxi- mately -0.72 in each of the horizontal, vertical, and diagonal directions of the image over scales ranging from 2000 to 256 000 m. The radiometric temperature data are also shown to exhibit self-similarity with slopes ranging from -0.9 in the diagonal direction to -1.0 in the vertical direction over the same scales. These slopes indicate long range behaviour and may imply a methodology for statistically assimilating remotely sensed data into large-scale meso and global climate models. 1. Introduction Satellite remote sensing potentially offers a valuable resource in the modelling and validation of terrestrial systems as well as the understanding of hydrological and meteorological processes, due in part to a large spatial coverage and varying temporal resolutions. Remote sensing could be used in several ways in conjunction with global and meso-scale modelling, the most likely of which is assimilating satellite data into such models (e.g. Houses et al. 1998). In another study, Santanello and Carlson (2001) showed some improvement in model (MM5) performance by incorp- orating a percentage cover of vegetation as derived from the Advanced Very High Resolution Radiometer (AVHRR) sensor and applying an aggregation scheme of the fields up to the resolution of the model. At the same time, this issue of assimilation exemplifies the lack of current understanding when translating data to larger spatial scales. Satellite data such as AVHRR are collected at a spatial scale of 1 km2 , whereas climate models generally run at a grid resolution of the order of 10–100 km2 . This is one aspect of the ‘scaling problem’ referred to as aggregation, and it is a direct International Journal of Remote Sensing ISSN 0143-1161 print/ISSN 1366-5901 online © 2003 Taylor & Francis Ltd http://www.tandf.co.uk/journals DOI: 10.1080/01431160210155983