Letter

Comparing roughness maps generated by five typical roughness descriptors for LiDAR-derived digital elevation models

  • Received: 22 December 2023 Revised: 13 March 2024 Accepted: 22 March 2024 Published: 02 April 2024
  • Terrain surface roughness, often described abstractly, poses challenges in quantitative characterization with various descriptors found in the literature. In this study, we compared five commonly used roughness descriptors, exploring correlations among their quantified terrain surface roughness maps across three terrains with distinct spatial variations. Additionally, we investigated the impacts of spatial scales and interpolation methods on these correlations. Dense point cloud data obtained through Light Detection and Ranging technique were used in this study. The findings highlighted both global pattern similarities and local pattern distinctions in the derived roughness maps, emphasizing the significance of incorporating multiple descriptors in studies where local roughness values play a crucial role in subsequent analyses. The spatial scales were found to have a smaller impact on rougher terrain, while interpolation methods had minimal influence on roughness maps derived from different descriptors.

    Citation: Lei Fan, Yang Zhao. Comparing roughness maps generated by five typical roughness descriptors for LiDAR-derived digital elevation models[J]. AIMS Geosciences, 2024, 10(2): 228-241. doi: 10.3934/geosci.2024013

    Related Papers:

    [1] H. M. Barakat, M. A. Alawady, I. A. Husseiny, M. Nagy, A. H. Mansi, M. O. Mohamed . Bivariate Epanechnikov-exponential distribution: statistical properties, reliability measures, and applications to computer science data. AIMS Mathematics, 2024, 9(11): 32299-32327. doi: 10.3934/math.20241550
    [2] Aisha Fayomi, Ehab M. Almetwally, Maha E. Qura . A novel bivariate Lomax-G family of distributions: Properties, inference, and applications to environmental, medical, and computer science data. AIMS Mathematics, 2023, 8(8): 17539-17584. doi: 10.3934/math.2023896
    [3] Fiaz Ahmad Bhatti, Azeem Ali, G. G. Hamedani, Mustafa Ç. Korkmaz, Munir Ahmad . The unit generalized log Burr XII distribution: properties and application. AIMS Mathematics, 2021, 6(9): 10222-10252. doi: 10.3934/math.2021592
    [4] Naif Alotaibi, A. S. Al-Moisheer, Ibrahim Elbatal, Salem A. Alyami, Ahmed M. Gemeay, Ehab M. Almetwally . Bivariate step-stress accelerated life test for a new three-parameter model under progressive censored schemes with application in medical. AIMS Mathematics, 2024, 9(2): 3521-3558. doi: 10.3934/math.2024173
    [5] M. Nagy, H. M. Barakat, M. A. Alawady, I. A. Husseiny, A. F. Alrasheedi, T. S. Taher, A. H. Mansi, M. O. Mohamed . Inference and other aspects for qWeibull distribution via generalized order statistics with applications to medical datasets. AIMS Mathematics, 2024, 9(4): 8311-8338. doi: 10.3934/math.2024404
    [6] Ammar M. Sarhan, Rabab S. Gomaa, Alia M. Magar, Najwan Alsadat . Bivariate exponentiated generalized inverted exponential distribution with applications on dependent competing risks data. AIMS Mathematics, 2024, 9(10): 29439-29473. doi: 10.3934/math.20241427
    [7] Guangshuai Zhou, Chuancun Yin . Family of extended mean mixtures of multivariate normal distributions: Properties, inference and applications. AIMS Mathematics, 2022, 7(7): 12390-12414. doi: 10.3934/math.2022688
    [8] Areej M. AL-Zaydi . On concomitants of generalized order statistics arising from bivariate generalized Weibull distribution and its application in estimation. AIMS Mathematics, 2024, 9(8): 22002-22021. doi: 10.3934/math.20241069
    [9] Rasha Abd-Elwahaab Attwa, Taha Radwan, Esraa Osama Abo Zaid . Bivariate q-extended Weibull morgenstern family and correlation coefficient formulas for some of its sub-models. AIMS Mathematics, 2023, 8(11): 25325-25342. doi: 10.3934/math.20231292
    [10] Heba S. Mohammed, Zubair Ahmad, Alanazi Talal Abdulrahman, Saima K. Khosa, E. H. Hafez, M. M. Abd El-Raouf, Marwa M. Mohie El-Din . Statistical modelling for Bladder cancer disease using the NLT-W distribution. AIMS Mathematics, 2021, 6(9): 9262-9276. doi: 10.3934/math.2021538
  • Terrain surface roughness, often described abstractly, poses challenges in quantitative characterization with various descriptors found in the literature. In this study, we compared five commonly used roughness descriptors, exploring correlations among their quantified terrain surface roughness maps across three terrains with distinct spatial variations. Additionally, we investigated the impacts of spatial scales and interpolation methods on these correlations. Dense point cloud data obtained through Light Detection and Ranging technique were used in this study. The findings highlighted both global pattern similarities and local pattern distinctions in the derived roughness maps, emphasizing the significance of incorporating multiple descriptors in studies where local roughness values play a crucial role in subsequent analyses. The spatial scales were found to have a smaller impact on rougher terrain, while interpolation methods had minimal influence on roughness maps derived from different descriptors.



    The use of probability distributions has been ages old in many areas of life. The probability distributions provide a way to model phenomenon in many fields that ranges from economics to finance and from physical sciences to astronomy. Various standard probability distributions are available that can be used to model various real-life phenomenon, but in several situations these standard probability distributions fail to capture the underlying trend of the data. In such situations, some extensions or generalizations are required. A method to extend any baseline probability distribution has been proposed by [1] by using the logit of beta distribution. Another simple method to extend any baseline distribution has been proposed by [2] and is known as the transmuted family of distribution. The cumulative distribution function (cdf) of this family of distributions is

    FX(x)=G(x)[1+λλG(x)], (1)

    where G(x) is the cdf of any baseline distribution and λ is the transmutation parameter such that 1λ1. The transmuted family of distributions has been explored by various authors for different baseline distributions, see for example [3].

    A general method of obtaining new families of distributions has been proposed by [4] by using two probability distributions. The proposed family of distributions is named as the T-X family of distributions. The cdf of the proposed family of distributions is

    FTX(x)=W[G(x)]ar(t)dt=R[W{G(x)}];xR, (2)

    where r(t) is density function of some random variable t with support on [a,b] and W[G(x)] is some function of G(x) such that W(0)a and W(1)b. The T-X family of distributions gives rise to several other families of distributions as a special case. The transmuted family of distributions can be obtained from the T-X family of distribution by using r(t)=1+λ2λt;t[0,1] and W[G(x)]=G(x) as shown by [5].

    Several situations arise where we are interested in joint modeling of two or more than two random variables. For example, we might be interested in simultaneous modeling of computer memory and processing time or we might be interested in joint modeling of various characteristics of plants. In such situations the bivariate and/or the multivariate distributions are needed. The development of bivariate distributions has attracted various authors but in most of the situations the bivariate and/or multivariate distributions are not unique. A simple way of obtaining the bivariate distributions from the univariate marginals has been proposed by [6] and is known as the Gumbel bivariate distributions. The joint cdf of this family of distributions is

    FX,Y(x,y)=G1(x)G2(y)[1+γ{1G1(x)}{1G2(y)}];(x,y)R2, (3)

    where G1(x) and G2(y) are marginal cdf's of X and Y respectively and γ is a measure of association such that 1γ1. The bivariate Gumbel distribution (3) has been used by various authors to propose some new bivariate distributions, see for example [7,8,9].

    Another method of generating the bivariate distributions from the univariate marginals is given by [10]. The joint cdf of this family of bivariate distributions is

    F(x,y)=G1(x)G2(y)1γ{1G1(x)}{1G2(y)};(x,y)R2,1γ1. (4)

    The bivariate family of distributions given in (4) includes the bivariate logistic distribution as a special case. The family of distributions given in (4) has not been studied in much details. The families of distributions given in (3) and (4) emerge by using the specialized technique of copulas. Further details about methods of generating bivariate and multivariate distributions by using copulas can be found in [11,12]. Some other bivariate families of distributions can be found in [13,14,15,16].

    Recently, a new method has been proposed by [17] to obtain the bivariate families of distributions by obtaining the bivariate version of the T-X family of distributions. The joint cdf of the proposed bivariate T-X family of distributions is

    FX,Y(x,y)=W2[G2(y)]a2W1[G1(x)]a1r(v1,v2)dv1dv2;a1<v1<b1;a2<v2<b2, (5)

    where r(v1,v2) is some joint density function with support on [a1,b1]×[a2,b2]. Also, W1[G1(x)] and W2[G2(y)] are some absolutely continuous functions of G1(x) and G2(y) such that W1(0)a1, W1(1)b1, W2(0)a2 and W2(1)b2. A simpler version of the family (5) is given by [18] when the support of r(v1,v2) is [0,1]×[0,1]. The joint cdf in this case is given as

    FX,Y(x,y)=G2(y)0G1(x)0r(v1,v2)dv1dv2;0<v1<1;0<v2<1. (6)

    It can be seen that the Gumbel bivariate distribution can be obtained from (6) by using

    r(v1,v2)=1+γ(12v1)(12v2);(v1,v2)[0,1]2.

    The family of distributions given in (6) has been used by [18] to propose a new bivariate family of distributions which is a simpler version of the Cambanis family of distributions, proposed by [19]. The proposed family of distributions has also been used by [20] to propose a bivariate transmuted Burr distribution by using the Burr-XII distribution, proposed by [21], as the baseline distribution. In this paper we have proposed a new bivariate family of distributions by using the bivariate T-X family of distributions given in (6). The research motivation and plan of this paper are given in the following.

    It often happens that the joint modeling of two or more variables is required. For example, a computer scientist might be interested in joint modeling of processing time and processor speed. In such cases the bivariate and/or multivariate distributions are required. Not much work has been done in developing the bivariate and/or multivariate families of distributions and this research is motivated by this fact. The research will attempt to propose new bivariate and multivariate families of distributions for joint modeling of several random variables. The plan of the paper follows.

    The new bivariate family of distributions is proposed in Section 3 and some of its properties are studied in Section 4. Section 5 is dedicated to the multivariate extension of the proposed family alongside some of its properties. In Section 6, a new bivariate Weibull distribution has been proposed by using the new bivariate family of distributions. Some properties of the proposed bivariate Weibull distribution are also studied in the same section. This section also contains a brief about a new multivariate Weibull distribution. Section 7 is reserved for simulation and real data applications of the proposed bivariate Weibull distribution. Conclusions and recommendations are given in Section 8.

    The bivariate family of distributions given in (6) can be used to obtain new bivariate families of distributions for different choices of r(v1,v2). Motivated by this fact, we have proposed a new bivariate family of distributions by using the following joint density of v1 and v2

    r(v1,v2)=1+λ1(12v1)+λ2(12v2)+λ3(14v1v2);(v1,v2)[0,1]2

    in (6) such that the new bivariate family of distribution will provide flexibility to model more complex bivariate data. The joint cdf of the proposed bivariate family of distributions is

    FX,Y(x,y)=G2(y)0G1(x)0{1+λ1(12v1)+λ2(12v2)+λ3(14v1v2)}dv1dv2,

    which on simplification becomes

    FX,Y(x,y)=G1(x)G2(y)[1+λ1{1G1(x)}+λ2{1G2(y)}+λ3{1G1(x)G2(y)}], (7)

    for (x,y)R2. The parameters (λ1,λ2,λ3) satisfy the conditions

    (λ1,λ2,λ3)[1,1];λ1+λ2+λ31;λ1+λ2+3λ31;1λ1+λ31

    and

    1λ2+λ31.

    The joint cdf given in (7) can also be written as

    FX,Y(x,y)=G1(x)G2(y)[1+λλ1G1(x)λ2G2(y)λ3G1(x)G2(y)],

    where λ=λ1+λ2+λ3. The joint density function of the proposed bivariate family of distributions is easily obtained from (7) and is

    fX,Y(x,y)=g1(x)g2(y)[1+λ2λ1G1(x)2λ2G2(y)4λ3G1(x)G2(y)];(x,y)R2. (8)

    The proposed bivariate family of distributions can be used to obtain new bivariate distributions for different baseline distributions G1(x) and G2(y). The marginal families of distributions of X and Y can be easily obtained from (7) and are

    fX(x)=g1(x)[1+(λ1+λ3){12G1(x)}]&FX(x)=G1(x)[1+(λ1+λ3){1G1(x)}];xR

    and

    fY(y)=g2(y)[1+(λ2+λ3){12G2(y)}]&FY(y)=G2(y)[1+(λ2+λ3){1G2(y)}];yR.

    We can see that the marginal families of distributions are transmuted families.

    The conditional families of distributions for X and Y can be easily obtained and are

    fX|y(x|y)=g1(x)Δ1(y)[1+λ2λ1G1(x)2λ2G2(y)4λ3G1(x)G2(y)];xR (9)

    and

    fY|x(y|x)=g2(y)Δ2(x)[1+λ2λ1G1(x)2λ2G2(y)4λ3G1(x)G2(y)];yR, (10)

    where Δ1(y)=1+(λ2+λ3){12G2(y)} and Δ2(x)=1+(λ1+λ3){12G1(x)}. The marginal and conditional distributions can be studied for any baseline distribution.

    In this section we will discuss some properties of the proposed bivariate family of distributions. These properties are discussed in the following subsections.

    The moments of a random variable are useful in studying its properties. In case of joint distribution of two random variables the joint and ratio moments can be computed. The (r, s)th order joint moment of two random variables is defined as

    μr,sX,Y=E(XrYs)=xrysfX,Y(x,y)dxdy.

    Now, using the joint density function of X and Y, given in (8), the (r, s)th order joint moment for the proposed bivariate family of distributions is

    μr,sX,Y=xrysg1(x)g2(y)[1+λ2λ1G1(x)2λ2G2(y)4λ3G1(x)G2(y)]dxdy,

    which on simplification becomes

    μr,sX,Y=(1+λ)μrXμsYλ1μrX(2:2)μsYλ2μrXμsY(2:2)λ3μrX(2:2)μsY(2:2), (11)

    where μrX is rth raw moment of X, μsY is sth raw moment of Y, μrX(2:2) is rth raw moment of larger observation in a sample of size 2 from G1(x) and μsY(2:2) is sth raw moment of larger observation in a sample of size 2 from G2(y); see for example [22,23].

    The ratio moments are also useful in some studies. The ratio moment for two random variables can be defined in two ways which are given as

    μrsX,Y=E(XrYs)=xrysfX,Y(x,y)dxdy and μsrX,Y=E(YsXr)=xrysfX,Y(x,y)dxdy.

    Using the joint distribution of X and Y the ratio moments μrsX,Y and μsrX,Y are

    μrsX,Y=(1+λ)μrXμsYλ1μrX(2:2)μsYλ2μrXμsY(2:2)λ3μrX(2:2)μsY(2:2), (12)

    and

    μsrX,Y=(1+λ)μrXμsYλ1μrX(2:2)μsYλ2μrXμsY(2:2)λ3μrX(2:2)μsY(2:2), (13)

    where μrX is rth inverse moment of X, μsY is sth inverse moment of Y, μrX(2:2) is rth inverse moment of larger observation in a sample of size 2 from G1(x) and μsY(2:2) is sth inverse moment of larger observation in a sample of size 2 from G2(y).

    The conditional moments are useful in studying the properties of the conditional distributions. In case of two random variables, we can obtain the conditional moments of X given Y = y and conditional moments of Y given X = x. These moments are defined as

    μrX|y=E(Xr|y)=xrfX|y(x|y)dx and μsY|x=E(Ys|x)=ysfY|x(y|x)dy.

    Now, using the conditional distribution of X given Y = y, the rth conditional moment of X given Y = y is

    μrX|y=E(Xr|y)=xrg1(x)Δ1(y)[1+λ2λ1G1(x)2λ2G2(y)4λ3G1(x)G2(y)]dx,

    which on simplification becomes

    μrX|y=1Δ1(y)[(1+λ)μrXλ1μrX(2:2)2{λ2μrX+λ3μrX(2:2)}G2(y)]. (14)

    Similarly, the sth conditional moment of Y given X = x can be obtained by using the conditional distribution of Y given X = x in

    μsY|x=E(Ys|x)=ysfY|x(y|x)dy

    and is given as

    μsY|x=1Δ2(x)[(1+λ)μsYλ1μsY(2:2)2{λ2μsY+λ3μsY(2:2)}G1(x)]. (15)

    The conditional means and conditional variances can be easily obtained from (14) and (15).

    The bivariate reliability and hazard rate functions play important role in reliability analysis, see for example [24] and [25]. The bivariate reliability function is defined as

    R(x,y)=1[FX(x)+FY(y)FX,Y(x,y)].

    Now, using the joint and marginal cdf's for the proposed bivariate family of distributions, the bivariate reliability function is given as

    R(x,y)=S1(x)S2(y)[1bG1(x)dG2(y)λ3G1(x)G2(y)], (16)

    where S1(x)=1G1(x), S2(y)=1G2(y), b=λ1+λ3 and d=λ2+λ3. The bivariate reliability function can be computed for different baseline distributions and different values of the parameters.

    The bivariate hazard rate function is defined as, see [26],

    hX,Y(x,y)=fX,Y(x,y)R(x,y).

    The bivariate hazard rate function for the proposed bivariate family of distributions can be obtained by using (8) and (16) in above equation and is given as

    hX,Y(x,y)=g1(x)g2(y)[1+λ2λ1G1(x)2λ2G2(y)4λ3G1(x)G2(y)]S1(x)S2(y)[1bG1(x)dG2(y)λ3G1(x)G2(y)], (17)

    where b and d are defined earlier.

    The dependence measures are useful to see the strength of dependence between two random variables. Different dependence measures can be computed for a joint probability density function, see for example [11,12,27]. In the following we will discuss some important dependence measures for the new bivariate family of distributions.

    The Kendall's Tau coefficient is a useful measure to see the dependence between two continuous random variables. The coefficient is defined as

    τ=4FX,Y(x,y)fX,Y(x,y)dxdy1.

    Using the joint density and distribution functions of the new bivariate family of distributions, we have

    τ=4G1(x)G2(y)[1+λ1{1G1(x)}+λ2{1G2(y)}+λ3{1G1(x)G2(y)}]×g1(x)g2(y)[1+λ2λ1G1(x)2λ2G2(y)4λ3G1(x)G2(y)]dxdy1.

    Making the transformation u1=G1(x) and u2=G2(y), we have

    τ=41010u1u2[1+λ1(1u1)+λ2(1u2)+λ3(1u1u2)](1+λ2λ1u12λ2u24λ3u1u2)du1du21.

    Solving the above double integral, the Kendall's Tau coefficient for the new bivariate family of distributions is given as

    τ=29[(λ1+λ3)(λ2+λ3)+λ3]. (18)

    The Kendall's Tau coefficient does not depend upon the underlying base distribution.

    The Spearman's Rho coefficient is another useful dependence measure to see the dependence between two random variables. The coefficient is defined as

    ρ=12[FX,Y(x,y)FX(x)FY(y)]fX(x)fY(y)dxdy.

    Using the joint and marginal cdf's and the marginal density functions of X and Y for the new bivariate family of distributions, we have

    ρ=12[{G1(x)G2(y)[1+λ1{1G1(x)}+λ2{1G2(y)}+λ3{1G1(x)G2(y)}]}{G1(x)[1+(λ1+λ3){1G1(x)}]}{G2(y)[1+(λ2+λ3){1G2(y)}]}]×g1(x)[1+(λ1+λ3){12G1(x)}]g2(y)[1+(λ2+λ3){12G2(y)}]dxdy.

    Making the transformation u1=G1(x), u2=G2(y) and solving the resulting integral, we have

    ρ=13[(λ1+λ3)(λ2+λ3)+λ3]. (19)

    We can see that the Spearman's Rho will always be smaller than the Kendall's Tau for the new bivariate family of distributions.

    The local dependence measure for a bivariate family of distributions is defined by [27] as

    γ(x,y)=2xylnfX,Y(x,y),

    which for the new bivariate family of distributions is

    γ(x,y)=4[(λ1+λ3)(λ2+λ3)+λ3]g1(x)g2(y)[1+λ2λ1G1(x)2λ2G2(y)4λ3G1(x)G2(y)]2, (20)

    where λ=λ1+λ2+λ3. We can see that the local dependence measure depends upon the baseline distribution.

    The random sample from the proposed bivariate family of distributions can be generated by using the conditional distribution approach. In this approach the random variate from a bivariate distribution is obtained by using the following steps:

    1) Generate a random observation from the marginal distribution of X, FX(x), and call it x.

    2) Generate a random observation from the conditional distribution of Y given X = x, FY|x(y|x), and call it y.

    3) The pair (x,y) is a random observation from the joint distribution FX,Y(x,y).

    Now, to use this method for the proposed bivariate family of distributions we have following steps:

    a) We first generate a random observation from the marginal distribution of X by solving

    FX(x)=u1orG1(x)[1+b{1G1(x)}]=u1;b=λ1+λ3,

    for x where u1 is a uniform random number. Solving above equation for x, a random observation from the marginal distribution of X is

    x=G1[12b{(1+b)±(1+b)24bu1}]=G11(u1), (21)

    where u1=12b{(1+b)±(1+b)24bu1} and choice between larger or smaller root is done to have an admissible solution.

    b) To apply the second stem, we first obtain the conditional cdf of Y given X = x as

    FY|x(y|x)=1Δ2(x)yg2(y)[1+λ2λ1G1(x)2λ2G2(y)4λ3G1(x)G2(y)]dy=G2(y)Δ2(x)[1+λ2λ1G1(x)λ2G2(y)2λ3G1(x)G2(y)];λ=λ1+λ2+λ3.

    Now, the random observation from the conditional distribution of Y given X = x can be obtained by solving

    G2(y)Δ2(x)[1+λ2λ1G1(x)λ2G2(y)2λ3G1(x)G2(y)]=u2

    for y where u2 is another uniform random observation. The solution of above equation for y is

    y=G12[12λ2(x){λ1(x)±λ21(x)4u2Δ2(x)λ2(x)}]=G12(u2), (22)

    where u2=12λ2(x){λ1(x)±λ21(x)4u2Δ2(x)λ2(x)}, λ1(x)=1+λ2λ1G1(x) and λ2(x)=λ2+2λ3G1(x).

    c) The pair (x,y) given in (21) and (22) is a random observation from the new bivariate family of distributions.

    d) Repeat steps 1–3 n times to obtain a random sample of size n from FX,Y(x,y) given in (7).

    A stress-strength reliability analysis is very useful in engineering and survival analysis, see for example [28]. The reliability of the stress-strength model is computed as R=P(Y<X) where Y is stress and X is strength. Now, if the stress (Y) and strength (X) are dependent and have joint bivariate distribution given in (8) then the reliability is given as

    R=P(Y<X)=0x0g1(x)g2(y)[1+λ2λ1G1(x)2λ2G2(y)4λ3G1(x)G2(y)]dydx,

    assuming that both X and Y are positive random variables. Simplifying above integral, the stress-strength reliability of two components X and Y having joint distribution given in (8) is

    R=0g1(x)G2(x)[(1+λ)2λ1G1(x)λ2G2(x)2λ3G1(x)G2(x)]dx. (23)

    Further, if X and Y have identical distributions, that is if G1(x)=G2(x)=G(x) then the reliability is

    R=0g(x)G(x)[(1+λ)2λ1G(x)λ2G(x)2λ3G2(x)]dx,

    which on simplification becomes

    R=16(3λ1+λ2). (24)

    We can see that the reliability in case of identical distributions of X and Y does not depend upon the baseline distribution.

    In this section we will discuss the maximum likelihood estimation of the parameters for the new bivariate family of distributions under the assumption that all the parameters of baseline distributions G1(x) and G2(y) are known. For this, suppose X1,X2,,Xn is a random sample of size n from the bivariate transmuted family of distributions. The likelihood function is

    L(λ1,λ2,λ3;x)=ni=1g1(xi)g2(yi)[1+λ2λ1G1(xi)2λ2G2(yi)4λ3G1(xi)G2(yi)],

    where λ=λ1+λ2+λ3. The log-likelihood function is

    =ni=1lng1(xi)+ni=1lng2(yi)+ni=1ln[1+λ2λ1G1(xi)2λ2G2(yi)4λ3G1(xi)G2(yi)]. (25)

    The derivatives of log-likelihood function with respect to λ1, λ2 and λ3 are

    λ1=ni=112G1(xi)[1+λ2λ1G1(xi)2λ2G2(yi)4λ3G1(xi)G2(yi)], (26)
    λ2=ni=112G2(yi)[1+λ2λ1G1(xi)2λ2G2(yi)4λ3G1(xi)G2(yi)] (27)

    and

    λ3=ni=114G1(xi)G2(yi)[1+λ2λ1G1(xi)2λ2G2(yi)4λ3G1(xi)G2(yi)]. (28)

    The maximum likelihood estimators of λ1, λ2 and λ3 are obtained by equating (26)–(28) to zero and numerically solving the resulting equations.

    In the following section we will give a multivariate extension of the bivariate family of distributions discussed above.

    A multivariate extension of the bivariate T-X family of distributions, given in (5), is

    Fx(x)=Wp[Gp(xp)]apW1[G1(x)]a1r(v1,,vp)dv1dv2;ai<vi<bi,i=1,2,,p,

    where x=(X1,X2,,Xp) is a vector of p random variables and r(v1,,vp) is any multivariate distribution with support on [a1,b1]××[ap,bp]. A simpler version of above family has the following form when domain of r(v1,,vp) is on [0,1]p

    Fx(x)=Gp(xp)0G1(x)0r(v1,,vp)dv1dv2;0<vi<1;i=1,2,,p. (29)

    The multivariate T-X family of distributions, given in (29), can be used to obtain new multivariate families of distributions for suitable choices of r(v1,,vp). We have proposed a new multivariate family of distributions by using

    r(v1,,vp)=1+pi=1λi(12vi)+λp+1(12ppi=1vi)

    in (29). The joint cdf of the new multivariate family of distribution is

    Fx(x)=Gp(xp)0G1(x)0{1+pi=1λi(12vi)+λp+1(12ppi=1vi)}dv1dv2

    which on simplification becomes

    Fx(x)={pi=1Gi(xi)}[1+pi=1λi{1Gi(xi)}+λp+1{1pi=1Gi(xi)}];xRp, (30)

    where Gi(xi) is marginal cdf of ith random variable Xi and (λ1,λ2,,λp+1) are the parameters such that p+1i=1λi1, pi=1λi+(p+1)λp+11, λi[1,1] and 1λi+λp+11 for i=1,2,,p. The joint density function of the new multivariate family of distributions is

    fx(x)={pi=1gi(xi)}[1+λ2pi=1λiGi(xi)2pλp+1pi=1Gi(xi)];xRp, (31)

    where gi(x) is the density function corresponding to Gi(xi) and λ=p+1i=1λi. It can be easily seen that the marginal density function of any random variable, Xi, for the joint density (31) is same as given in (1) with parameter λi+λp+1. We can also see that the joint marginal distribution for any pair of random variables, (Xi,Xm), for the joint density (31) is the same as given in (8).

    The marginal density function of a subset of random variables x1=(X1,X2,,Xt) for the new multivariate family of distributions is readily written from (31) as

    fx1(x1)=ti=1gi(xi)[1+λ2ti=1λiGi(xi)2tλp+1ti=1Gi(xi)];x1Rt. (32)

    The conditional density function of any random variable Xi given the information of all other random variables is easily obtained by using

    fXi|x(i)(xi|x(i))=fx(x)fx(i)(x(i)),

    where x(i) is the random vector x without Xi. The conditional distribution of Xi given the information of all other variables for the proposed multivariate family of distributions is

    fXi|x(i)(xi|x(i))=gi(xi)Δi(x(i))[1+λ2pi=1λiGi(xi)2pλp+1pi=1Gi(xi)];xiR, (33)

    where Δi(x(i))=[1+phi=1λh{12Gh(xh)}+λp+1{12p1phi=1Gh(xh)}].

    The joint conditional distribution of any pair of random variables (Xi,Xm) given the information of all other variables is obtained by using

    fXi,Xm|x(i,m)(xi,xm|x(i,m))=fx(x)fx(i,m)(x(i,m)),

    where x(i,m) is the random vector x without Xi and Xm. The joint conditional distribution of (Xi,Xm) given the information of all other variables for the proposed multivariate family of distributions is

    fXi,Xm|x(i,m)(xi,xm|x(i,m))=gi(xi)gm(xm)Δi,m(x(i,m))[1+λ2pi=1λiGi(xi)2pλp+1pi=1Gi(xi)], (34)

    for (xi,xm)R2 where

    Δi,m(x(i,m))=[1+phi,m=1λh{12Gh(xh)}+λp+1{12p2phi,m=1Gh(xh)}].

    In general, the conditional distribution of one subset of variables x1 given the information of other subset of variables x2 for the new multivariate family of distributions is

    fx1|x2(x1|x2)=1Δ2(x2){ti=1gi(xi)}[1+λ2pi=1λiGi(xi)2pλp+1pi=1Gi(xi)], (35)

    where Δ2(x2)=[1+ph=t+1λh{12Gh(xh)}+λp+1{12ptph=t+1Gh(xh)}].

    Some properties of the new multivariate family of distributions are given in the following.

    The marginal and joint moments for the proposed multivariate family of distributions can be easily obtained in the same way as discussed above in the case of the new bivariate family of distributions. The conditional moments of a single random variable given the information of all other variables can be obtained by using

    μriXi|x(i)=E(Xrii|x(i))=xriifXi|x(i)(xi|x(i))dxi.

    Now, using the conditional distribution of xi given the information of all other variables from (33), we have

    μriXi|x(i)=1Δi(x(i))xriigi(xi)[1+λ2pi=1λiGi(xi)2pλp+1pi=1Gi(xi)]dxi

    or

    μriXi|x(i)=1Δi(x(i))[(1+λ2phi=1λhGh(xh))μriXi{λi2p1λp+1phi=1Gh(xh)}μriXi(2:2)]. (36)

    The joint conditional moments of (Xi,Xm) given the information of other random variables can be obtained by using

    μri,rmXi,Xm=xriixrmmfXi,Xm|x(i,m)(xi,xm|x(i,m))dxidxm.

    Now, using the joint conditional distribution from (34), we have

    μri,rmXi,Xm=1Δi,m(x(i,m))xriixrmmgi(xi)gm(xm)[1+λ2pi=1λiGi(xi)2pλp+1pi=1Gi(xi)]dxidxm.

    Simplifying above integral, the joint conditional moment of (Xi,Xm) given the information of other random variables is

    μri,rmXi,Xm=1Δi,m(x(i,m))[(1+λ2phi,m=1λhGh(xh))μriXiμrmxmλiμriXi(2:2)μrmXmλmμriXiμrmXm(2:2)2p2λp+1μriXi(2:2)μrmXm(2:2)phi,m=1Gi(xi)]. (37)

    The conditional means, variances and conditional covariances can be obtained from (36) and (37).

    In the following we have given the multivariate dependence measures for the proposed multivariate family of distributions. We have discussed two multivariate dependence measures which are the Kendall's Tau and Spearman's Rho. These measures are given in the following.

    The multivariate extension of Kendall's Tau coefficient is given as see [29],

    τp=2p2p11Fx(x)fx(x)dx1dxp12p11.

    Now, using the joint distribution and density functions of the proposed multivariate family of distributions, from (30) and (31), we have

    τp=2p2p11{pi=1Gi(xi)}[1+pi=1λi{1Gi(xi)}+λp+1{1pi=1Gi(xi)}]×{pi=1gi(xi)}[1+λ2pi=1λiGi(xi)2pλp+1pi=1Gi(xi)]dx1dxp12p11.

    Making the transformation Gi(xi)=ui, we have

    τp=2p2p111010(pi=1ui)[1+pi=1λi(1ui)+λp+1(1pi=1ui)]×[1+pi=1λi(1ui)2pλp+1pi=1ui]dx1dxp12p11.

    Solving above multiple integrals, the Kendall's Tau coefficient for the proposed multivariate family of distributions is

    τp=12p11[29pi=1pk=i+1λiλk2pλp+14(2p23p)pi=1λi(2p2×3p+22p3p)λp+1(1+λp+1)]. (38)

    We can see that for p = 2 the expression (38) reduces to (18) as it should be.

    The multivariate extension of Spearman's Rho coefficient is given as see [30],

    ρp=2p(p+1)2p(p+1)Fx(x){pi=1fi(xi)}dxidxp(p+1)2p(p+1).

    Now, using the joint distribution function and marginal density functions of individual variables we have

    ρp=2p(p+1)2p(p+1){pi=1Gi(xi)}[1+pi=1λi{1Gi(xi)}+λp+1{1pi=1Gi(xi)}]×pi=1[gi(xi){1+(λi+λp+1)[12Gi(xi)]}]dx1dxp(p+1)2p(p+1)

    Making the transformation ui=Gi(xi), we have

    ρp=2p(p+1)2p(p+1)1010(pi=1ui)[1+pi=1λi(1ui)+λp+1(1pi=1ui)]×pi=11+(λi+λp+1)(12ui)dx1dxp(p+1)2p(p+1)

    Simplifying above multiple integral, the Spearman's Rho for the proposed multivariate family of distributions is

    ρp=p+13p[2p(p+1)][pi=1(3λiλp+1){1+pi=1λi+λp+13λiλp+1}3p]. (39)

    It can be easily seen that (39) reduces to (19) for p = 2 as it should be.

    In the following we have given maximum likelihood estimator for parameters of the proposed multivariate family of distributions when all the parameters of the base distribution are known. For this, we assume that a random sample of n vector observations is available from the joint density function (31). The likelihood function is

    L(λ;x1,x2,,xn)=nj=1{pi=1gi(xij)}nj=1[1+pi=1λi{12Gi(xij)}+λp+1{12ppi=1Gi(xij)}],

    where λ=(λ1,λ2,,λp+1) is vector of unknown parameters. The log of likelihood function is

    (λ)=nj=1pi=1lngi(xij)+nj=1ln[1+pi=1λi{12Gi(xij)}+λp+1{12ppi=1Gi(xij)}].

    The derivatives of log of likelihood function for λi;i=1,2,,p and λp+1 are

    (λ)λi=nj=112Gi(xij)[1+pi=1λi{12Gi(xij)}+λp+1{12ppi=1Gi(xij)}];i=1,2,,p (40:1–p)

    and

    (λ)λp+1=nj=112ppi=1Gi(xij)[1+pi=1λi{12Gi(xij)}+λp+1{12ppi=1Gi(xij)}]. (41)

    The maximum likelihood estimators can be obtained by numerically solving above (p+1) equations.

    We will now study a special case of the proposed bivariate and multivariate families of distributions for Weibull baseline distribution.

    In this section we have discussed new bivariate and multivariate Weibull distributions by using Weibull baseline distribution, proposed by [31], in the new bivariate and multivariate families of distributions proposed above.

    In the following we have given a new bivariate Weibull distribution by using following baseline Weibull distributions for X and Y in (7) and (8)

    G1(x)=1e(xθ1)α1;x,θ1,α1>0 and G2(y)=1e(yθ2)α2;y,θ2,α2>0.

    The joint distribution function of the new bivariate Weibull distribution is

    FX,Y(x,y)=[1e(xθ1)α1][1e(yθ2)α2][1+be(xθ1)α1+de(yθ2)α2λ3e(xθ1)α1(yθ2)α2], (42)

    where (x,y,α1,α2,θ1,θ2)>0. Also (λ1,λ2,λ3)[1,1], λ1+λ2+λ31, λ1+λ2+3λ31, 1λ1+λ31 and 1λ2+λ31. The density function corresponding to (42) is

    fX,Y(x,y)=α1α2θα11θα22xα11yα21e(xθ1)α1e(yθ2)α2[1+λ1{2e(xθ1)α11}+λ2{2e(yθ2)α21}λ3{34e(xθ1)α14e(yθ2)α2+4e(xθ1)α1(yθ2)α2}];x,y>0,

    or

    fX,Y(x,y)=α1α2θα11θα22xα11yα21e(xθ1)α1e(yθ2)α2ΔBW(x,y);x,y>0, (43)

    where ΔBW(x,y)=[1λ+2λ13e(xθ1)α1+2λ23e(yθ2)α22λ3{1+2e(xθ1)α1(yθ2)α2}], λ13=(λ1+2λ3) and λ23=(λ2+2λ3). The marginal distributions of X and Y are immediately written from (43) as

    fX(x)=α1θα11xα11e(xθ1)α1[1(λ1+λ3)+2(λ1+λ3)e(xθ1)α1];x>0

    and

    fY(y)=α2θα22yα21e(yθ2)α2[1(λ2+λ3)+2(λ2+λ3)e(yθ2)α2];y>0.

    The conditional distributions of X given Y = y and of Y given X = x are readily obtained from (9) and (10) and are

    fX|y(x|y)=1ΔBW(y)α1θα11xα11e(xθ1)α1ΔBW(x,y) (44)

    and

    fY|x(y|x)=1ΔBW(x)α2θα22yα21e(yθ2)α2ΔBW(x,y), (45)

    where ΔBW(x)=1+b2be(xθ1)α1 and ΔBW(y)=1+d2de(yθ2)α2. The conditional distributions are useful in studying the conditional behavior of one random variable given the information of other.

    Some plots of the joint density function are given in Table 1 below for θ1=θ2=1 and for different combinations of other parameters.

    Table 1.  The density function of new bivariate Weibull distribution.
    α1=0.5,α2=0.5 α1=0.5,α2=1.5 α1=1.5,α2=0.5 α1=1.5,α2=1.5
    λ1=0.35λ2=0.35λ3=0.15
    λ1=0.25λ2=0.30λ3=0.50
    λ1=0.50λ2=0.35λ3=0.25
    λ1=0.35λ2=0.45λ3=0.45

     | Show Table
    DownLoad: CSV

    From above graph we can see that the shape is controlled by parameters α1 and α2. The joint hazard rate function of the distribution is immediately written from (17) as

    hX,Y(x,y)=α1α2xα11yα21ΔBW(x,y)θα11θα22ΔBW(x,y), (46)

    where ΔBW(x,y)=[1λ+λ13e(xθ1)α1+λ23e(yθ2)α2λ3{2+e(xθ1)α1(yθ2)α2}]. The hazard rate function can be plotted for different combinations of the parameters.

    Some important properties of the proposed bivariate Weibull distribution are in the following.

    We have seen that expressions for joint, marginal and conditional moments of the new bivariate family of distributions involve moments of baseline distribution and moments of larger observation in a sample of size 2 from the baseline distribution. It is known that the rth moment and rth moment of larger observation in a sample of size 2 from Weibull distribution with density

    f(w)=αθαwα1e(wθ)α;w>0

    are given as

    μrW=θrΓ(1+rα) and μrW(2:2)=2θrΓ(rα+1)[12(rα+1)].

    Using these expressions in (11) the joint moments for the new bivariate Weibull distribution is

    μr,sX,Y=θr1θs2Γ(1+rα1)Γ(1+sα2)[1λ+λ132rα1+λ232sα2λ3{2+2(rα1+sα2)}]. (47)

    Again, using simple moments and moments of order statistics of Weibull distribution in (14) and (15), the conditional moments for the new Weibull distribution are

    μrX|y=1ΔBW(y)θr1Γ(1+rα1)[1λ+λ132rα1+2λ23e(yθ2)α22λ3{1+2rα1e(yθ2)α2}] (48)

    and

    μsY|x=1ΔBW(x)θs2Γ(1+sα2)[1λ+λ232sα2+2λ13e(xθ1)α12λ3{1+2sα2e(xθ1)α1}]. (49)

    The conditional means and conditional variances can be obtained from (48) and (49).

    We have discussed different dependence measures in Section 4.4. We have seen that two of the dependence measures, namely Kendall's Tau and Spearman's Rho do not depend upon a specific baseline distribution. We have also seen that the local dependence measure, given (20), depends upon a specific distribution. The local dependence measure for the new bivariate Weibull distribution can be obtained by using the density and distribution functions of Weibull random variables in (20) and is

    γ(x,y)=4[(λ1+λ3)(λ2+λ3)+λ3]α1α2xα11yα21e(xθ1)α1e(yθ2)α2θα11θα22Δ2BW(x,y). (50)

    The local dependence measure can be computed for specific values of the parameters. We can see that γ(x,y) is positive if (λ1+λ3)(λ2+λ3)<λ3 and is negative otherwise.

    In the following we have obtained the stress-strength reliability if the stress (Y) and strength (X) have joint bivariate distribution given in (43). We have seen that the stress-strength reliability of for the new bivariate family of distributions is given in (23) as

    R=0g1(x)G2(x)[(1+λ)2λ1G1(x)λ2G2(x)2λ3G1(x)G2(x)]dx.

    Now, using G1(x)=1e(xθ1)α1 and G2(x)=1e(xθ2)α2, the reliability for the new bivariate Weibull distribution is

    R=0α1θα11xα11e(xθ1)α1{1e(xθ2)α2}[(1+λ)2λ1{1e(xθ1)α1}λ2{1e(xθ2)α2}2λ3{1e(xθ1)α1}{1e(xθ2)α2}]dx

    or

    R=0α1θα11xα11e(xθ1)α1{1e(xθ2)α2}[(1λ1λ3)2λ1e(xθ1)α1λ2e(xθ2)α2+2λ3e(xθ1)α1+2λ3e(xθ2)α22λ3e(xθ1)α1e(xθ2)α2]dx.

    Simplifying the above integral, the reliability coefficient for the new bivariate Weibull distribution is

    R=λ3λ1+r=0(θ1θ2)rα2Γ(1+α2rα1)[(2)rr!(λ22λ3+λ32α2rα1)(1)rr!(λ22λ3+λ32α2rα1+1λ12α2rα1)]. (51)

    The reliability coefficient can be computed for specific values of the parameters.

    A random sample from the proposed bivariate Weibull distribution can be obtained by using the algorithm given in Section 4.5. An algorithm to generate a random sample from the proposed bivariate Weibull distribution is listed below.

    1) Generate a random observation from the marginal distribution of X by using

    x=θ1[ln(1u1)]1α1, (52)

    where u1=12b{(1+b)±(1+b)24bu1} and choice between larger or smaller root is done to have an admissible solution.

    2) Generate a random observation from the conditional distribution of Y given X = x by using

    y=θ2[ln(1u2)]1α2, (53)

    where u2=12λ2(x){λ1(x)±λ21(x)4u2Δ2(x)λ2(x)}, λ1(x)=1+λ2λ1{1e(xθ1)α1}, λ=λ1+λ2+λ3 and λ2(x)=λ2+2λ3{1e(xθ1)α1}.

    3) The pair (x, y) in (52) and (53) is a random observation from the proposed bivariate Weibull distribution.

    In the following we will discuss maximum likelihood estimation for parameters of the new bivariate Weibull distribution. For this suppose that a random sample of n observations is available from the joint density given in (43). The likelihood function is

    L(x,y;θ)=αn1αn2θnα11θnα22nj=1xα11jnj=1yα21jenj=1(xjθ1)α1enj=1(yjθ2)α2nj=1ΔBW(xj,yj).

    where θ=(α1,α2,θ1,θ2,λ1,λ2,λ3) and

    ΔBW(xj,yj)=[1λ+2λ13e(xjθ1)α1+2λ23e(yjθ2)α22λ3{1+2e(xjθ1)α1(yjθ2)α2}].

    The log of likelihood function is

    (θ)=nlnα1+nlnα2nα1lnθ1nα2lnθ2+(α11)nj=1lnxj+(α21)nj=1lnyjnj=1(xjθ1)α1nj=1(yjθ2)α2+nj=1lnΔBW(xj,yj). (54)

    The derivatives of log-likelihood function with respect to the unknown parameters are

    (θ)θ1=nα1θ1α1θ21nj=1xj(xjθ1)α11+2α1θ1nj=1(xjθ1)α1e(xjθ1)α1ΔBW(xj,yj)[λ1+2λ32λ3e(yjθ2)α2], (55)
    (θ)θ2=nα2θ2α2θ22nj=1yj(yjθ2)α21+2α2θ2nj=1(yjθ2)α2e(yjθ2)α2ΔBW(xj,yj)[λ2+2λ32λ3e(xjθ1)α1], (56)
    (θ)α1=nα1+nj=1lnxjnlnθ1nj=1(xjθ1)α1ln(xjθ1)2nj=1(xjθ1)α1ln(xjθ1)ΔBW(xj,yj)×e(xjθ1)α1[(λ1+2λ3)2λ3e(yjθ2)α2], (57)
    (θ)α2=nα2+nj=1lnyjnlnθ2nj=1(yjθ2)α2ln(yjθ2)2nj=1(yjθ2)α2ln(yjθ2)ΔBW(xj,yj)×e(yjθ2)α2[(λ2+2λ3)2λ3e(xjθ1)α1], (58)
    (θ)λ1=nj=11ΔBW(xj,yj)[2e(xjθ1)α11], (59)
    (θ)λ2=nj=11ΔBW(xj,yj)[2e(yjθ2)α21] (60)

    and

    (θ)λ3=nj=11ΔBW(xj,yj)[4{e(xjθ1)α1+e(yjθ2)α2e(xjθ1)α1(yjθ2)α2}3]. (61)

    The maximum likelihood estimator of parameter vector θ=(α1,α2,θ1,θ2,λ1,λ2,λ3) can be obtained by numerically solving (55)–(61).

    We will, now, briefly discuss a new multivariate Weibull distribution by using the proposed multivariate family of distribution.

    In the following we have given a new multivariate Weibull distribution by using following baseline Weibull distributions for Xi in (30) and (31)

    Gi(xi)=1e(xiθi)αi;xi,θi,αi>0.

    The joint distribution function of the new multivariate Weibull distribution is

    Fx(x)=[pi=1{1e(xiθi)αi}][1+pi=1λie(xiθi)αi+λp+1(1pi=1{1e(xiθi)αi})];xRp+. (62)

    The density function corresponding to (60) is

    fx(x)=[pi=1αiθαiixαi1ie(xiθi)αi][1pi=1λi2pi=1λie(xiθi)αi+λp+1(12ppi=1{1e(xiθi)αi})]. (63)

    The marginal and conditional distributions can be easily obtained by using the methods given in Section 5. In particular the joint marginal distribution of any pair of random variables is same as given in (43). Further, the conditional distribution of any random variable given the information of other random variables can be obtained by using (33). In particular, the joint conditional distribution of any two random variables given the information of all other variables can be obtained by using (34) and is

    fXi,Xm|x(i,m)(xi,xm|x(i,m))=1Δi,m(x(i,m))[αiαmθαiiθαmmxαi1ixαm1me(xiθi)αi(xmθm)αm]ΔMW(x);xi,xm>0, (64)

    where

    ΔMW(x)=[1pi=1λi2pi=1λie(xiθi)αi+λp+1(12ppi=1{1e(xiθi)αi})]

    and

    Δi,m(x(i,m))=[1+phi,m=1λh{2e(xhθh)αh1}+λp+1{12p2phi,m=1[1e(xhθh)αh]}].

    The joint conditional moments for any two random variables given the information of all other variables can be obtained by using (37). Now, using the raw moments of a Weibull random variable and moments of maximum in a sample of size 2 from the Weibull distribution, the joint conditional moments between Xi and Xm given the information of other random variables for the proposed multivariate Weibull distribution is

    μri,rmXi,Xm==1Δi,m(x(i,m))θriiθrmmΓ(1+riαi)Γ(1+rmαm)[Δi,m(x(i,m))2λi{12(riαi+1)}2λm{12(rmαm+1)}2p2λp+1{12(riαi+1)}{12(rmαm+1)}phi,m=1{1e(xhθh)αh}], (65)

    where Δi,m(x(i,m))=1+λ2phi,m=1λh{1e(xhθh)αh}.

    The conditional means, conditional variances and conditional covariances can be obtained from (65). We will now give some numerical study about the proposed bivariate Weibull distribution.

    In this section we have given some numerical study for the proposed bivariate Weibull distribution. The study contains simulation study and real data application. These are given in the following subsections.

    In this section we will give a simulation study for the proposed bivariate Weibull distribution. The simulation algorithm is given below.

    1) Generate a random sample of specific size n, by using different values of the parameters, from the bivariate Weibull distribution by using the steps given in Sub-section 6.1.4.

    2) Compute maximum likelihood estimates of the parameters by using the sample obtained in Step 1.

    3) Repeat Steps 1 and 2 for a specific number of times, say M.

    4) Obtain the simulated estimate and simulated mean square error of the estimate by using

    ˆθk=M1Mj=1ˆθkjandMSE(ˆθk)=(M1)1Mj=1(ˆθkjˆθk)2,

    where ˆθk is kth parameter and ˆθkj is kth parameter obtained from jth sample.

    We have conducted simulation study by using samples of sizes 20, 50,100,200 and 500 and the number of simulations used are 20000. The results are given in Table 2 below.

    Table 2.  Simulation results for the new bivariate Weibull distribution.
    θ1=1.5 θ2=2.5 α1=1.25 α2=1.75 λ1=0.25 λ2=0.35 λ3=0.45
    20 1.496 2.5 1.253 1.752 0.284 0.328 -0.455
    (0.658) (1.132) (0.551) (0.778) (0.107) (0.149) (0.194)
    50 1.498 2.497 1.242 1.751 0.283 0.322 -0.455
    (0.270) (0.443) (0.215) (0.316) (0.043) (0.061) (0.078)
    100 1.497 2.496 1.250 1.745 0.212 0.369 -0.451
    (0.137) (0.217) (0.108) (0.152) (0.021) (0.032) (0.038)
    200 1.497 2.501 1.248 1.745 0.281 0.353 -0.446
    (0.068) (0.108) (0.055) (0.077) (0.011) (0.016) (0.02)
    500 1.504 2.499 1.247 1.745 0.237 0.334 -0.465
    (0.026) (0.044) (0.023) (0.03) (0.005) (0.006) (0.008)
    1000 1.503 2.502 1.249 1.751 0.243 0.344 -0.457
    (0.025) (0.043) (0.023) (0.029) (0.005) (0.005) (0.008)
    θ1=2.5 θ2=1.7 α1=2.25 α2=1.5 λ1=0.25 λ2=0.45 λ3=0.25
    20 2.496 1.695 2.247 1.505 -0.282 0.428 0.272
    (1.064) (0.756) (0.938) (0.679) (0.107) (0.199) (0.113)
    50 2.498 1.697 2.252 1.495 -0.262 0.439 0.242
    (0.423) (0.289) (0.399) (0.273) (0.046) (0.076) (0.044)
    100 2.5 1.705 2.246 1.506 -0.266 0.435 0.27
    (0.218) (0.147) (0.197) (0.136) (0.023) (0.04) (0.021)
    200 2.501 1.702 2.251 1.495 -0.213 0.453 0.212
    (0.112) (0.076) (0.102) (0.064) (0.012) (0.02) (0.011)
    500 2.499 1.698 2.252 1.503 -0.217 0.452 0.266
    (0.042) (0.03) (0.041) (0.027) (0.005) (0.008) (0.005)
    1000 2.502 1.703 2.251 1.504 -0.223 0.451 0.258
    (0.043) (0.03) (0.039) (0.024) (0.005) (0.007) (0.005)

     | Show Table
    DownLoad: CSV

    From results of above table, we can see that the estimated values are close to pre-specified values of the parameters and hence the maximum likelihood estimates are consistent. We can also see that the mean square error (in parenthesis) of the estimates decreases with increase in the sample size.

    We will now give a real data application to compare the proposed bivariate Weibull distribution with some available bivariate distributions.

    In the following a real data application will be given where the proposed bivariate Weibull distribution is compared with some available bivariate distributions. We have used two data sets for comparison of the proposed bivariate distribution with some existing bivariate distributions. The first data set that we have used is the mammal brain data (Data 1) that has been obtained from [32]. This data set contains body weight of mammals (X) and the brain weigh (Y) of 84 mammals. The second data set that we have used is the child smoker data (Data 2) that has been obtained from [34]. This data set contains information about forced expiratory volume (X) and height (Y) of 655 children. Some summary measures of the data are given in Table 3 below.

    Table 3.  Summary measures for the two data sets.
    Data Min Q1 Median Mean Q3 Max Variance Skewness
    1 X 0.02 0.96 6.00 38.79 45.00 250.00 4053.22 1.886
    Y 0.45 9.82 50.00 106.92 183.50 442.00 15127.71 1.102
    2 X 0.79 1.98 2.55 2.64 3.12 5.79 0.75 0.66
    Y 117.00 145.00 156.00 155.30 166.0 188.00 209.43 –0.21

     | Show Table
    DownLoad: CSV

    The summary measures indicates that both of the variables in data 1 are positively skewed whereas for second data one variable is positively skewed and one is negatively skewed. We have fitted the proposed bivariate Weibull distribution on this data alongside three more bivariate distributions. The other distributions fitted on the data are a bivariate exponential distribution by [33], a bivariate Weibull distribution by [34] and a bivariate Weibull distribution obtained by using the Gumbel copula. The joint density functions of these bivariate distributions are Bivariate Exponential [33]

    f(x,y)=αx exp[x(α+y)];x,y,α>0. (66)

    Bivariate Weibull [34]

    f(x,y)=α1α2x2α11yα21exp[xα1(1+yα2)];x,y,α1,α2>0. (67)

    Bivariate Weibull with Gumbel copula

    f(x,y)=α1α2β1β2xα11yα21exp[(xα1β1+yα2β2)][1+γ{2exp(xα1β1)1}{2exp(yα2β2)1}], (68)

    where x>0,y>0,α1>0,α2>0,β1>0,β2>0 and 1γ1.

    The distributions are fitted by computing the maximum likelihood estimates of the parameters of various distributions. The maximum likelihood estimation has been done by using "maxLik" package of R, [35]. The maximum likelihood estimates alongside the standard errors are given in the Table 4 below.

    Table 4.  Maximum likelihood estimates for various distributions.
    Distribution Parameter Estimate SE Estimate SE
    Brains Data Smoker's Data
    The New Bivariate Weibull θ1 16.3241 2.1523 2.9194 0.0234
    θ2 47.3533 1.2427 119.2692 1.1326
    α1 0.4767 0.0307 3.5293 0.1432
    α2 0.5873 0.0407 1.7304 0.0605
    λ1 0.9995 0.2514 0.9970 0.6769
    λ2 0.2206 0.0842 –0.2819 0.0844
    λ3 –0.9998 0.3800 –0.7178 0.0756
    Bivariate Exponential, [33] α 0.0258 0.0028 0.3793 0.0148
    Bivariate Weibull, [34] α1 0.1378 0.0118 0.4433 0.0136
    α2 0.1375 0.0114 0.1152 0.0035
    Gumbel Bivariate Weibull α1 0.4712 0.0438 3.2904 0.0308
    α2 0.6643 0.0310 1.2238 0.0113
    β1 4.1226 0.8869 36.3390 3.2536
    β2 19.6784 2.5634 447.3664 0.6409
    γ 0.9999 0.6035 0.9958 0.5890

     | Show Table
    DownLoad: CSV

    We have next computed some goodness of fit measures to decide about the most suitable model for the data. These summary measures are given in Table 5 below.

    Table 5.  Goodness of fit measures for various distributions.
    Data Set Distribution Log Likelihood AIC BIC
    1 The New Bivariate Weibull –772.7114 1559.423 1558.893
    Bivariate Exponential, [33] –86512.922 173027.8 173027.8
    Bivariate Weibull, [34] –996.2986 1996.597 1996.446
    Gumbel Bivariate Weibull –782.7159 1575.432 1575.053
    2 The New Bivariate Weibull –4214.01 8442.02 8473.41
    Bivariate Exponential, [33] –275602.100 551206.20 551210.68
    Bivariate Weibull, [34] –7697.799 15399.60 15408.57
    Gumbel Bivariate Weibull –4594.979 9199.96 9222.38

     | Show Table
    DownLoad: CSV

    From Table 5, we can see that the proposed bivariate Weibull distribution is the best fit for both of the data sets as it has smallest value of AIC and BIC. We have also plotted the bivariate histograms of two data sets alongside the fitted distribution and are given in Table 6, below.

    Table 6.  Bivariate histograms and fitted distributions for two data sets.
    Data Observed Histogram Fitted Distribution
    Mammal Brain Data
    Child Smoker Data

     | Show Table
    DownLoad: CSV

    From above figures in Table 6 we can see that the fitted distributions are reasonable fit to the two data sets.

    In this paper we have proposed new bivariate and multivariate families of distributions. These families of distributions are useful to generate the new bivariate and multivariate distributions from the univariate marginals. We have studied some important properties of the proposed families of distributions. We have seen that the moments of the proposed families of distributions depend upon the raw moments and moments of order statistics for the baseline distributions. The measures of dependence of the proposed families have also been discussed. We have also given the stress-strength reliability analysis for the proposed bivariate family of distributions. We have studied the proposed bivariate and multivariate families of distributions for the baseline Weibull distribution and have proposed new bivariate and multivariate Weibull distributions. We have studied some useful properties of the proposed bivariate Weibull distribution including moments, reliability and maximum likelihood estimation of the parameters. The simulation study and real data application of the proposed bivariate Weibull distribution has been done by using the mammal brain data. We have seen that the proposed bivariate Weibull distribution is the best fit for the data used. The proposed bivariate and multivariate families of distributions can be explored for different baseline distributions.

    The authors declare that they have no conflicts of interest.



    [1] Fernando JA, Francisco A, Manuel AA, et al. (2005) Effects of terrain morphology, sampling density, and interpolation methods on grid DEM accuracy. Photogramm Eng Remote Sens 71: 805–816. https://doi.org/10.14358/pers.71.7.805 doi: 10.14358/pers.71.7.805
    [2] Grohmann CH, Smith MJ, Riccomini C (2011) Multiscale analysis of topographic surface roughness in the Midland Valley, Scotland. IEEE Trans Geosci Remote Sens 49: 1200–1213. https://doi.org/10.1109/tgrs.2010.2053546 doi: 10.1109/tgrs.2010.2053546
    [3] Peter JB, Vernon HS, Jiro S, et al. (2004) Methods for Remote Engineering Geology Terrain Analysis in Boreal Forest Regions of Ontario, Canada. Environ Eng Geosci 10: 229–241. https://doi.org/10.2113/10.3.229 doi: 10.2113/10.3.229
    [4] Fan L, Atkinson PM (2015) Accuracy of digital elevation models derived from terrestrial laser scanning data. IEEE Geosci Remote Sens Lett 12: 1923-1927. https://doi.org/10.1109/LGRS.2015.2438394 doi: 10.1109/LGRS.2015.2438394
    [5] Fan L (2014) Uncertainty in terrestrial laser scanning for measuring surface movements at a local scale. University of Southampton: Southampton, UK.
    [6] Fan L, Powrie W, Smethurst JA, et al. (2014) The effect of short ground vegetation on terrestrial laser scans at a local scale. ISPRS J Photogramm Remote Sens 95: 42–52. https://doi.org/10.1016/j.isprsjprs.2014.06.003 doi: 10.1016/j.isprsjprs.2014.06.003
    [7] Frankel KL, Dolan JF (2007) Characterizing arid region alluvial fan surface roughness with airborne laser swath mapping digital topographic data. J Geophys Res 112: F02025. https://doi.org/10.1029/2006JF000644 doi: 10.1029/2006JF000644
    [8] Nield JM, Wiggs GFS (2011) The application of terrestrial laser scanning to aeolian saltation cloud measurement and its response to changing surface moisture. Earth Surf Processes Landforms 36: 273–278. https://doi.org/10.1002/esp.2102 doi: 10.1002/esp.2102
    [9] Kristen MB, Wayne LM, Patrick JD, et al. (2013) The Use of LiDAR Terrain Data in Characterizing Surface Roughness and Microtopography. Appl Environ Soil Sci 2013: 891534. https://doi.org/10.1155/2013/891534 doi: 10.1155/2013/891534
    [10] Milenković M, Pfeifer N, Glira P (2015) Applying Terrestrial Laser Scanning for Soil Surface Roughness Assessment. Remote Sens 7: 2007–2045. https://doi.org/10.3390/rs70202007 doi: 10.3390/rs70202007
    [11] Fan L (2020) A comparison between structure-from-motion and terrestrial laser scanning for deriving surface roughness: A case study on a sandy terrain surface. Int Arch Photogramm Remote Sens Spatial Inf Sci 42: 1225–1229. https://doi.org/10.5194/isprs-archives-XLⅡ-3-W10-1225-2020 doi: 10.5194/isprs-archives-XLⅡ-3-W10-1225-2020
    [12] Glenn NF, Streutker DR, Chadwick DJ, et al. (2006) Analysis of LiDAR-derived topographic information for characterizing and differentiating landslide morphology and activity. Geomorphology 73: 131–148. https://doi.org/10.1016/j.geomorph.2005.07.006 doi: 10.1016/j.geomorph.2005.07.006
    [13] Brightman N, Fan L, Zhao Y (2023) Point cloud registration: a mini-review of current state, challenging issues and future directions. AIMS Geosci 9: 68–85. https://doi.org/10.3934/geosci.2023005 doi: 10.3934/geosci.2023005
    [14] Zhao Y, Fan L (2023) Review on deep learning algorithms and benchmark datasets for pairwise global point cloud registration. Remote Sens. 15: 2060. https://doi.org/10.3390/rs15082060 doi: 10.3390/rs15082060
    [15] Trevisani S, Teza G, Guth PL (2023) Hacking the topographic ruggedness index. Geomorphology 439: 108838. https://doi.org/10.1016/j.geomorph.2023.108838 doi: 10.1016/j.geomorph.2023.108838
    [16] Andrle R, Abrahams AD (1989) Fractal techniques and the surface roughness of talus slopes. Earth Surf Processes Landforms 14: 197–209. https://doi.org/10.1002/esp.3290140303 doi: 10.1002/esp.3290140303
    [17] Dusséaux R, Vannier E (2022) Soil surface roughness modelling with the bidirectional autocorrelation function. Biosyst Eng 220: 87–102. https://doi.org/10.1016/j.biosystemseng.2022.05.012 doi: 10.1016/j.biosystemseng.2022.05.012
    [18] Huang CH, Bradford JM (1992) Applications of a Laser Scanner to Quantify Soil Microtopography. Soil Sci Soc Am J 56: 14–21. https://doi.org/10.2136/sssaj1992.03615995005600010002x doi: 10.2136/sssaj1992.03615995005600010002x
    [19] Trevisani S, Teza G, Guth P (2023) A Simplified Geostatistical Approach for Characterizing Key Aspects of Short-Range Roughness. CATENA 223: 106927. https://doi.org/10.1016/j.catena.2023.106927 doi: 10.1016/j.catena.2023.106927
    [20] Shepard MK, Campbell BA, Bulmer MH, et al. (2001) The roughness of natural terrain: A planetary and remote sensing perspective. J Geophys Res Planets 106: 32777–32795. https://doi.org/10.1029/2000JE001429 doi: 10.1029/2000JE001429
    [21] Fan L, Atkinson PM (2019) An Iterative Coarse-to-Fine Sub-Sampling Method for Density Reduction of Terrain Point Clouds. Remote Sens 11: 947. https://doi.org/10.3390/rs11080947 doi: 10.3390/rs11080947
    [22] LiDAR data access was based on[LiDAR, ground] services provided by the OpenTopography Facility. Lidar data acquisition completed by the National Centre for Airborne Laser Mapping. Available from: https://doi.org/10.5069/G9PR7SX0
    [23] Fan L, Atkinson PM (2018) A new multi-resolution based method for estimating local surface roughness from point clouds. ISPRS J Photogramm Remote Sens 144,369–378. https://doi.org/10.1016/j.isprsjprs.2018.08.003 doi: 10.1016/j.isprsjprs.2018.08.003
    [24] Zevenbergen LW, Thorne CR (1987) Quantitative analysis of land surface topography. Earth Surf Processes Landforms 12: 47–56. https://doi.org/10.1002/esp.3290120107 doi: 10.1002/esp.3290120107
    [25] Moore ID, Grayson RB, Ladson AR (1991) Digital terrain modelling: A review of hydrological, geomorphological, and biological applications. Hydrol Processes 5: 3–30. https://doi.org/10.1002/hyp.3360050103 doi: 10.1002/hyp.3360050103
    [26] Lindsay JB, Newman DR, Francioni A (2019) Scale-optimized surface roughness for topographic analysis. Geosciences 9: 322. https://doi.org/10.3390/geosciences9070322 doi: 10.3390/geosciences9070322
    [27] Fan L, Smethurst JA, Atkinson PM, et al. (2014) Propagation of vertical and horizontal source data errors into a TIN with linear interpolation. Int J Geogr Inf Sci 28: 1378–1400. https://doi.org/10.1080/13658816.2014.889299 doi: 10.1080/13658816.2014.889299
    [28] Pollyea RM, Fairley JP (2011) Estimating surface roughness of terrestrial laser scan data using orthogonal distance regression. Geology 39: 623–626. https://doi.org/10.1130/G32078.1 doi: 10.1130/G32078.1
  • This article has been cited by:

    1. Rohan D. Koshti, Kirtee K. Kamalja, A review on concomitants of order statistics and its application in parameter estimation under ranked set sampling, 2024, 53, 1226-3192, 65, 10.1007/s42952-023-00235-2
  • Reader Comments
  • © 2024 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(1598) PDF downloads(78) Cited by(0)

Figures and Tables

Figures(6)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog