Gas Diffusivity in Undisturbed Volcanic Ash Soils: Test of Soil-Water-Characteristic- Based Prediction Models Per Moldrup,* Seiko Yoshikawa, Torben Olesen, Toshiko Komatsu, and Dennis E. Rolston ABSTRACT Andisols are found on ≈7.4 million ha in Southeast Asia (Fumoto and Sverdrup, 2000). The major areas with Soil-water-characteristic-dependent (SWC-dependent) models to volcanic ash soils are located in the circum-Pacific region predict the gas diffusion coefficient, D P , in undisturbed soil have only been tested within limited ranges of pore-size distribution and total including Japan, Korea, New Zealand, the Philippines, porosity. Andisols (volcanic ash soils) exhibit unusually high porosities and the western coast of North America (Bullard, 1984), and water retention properties. The Campbell SWC model and two but some are also found in Europe (Italy and France). Campbell SWC-based models for predicting D P in undisturbed soil The soils have a high potential for agricultural produc- were tested against SWC and D P data for 18 Andisols and four Gray- tion because of their unique physical and chemical prop- lowland (paddy field) soils from Japan. The Campbell model accu- erties, including high water retention, good drainage, rately described SWC data for all 22 soils within the matric potential and high nutrient availability (Shoji et al., 1993). A com- range from ≈ 10 to 15 000 cm H 2 O. The SWC-dependent Bucking- prehensive description of the genesis, properties, and ham-Burdine-Campbell (BBC) gas diffusivity model predicted D P utilization of volcanic ash soils is provided by Shoji et data well within the same matric potential range for the 18 Andisols. al. (1993). The BBC model showed a minor but systematic underprediction of D P for three out of the four Gray-lowland soils, likely due to a blocky Andisols show a wide variation in soil texture, but soil structure with internal fissures. A recent D P model that also takes exact grain size distribution is difficult to determine, into account macroporosity performed nearly as well as the BBC due to the typically high content of noncrystalline Al- model. However, D P in the macropore region (air-filled pores 30 and Fe-based materials that inhibit dispersion of mineral m) was consistently underpredicted, likely due to high continuity particles during analysis (Shoji et al., 1993). Thus, grain of the macropore system in both Andisols and Gray-lowland soils. In size distributions are of limited value in characterizing agreement with previous model tests for 21 European soils (represent- Andisols since sand, silt, and clay contents do not have ing lower porosities and water retention properties), both SWC-de- as precise a meaning for Andisols as for soils consisting pendent D P models gave better predictions for the 22 Japanese soils largely of crystalline minerals (Warkentin and Maeda, than soil-type independent models. Combining D P and SWC data, a 1974). The physical characteristics of Andisols may so-called gas diffusion fingerprint (GDF) plot to describe soil aeration potential is proposed. therefore be more stringently defined by their soil-water characteristic curve (pore-size distribution) within zero to 15 000 cm H 2 O of matric potential (Osozawa et al., 1990). Ito et al. (1991) observed a highly significant M ost transport parameter models have at present relationship between water retention at a matric poten- only been tested within relatively limited ranges tial of 15 000 cm H 2 O and the content of both allopha- of soil pore-size distributions and total porosities nic clays and humus. Drying to matric potentials below (Moldrup et al., 2001). Since volcanic ash soils exhibit 15 000 cm H 2 O makes volcanic ash soils less dispersible pore-size properties quite different from normal mineral through irreversible aggregation (Kubota, 1976), a phe- soils, including larger total porosities and broader pore nomenon not seen in other mineral soils (Iwata et al., size distributions, data for Andisols should prove highly 1995), and further illustrating the unique physical and valuable in testing the general validity of predictive chemical characteristics. models for the main gaseous and liquid phase transport Andisols typically have a well-developed soil struc- parameters (the gas and solute diffusion coefficients, ture with a wide range of pore sizes that retain a large and the air and water permeabilities) in unsaturated, amount of water with varying matric potentials (Furu- undisturbed soils. However, only limited information hata and Hayashi, 1980; Saigusa et al., 1987). Egashira about transport properties of Andisols are available in et al. (1983) showed that water-stable aggregates in Jap- English-language journals. anese Andisols typically contain high amounts of clay- Volcanic ash soils are distributed across ≈0.84% of size particles (12–71%) and organic matter (3–35%). the earth’s land surface (Leamy, 1984). For example, The high content of noncrystalline and organic materials affects the soil structure that typically exhibits either P. Moldrup, Dep. of Environ. Engineering, Aalborg Univ., Sohngaard- granular, angular blocky, or subangular blocky structure sholmsvej 57; T. Olesen, City and Environment Section, Aalborg (Shoji et al., 1993). The high degree of soil structure Municipality, Vesterbro 14, DK-9000 Aalborg, Denmark; S. Yoshi- kawa (formerly, Osozawa), Dep. of Regional Crops Science, Natl. normally found in Andisols and its dramatic effect on Agric. Res. Center for Western Region, Senyu 1-3-1, Zentsuji, Ka- transport properties can be illustrated by the observa- gawa, 765-8508 Japan; T. Komatsu (formerly, Yamaguchi), Graduate tion that light clay soils generally have saturated hydrau- School of Science and Engineering, Saitama University, 255 Shimo- okub, Saitama, 338-8570 Japan; D.E. Rolston, Soils and Biogeo- chemistry, Dep. of Land, Air, and Water Resources, Univ. of California, Abbreviations: BBC, Buckingham-Burdine-Campbell; GDF, gas dif- Davis, CA 95616. Received 29 June 2001. *Corresponding author fusion fingerprint; IRC, incremental relative change in gas diffusivity; (i5pm@civil.auc.dk). REV, representative elementary volume; RMSE, root mean square error; SWC, soil-water characteristic. Published in Soil Sci. Soc. Am. J. 67:41–51 (2003). 41 Published January, 2003