
Floods have been identified as one of the world's most common and widely distributed natural disasters over the last few decades. Floods' negative impacts could be significantly reduced if accurately predicted or forecasted in advance. Apart from large-scale spatiotemporal data and greater attention to data from the Internet of Things, the worldwide volume of digital data is increasing. Artificial intelligence plays a vital role in analyzing and developing the corresponding flood mitigation plan, flood prediction, or forecast. Machine learning (ML)-based models have recently received much attention due to their self-learning capabilities from data without incorporating any complex physical processes. This study provides a comprehensive review of ML approaches used in flood prediction, forecasting, and classification tasks, serving as a guide for future challenges. The importance and challenges of applying these techniques to flood prediction are discussed. Finally, recommendations and future directions of ML models in flood analysis are presented.
Citation: Vijendra Kumar, Kul Vaibhav Sharma, Nikunj K. Mangukiya, Deepak Kumar Tiwari, Preeti Vijay Ramkar, Upaka Rathnayake. Machine learning applications in flood forecasting and predictions, challenges, and way-out in the perspective of changing environment[J]. AIMS Environmental Science, 2025, 12(1): 72-105. doi: 10.3934/environsci.2025004
[1] | Thuy Hien Nguyen, Catherine C. Moore, Preston B. Moore, Zhiwei Liu . Molecular dynamics study of homo-oligomeric ion channels: Structures of the surrounding lipids and dynamics of water movement. AIMS Biophysics, 2018, 5(1): 50-76. doi: 10.3934/biophy.2018.1.50 |
[2] | Carl S. Helrich . Studies of cholesterol structures in phospholipid bilayers. AIMS Biophysics, 2017, 4(3): 415-437. doi: 10.3934/biophy.2017.3.415 |
[3] | Chia-Wen Wang, Meng-Han Lin, Wolfgang B. Fischer . Cholesterol affected dynamics of lipids in tailor-made vesicles by ArcVes software during multi micro second coarse grained molecular dynamics simulations. AIMS Biophysics, 2023, 10(4): 482-502. doi: 10.3934/biophy.2023027 |
[4] | Iva Valkova, Petar Todorov, Victoria Vitkova . VV-hemorphin-5 association to lipid bilayers and alterations of membrane bending rigidity. AIMS Biophysics, 2022, 9(4): 294-307. doi: 10.3934/biophy.2022025 |
[5] | Jacques Fantini, Francisco J. Barrantes . How membrane lipids control the 3D structure and function of receptors. AIMS Biophysics, 2018, 5(1): 22-35. doi: 10.3934/biophy.2018.1.22 |
[6] | Rianne Bartelds, Jonathan Barnoud, Arnold J. Boersma, Siewert J. Marrink, Bert Poolman . Lipid phase separation in the presence of hydrocarbons in giant unilamellar vesicles. AIMS Biophysics, 2017, 4(4): 528-542. doi: 10.3934/biophy.2017.4.528 |
[7] | Sudarat Tharad, Chontida Tangsongcharoen, Panadda Boonserm, José L. Toca-Herrera, Kanokporn Srisucharitpanit . Local conformations affect the histidine tag-Ni2+ binding affinity of BinA and BinB proteins. AIMS Biophysics, 2020, 7(3): 133-143. doi: 10.3934/biophy.2020011 |
[8] | Shan-Yang Lin . Salmon calcitonin: conformational changes and stabilizer effects. AIMS Biophysics, 2015, 2(4): 695-723. doi: 10.3934/biophy.2015.4.695 |
[9] | Mathieu F. M. Cellier . Evolutionary analysis of Slc11 mechanism of proton-coupled metal-ion transmembrane import. AIMS Biophysics, 2016, 3(2): 286-318. doi: 10.3934/biophy.2016.2.286 |
[10] | Nicola Gaetano Gatta, Gaetano Cammarota, Vittorio Gentile . Possible roles of transglutaminases in molecular mechanisms responsible for human neurodegenerative diseases. AIMS Biophysics, 2016, 3(4): 529-545. doi: 10.3934/biophy.2016.4.529 |
Floods have been identified as one of the world's most common and widely distributed natural disasters over the last few decades. Floods' negative impacts could be significantly reduced if accurately predicted or forecasted in advance. Apart from large-scale spatiotemporal data and greater attention to data from the Internet of Things, the worldwide volume of digital data is increasing. Artificial intelligence plays a vital role in analyzing and developing the corresponding flood mitigation plan, flood prediction, or forecast. Machine learning (ML)-based models have recently received much attention due to their self-learning capabilities from data without incorporating any complex physical processes. This study provides a comprehensive review of ML approaches used in flood prediction, forecasting, and classification tasks, serving as a guide for future challenges. The importance and challenges of applying these techniques to flood prediction are discussed. Finally, recommendations and future directions of ML models in flood analysis are presented.
Over the past decade, scientists have studied several indicators of the health status of individuals. Chronological age, the time since birth, can be considered one of them. Several other indicators are required to improve the understanding of the health status of individuals. These indicators will combine CpG, DNA methylation (DNAm) and age (chronological age), health, and lifestyle outcomes. In the present article, we will call this indicator biological age, or epigenetic age [1,2].
Aging is generally considered irreversible due to the severe damage to the cell resulting in a gradual loss of functions and increasing fragility until final death. Recent publications proved that this process can be reversible. For example, biological age increases due to stress or traumatism and decreases in the recuperation phase during post-partum or after stress-induced or surgery-induced cell aging [3] (for mice).
This possibility of forward and backward changes in biological age due to specific events during an individual's life (pregnancy, stress, surgery, traumatism, etc.) has been considered in various approaches. For example, in [4], the calculation of biological age is based on estimating the time left to live depending on the number of undifferentiated cells remaining in the stem cell reservoir of the organs providing the patient's vital functions.
The present article aims to present a model for a population of patients. Our model mainly consists of continuous growth of the biological age together with some models for jumps (forward and backward) in the biological age to derive a mean behavior at a population's level. The continuous growth of the biological age corresponds to the biological age when no accidents or jumps occur. Therefore, in this case, the biological age grows at the same speed as the chronological age. In our model, the difference between biological age and chronological age consists of modeling random jumps forward or backward when the patient becomes sick or recovers from a disease such as cancer. In our model, the disease's severity corresponds to the jumps' amplitude. Here it is assumed to increase linearly with biological age. In conclusion, we will propose a larger model class for the jumps. The general framework of models is the so-called Levy process [5,6]. In Huang et al.[7], a mathematical model based on a stochastic differential equation is proposed to model the dynamic of biological age at the level of a single patient.
The plan of the paper is the following. Section 2 is devoted to the biological background of aging. In section 3, we present an aging model with a rejuvenation mechanism only. In section 4, we present an aging model with a premature aging mechanism only. In section 5, we combine rejuvenation and premature aging mechanisms. Section 6 is devoted to numerical simulation, and section 7 is to the discussion.
Due to microscopic cellular events, the health status of patients the life is first able to regenerate the tissue destroyed. When the chronological age increases, the patients are less capable of repairing such a tissue, leading to phenotypic medical symptoms. The severity of the medical symptoms is also more and more severe when patients' chronological age increases. The medical symptoms result from an accumulation threshold at the microscopic level of cell delations. This justifies the model with jumps (aging and rejuvenating jump) in biological age with an amplitude that increases in the life course.
We can consider that, even if in fine a loss of function is due to the dysfunctioning (or even death) of organs or even to the death of the whole organism, it results from a continuous disappearance of differentiated cells not replaced by undifferentiated cells (which are finite in number at birth and whose proliferation is bounded by Hayflick's limit), the event signaling accelerated aging or rejuvenation is a discrete phenomenon linked to a phenotypic symptom appearing at a precise moment corresponding to a biological age jump in the proposed mathematical models.
Due to microscopic cellular events, the health status of patients the life is first able to regenerate the tissue destroyed. When the chronological age increases, the patients are less capable of repairing such tissue, leading to phenotypic medical symptoms. The gravity of these medical symptoms is also more and more severe when patients' chronological age increases. The medical symptoms result from an accumulation at the microscopic level of cell deletions.
The determination of longevity is a problem that affects all species. The determining factors are multiple and are of two kinds, endogenous and exogenous. Among the endogenous determinants, we find, for example, ⅰ) the microbiome, whose importance has been demonstrated for species that have longevity comparable to the human species, such as crocodiles with a maximal life span of about 100 years [8] and ⅱ) the genome, whose influence is decisive in family diseases such as progeria, a rare genetic disorder [9]. Progeria accelerates aging, reduces the maximal human life span, and presents early symptoms like osteoporosis and hair loss. The exogenous factors are due to the environment and come from the food, which often contains pesticides and oxidizing components, the stress (especially occupational stress and emotional stress), infectious diseases, chronic diseases (namely diabetes, neurodegenerative and cardiovascular diseases), cancer, accidental musculoskeletal trauma, and surgical operations, etc.) [3]. For example, an acute viral disease can cause a loss of 3 109 cells in 1 day, equivalent to 2 weeks of additional aging, because normal aging causes only a loss of 2 108 cells per day [10]. Another example is given by the pre-dementia, in which the observed loss of cells can be caused by multiple etiologies, among which the loss of limbic and hypothalamic cell connections contributing to altering the sense of satiety (causing the adipocytes depletion), alterations in cardiomyocytes insulin sensitivity, or age-related regulatory changes in carbohydrate metabolism of the liver cells [11]. The loss of cells due to apoptosis (not compensated by the mitoses remaining before the Hayflick's limit [4][12]) in the different organs (brain, heart, liver, etc.) involved in these dysfunctions causes not the marginal organ death, but the whole organism death due to an involution (in cell size and number) of these organs incompatible with the survival of the whole organism.
As the accelerating aging factors, the rejuvenation factors also have two sources: the endogenous origin comes from cellular repair processes, in particular DNA repair mechanisms [13,14] which prevent the abnormal apoptosis due to DNA damages (caused by radiations, abortive mitoses, etc.). The heterogeneous origin is due to recovery processes during the healing time after a disease or an exhausting physiological event like pregnancy [15,3,16].
The chronological age is the commonly used age which is the time since birth. To describe a more realistic lifetime expectancy, we now focus on the biological age, which is much more difficult to define since it is multi-factorial.
According to Gerdhem et al. [17], biological age is a commonly used term without clear definitions but is routinely used to describe patients. Biological age may differ from chronological age and correlates to gait, muscle strength, and balance. The general appearance is considered when estimating the biological age.
In Demongeot [4], the biological age is defined as the real age of interface tissues (like intestinal endothelium, alveoli epithelium, and skin epithelium) and differs from the chronological age classically used in demography, but unable to give useful information about the exact stage in development or aging an organ or more globally an organism has reached. Biological age is essentially determined by the number of divisions remaining before Hayflick's limit (the maximal number of mitoses for a cell lineage inside a dedicated differentiated cellular population) of the tissues of a critical organ (critical in the sense that its loss causes the death of the whole organism). We refer to Shay and Wright [18] for more information. The critical organs have vital functions in interaction with the environment (protection, homeothermy, nutrition, and respiration) and present a rapid turnover (the total cell renewal time is in mice equal to 3 weeks for the skin, 1.5 days for the intestine, 4 months for the alveolate and 11 days for mitochondrial inner membrane) conditioning their biological age.
Another estimation of the biological age is the epigenetic age based on the expression level of certain genes like ELMSAN1 (also known as MIDEAS) and their transcription factors (e.g., mir4505 for MIDEAS) [1]. Figure 2 gives the statistical relationship between the epigenetic age and the chronological age showing that the probability of observing backward or forward jumps in epigenetic age and the intensity of these jumps increase linearly with chronological age. We refer to [19] and [20] for more results on this subject.
To describe the rejuvenation of a population of individuals structured with respect to biological age b∈[0,+∞), we consider b↦n(t,b) the population density at time t. That is
∫b2b1u(t,b)db |
is the number of individuals at time t with biological age between b1 and b2.
To model the rejuvenation, we will use the following system of partial differential equations
{∂tu(t,b)+∂bu(t,b)=−τ+u(t,b)+τ+(1+δ+)u(t,(1+δ+)b),u(t,0)=0,u(0,b)=u0(b)∈L1+((0,∞),R), | (3.1) |
where τ+>0 is the rate of rejuvenation (i.e. 1τ+ is for a single individual, the average time between two rejuvenations jumps), δ+≥0 is the fraction of biological age b (after rejuvenation) that should be added to b to obtain ˆb=(1+δ+)b the biological age before rejuvenation.
In this above model, the term ∂bu(t,b) corresponds to a drift term with a constant velocity (which represents the classical chronological aging in the absence of perturbations jumps), the term τ+u(t,b) corresponds to the flow of individuals which rejuvenate (i.e. having a jump in biological age) at time t. This flow is given by τ+∫∞0u(t,b)db. That is,
∫t2t1τ+∫∞0u(t,b)dbdt |
is the number of individuals which rejuvenate in between t1 and t2.
More precisely, when a rejuvenation occurs an individual having a biological age b after a rejuvenation's jump, its biological age was
ˆb=(1+δ+)b>b+δ+b>b, |
before the rejuvenation's jump.
In other words, a rejuvenating individual with a biological age b after a rejuvenation's jump, was an older individual with biological age ˆb=(1+δ+)b before rejuvenation's jump. That is also equivalent to say that, rejuvenating individual starting from the biological age ˆb end-ups with a biological age b=11+δ+ˆb after rejuvenation. So this individual looses a fraction
1−11+δ+=δ+1+δ+ |
of its biological age ˆb before rejuvenation's jump.
Setting
g+=1+δ+, | (3.2) |
the system (3.1) becomes
{∂tu(t,b)+∂bu(t,b)=−τ+u(t,b)+τ+g+u(t,g+b),u(t,0)=0,u(0,b)=u0(b)∈L1+((0,∞),R). | (3.3) |
Integral formulation of the solutions: By integrating the first equation of (3.3) along the characteristics (i.e. t−a constant) we obtain the following
u(t,b)={e−τ+tu0(b−t)+v(t,b),if t<b,v(t,b),if t>b, | (3.4) |
where
v(t,b)={∫t0e−τ+(t−σ)τ+g+u(σ,g+(b−t+σ))dσ,if t<b,∫b0e−τ+(b−σ)τ+g+u(t−b+σ,g+σ)dσ,if t>b. | (3.5) |
As a consequence of the above formula (3.4) and (3.5), and by applying fixed argument in suitable subspaces of L1 with compact supports, we obtain the following lemma.
Lemma 3.1. If
Support(u0)⊂[0,b⋆], |
then
Support(u(t,.))⊂[0,b⋆+t],∀t>0. |
Abstract Cauchy problem formulation: We refer to [21] [22] [23] [24] [25] for more results on semigroup theory and their application to age structured models. We consider
X=L1((0,∞),R), |
endowed its standard norm
‖ϕ‖L1=∫∞0|ϕ(σ)|dσ. |
We consider A:D(A)⊂X→X the linear operator defined by
Aϕ=−ϕ′ |
with
D(A)={ϕ∈W1,1((0,∞),R):ϕ(0)=0}. |
We consider B:X→X the bounded linear operator defined by
Bgϕ(x)=gϕ(gx), for x≥0., |
where g>0. Then Bg is an isometric bounded linear operator. That is
‖Bgϕ‖L1=‖ϕ‖L1,∀ϕ∈L1((0,∞),R). |
The problem (3.3) can be reformulated as an abstract Cauchy problem
{u′(t)=Au(t)−τ+u(t)+τ+Bg+u(t), for t≥0,withu(0)=u0∈L1+((0,∞),R). | (3.6) |
Lemma 3.2. The linear operator A is the infinitesimal general of {TA(t)}t≥0 the strongly continuous semigroup of linear operators, defined by
TA(t)(ϕ)(a)={ϕ(a−t),ifa>t,0,ifa<t. | (3.7) |
Definition 3.3. We will say that u∈C([0,∞),L1+((0,∞),R)) is a mild solution of (3.6) if
∫t0u(σ)dσ∈D(A),∀t≥0, |
and
u(t)=u0+A∫t0u(σ)dσ+∫t0−τ+u(σ)+τ+Bg+u(σ)dσ,∀t≥0. |
Theorem 3.4. For each u0∈L1+((0,∞),R), the Cauchy problem (3.6) admits a unique mild solution which is the unique continuous function u∈C([0,∞),L1+((0,∞),R)) satisfying the fixed point problem
u(t)=TA−τ+I(t)u0+∫t0TA−τ+I(t−σ)τ+Bg+u(σ)dσ,∀t≥0, | (3.8) |
where
TA−τ+I(t)=e−τ+tTA(t),∀t≥0. | (3.9) |
Remark 3.5. The variation of constant formula (3.8) is an abstract reformulation of the formulas (3.4) and (3.5).
Moments' formulation of the PDE: Define
Ek(u(t))=∫∞0σku(t,σ)dσ,∀k≥0, |
with
E0(u(t))=∫∞0u(t,σ)dσ, |
assuming that the integrals are well defined.
Theorem 3.6. The rejuvenation model (3.1) has a unique non-negative mild solution. We have
ddtE0(u(t))=0, | (3.10) |
and the model preserves the total mass (number) of individuals. That is
∫∞0u(t,σ)dσ=∫∞0u0(σ)dσ,∀t≥0, | (3.11) |
Moreover, the higher moment satisfies the following system of ordinary differential equations
ddtEk(u(t))=kEk−1(u(t))−χkEk(u(t)),∀k≥1, | (3.12) |
where
χk=τ+(1−1(1+δ+)k)>0,∀k≥1, |
and
limk→+∞χk=τ+>0. |
If we denote the equilibrium solution of (3.12) by
¯Ek=(k∏j=1jχj)∫∞0u0(σ)dσ,∀k≥1, |
consequently from (3.12), we have the following convergence result
limt→∞Ek(u(t))=¯Ek,∀k≥1. |
Proof.
∫∞0σk∂tu(t,σ)dσ+∫∞0σk∂bu(t,σ)dσ=−τ+∫∞0σku(t,σ)dσ+τ+∫∞0σku(t,g+σ)g+dσ, |
by integrating by parts the second integral, and by making a change of variable in the last integral, we obtain
ddtEk(u(t))=kEk−1(u(t))−τ+(1−1(1+δ+)k)∫∞0Ek(u(t)). |
To model the premature aging, we will use the following system of partial differential equations
{∂tu(t,b)+∂bu(t,b)=−τ−u(t,b)+τ−(1−δ−)u(t,(1−δ−)b),u(t,0)=0,u(0,b)=u0(b)∈L1+((0,∞),R), | (4.1) |
where τ−>0 is the rate of premature aging (i.e. 1τ− is for a single individual, the average time between two premature aging jumps), δ−∈(0,1) is the fraction of biological age b (after premature aging) that should be subtracted to b to obtain ˆb=(1−δ−)b the biological age before premature aging.
In this above model, the flow of individuals with premature aging (i.e. having a jump in biological age) at time t is given by τ−∫∞0u(t,b)db. That is,
∫t2t1τ−∫∞0u(t,b)dbdt |
is the number of individual with premature aging in between t1 and t2.
More precisely, when a rejuvenation occurs an individual having a biological age b after a premature aging's jump, its biological age was
ˆb=(1−δ−)b>b−δ−b<b. |
before the premature aging's jump.
In other words, a premature aging individual with a biological age b after a premature aging's jump, was an younger individual with biological age ˆb=(1−δ−)b before premature aging's jump. That is also equivalent to say that, premature aging individual starting from the biological age ˆb end-ups with a biological age b=11−δ−ˆb after premature aging. So this individual gains a fraction
11−δ−−1=δ−1−δ− |
of its biological age ˆb before premature aging's jump.
Setting
g−=1−δ−, | (4.2) |
the system (4.1) becomes
{∂tu(t,b)+∂bu(t,b)=−τ−u(t,b)+τ−g−u(t,g−b),u(t,0)=0,u(0,b)=u0(b)∈L1+((0,∞),R). | (4.3) |
The full model with both rejuvenation and premature aging reads as follows
{∂tu(t,b)+∂bu(t,b)= −(τ−+τ+)u(t,b)+τ−g−u(t,g−b)+τ+g+u(t,g+b),u(t,0)=0,u(0,b)=u0(b)∈L1+((0,∞),R). | (5.1) |
and we make the following assumption on the parameters of the system.
Assumption 5.1. We assume that
τ+>0,τ−>0,andg+>1>g−>0. |
Volterra integral formulation: By integrating the first equation of (5.1) along the characteristics (i.e. t−a constant) we obtain the following
u(t,b)={e−(τ−+τ+)tu0(b−t)+v+(t,b)+v−(t,b),if t<b,v+(t,b)+v−(t,b),if t>b, |
where
v+(t,b)={∫t0e−(τ−+τ+)(t−σ)τ+g+u(σ,g+(b−t+σ))dσ,if t<b,∫b0e−(τ−+τ+)(b−σ)τ+g+u(t−b+σ,g+σ)dσ,if t>b, |
and
v−(t,b)={∫t0e−(τ−+τ+)(t−σ)τ−g−u(σ,g−(b−t+σ))dσ,if t<b,∫b0e−(τ−+τ+)(b−σ)τ−g−u(t−b+σ,g−σ)dσ,if t>b. |
Abstract Cauchy problem reformulation: The problem (5.1) can be reformulated as an abstract Cauchy problem
{u′(t)=Au(t)−(τ−+τ+)u(t)+(τ−Bg−+τ+Bg+)u(t), for t≥0,withu(0)=u0∈L1+((0,∞),R). | (5.2) |
Definition 5.2. We will say that u∈C([0,∞),L1+((0,∞),R)) is a mild solution of (5.2) if
∫t0u(σ)dσ∈D(A),∀t≥0, |
and for each t≥0,
u(t)=u0+A∫t0u(σ)dσ+∫t0−(τ−+τ+)u(σ)+ (τ−Bg−+τ+Bg+)u(σ)dσ. |
Theorem 5.3. Let Assumption 5.1 be satisfied. For each u0∈L1+((0,∞),R), the Cauchy problem (5.2) admits a unique mild solution u∈C([0,∞),L1+((0,∞),R)) which is the unique continuous function satisfying the fixed point problem for each t≥0,
u(t)=TA−(τ−+τ+)I(t)u0+∫t0TA−(τ−+τ+)I(t−σ)(τ−Bg−+τ+Bg+)u(σ)dσ, |
where
TA−(τ−+τ+)I(t)=e−(τ−+τ+)tTA(t),∀t≥0. |
Moments formulation: We obtain the following result using similar arguments in for Theorem 3.6.
Theorem 5.4. Let Assumption 5.1 be satisfied. The rejuvenation and premature aging model (5.1) has a unique non-negative mild solution. We have
ddtE0(u(t))=0, |
and the model preserves the total mass (number) of individuals. That is
∫∞0u(t,σ)dσ=∫∞0u0(σ)dσ,∀t≥0. | (5.3) |
Moreover, the higher moment satisfies the following system of ordinary differential equations
ddtEk(u(t))=kEk−1(u(t))−χkEk(u(t)),∀k≥1, | (5.4) |
where
χk=τ+(1−1(1+δ+)k)+τ−(1−1(1−δ−)k),∀k≥1, | (5.5) |
and
limk→+∞χk=−∞. | (5.6) |
Remark 5.5. For k=1, and τ+=τ−=τ/2, and δ+=δ−=δ∈(0,1), then by using the formula (5.5) we obtain
χ1=τ+(1−1(1+δ+))+τ−(1−1(1−δ−))=τ/2[δ1+δ−δ1−δ]=τ/2[δ(1−δ)−δ(1+δ)1−δ2] |
and
χ1=−τδ21−δ2. |
Next, by using formula (5.7), we obtain
ddtE1(u(t))=E0(u0)+τδ21−δ2E1(u(t)),∀k≥1. | (5.7) |
The means value of the distribution is
M(u(t))=E1(u(t))E0(u(t)) |
and since t→E0(u(t)) is a constant function we obtain consequently, we obtain
M(u(t))′=1+τδ21−δ2M(u(t)). |
We conclude that,
limt→∞M(u(t))=+∞, |
and the smaller δ is, the closer the mean value M(u(t)) growth like the time t.
Lemma 5.6. Let Assumption 5.1 be satisfied. Define
xmax=ln(τ+lnq+)−ln(−τ−lnq−)lng+−lng−. |
Consider the function χ:[0,+∞)→R defined by
χ(x)=τ+(1−g−x+)+τ−(1−g−x−). |
There we have the following alternative
(ⅰ) If τ+lng++τ−lng−>0, then the map χ is first increases from χ(0)=0 to χ(xmax)>0 on [0,xmax], and then decreases from χ(xmax)>0 to −∞ on [xmax,∞).
(ⅱ) If τ+lng++τ−lng−≤0, then the map χ decreases from 0 to −∞ on [0,∞).
Proposition 5.7. Assume that there exists k0∈N, such that
χk>0,∀k=1,…,k0, |
and
χk<0,∀k>k0. |
If we denote the equilibrium solution of (5.7) by
¯Ek=(k∏j=1jχj)∫∞0u0(σ)dσ,∀k=1,…,k0, | (5.8) |
consequently from (5.7), we have the following convergence result
limt→+∞Ek(u(t))=¯Ek,∀k=1,…,k0, | (5.9) |
and
limt→+∞Ek(u(t))=+∞,∀k>k0. | (5.10) |
Equilibrium solution: An equilibrium solution satisfies some delay equation with both advance and retarded delay. That is,
{¯u′(b)= −(τ−+τ+)¯u(b)+τ−g−¯u(g−b)+τ+g+¯u(g+b),¯u(0)=0, | (5.11) |
and the difficulty of solving such an equation comes from the following
g+b>b>g−b,∀b>0. |
It follows that, even the existence of equilibrium solution is not a classical problem to investigate.
Non existence result for exponentially decreasing equilibrium solution of (5.1): Assume that b↦¯u(b) is a non-negative continuously differentiable map satisfying system (5.11).
Assumption 5.8. Assume that the map b↦¯u(b) is non-negative, and non null, and continuously differentiable map, and satisfies the system (5.11). Assume in addition that ¯u(b) is exponentially decreasing. That is, there exist two constants M>0, and γ>0, such that
¯u(b)≤Me−γb,∀b≥0. |
By using the first equation of (5.11), we deduce that
|¯u′(b)|≤˜Me−γmin(1,g+,g−)b,∀b≥0, |
for some suitable ˜M>0.
So, under Assumption 5.8, all the moments of ¯u(b) and ¯u′(b) are well defined. Moreover, by using the first equation of (5.11), we obtain for each k≥1
∫∞0σk¯u′(σ)dσ=−(τ−+τ+)∫∞0σk¯u(σ)dσ+τ+∫∞0σk¯u(t,g+σ)g+dσ+τ−∫∞0σk¯u(t,g−σ)g−dσ, |
and since ¯u(0)=0, we obtain by integrating by parts
−k∫∞0σk−1¯u(σ)dσ=−(τ−+τ+)∫∞0σk¯u(σ)dσ+τ+∫∞0σk¯u(t,g+σ)g+dσ+τ−∫∞0σk¯u(t,g−σ)g−dσ, |
and we obtain
kEk−1(¯u)=χkEk(¯u),∀k≥1,∀k≥1, | (5.12) |
where χk is define by (5.5).
Under Assumption 5.8, we must have
Ek−1(¯u)>0,∀k≥0, |
and by using (5.6), we deduce that (5.12) can not be satisfied for all k≥1 large enough. Therefore we obtain the following proposition.
Proposition 5.9. The system (5.11) has no exponential decreasing solution. That is a solution of (5.11) satisfying the Assumption 5.8.
By setting
τ+=τp, and τ−=τ(1−p), |
where τ is the rate at which individual either rejuvenates of prematurely ages, and p∈[0,1] is the probability of rejuvenation and (1−p)∈[0,1] is the probability of premature aging. The parameter p can be interpreted as the probability of being cured in the event of illness or injury. The parameter 1−p at the opposite is the probability of getting injured or sick.
By using these new parameters, we obtain a probabilistic interpretation of the model (5.1), and the model (5.1) becomes
{∂tu(t,b)+∂bu(t,b)=−τu(t,b)+τ(1−p)g−u(t,g−b)+τpg+u(t,g+b)u(t,0)=0,u(0,b)=u0(b)∈L1+((0,∞),R), | (6.1) |
and
g+>1>g−>0. |
In order to run a simulation of model (6.1), we use stochastic simulations. We consider a population composed of a finite number N=100000 of individuals. We start the simulation a time t=0 with all individuals with the same age a=20 years. The time to the next event (rejuvenation or premature aging) follows an exponential law with parameters 1/τ. The principles of the simulations are as follows: When an event occurs we choose randomly one individuals; and we compute a random value between 0 and 1. If this value is less or equal to p, rejuvenation occurs, and premature aging occurs otherwise. At each time step, the biological age increases by one-time step.
In Figure 3, we present some simulation of the model (6.1) whenever p=0.5, 1/τ=1 years, g+=1+δ+=1.1 and g−=1−δ−=0.9. In Figure 3, we can observe that starting from a single cohort of individuals with biological age 20 at time t=0, the mean value of the distribution follows the chronological age (thanks to the fact that p=1/2, and δ+=δ−=0.1, and the remark 5.5). But the density of the population deviates more and more with time. One also needs to interpret the biological age by saying that the larger the biological age is, the more people are likely to die. Therefore the more the population is dispersed around the means value, the more people (with larger ages) are likely to die earlier. People with a large biological age can be understood as people suffering from a lack of treatment for their illnesses or injuries.
In the Figures 4 and 5, we investigate the influence of the parameter p=0.25 in Figure 4, and p=0.75 in Figure 5. In Figures 4 and 5, we can see that due to the dissymmetric value of p the mean value no longer follows the chronological age. We can observe that the parameter p plays very important in the aging process.
We choose
τ+=τ−=0.1,δ+=0.1,δ+=0.01, |
and we obtain
g+=1+δ+=1.1, and g−=1−δ−=0.99. |
We use the initial distribution with compact support
u0(b)=25max(0,b(1−b/10)). |
In Figure 6, the initial values k↦Ek(u0) increase with k (see (b)). Moreover, the components k↦Ek(u(t)) keep the same order for all t>0 (see (d)). Moreover by using (c) and the Theorem 5.3, we deduce that k0∈[60,80], such that if k is below k0 the components converge to ¯Ek, and if k is above k0 the components go to +∞.
In the case of chronic diseases or environmental processes that continuously degrade patients' health, the implementation of therapy or the cessation of risk factors, even if they begin at a specific moment, only have delayed effects in the short or medium term. A good example is the smoking cessation. After stopping smoking, there is a rapid (one or two weeks) improvement, at the genetic level, of the pathologic cell markers involved in the respiratory system [26], and, at the physiological level, an increase in muscle resistance and a reduction in tissue inflammatory signs [27]. This early short-term effect is consistent with the model of a very rapid rejuvenation, as in a jumping process. The return to a completely normal state, with a very significant reduction in the risks linked to smoker's pathologies, takes about a year longer [28]. This middle-term normalization concerns the reduction of risk of numerous diseases and pathologic symptoms as the disappearance of cravings, restlessness, weight gain, and fatigue [29], a decline of chronic obstructive pulmonary disease, lung emphysema, inflammation, oxidative stress [30], and pulmonary cancers [31]. Stopping smoking speeds up recovery surgical operations [32]. Therefore, the rejuvenation is accelerated.
Despite this, an idealization of the phenomenon consists of marking the beginning of the rehabilitation phase with a jump and then, in a more complicated model whose parameters should evolve over time, taking into account the favorable evolution, particularly the reduction risk factors, on an ongoing basis. The example of maternity allows us to understand this process better. Hormonal impregnation [33] provides a rapid tissue rejuvenation well described in the literature. The muscle regeneration benefit effect appears immediately from the start of the pregnancy. But, this effect is transient, lasting up to 2 months after delivery. This is consistent with the hypothesis that pregnancy induces activation of muscle progenitor cells through the Notch genetic pathway [34,35]. The short and long-term effects can be measured by the extension of longevity of the multiparous women (e.g., in a study on more than 20,000 women [36]), even if some works show a short-term aging effect on certain blood-derived cell lines by shortening their telomere length and accelerating DNA damage [37].
Organ transplantations give an example of immediate rejuvenation. For example, thymus rejuvenation is brought by autologous hematopoietic stem cell transplantation and considerably increases the life expectancy of patients with autoimmune diseases [38]. Conversely, an infectious disease due to a virus like HIV drastically reduces the lifespan of affected patients [39].
Consequently, an improvement of the model requires taking into account adapted to each rejuvenation phenomenon, the short- and medium-term effects of the founding event causing improvement or damage of genetic expression or physiological function. The continuous part of the rejuvenation equation, dependent on time through the model parameters, can only be very specific to the problem addressed and will be considered in subsequent works dedicated to precise rejuvenation phenomena for which the data concerning short- and medium-term improvements are available.
This article proposes a new class of models describing the aging process. The model is based on the notion of biological age, which is a quantity reflecting the aging due to cells failing to repair DNA damages, illness, injuries, or, more generally speaking, corresponding to the body's aging. The key features of the model are the following:
1) a drift term with a constant velocity describing the aging process at the cellular level;
2) a rejuvenation operator describing the repair, recovery and healing processes during life;
3) a premature aging operator corresponds to the medical care of injuries and illness.
In this work, we consider the simplest version of the model. The model can be extended in several directions. To conclude the paper, we propose some possible extensions.
The full model with both rejuvenation and premature aging with birth and death processes is the following
{∂tu(t,b)+∂bu(t,b)=−μ(b)u(t,b)+β(t)Γ(b),−(τ−+τ+)u(t,b)+τ−g−u(t,g−b)+τ+g+u(t,g+b)u(t,0)=0,u(0,b)=u0(b)∈L1+((0,∞),R). | (7.1) |
In the above model, the function −μ(b) is the mortality rate for individuals with biological age b. That is
exp(−∫b2b1μ(b)db) |
is the probability for an individual to survive from the biological b1 to the biological age b2.
The term β(t) is the flow of new born at time t, that is
∫t2t1β(σ)dσ |
is the number of newborn between t1 and t2.
The map b→Γ(b) is the density of probability to have a biological age b at birth. That is
∫b2b1Γ(b)db |
is the probability to obtain a newborn with biological age b between b1 and b2. Moreover,
∫∞0Γ(b)db=1. |
Assumption 7.1. We assume that b↦μ(b) and t↦β(t) are constant functions. We also assume that
μ>0andβ>0. |
We assume that the biological age of newborns follows a gamma density of probably. That is,
Γ(b)=δαbα−1exp(−δb)(α−1)!forb>0, |
where α>1, δ>0, and the gamma function (α−1)! corresponds here to a constant of normalization, and is defined by
(α−1)!=∫∞0bα−1exp(−b)db. |
In Appendix 7.3, we discuss a aging model with birth and death processes.
The full model with both rejuvenation and premature aging reads as follows
{∂tu(t,b)+∂bu(t,b)= −(τ−+τ+)u(t,b)+τ−f′−(b)u(t,f−(b))+τ+f′+(b)u(t,f+(b))u(t,0)=0,u(0,b)=u0(b)∈L1+((0,∞),R). | (7.2) |
We can extend the above model by setting
f−(b)=(1−δ−(b))b≤b≤(1+δ+(b))b=f+(b). |
To assure the total mass preservation of the model we assume that
f−(0)=f+(0)=0 |
and to preserve the positivity of the solutions we assume that
f′−(b)≥0 and f′+(b)≥0,∀b≥0. |
For example, we could use
f+(b)=(1+δ+bm)b, with m≥0, |
where δ+(b)=δ+bm could be any positive polynomial in b, that would model the average amplitude of rejuvenation jumps.
The premature aging jumps f−(b) must remain below b (the biological age after the jump) therefore, we could use
f−(b)=(1−δ−1+χbm)b, with m≥0 and 0≤χ≤δ−<1. |
It is unrealistic to get younger than 20 years because puberty and post-pubertal growth are irreversible events, nor after 80 years when the stock of undifferentiated stem cells is exhausted. Likewise, it is unreasonable to exceed the age of more than 120 years, which is the physiological limit of the human lifespan. Hence the rate of rejuvenation or premature aging should not be constant in the function of the chronological age. Therefore it would be important for this problem to consider both the biological and chronological ages.
We could also combine the biological and chronological age. Consider a as the chronological age (i.e. the time since birth) then we can combine both the chronological and the biological age and we obtain
{∂tu+∂au+∂bu=−μ(a,b)u(t,a,b)−τ+u(t,a,b)+τ+(1+δ+)u(t,a,(1+δ+)b)−τ−u(t,a,b)+τ−(1−δ−)u(t,a,(1−δ−)b)u(t,0,b)=b(t)Γ(b), for b≥0,u(t,a,0)=0, for a≥0, | (7.3) |
with an initial distribution
u(0,a,b)=u0(a,b)∈L1+((0,∞)2,R). |
The function u(t,a,b) is the density of population at time t with respect to the chronological age a and the biological age b. This means that if 0≤a1≤a2 and 0≤b1≤b2 then
∫a2a1∫b2b1u(t,a,b)dadb |
is the number of individuals with chronological age a in between a1 and a2 and the biological b in between b1 and b2.
This problem can be reformulated as follows
{∂tu+∂au=Au(t,a,b)−μ(a,b)u(t,a,b)−τ+u(t,a,b)+τ+(1+δ+)u(t,a,(1+δ+)b)−τ−u(t,a,b)+τ−(1−δ−)u(t,a,(1−δ−)b)u(t,0,b)=b(t)Γ(b), for b≥0, | (7.4) |
We refer to Magal and Ruan [25] for more results about age-structured models combined with an extra structuring variable.
The authors declare they have not used Artificial Intelligence tools in the creation of this article.
Pierre Magal is an editorial board member for Mathematical Biosciences and Engineering and was not involved in the editorial review or the decision to publish this article. All authors declare that there are no competing interests.
Aging model with birth and death processes
Abstract Cauchy problem reformulation: The problem (7.1) can be reformulated as an abstract Cauchy problem
{u′(t)=Au(t)+Bu(t)−μu(t)+βΓ(.), for t≥0,withu(0)=u0∈L1+((0,∞),R). | (7.5) |
where
Bu=τ−Bg−u+τ+Bg+u−(τ−+τ+)u. |
Theorem 7.2. Let Assumption 5.1 and Assumption 7.1 be satisfied. Then the mild solution (7.5) is given by
u(t)=TA+B−μI(t)u0+∫t0TA+B−μI(t−σ)βΓ(.)dσ,∀t≥0, |
where
TA+B−μI(t)=e−μtTA+B(t),∀t≥0. |
Moreover
limt→∞u(t)=¯u≥0, |
where
¯u=∫∞0TA+B−μI(σ)βΓ(.)dσ. | (7.6) |
Moments formulation: We obtain the following result using similar arguments to Theorem 3.6.
By using change of variable, we obtain
Ek(Γ)=∫∞0δαbα+k−1exp(−δb)(α−1)!db=δ−k(α+k−1)!(α−1)!. |
Theorem 7.3. Let Assumption 5.1 be satisfied. The rejuvenation and premature aging model (5.1) has a unique non-negative mild solution. We have
ddtE0(u(t))=−μE0(u(t))+β, |
and the model preserves the total mass (number) of individuals. That is
limt→+∞E0(u(t))=βμ. | (7.7) |
Moreover, the higher moment satisfies the following system of ordinary differential equations for each k≥1,
ddtEk(u(t))=kEk−1(u(t))−(μ+χk)Ek(u(t))+βδ−k(α+k−1)!(α−1)!. | (7.8) |
Proposition 7.4. Let k1 be an integer such that
χk<−μ,∀k>k1. |
Then
limt→+∞Ek(u(t))=+∞,∀k>k1, | (7.9) |
whenever Ek(u0)<+∞.
Equilibrium solution: An equilibrium solution satisfies some delay equation with both advance and retarded delay. That is,
{¯u′(b)=βΓ(b)−(μ+τ−+τ+)¯u(b)+τ−g−¯u(g−b)+τ+g+¯u(g+b),¯u(0)=0, | (7.10) |
and the difficulty of solving such an equation comes from the following
g+b>b>g−b,∀b>0. |
It follows that, even the existence of equilibrium solution is not a classical problem to investigate.
Non existence result for exponentially decreasing equilibrium solution of (5.1): Assume that b↦¯u(b) a non-negative continuously differentiable map satisfying system (5.11).
Assumption 7.5. Assume that the map b↦¯u(b) is non-negative, and non null, and continuously differentiable map, and satisfies the system (7.10). Assume in addition that ¯u(b) is exponentially decreasing. That is, there exist two constants M>0, and γ>0, such that
¯u(b)≤Me−γb,∀b≥0. | (7.11) |
By using the first equation of (5.11), we deduce that
|¯u′(b)|≤˜Me−γmin(1,g+,g−,δ/2)b,∀b≥0, |
for some suitable ˜M>0.
So, under Assumption 5.8, all the moments of ¯u(b) and ¯u′(b) are well defined. Moreover, by using the first equation of (5.11), we obtain for each k≥1
∫∞0σk¯u′(σ)dσ=∫∞0βΓ(b)db−(μ+τ−+τ+)∫∞0σk¯u(σ)dσ+τ+∫∞0σk¯u(t,g+σ)g+dσ+τ−∫∞0σk¯u(t,g−σ)g−dσ, |
and since ¯u(0)=0, we obtain by integrating by parts
−k∫∞0σk−1¯u(σ)dσ=βδ−k(α+k−1)!(α−1)!−(μ+τ−+τ+)∫∞0σk¯u(σ)dσ+τ+∫∞0σk¯u(t,g+σ)g+dσ+τ−∫∞0σk¯u(t,g−σ)g−dσ, |
and we obtain
0=βδ−k(α+k−1)!(α−1)!+kEk−1(¯u)−(χk+μ)Ek(¯u),∀k≥1,∀k≥1, | (7.12) |
where χk is defined by (5.5).
Under Assumption 7.5, we must have
Ek−1(¯u)>0, and Ek(¯u)>0,∀k≥0, |
and by using (5.6), we deduce that (7.12) can not be satisfied for all k≥1 large enough. Because −(χk+μ)>0, for all k≥1 large enough.
Therefore we obtain the following proposition.
Proposition 7.6. The equilibrium ¯u defined by (7.6) is not an exponentially decreasing function. That is, for each M>0, and γ>0,, the function ¯u does not satisfy (7.11).
The above proposition is surprising because the proposition shows that it is sufficient to perturb an age-structured model with a non-local term −τ−u(t,b)+τ−(1−δ−)u(t,(1−δ−)b) to obtain an equilibrium solution which is not exponential bounded.
[1] |
Tehrany MS, Pradhan B, Jebur MN (2015) Flood susceptibility analysis and its verification using a novel ensemble support vector machine and frequency ratio method. Stoch Environ Res Risk Assess 29: 1149–1165. https://doi.org/10.1007/s00477-015-1021-9 doi: 10.1007/s00477-015-1021-9
![]() |
[2] |
Lin Q, Leandro J, Gerber S, et al. (2020) Multistep flood inundation forecasts with resilient backpropagation neural networks: Kulmbach case study. Water 12. https://doi.org/10.3390/w12123568 doi: 10.3390/w12123568
![]() |
[3] |
Ritter J, Berenguer M, Corral C, et al. (2020) ReAFFIRM: Real-time Assessment of Flash Flood Impacts – a Regional high-resolution Method. Environ Int 136: 105375. https://doi.org/10.1016/j.envint.2019.105375 doi: 10.1016/j.envint.2019.105375
![]() |
[4] |
Abdelhady AU, Xu D, Ouyang Z, et al. (2022) A framework for estimating water ingress due to hurricane rainfall. J Wind Eng Ind Aerodyn 221: 104891. https://doi.org/10.1016/j.jweia.2021.104891 doi: 10.1016/j.jweia.2021.104891
![]() |
[5] |
Sankaranarayanan S, Prabhakar M, Satish S, et al. (2020) Flood prediction based on weather parameters using deep learning. J Water Clim Change 11: 1766–1783. https://doi.org/10.2166/wcc.2019.321 doi: 10.2166/wcc.2019.321
![]() |
[6] |
Kolen B, Slomp R, Jonkman SN (2013) The impacts of storm Xynthia February 27-28, 2010 in France: Lessons for flood risk management. J Flood Risk Manag 6: 261–278. https://doi.org/10.1111/jfr3.12011 doi: 10.1111/jfr3.12011
![]() |
[7] |
Berndtsson R, Becker P, Persson A, et al. (2019) Drivers of changing urban flood risk: A framework for action. J Environ Manag 240: 47–56. https://doi.org/10.1016/j.jenvman.2019.03.094 doi: 10.1016/j.jenvman.2019.03.094
![]() |
[8] |
Kwon SH, Kim JH (2021) Machine learning and urban drainage systems: State-of-the-art review. Water (Switzerland) 13: 1–14. https://doi.org/10.3390/w13243545 doi: 10.3390/w13243545
![]() |
[9] |
Jain SK, Mani P, Jain SK, et al. (2018) A Brief review of flood forecasting techniques and their applications. Int J River Basin Manag 16: 329–344. https://doi.org/10.1080/15715124.2017.1411920 doi: 10.1080/15715124.2017.1411920
![]() |
[10] |
Moore RJ, Bell VA, Jones DA (2005) Forecasting for flood warning. C R Geosci 337: 203–217. https://doi.org/10.1016/j.crte.2004.10.017 doi: 10.1016/j.crte.2004.10.017
![]() |
[11] |
Difrancesco KN, Tullos DD (2014) Flexibility in Water Resour Manag: Review of Concepts and Development of Assessment Measures for Flood Management Systems. J Am Water Resour Assoc 50: 1527–1539. https://doi.org/10.1111/jawr.12214 doi: 10.1111/jawr.12214
![]() |
[12] |
Zounemat-Kermani M, Matta E, Cominola A, et al. (2020) Neurocomputing in surface water hydrology and hydraulics: A review of two decades retrospective, current status and future prospects. J Hydrol 588: 125085. https://doi.org/10.1016/j.jhydrol.2020.125085 doi: 10.1016/j.jhydrol.2020.125085
![]() |
[13] |
Dazzi S, Vacondio R, Mignosa P (2021) Flood stage forecasting using machine-learning methods: A case study on the parma river (italy). Water 13. https://doi.org/10.3390/w13121612 doi: 10.3390/w13121612
![]() |
[14] |
Kratzert F, Klotz D, Brenner C, et al. (2018) Rainfall–runoff modelling using Long Short-Term Memory (LSTM) networks. Hydrol Earth Syst Sci 22: 6005–6022. https://doi.org/10.5194/hess-22-6005-2018 doi: 10.5194/hess-22-6005-2018
![]() |
[15] |
Mosavi A, Ozturk P, Chau K (2018) Flood Prediction Using Machine Learning Models: Literature Review. Water 10: 1536. https://doi.org/10.3390/w10111536 doi: 10.3390/w10111536
![]() |
[16] |
Han S, Coulibaly P (2017) Bayesian flood forecasting methods: A review. J Hydrol 551: 340–351. https://doi.org/10.1016/j.jhydrol.2017.06.004 doi: 10.1016/j.jhydrol.2017.06.004
![]() |
[17] |
Badjana HM, Fink M, Helmschrot J, et al. (2017) Hydrological system analysis and modelling of the Kara River basin (West Africa) using a lumped metric conceptual model. Hydrol Sci J 62: 1094–1113. https://doi.org/10.1080/02626667.2017.1307571 doi: 10.1080/02626667.2017.1307571
![]() |
[18] |
Dal Molin M, Schirmer M, Zappa M, et al. (2020) Understanding dominant controls on streamflow spatial variability to set up a semi-distributed hydrological model: The case study of the Thur catchment. Hydrol Earth Syst Sci 24: 1319–1345. https://doi.org/10.5194/hess-24-1319-2020 doi: 10.5194/hess-24-1319-2020
![]() |
[19] |
Wang J, Shi P, Jiang P, et al. (2017) Application of BP neural network algorithm in traditional hydrological model for flood forecasting. Water 9: 1–16. https://doi.org/10.3390/w9010048 doi: 10.3390/w9010048
![]() |
[20] |
Sarker IH (2021) Machine Learning: Algorithms, Real-World Applications and Research Directions. SN Computer Science 2: 1–21. https://doi.org/10.1007/s42979-021-00592-x doi: 10.1007/s42979-021-00592-x
![]() |
[21] |
Liakos KG, Busato P, Moshou D, et al. (2018) Machine learning in agriculture: A review. Sensors 18: 1–29. https://doi.org/10.3390/s18082674 doi: 10.3390/s18082674
![]() |
[22] | Sene K (2016) Hydrometeorology, Cham, Springer International Publishing. https://doi.org/10.1007/978-3-319-23546-2 |
[23] |
Mosavi A, Ozturk P, Chau KW (2018) Flood prediction using machine learning models: Literature review. Water (Switzerland) 10: 1–40. https://doi.org/10.3390/w10111536 doi: 10.3390/w10111536
![]() |
[24] |
Ighile EH, Shirakawa H, Tanikawa H (2022) A Study on the Application of GIS and Machine Learning to Predict Flood Areas in Nigeria. Sustainability14. https://doi.org/10.3390/su14095039 doi: 10.3390/su14095039
![]() |
[25] |
Nayak M, Das S, Senapati MR (2022) Improving Flood Prediction with Deep Learning Methods. J Inst Eng India Ser B. https://doi.org/10.1007/s40031-022-00720-y doi: 10.1007/s40031-022-00720-y
![]() |
[26] |
Sankaranarayanan S, Prabhakar M, Satish S, et al. (2020) Flood prediction based on weather parameters using deep learning. J Water Clim Change 11: 1766–1783. https://doi.org/10.2166/wcc.2019.321 doi: 10.2166/wcc.2019.321
![]() |
[27] |
Jabbari A, Bae DH (2018) Application of Artificial Neural Networks for accuracy enhancements of real-time flood forecasting in the Imjin basin. Water10. https://doi.org/10.3390/w10111626 doi: 10.3390/w10111626
![]() |
[28] |
Elsafi SH (2014) Artificial Neural Networks (ANNs) for flood forecasting at Dongola Station in the River Nile, Sudan. Alex Eng J 53: 655–662. https://doi.org/10.1016/j.aej.2014.06.010 doi: 10.1016/j.aej.2014.06.010
![]() |
[29] |
Chen JC, Ning SK, Chen HW, et al. (2008) Flooding probability of urban area estimated by decision tree and artificial neural networks. J Hydroinform 10: 57–67. https://doi.org/10.2166/hydro.2008.009 doi: 10.2166/hydro.2008.009
![]() |
[30] |
Tehrany MS, Pradhan B, Jebur MN (2013) Spatial prediction of flood susceptible areas using rule based decision tree (DT) and a novel ensemble bivariate and multivariate statistical models in GIS. J Hydrol 504: 69–79. https://doi.org/10.1016/j.jhydrol.2013.09.034 doi: 10.1016/j.jhydrol.2013.09.034
![]() |
[31] |
Tehrany MS, Pradhan B, Mansor S, et al. (2015) Flood susceptibility assessment using GIS-based support vector machine model with different kernel types. Catena 125: 91–101. https://doi.org/10.1016/j.catena.2014.10.017 doi: 10.1016/j.catena.2014.10.017
![]() |
[32] |
Liu M, Huang Y, Li Z, et al. (2020) The applicability of lstm-knn model for real-time flood forecasting in different climate zones in China. Water 12: 1–21. https://doi.org/10.3390/w12020440 doi: 10.3390/w12020440
![]() |
[33] |
Boateng EY, Otoo J, Abaye DA (2020) Basic Tenets of Classification Algorithms K-Nearest-Neighbor, Support Vector Machine, Random Forest and Neural Network: A Review. J Data Anal Inf Process 08: 341–357. https://doi.org/10.4236/jdaip.2020.84020 doi: 10.4236/jdaip.2020.84020
![]() |
[34] |
Modaresi F, Araghinejad S, Ebrahimi K (2018) A Comparative Assessment of Artificial Neural Network, Generalized Regression Neural Network, Least-Square Support Vector Regression, and K-Nearest Neighbor Regression for Monthly Streamflow Forecasting in Linear and Nonlinear Conditions. Water Resour Manag 32: 243–258. https://doi.org/10.1007/s11269-017-1807-2 doi: 10.1007/s11269-017-1807-2
![]() |
[35] |
Ghorbani MA, Zadeh HA, Isazadeh M, et al. (2016) A comparative study of artificial neural network (MLP, RBF) and support vector machine models for river flow prediction. Environ Earth Sci 75: 1–14. https://doi.org/10.1007/s12665-015-5096-x doi: 10.1007/s12665-015-5096-x
![]() |
[36] |
Sarker IH, Kayes ASM, Badsha S, et al. (2020) Cybersecurity data science: an overview from machine learning perspective. J Big Data 7. https://doi.org/10.1186/s40537-020-00318-5 doi: 10.1186/s40537-020-00318-5
![]() |
[37] |
Janiesch C, Zschech P, Heinrich K (2021) Machine learning and deep learning. Electron Mark 31: 685–695. https://doi.org/10.1007/s12525-021-00475-2 doi: 10.1007/s12525-021-00475-2
![]() |
[38] | Xu T, Liang F (2021) Machine learning for hydrologic sciences: An introductory overview. Wiley Interdisciplinary Reviews: Water 8: 1–29. https://doi.org/10.1002/wat2.1533 |
[39] | Mohammed M, Khan MB, Bashier EBM (2016) Machine Learning algorithms and applications, Dordrecht, CRC Press. https://doi.org/10.1201/9781315371658 |
[40] |
Shen C (2018) A Transdisciplinary Review of Deep Learning Research and Its Relevance for Water Resources Scientists. Water Resour Res 54: 8558–8593. https://doi.org/10.1029/2018WR022643 doi: 10.1029/2018WR022643
![]() |
[41] |
Gnecco G, Morisi R, Roth G, et al. (2017) Supervised and semi-supervised classifiers for the detection of flood-prone areas. Soft Comput 21: 3673–3685. https://doi.org/10.1007/s00500-015-1983-z doi: 10.1007/s00500-015-1983-z
![]() |
[42] |
Wagenaar D, De Jong J, Bouwer LM (2017) Multi-variable flood damage modelling with limited data using supervised learning approaches. Nat Hazards Earth Syst Sci 17: 1683–1696. https://doi.org/10.5194/nhess-17-1683-2017 doi: 10.5194/nhess-17-1683-2017
![]() |
[43] |
Oppel H, Fischer S (2020) A New Unsupervised Learning Method to Assess Clusters of Temporal Distribution of Rainfall and Their Coherence with Flood Types. Water Resour Res 56. https://doi.org/10.1029/2019WR026511 doi: 10.1029/2019WR026511
![]() |
[44] | Gentleman R, Carey VJ (2008) Unsupervised Machine Learning, Bioconductor Case Studies, New York. NY, Springer., 137–157. https://doi.org/10.1007/978-0-387-77240-0_10 |
[45] |
Usama M, Qadir J, Raza A, et al. (2019) Unsupervised Machine Learning for Networking: Techniques, Applications and Research Challenges. IEEE Access 7: 65579–65615. https://doi.org/10.1109/ACCESS.2019.2916648 doi: 10.1109/ACCESS.2019.2916648
![]() |
[46] |
Vamplew P, Dazeley R, Berry A, et al. (2011) Empirical evaluation methods for multiobjective reinforcement learning algorithms. Mach Learn 84: 51–80. https://doi.org/10.1109/ACCESS.2019.2916648 doi: 10.1109/ACCESS.2019.2916648
![]() |
[47] |
Santiago Júnior VA de, Özcan E, Carvalho VR de (2020) Hyper-Heuristics based on Reinforcement Learning, Balanced Heuristic Selection and Group Decision Acceptance. Appl Soft Comput 97: 106760. https://doi.org/10.1007/s10994-010-5232-5 doi: 10.1007/s10994-010-5232-5
![]() |
[48] | Babbar-sebens M, Mukhopadhyay S (2009) Reinforcement Learning for Human-Machine Collaborative Optimization. Appl Ground Water Monit October 3563–3568. |
[49] |
Jain SK, Mani P, Jain SK, et al. (2018) A Brief review of flood forecasting techniques and their applications. Int J River Basin Manag 16: 329–344. https://doi.org/10.1080/15715124.2017.1411920 doi: 10.1080/15715124.2017.1411920
![]() |
[50] |
Rözer V, Müller M, Bubeck P, et al. (2016) Coping with pluvial floods by private households. Water 8. https://doi.org/10.3390/W8070304 doi: 10.3390/W8070304
![]() |
[51] |
Nachappa T, Meena SR (2020) A novel per pixel and object-based ensemble approach for flood susceptibility mapping. Geom Nat Hazards Risk 11: 2147–2175. https://doi.org/10.1080/19475705.2020.1833990 doi: 10.1080/19475705.2020.1833990
![]() |
[52] |
Wu J, Liu H, Wei G, et al. (2019) Flash flood forecasting using support vector regression model in a small mountainous catchment. Water 11. https://doi.org/10.3390/w11071327 doi: 10.3390/w11071327
![]() |
[53] |
Yariyan P, Janizadeh S, Van Phong T, et al. (2020) Improvement of Best First Decision Trees Using Bagging and Dagging Ensembles for Flood Probability Mapping. Water Resour Manag 34: 3037–3053. https://doi.org/10.1007/s11269-020-02603-7 doi: 10.1007/s11269-020-02603-7
![]() |
[54] |
Landuyt L, Verhoest NEC, Van Coillie FMB (2020) Flood mapping in vegetated areas using an unsupervised clustering approach on sentinel-1 and-2 imagery. Remote Sens 12: 1–20. https://doi.org/10.3390/rs12213611 doi: 10.3390/rs12213611
![]() |
[55] |
Li W, Kiaghadi A, Dawson C (2021) Exploring the best sequence LSTM modeling architecture for flood prediction. Neural Comput Appl 33: 5571–5580. https://doi.org/10.1007/s00521-020-05334-3 doi: 10.1007/s00521-020-05334-3
![]() |
[56] |
Ahmed AN, Van Lam T, Hung ND, et al. (2021) A comprehensive comparison of recent developed meta-heuristic algorithms for streamflow time series forecasting problem. Appl Soft Comput 105: 107282. https://doi.org/10.1016/j.asoc.2021.107282 doi: 10.1016/j.asoc.2021.107282
![]() |
[57] |
Liu K, Li Z, Yao C, et al. (2016) Coupling the k-nearest neighbor procedure with the Kalman filter for real-time updating of the hydraulic model in flood forecasting. Int J Sediment Res 31: 149–158. https://doi.org/10.1016/j.ijsrc.2016.02.002 doi: 10.1016/j.ijsrc.2016.02.002
![]() |
[58] |
Kabir S, Patidar S, Xia X, et al. (2020) A deep convolutional neural network model for rapid prediction of fluvial flood inundation. J Hydrol 590: 125481. https://doi.org/10.1016/j.jhydrol.2020.125481 doi: 10.1016/j.jhydrol.2020.125481
![]() |
[59] |
Costache R, Arabameri A, Blaschke T, et al. (2021) Flash-flood potential mapping using deep learning, alternating decision trees and data provided by remote sensing sensors. Sensors 21: 1–21. https://doi.org/10.3390/s21010280 doi: 10.3390/s21010280
![]() |
[60] |
Ateeq-ur-Rauf, Ghumman AR, Ahmad S, et al. (2018) Performance assessment of artificial neural networks and support vector regression models for stream flow predictions. Environ Monit Assess 190. https://doi.org/10.1007/s10661-018-7012-9 doi: 10.1007/s10661-018-7012-9
![]() |
[61] | Al-Fawa'reh M, Hawamdeh A, Alrawashdeh R, et al. (2021) Intelligent Methods for flood forecasting in Wadi al Wala, Jordan. International Congress of Advanced Technology and Engineering, ICOTEN. https://doi.org/10.1109/ICOTEN52080.2021.9493425 |
[62] |
Parizi E, Bagheri-Gavkosh M, Hosseini SM, et al. (2021) Linkage of geographically weighted regression with spatial cluster analyses for regionalization of flood peak discharges drivers: Case studies across Iran. J Clean Prod 310. https://doi.org/10.1016/j.jclepro.2021.127526 doi: 10.1016/j.jclepro.2021.127526
![]() |
[63] |
Tsakiri K, Marsellos A, Kapetanakis S (2018) Artificial neural network and multiple linear regression for flood prediction in Mohawk River, New York. Water 10. https://doi.org/10.3390/w10091158 doi: 10.3390/w10091158
![]() |
[64] |
Wu CL, Chau KW (2010) Data-driven models for monthly streamflow time series prediction. Eng. Appl. Artif. Intell. 23: 1350–1367. https://doi.org/10.1016/j.engappai.2010.04.003 doi: 10.1016/j.engappai.2010.04.003
![]() |
[65] |
Zounemat-Kermani M, Kisi O, Rajaee T (2013) Performance of radial basis and LM-feed forward artificial neural networks for predicting daily watershed runoff. Appl Soft Comput 13: 4633–4644. https://doi.org/10.1016/j.asoc.2013.07.007 doi: 10.1016/j.asoc.2013.07.007
![]() |
[66] | Sulaiman J, Wahab SH (2018) Heavy Rainfall Forecasting Model Using Artificial Neural Network for Flood Prone Area, In: Kim KJ, Kim H, Baek N (Eds.), IT Convergence and Security. Singap, Springer, 68–76. https://doi.org/10.1007/978-981-10-6451-7_9 |
[67] |
Jain A, Indurthy SKVP (2003) Comparative Analysis of Event-based Rainfall-runoff Modeling Techniques—Deterministic, Statistical, and Artificial Neural Networks. J Hydrol Eng 8: 93–98. https://doi.org/10.1061/(asce)1084-0699(2003)8:2(93) doi: 10.1061/(asce)1084-0699(2003)8:2(93)
![]() |
[68] | Cruz FRG, Binag MG, Ga MRG, et al. (2019) Flood Prediction Using Multi-Layer Artificial Neural Network in Monitoring System with Rain Gauge, Water Level, Soil Moisture Sensors. IEEE Region 10 Annual International Conference, Proceedings/TENCON 2499–2503. https://doi.org/10.1109/TENCON.2018.8650387 |
[69] |
Kim H Il, Han KY (2020) Urban flood prediction using deep neural network with data augmentation. Water 12. https://doi.org/10.3390/w12030899 doi: 10.3390/w12030899
![]() |
[70] |
Ni JR, Xue A (2003) Application of artificial neural network to the rapid feedback of potential ecological risk in flood diversion zone. Eng Appl Artif Intell 16: 105–119. https://doi.org/10.1016/S0952-1976(03)00059-9 doi: 10.1016/S0952-1976(03)00059-9
![]() |
[71] |
Wu CL, Chau KW, Fan C (2010) Prediction of rainfall time series using modular artificial neural networks coupled with data-preprocessing techniques. J Hydrol 389: 146–167. https://doi.org/10.1016/j.jhydrol.2010.05.040 doi: 10.1016/j.jhydrol.2010.05.040
![]() |
[72] |
Feng LH, Lu J (2010) The practical research on flood forecasting based on artificial neural networks. Expert Syst Appl 37: 2974–2977. https://doi.org/10.1016/j.eswa.2009.09.037 doi: 10.1016/j.eswa.2009.09.037
![]() |
[73] |
Dtissibe FY, Ari AAA, Titouna C, et al. (2020) Flood forecasting based on an artificial neural network scheme. Natural Hazards 104: 1211–1237. https://doi.org/10.1007/s11069-020-04211-5 doi: 10.1007/s11069-020-04211-5
![]() |
[74] |
Deo RC, Şahin M (2015) Application of the Artificial Neural Network model for prediction of monthly Standardized Precipitation and Evapotranspiration Index using hydrometeorological parameters and climate indices in eastern Australia. Atmos Res 161–162: 65–81. https://doi.org/10.1016/j.atmosres.2015.03.018 doi: 10.1016/j.atmosres.2015.03.018
![]() |
[75] |
Sahoo A, Samantaray S, Ghose DK (2022) Multilayer perceptron and support vector machine trained with grey wolf optimiser for predicting floods in Barak river, India. J Earth Syst Sci 131. https://doi.org/10.1007/s12040-022-01815-2 doi: 10.1007/s12040-022-01815-2
![]() |
[76] |
Linh NTT, Ruigar H, Golian S, et al. (2021) Flood prediction based on climatic signals using wavelet neural network. Acta Geophys 69: 1413–1426. https://doi.org/10.1007/s11600-021-00620-7 doi: 10.1007/s11600-021-00620-7
![]() |
[77] |
Panahi M, Jaafari A, Shirzadi A, et al. (2021) Deep learning neural networks for spatially explicit prediction of flash flood probability. Geosci Front 12: 101076. https://doi.org/10.1016/j.gsf.2020.09.007 doi: 10.1016/j.gsf.2020.09.007
![]() |
[78] |
Wang JH, Lin GF, Chang MJ, et al. (2019) Real-Time Water-Level Forecasting Using Dilated Causal Convolutional Neural Networks. Water Resour Manag 33: 3759–3780. https://doi.org/10.1007/s11269-019-02342-4 doi: 10.1007/s11269-019-02342-4
![]() |
[79] |
Song T, Ding W, Wu J, et al. (2020) Flash flood forecasting based on long short-term memory networks. Water 12. https://doi.org/10.3390/w12010109 doi: 10.3390/w12010109
![]() |
[80] |
Wang HW, Lin GF, Hsu CT, et al. (2022) Long-Term Temporal Flood Predictions Made Using Convolutional Neural Networks. Water 14. https://doi.org/10.3390/w14244134 doi: 10.3390/w14244134
![]() |
[81] |
Cho M, Kim C, Jung K, et al. (2022) Water Level Prediction Model Applying a Long Short-Term Memory (LSTM)–Gated Recurrent Unit (GRU) Method for Flood Prediction. Water 14: 2221. https://doi.org/10.3390/w14142221 doi: 10.3390/w14142221
![]() |
[82] |
De Vos NJ (2013) Echo state networks as an alternative to traditional artificial neural networks in rainfall-runoff modelling. Hydrol Earth Syst Sci 17: 253–267. https://doi.org/10.5194/hess-17-253-2013 doi: 10.5194/hess-17-253-2013
![]() |
[83] |
Raghavendra S, Deka PC (2014) Support vector machine applications in the field of hydrology: A review. Appl Soft Comput 19: 372–386. https://doi.org/10.1016/j.asoc.2014.02.002 doi: 10.1016/j.asoc.2014.02.002
![]() |
[84] |
Xiang Y, Gou L, He L, et al. (2018) A SVR–ANN combined model based on ensemble EMD for rainfall prediction. Appl Soft Comput 73: 874–883. https://doi.org/10.1016/j.asoc.2018.09.018 doi: 10.1016/j.asoc.2018.09.018
![]() |
[85] |
Üstün B, Melssen WJ, Buydens LMC (2007) Visualisation and interpretation of Support Vector Regression models. Anal Chim Acta 595: 299–309. https://doi.org/10.1016/j.aca.2007.03.023 doi: 10.1016/j.aca.2007.03.023
![]() |
[86] |
Mosavi A, Rabczuk T, Varkonyi-Koczy AR (2018) Reviewing the novel machine learning tools for materials design. Adv Intell Syst Comput 660: 50–58. https://doi.org/10.1007/978-3-319-67459-9_7 doi: 10.1007/978-3-319-67459-9_7
![]() |
[87] |
Choubin B, Moradi E, Golshan M, et al. (2019) An ensemble prediction of flood susceptibility using multivariate discriminant analysis, classification and regression trees, and support vector machines. Sci Total Environ 651: 2087–2096. https://doi.org/10.1016/j.scitotenv.2018.10.064 doi: 10.1016/j.scitotenv.2018.10.064
![]() |
[88] |
Panahi M, Dodangeh E, Rezaie F, et al. (2021) Flood spatial prediction modeling using a hybrid of meta-optimization and support vector regression modeling. Catena 199: 105114. https://doi.org/10.1016/j.catena.2020.105114 doi: 10.1016/j.catena.2020.105114
![]() |
[89] |
Liu Y, Pender G (2015) A flood inundation modelling using v-support vector machine regression model. Eng Appl Artif Intell 46: 223–231. https://doi.org/10.1016/j.engappai.2015.09.014 doi: 10.1016/j.engappai.2015.09.014
![]() |
[90] |
Shirzadi A, Asadi S, Shahabi H, et al. (2020) A novel ensemble learning based on Bayesian Belief Network coupled with an extreme learning machine for flash flood susceptibility mapping. Eng Appl Artif Intell 96: 103971. https://doi.org/10.1016/j.engappai.2020.103971 doi: 10.1016/j.engappai.2020.103971
![]() |
[91] |
Young CC, Liu WC, Wu MC (2017) A physically based and machine learning hybrid approach for accurate rainfall-runoff modeling during extreme typhoon events. Appl Soft Comput 53: 205–216. https://doi.org/10.1016/j.asoc.2016.12.052 doi: 10.1016/j.asoc.2016.12.052
![]() |
[92] |
Yan J, Jin J, Chen F, et al. (2018) Urban flash flood forecast using support vector machine and numerical simulation. J Hydroinform 20: 232–245. https://doi.org/10.2166/hydro.2017.175 doi: 10.2166/hydro.2017.175
![]() |
[93] |
Bermúdez M, Cea L, Puertas J (2019) A rapid flood inundation model for hazard mapping based on least squares support vector machine regression. J Flood Risk Manag 12: 1–14. https://doi.org/10.1111/jfr3.12522 doi: 10.1111/jfr3.12522
![]() |
[94] |
Cervantes J, Garcia-Lamont F, Rodríguez-Mazahua L, et al. (2020) A comprehensive survey on support vector machine classification: Applications, challenges and trends. Neurocomputing. https://doi.org/10.1016/j.neucom.2019.10.118 doi: 10.1016/j.neucom.2019.10.118
![]() |
[95] |
Li PH, Kwon HH, Sun L, et al. (2010) A modified support vector machine based prediction model on streamflow at the Shihmen Reservoir, Taiwan. Int J Climatol 30: 1256–1268. https://doi.org/10.1002/joc.1954 doi: 10.1002/joc.1954
![]() |
[96] |
Sahoo A, Samantaray S, Ghose DK (2021) Prediction of Flood in Barak River using Hybrid Machine Learning Approaches: A Case Study. J Geol Soc India 97: 186–198. https://doi.org/10.1007/s12594-021-1650-1 doi: 10.1007/s12594-021-1650-1
![]() |
[97] |
Costache R (2019) Flash-flood Potential Index mapping using weights of evidence, decision Trees models and their novel hybrid integration. Stoch Environ Res Risk Assess33: 1375–1402. https://doi.org/10.1007/s00477-019-01689-9 doi: 10.1007/s00477-019-01689-9
![]() |
[98] | Lawal ZK, Yassin H, Zakari RY (2021) Flood Prediction Using Machine Learning Models: A Case Study of Kebbi State Nigeria. IEEE Asia-Pacific Conference on Computer Science and Data Engineering, CSDE. https://doi.org/10.1109/CSDE53843.2021.9718497 |
[99] |
De'Ath G, Fabricius KE (2000) Classification and regression trees: A powerful yet simple technique for ecological data analysis. Ecology 81: 3178–3192. https://doi.org/10.1890/0012-9658(2000)081[3178:CARTAP]2.0.CO; 2 doi: 10.1890/0012-9658(2000)081[3178:CARTAP]2.0.CO;2
![]() |
[100] |
Chen W, Li Y, Xue W, et al. (2020) Modeling flood susceptibility using data-driven approaches of naï ve Bayes tree, alternating decision tree, and random forest methods. Sci Total Environ 701: 134979. https://doi.org/10.1016/j.scitotenv.2019.134979 doi: 10.1016/j.scitotenv.2019.134979
![]() |
[101] |
Yariyan P, Janizadeh S, Van Phong T, et al. (2020) Improvement of Best First Decision Trees Using Bagging and Dagging Ensembles for Flood Probability Mapping. Water Resour Manag 34: 3037–3053. https://doi.org/10.1007/s11269-020-02603-7 doi: 10.1007/s11269-020-02603-7
![]() |
[102] |
Zahiri A, Azamathulla HM (2014) Comparison between linear genetic programming and M5 tree models to predict flow discharge in compound channels. Neural Comput Appl 24: 413–420. https://doi.org/10.1007/s00521-012-1247-0 doi: 10.1007/s00521-012-1247-0
![]() |
[103] |
Singh KK, Pal M, Singh VP (2010) Estimation of mean annual flood in indian catchments using backpropagation neural network and M5 model tree. Water Resour Manag 24: 2007–2019. https://doi.org/10.1007/s11269-009-9535-x doi: 10.1007/s11269-009-9535-x
![]() |
[104] |
Nguyen DT, Chen ST (2020) Real-time probabilistic flood forecasting using multiple machine learning methods. Water 12: 1–13. https://doi.org/10.3390/w12030787 doi: 10.3390/w12030787
![]() |
[105] |
Alizadeh Z, Yazdi J, Kim JH, et al. (2018) Assessment of machine learning techniques for monthly flow prediction. Water 10: 1–24. https://doi.org/10.3390/w10111676 doi: 10.3390/w10111676
![]() |
[106] |
Hou J, Zhou N, Chen G, et al. (2021) Rapid forecasting of urban flood inundation using multiple machine learning models. Nat Hazards 108: 2335–2356. https://doi.org/10.1007/s11069-021-04782-x doi: 10.1007/s11069-021-04782-x
![]() |
[107] |
Sankaranarayanan S, Prabhakar M, Satish S, et al. (2020) Flood prediction based on weather parameters using deep learning. J Water Clim Change 11: 1766–1783. https://doi.org/10.2166/wcc.2019.321 doi: 10.2166/wcc.2019.321
![]() |
[108] |
El-Magd SAA, Pradhan B, Alamri A (2021) Machine learning algorithm for flash flood prediction mapping in Wadi El-Laqeita and surroundings, Central Eastern Desert, Egypt. Arab J Geosci 14. https://doi.org/10.1007/s12517-021-06466-z doi: 10.1007/s12517-021-06466-z
![]() |
[109] |
Huang M, Lin R, Huang S, et al. (2017) A novel approach for precipitation forecast via improved K-nearest neighbor algorithm. Adv Eng Inform 33: 89–95. https://doi.org/10.1016/j.aei.2017.05.003 doi: 10.1016/j.aei.2017.05.003
![]() |
[110] |
Cuomo S, Di Cola VS, Giampaolo F, et al. (2022) Scientific Machine Learning Through Physics–Informed Neural Networks: Where we are and What's Next. J Sci Comput 92: 88. https://doi.org/10.1007/s10915-022-01939-z doi: 10.1007/s10915-022-01939-z
![]() |
[111] |
Michele A, Colin V, Santika DD (2019) MobileNet Convolutional Neural Networks and Support Vector Machines for Palmprint Recognition. Procedia Comput Sci 157: 110–117. https://doi.org/10.1016/j.procs.2019.08.147 doi: 10.1016/j.procs.2019.08.147
![]() |
[112] |
Gao X, Shan C, Hu C, et al. (2019) An Adaptive Ensemble Machine Learning Model for Intrusion Detection. IEEE Access 7: 82512–82521. https://doi.org/10.1109/ACCESS.2019.2923640 doi: 10.1109/ACCESS.2019.2923640
![]() |
[113] |
Shanmugasundar G, Vanitha M, Čep R, et al. (2021) A Comparative Study of Linear, Random Forest and AdaBoost Regressions for Modeling Non-Traditional Machining. Processes 9: 2015. https://doi.org/10.3390/pr9112015 doi: 10.3390/pr9112015
![]() |
[114] |
Triguero I, García‐Gil D, Maillo J, et al. (2019) Transforming big data into smart data: An insight on the use of the k‐nearest neighbors algorithm to obtain quality data. WIREs Data Min Knowl Discov. https://doi.org/10.1002/widm.1289 doi: 10.1002/widm.1289
![]() |
[115] |
Senthilnath J, Shreyas PB, Rajendra R, et al. (2019) Hierarchical clustering approaches for flood assessment using multi-sensor satellite images. Int J Image Data Fusion 10: 28–44. https://doi.org/10.1080/19479832.2018.1513956 doi: 10.1080/19479832.2018.1513956
![]() |
[116] |
Rahman AS, Rahman A (2020) Application of principal component analysis and cluster analysis in regional flood frequency analysis: A case study in new South Wales, Australia. Water 12: 1–26. https://doi.org/10.3390/w12030781 doi: 10.3390/w12030781
![]() |
[117] |
Engelen JE, Hoos HH (2020) A survey on semi-supervised learning. Mach Learn 109: 373–440. https://doi.org/10.1007/s10994-019-05855-6 doi: 10.1007/s10994-019-05855-6
![]() |
[118] |
Inyang UG, Akpan EE, Akinyokun OC (2020) A Hybrid Machine Learning Approach for Flood Risk Assessment and Classification. Int J Comput Intell Appl 19: 1–20. https://doi.org/10.1142/S1469026820500121 doi: 10.1142/S1469026820500121
![]() |
[119] |
Devi G, Sharma M, Sarma P, et al. (2022) Flood Frequency Modeling and Prediction of Beki and Pagladia Rivers Using Deep Learning Approach. Neural Process Lett. https://doi.org/10.1007/s11063-022-10773-1 doi: 10.1007/s11063-022-10773-1
![]() |
[120] |
He W, Jiang Z (2022) Semi-Supervised Learning With the EM Algorithm: A Comparative Study Between Unstructured and Structured Prediction. IEEE Trans Knowl Data Eng 34: 2912–2920. https://doi.org/10.1109/TKDE.2020.3019038 doi: 10.1109/TKDE.2020.3019038
![]() |
[121] |
Zhao G, Pang B, Xu Z, et al. (2019) Assessment of urban flood susceptibility using semi-supervised machine learning model. Sci Total Environ 659: 940–949. https://doi.org/10.1016/j.scitotenv.2018.12.217 doi: 10.1016/j.scitotenv.2018.12.217
![]() |
[122] |
Silver D, Singh S, Precup D, et al. (2021) Reward is enough. Artificial Intelligence 299: 103535. https://doi.org/10.1016/j.artint.2021.103535 doi: 10.1016/j.artint.2021.103535
![]() |
[123] |
Serrano W (2022) Deep Reinforcement Learning with the Random Neural Network. Eng Appl Artif Intell 110: 104751. https://doi.org/10.1016/j.engappai.2022.104751 doi: 10.1016/j.engappai.2022.104751
![]() |
[124] |
Bowes BD, Tavakoli A, Wang C, et al. (2021) Flood mitigation in coastal urban catchments using real-time stormwater infrastructure control and reinforcement learning. J Hydroinform 23: 529–547. https://doi.org/10.2166/HYDRO.2020.080 doi: 10.2166/HYDRO.2020.080
![]() |
[125] |
Baldazo D, Parras J, Zazo S (2019) Decentralized multi-agent deep reinforcement learning in swarms of drones for flood monitoring. Eur Signal Process Conf. https://doi.org/10.23919/EUSIPCO.2019.8903068 doi: 10.23919/EUSIPCO.2019.8903068
![]() |
[126] |
Hapuarachchi HAP, Wang QJ, Pagano TC (2011) A review of advances in flash flood forecasting. Hydrol Process 25: 2771–2784. https://doi.org/10.1002/hyp.8040 doi: 10.1002/hyp.8040
![]() |
[127] |
Sood A, Smakhtin V (2015) Global hydrological models: a review. Hydrol Sci J 60: 549–565. https://doi.org/10.1080/02626667.2014.950580 doi: 10.1080/02626667.2014.950580
![]() |
[128] |
Grimaldi S, Li Y, Pauwels VRN, et al. (2016) Remote Sensing-Derived Water Extent and Level to Constrain Hydraulic Flood Forecasting Models: Opportunities and Challenges. Surv Geophys 37: 977–1034. https://doi.org/10.1007/s10712-016-9378-y doi: 10.1007/s10712-016-9378-y
![]() |
[129] |
Munawar HS, Hammad AWA, Waller ST (2022) Remote Sensing Methods for Flood Prediction: A Review. Sensors 22. https://doi.org/10.3390/s22030960 doi: 10.3390/s22030960
![]() |
[130] |
Yuan Q, Shen H, Li T, et al. (2020) Deep learning in environmental remote sensing: Achievements and challenges. Remote Sens Environ 241: 111716. https://doi.org/10.1016/j.rse.2020.111716 doi: 10.1016/j.rse.2020.111716
![]() |
[131] |
Maggioni V, Massari C (2018) On the performance of satellite precipitation products in riverine flood modeling: A review. J Hydrol 558: 214–224. https://doi.org/10.1016/j.jhydrol.2018.01.039 doi: 10.1016/j.jhydrol.2018.01.039
![]() |
[132] |
Olson D, Anderson J (2021) Review on unmanned aerial vehicles, remote sensors, imagery processing, and their applications in agriculture. Agron J 113: 971–992. https://doi.org/10.1002/agj2.20595 doi: 10.1002/agj2.20595
![]() |
[133] |
Justice CO, Townshend JRG, Vermote EF, et al. (2002) An overview of MODIS Land data processing and product status. Remote Sens Environ 83: 3–15. https://doi.org/10.1016/S0034-4257(02)00084-6 doi: 10.1016/S0034-4257(02)00084-6
![]() |
[134] |
Qi W, Ma C, Xu H, et al. (2021) A review on applications of urban flood models in flood mitigation strategies. Nat Hazards 108: 31–62. https://doi.org/10.1007/s11069-021-04715-8 doi: 10.1007/s11069-021-04715-8
![]() |
[135] |
Tien Bui D, Hoang ND, Martínez-Álvarez F, et al. (2020) A novel deep learning neural network approach for predicting flash flood susceptibility: A case study at a high frequency tropical storm area. Sci Total Environ 701: 134413. https://doi.org/10.1016/j.scitotenv.2019.134413 doi: 10.1016/j.scitotenv.2019.134413
![]() |
[136] | Xu T, Liang F (2021) Machine learning for hydrologic sciences: An introductory overview. Wiley Interdisciplinary Reviews: Water 8: 1–29. https://doi.org/10.1002/wat2.1533 |
[137] | Shen C, Lawson K (2021) Applications of Deep Learning in Hydrology. Deep Learn Earth Sci, Wiley 283–297. https://doi.org/10.1002/9781119646181.ch19 |
[138] |
Sit M, Demiray BZ, Xiang Z, et al. (2020) A comprehensive review of deep learning applications in hydrology and water resources. Water Sci Technol 82: 2635–2670. https://doi.org/10.2166/wst.2020.369. doi: 10.2166/wst.2020.369
![]() |
[139] |
Zounemat-Kermani M, Batelaan O, Fadaee M, et al. (2021) Ensemble machine learning paradigms in hydrology: A review. J Hydrol 598: 126266. https://doi.org/10.1016/j.jhydrol.2021.126266 doi: 10.1016/j.jhydrol.2021.126266
![]() |
[140] |
Merz B, Kreibich H, Schwarze R, et al. (2010) Review article Assessment of economic flood damage. Nat Hazards Earth Syst Sci 10: 1697–1724. https://doi.org/10.5194/nhess-10-1697-2010 doi: 10.5194/nhess-10-1697-2010
![]() |
[141] |
Tarasova L, Merz R, Kiss A, et al. (2019) Causative classification of river flood events. WIREs Water 6. https://doi.org/10.1002/wat2.1353 doi: 10.1002/wat2.1353
![]() |
[142] |
Mahmoodi N, Wagner PD, Kiesel J, et al. (2021) Modeling the impact of climate change on streamflow and major hydrological components of an Iranian Wadi system. J Water Clim Change 12: 1598–1613. https://doi.org/10.2166/wcc.2020.098 doi: 10.2166/wcc.2020.098
![]() |
[143] |
Mangukiya NK, Sharma A (2022) Flood risk mapping for the lower Narmada basin in India: a machine learning and IoT-based framework. Nat Hazards 113: 1285–1304. https://doi.org/10.1007/s11069-022-05347-2 doi: 10.1007/s11069-022-05347-2
![]() |
[144] |
Alipour MH (2015) Risk-Informed Decision Making Framework for Operating a Multi-Purpose Hydropower Reservoir During Flooding and High Inflow Events, Case Study: Cheakamus River System. Water Resour Manag 29: 801–815. https://doi.org/10.1007/s11269-014-0844-3 doi: 10.1007/s11269-014-0844-3
![]() |
[145] |
Coates G, Li C, Ahilan S, et al. (2019) Agent-based modeling and simulation to assess flood preparedness and recovery of manufacturing small and medium-sized enterprises. Eng Appl Artif Intell 78: 195–217. https://doi.org/10.1016/j.engappai.2018.11.010 doi: 10.1016/j.engappai.2018.11.010
![]() |
[146] |
Kienzler S, Pech I, Kreibich H, et al. (2015) After the extreme flood in 2002: changes in preparedness, response and recovery of flood-affected residents in Germany between 2005 and 2011. Nat Hazards Earth Syst Sci 15: 505–526. https://doi.org/10.5194/nhess-15-505-2015 doi: 10.5194/nhess-15-505-2015
![]() |
[147] | Packer C, Gao K, Kos J, et al. (2018) Assessing Generalization in Deep Reinforcement Learning. |
[148] |
Ma K, Feng D, Lawson K, et al. (2021) Transferring Hydrologic Data Across Continents – Leveraging Data‐Rich Regions to Improve Hydrologic Prediction in Data‐Sparse Regions. Water Resour Res 57: e2020WR028600. https://doi.org/10.1029/2020WR028600 doi: 10.1029/2020WR028600
![]() |
[149] |
Weiss K, Khoshgoftaar TM, Wang D (2016) A survey of transfer learning. J Big Data 3: 9. https://doi.org/10.1186/s40537-016-0043-6 doi: 10.1186/s40537-016-0043-6
![]() |
[150] |
Boelee L, Lumbroso DM, Samuels PG, et al. (2019) Estimation of uncertainty in flood forecasts—A comparison of methods. J Flood Risk Manag 12. https://doi.org/10.1111/jfr3.12516 doi: 10.1111/jfr3.12516
![]() |
[151] | Nevo S, Anisimov V, Elidan G, et al. (2019) ML for Flood Forecasting at Scale. |
[152] |
Hardy J, Gourley JJ, Kirstetter P-E, et al. (2016) A method for probabilistic flash flood forecasting. J Hydrol 541: 480–494. https://doi.org/10.1016/j.jhydrol.2016.04.007 doi: 10.1016/j.jhydrol.2016.04.007
![]() |
[153] |
Han S, Coulibaly P (2019) Probabilistic Flood Forecasting Using Hydrologic Uncertainty Processor with Ensemble Weather Forecasts. J Hydrometeorol 20: 1379–1398. https://doi.org/10.1175/JHM-D-18-0251.1 doi: 10.1175/JHM-D-18-0251.1
![]() |
[154] |
Zhan X, Qin H, Liu Y, et al. (2020) Variational Bayesian Neural Network for Ensemble Flood Forecasting. Water 12: 2740. https://doi.org/10.3390/w12102740 doi: 10.3390/w12102740
![]() |
[155] |
Ivanov VY, Xu D, Dwelle MC, et al. (2021) Breaking Down the Computational Barriers to Real‐Time Urban Flood Forecasting. Geophys Res Lett 48. https://doi.org/10.1029/2021GL093585 doi: 10.1029/2021GL093585
![]() |
[156] |
Wang H, Chen Y (2019) Identifying Key Hydrological Processes in Highly Urbanized Watersheds for Flood Forecasting with a Distributed Hydrological Model. Water 11: 1641. https://doi.org/10.3390/w11081641 doi: 10.3390/w11081641
![]() |
[157] |
Liu Z, Felton T, Mostafavi A (2024) Interpretable machine learning for predicting urban flash flood hotspots using intertwined land and built-environment features. Comput Environ Urban Syst110: 102096. https://doi.org/10.1016/j.compenvurbsys.2024.102096 doi: 10.1016/j.compenvurbsys.2024.102096
![]() |
[158] |
Ding Y, Zhu Y, Feng J, et al. (2020) Interpretable spatio-temporal attention LSTM model for flood forecasting. Neurocomputing 403: 348–359. https://doi.org/10.1016/j.neucom.2020.04.110 doi: 10.1016/j.neucom.2020.04.110
![]() |
[159] | Vollert S, Atzmueller M, Theissler A (2021) Interpretable Machine Learning: A brief survey from the predictive maintenance perspective, 26th IEEE Int Conf Emerg Technol Fact Autom (ETFA) 01–08. https://doi.org/10.1109/ETFA45728.2021.9613467 |
[160] |
Motta M, de Castro Neto M, Sarmento P (2021) A mixed approach for urban flood prediction using Machine Learning and GIS. Int J Disaster Risk Reduct 56: 102154. https://doi.org/10.1016/j.ijdrr.2021.102154 doi: 10.1016/j.ijdrr.2021.102154
![]() |
[161] |
Qiao L, Livsey D, Wise J, et al. (2024) Predicting flood stages in watersheds with different scales using hourly rainfall dataset: A high-volume rainfall features empowered machine learning approach. Sci Total Environ 950: 175231. https://doi.org/10.1016/j.scitotenv.2024.175231 doi: 10.1016/j.scitotenv.2024.175231
![]() |
[162] |
Khaniya B, Gunathilake MB, Rathnayake U (2021) Ecosystem-Based adaptation for the impact of climate change and variation in the water management sector of Sri Lanka. Math Probl Eng, 2021: 1–10. https://doi.org/10.1155/2021/8821329 doi: 10.1155/2021/8821329
![]() |
[163] |
Islam ARMT, Talukdar S, Mahato S, et al. (2021) Flood susceptibility modelling using advanced ensemble machine learning models. Geosci Front, 12(3): 101075. https://doi.org/https://doi.org/10.1016/j.gsf.2020.09.006 doi: 10.1016/j.gsf.2020.09.006
![]() |
1. | Marco Diociaiuti, Cecilia Bombelli, Laura Zanetti-Polzi, Marcello Belfiore, Raoul Fioravanti, Gianfranco Macchia, Cristiano Giordani, The Interaction between Amyloid Prefibrillar Oligomers of Salmon Calcitonin and a Lipid-Raft Model: Molecular Mechanisms Leading to Membrane Damage, Ca2+-Influx and Neurotoxicity, 2019, 10, 2218-273X, 58, 10.3390/biom10010058 | |
2. | Richard Lantz, Brian Busbee, Ewa P. Wojcikiewicz, Deguo Du, Effects of disulfide bond and cholesterol derivatives on human calcitonin amyloid formation, 2020, 111, 0006-3525, 10.1002/bip.23343 | |
3. | Igor Krivoi, Alexey Petrov, Cholesterol and the Safety Factor for Neuromuscular Transmission, 2019, 20, 1422-0067, 1046, 10.3390/ijms20051046 | |
4. | Lipika Mirdha, Aggregation Behavior of Amyloid Beta Peptide Depends Upon the Membrane Lipid Composition, 2024, 257, 0022-2631, 151, 10.1007/s00232-024-00314-3 |