Haar wavelet in estimating depth profile of soil temperature

https://doi.org/10.1016/j.amc.2008.12.036Get rights and content

Abstract

A Haar wavelet based method for the estimation of soil temperature at different depths is described in this paper. Diurnal variation in the hourly soil temperature is estimated at different depths varying from 0 to 45 cm. This estimation is compared with observed data available for Trombay site at the depths 5, 10, and 20 cm. This Haar technique can be interpreted from incremental and multi-resolution viewpoint. The estimated values show and excellent agreement with the observed values. The Haar based estimated values are found to be more accurate than the values obtained by FDM approach. More accurate solutions can be obtained by changing the time scale in Haar wavelet; at the same time main features of the solution are preserved. Sensitivity analyses indicate that soil temperature was no so sensitive to changes of soil thermal parameters. Moreover the use of Haar wavelets is found to be accurate, simple, fast, flexible, convenient, small computation costs and computationally attractive.

Introduction

The application of reaction–diffusion equations is day by day increasing. Its application in determining the depth profile of soil temperature was first discussed by Carslaw and Jaeger [1] and the analytical technique for obtaining the diurnal variation of soil temperature with depth by solving one-dimensional Fickian diffusion equation is well documented in literature [13], [14], [16], [17], [18], [19]. Since this method involved so many assumptions, which made the applications so restrictive, Dutta and Daoo [2] used an upwind finite difference technique, solved the same problem and some improvement in the results were reported to literature. Lepik [7], [20] used the Haar wavelet method for solving nonlinear integral and differential equations. The authors used the same boundary conditions, which were used in the analytical method, and good agreements in the results with the observed data were recorded. However the mean percentage of errors were observed to be 25.1 at 5 cm depth, 15.7 at 10 cm and 9.4 at 20 cm. Dutta in his paper used 5 cm spacing for z and 1 h for Δt. In the sensitivity analysis, it has been shown a small change in soil depth z < 5 cm (similarly a small change in Δt < 15 min) affect the soil temperature. In this paper we applied Haar wavelet technique to solve this problem in order to (i) minimize the error propagation as soil depth increases and (ii) have computational complexity under control. We were able to arrive better results mainly because, by nature, the surface temperature and the soil temperature at any depth follow the diurnal pattern of the ground heat flux, which is of sinusoidal waveform. This paper is organized as follows. In Section 2, we present the derivation of one-dimensional unsteady state partial differential equation for heat diffusion in soil related boundary conditions and its analytic solution. In Section 3, we explain the fundamentals of Haar wavelet and the construction of operational matrices. In Section 4, we explain the solution by wavelet transform.We discuss the results by comparing with observed data in Section 5. Sensitivity analysis is carried out in Section 6. Concluding remarks are given in Section 7.

Section snippets

Heat diffusion equation in soil

Hydraulic properties of unsaturated soils are adequately described by Van Geunuchten and Nielsen [3]. Kimball et al. [4] describes soil heat flux determination through temperature gradient method with computed thermal conductivities Messman [5] corrected errors associated with soil heat flux measurements. Dutta et al. [2] constructed a numerical model for determining depth profile of soil temperature. A brief description about the soil characteristics and methods determining soil heat flux are

Haar wavelet preliminaries

The orthogonal set of Haar wavelets hi(t) is a group of square waves [6] with magnitude +1 or −1 in some intervals and zeros elsewhereh0(t)=1,0t<1h1(t)=10t<1/2,-11/2t<1,hn(t)=h1(2jt-k),n=2j+k,j0,0k<2j,n,j,kZ.

Just these zeros make Haar wavelets be local, faster than other square functions such as Walsh and very useful in solving stiff systems.

Any function y(t) which is square integrable in the interval [0, 1) can be expanded in a Haar series with an infinite number of termsy(t)=0cihi(t),

Solution of diffusion equation by Haar wavelet

In the Haar domain, we assume that ut can be expanded as a Haar series asut=at(x)H(t).Integrating (26) and applying the integration matrix P of Section 3, we haveu(x,t)=at0tH(t)dt=atPH(t).Using (26), (27) in (7), we haveatH(t)=K(a¨tPH(t))a¨t=katP-1,wherek=1K,a¨d2adx2.The differential equation (29) is solved using the conventional method. After dropping all the eigen values of (29) with positive real parts we haveat=a0te(-x12k)P-12,u(x,t)=a0te(-x1K)PH(t).Let us seta0t=[0,0,0,2m,,2m],a0tP

Results and discussion

Soil temperature measurements at five specific depths namely 0, 5, 10, 20 and 40 cm at Trombay were made during January 2001 by Daoo and subsequently used by Dutta [2]. Here the same set of data is taken into account for comparison with the Haar wavelet solution for time ranging from 1 to 24 h and z values ranging from 0 to 45 cm. The values of k, u0 and Δu0 calculated from the data set for one day are used for comparison of the predicted soil temperature using Haar wavelet procedure at the

Sensitivity analyses

Sensitivity analyses were used to determine the significance of: (1) the use of approximate thermal properties and (2) the use of fixed k values with respect to temperature. A sensitivity analysis of the Haar scheme was also performed with respect to the parameters k, u0 and Δu0 in order to substantiate their relative importance. We found that changes in thermal conductivity had little effect on u0 but caused some differences in u(z, t) in deep layers whereas changes in Δu0 (amplitude of soil

Conclusion

The use of Haar wavelets in the engineering produces exciting results. The characteristics of wavelet transforms make them particularly useful for the approximation of functions with steep gradients or sharp spikes. Certainly the orthogonal and orthonormal properties of wavelet basis allow us to simplify the calculation of integrals. The Haar wavelet model described in this study provides reasonable estimates of soil temperature at any depth and time. If few more parameters like specific heat

Acknowledgement

Our Thanks are due to Dr. T.R. Neelakandan, Professor of Civil Engineering, SASTRA University, Thanjavur 613 402, India for the useful discussions in the subject.

References (21)

  • H.S. Carslaw et al.

    Conduction of Heat in Solids

    (1959)
  • M. Dutta, V.J. Daoo, A numerical model to estimate depth profile of soil temperature, in: National Symposium on...
  • M. Van Geunuchten et al.

    On describing and predicting hydraulic properties of unsaturated soils

    Ann. Geophys.

    (1985)
  • B.A. Kimball et al.

    Soil heat flux determination: temperature gradient method with computed thermal conductivities

    Soil Sci. Soc. Am. J.

    (1975)
  • W.J. Messman

    Correcting errors associated with soil heat flux measurements and estimating soil thermal properties from soil temperature and heat flux plate data

    Agric. Forest Meteorol.

    (1992)
  • A. Haar

    Zur theorie der orthogonalen funktionen systeme

    Math. Ann.

    (1910)
  • Ulo. Lepik

    Numerical solution of differential equations using Haar wavelets

    Math. Comput. Simulat.

    (2005)
  • Ulo. Lepik

    Application of the Haar wavelet transform to solving integral and differential Equations

    Proc. Estonian Acad. Sci. Phys. Math.

    (2007)
  • T.J. Kuhaida et al.

    The prediction of mean monthly soil temperature from mean monthly air temperature

    Soil Sci.

    (1978)
  • C.F. Chen et al.

    Wavelet approach to optimizing dynamical systems

    IEEE Proc. Control Theory Appl.

    (1999)
There are more references available in the full text version of this article.

Cited by (0)

View full text