Abstract
Landslides are destructive natural disasters that cause human and economic losses. Although many studies report the effects of forest age on landslide susceptibility, especially for shallow landslides, no studies have examined the effects at a national scale. We assumed that temporal variations in the annual number of rainfall-triggered landslides in Japan were determined by variations in rainfall and forest age distribution. By this assumption, this study aimed to quantify the decrease in the frequency of rainfall-induced landslides owing to the increasing maturity of forests in Japan. Data were collated from 21 studies covering 11 sites in three countries that reported a landslide susceptibility index (i.e., frequency ratio or landslide density) and the relation between forest age and the normalized landslide susceptibility index (NLSI) was modeled. Using this relation and the area for each forest age class, the change in landslide susceptibility at a national scale (NLSIJpn) was quantified during 1966–2017. The authors developed generalized linear models (GLMs) using the annual number of landslides as the response variable and the NLSIJpn and a rainfall index for each year as the explanatory variables. The number of rainfall-induced landslides was simulated in the GLMs in 15 scenarios with different forest age distributions and rainfall amounts. The number of landslides in young-age-dominated and middle-age-dominated forests was estimated to be 2.4 and 1.1 times, respectively, that in mature-age-dominated forests. The change in the number of landslides from young-age-dominated to mature-age-dominated forests was larger than that from an increase in the rainfall amount of 20%. We conclude that increasing the maturity of forests greatly reduces landslide frequency in Japan. In a changing climate with potentially threatening increases in rainfall, preserving mature forests is important to avoid amplifying landslide susceptibility on a national scale.
Similar content being viewed by others
1 Introduction
Landslides are categorized as destructive natural disasters and are heterogeneously distributed in the world (Petley 2012; Froude and Petley 2018). Although floods are the deadliest natural disaster (Jonkman and Kelman 2005), the number of deaths from landslides is greater than that from floods in some regions and countries such as East Uganda (Agrawal et al. 2013), Italy (Guzzetti et al. 2005), and Japan (Shinohara and Kume 2022). Landslide susceptibility is the likelihood of a landslide occurrence in an area based on local and environmental conditions (Reichenbach et al. 2018). The evaluation of causative factors affecting landslide susceptibility is an initial step for reducing landslides.
Land use is a factor affecting landslide susceptibility that may change considerably owing to human activity and natural disasters (Guns and Vanacker 2013), unlike other geo-environmental factors (e.g., morphology and geology). The presence of forests and trees contributes to slope stability owing to both hydrological and mechanical processes (Cislaghi et al. 2021). In most temperate regions, the latter contribution is much larger than the former (Stokes et al. 2009). Soil shear strength is determined by cohesion, friction angle, and total stress on the shear surface (Zhang et al. 2001). Root reinforcement, represented as a function of the empirical root-soil interaction factor, root-area ratio, and root tensile strength (Waldron 1977; Wu et al. 1979), increases soil shear strength, commonly by increasing cohesion (O’loughlin and Ziemer 1982). Therefore, landslide susceptibility in forested areas is generally lower than that in other land-use types (Shu et al. 2019; Lusiana and Shinohara 2022).
Landslide susceptibility may also differ depending on forest characteristics, such as species (Fattet et al. 2011; Ji et al. 2012), density (Douglas et al. 2013; Spiekermann et al. 2022), height (Evans et al. 2020), age (Genet et al. 2008), and forest management strategies (Cislaghi et al. 2021). Landslide susceptibility in mature forests is generally smaller than that in younger forests owing to thicker, deeper, and greater amounts of roots. This indicates that changes in root strength caused by changes in forest age after harvesting will affect landslide susceptibility (Brardinoni et al. 2002). Root reinforcement gradually reduces after harvesting owing to root decay (Cislaghi et al. 2021). The increase in landslide susceptibility occurs during the period of minimum root strength owing to root decay after harvesting and prior to substantial regeneration (Sidle and Ochiai 2006). The susceptibility gradually decreases over time as root strength increases during regrowth of planted and natural forests (Gorsevski et al. 2006; Turner et al. 2010).
Many studies have reported changes in landslide susceptibility owing to changes in forest age distribution on the local scale (Lee and Lee 2006; Imaizumi et al. 2008; Turner et al. 2010); some of them (e.g., Zhang et al. 2022) have analyzed multiple causative factors. On the national scale, Shinohara and Kume (2022) examined the trends in the number of landslide fatalities and underlying factors, including forest maturity, and found that increasing forest maturity reduced landslide numbers and fatalities in Japan, especially from 1960 to 1989. Sato and Shuin (2022) obtained a similar result using the data for forest growing stocks and flooded areas. However, these findings were obtained based on comparing trends, and no studies since have quantified the effect of increasing forest maturity on landslides on a national scale. The contribution of forest maturity to reducing landslide susceptibility on a national scale should be determined because this information is of fundamental importance in designing and implementing risk management policies (Malet et al. 2013), particularly in providing proper recommendations for future forest management-affected landslides over the whole of Japan.
Forest harvesting and rainfall are key factors affecting slope stability owing to the removal of root reinforcement and the increase in soil saturation, respectively (Malek et al. 2015; Barik et al. 2017). Because a change in rainfall regimes (i.e., an increase in rainfall) is expected in the future owing to the warmer climate, the rates of rainfall-triggered landslides are consequently predicted to increase (Gariano and Guzzetti 2016). To control rising landslide frequency, we hypothesized that appropriate forest management by adjusting age distribution may effectively mitigate the impact of climate change on landslide hazards. This study aimed to quantify the effects of an increase in forest maturity in different forest age distributions on rainfall-triggered landslides in Japan. In Japan, the forested area has been constant since from 1966 to 2017 and forests have become more mature. The authors first collected case studies reporting landslide susceptibility in several forest age classes and developed a general relation between landslide susceptibility and forest age. Then, we developed models to estimate the annual number of rainfall-triggered landslides in Japan using a rainfall index and forest age distributions. Finally, we simulated potential changes in the number of landslides owing to changes in forest age distributions and rainfall.
2 Materials and methods
2.1 Developing the relation between the forest age and landslide susceptibility
Landslide susceptibility is usually represented as a frequency ratio (FR) (i.e., the ratio of the percentage landslide area for a class to the percentage landslide area for the site) or landslide density (the number of landslides / the total area for a class). Assuming the mean of a landslide area (i.e., the total landslide area / the number of landslides) does not change owing to the forest age, both FR and the landslide density can be treated as equal when examining the relation between the forest age and landslide susceptibility. In this study, FR and the landslide density are collectively referred to as the landslide susceptibility index (LSI).
Case studies reporting FR or landslide density in several forest age classes were collected by referring to the SCOPUS and ISI Web of Knowledge databases using landslide susceptibility, landslide density, and forest age as keyword searches. It is known that landslide susceptibility generally varies when the forest age is less than 20 years (Sidle and Wu 1999; Dhakal and Sidle 2003; Turner et al. 2010). Considering this and the data availability from previous studies, the authors set the maximum forest age as 60 years and established six age classes with 10-year intervals (i.e., 1–10, 11–20, 21–30, 31–40, 41–50, and 51–60 years). Therefore, only studies with an interval of 5 or 10 years were collected. For studies with an interval of 5 years, the LSI for each forest age class with an interval of 10 years was calculated by averaging LSI for two subject classes. In total, 22 datasets were collected. Some datasets targeted the same rainfall event and location as other studies (or another study). For these cases, the average LSI for each forest age category was used for analysis.
The magnitude of LSI was different because of site-specific causative factors. Therefore, the normalized LSI (NLSI) was calculated for each age class and the LSI of 21–30 years, available in all datasets, was set as 1. Using all data available for each forest age class, the average of the NLSI was calculated. The relations between the forest age and average NLSI were examined using four types of regression (i.e., power, logarithmic, exponential, and linear regression). When examining the regression models, the forest age classes of 1–10, 11–20, 21–30, 31–40, 41–50, and 51–60 years were set at 5, 15, 25, 35, 45, and 55 years, respectively. Using the regressions, NLSI was calculated for each year between 1 and 60 years. The NLSIs based on power, logarithmic, exponential, and linear regressions are referred to as NLSIpower, NLSIlog, NLSIexp, and NLSIlinear, respectively.
2.2 Developing models to estimate the annual number of rainfall-triggered landslides in Japan
Generalized linear models (GLMs) were developed to estimate variations in the annual number of rainfall-triggered landslides in Japan from 1966 to 2017. In Japan, 66% of the total land is covered by forests and the forested area remained almost constant during the analysis period. Geologically, the accretionary complexes, metamorphic rocks, plutonic and volcanic rocks, and surface sediments have an extremely complex distribution. The annual precipitation and average temperature are 1,650 mm and 15.5 °C, respectively. Detailed geological and meteorological conditions are available in Shinohara and Watanabe (2023). When developing the models, we assumed that the temporal change in the number of landslides was determined by forest age distributions and rainfall. The validity of this assumption is discussed in Sect. 4. Although the definition of forest age was not available for most datasets used in this study, the forest age for planted and natural forests were expected to be the number of years after planting and after clearcutting, respectively.
The area of forest age classes at 5-year intervals was available for planted forests in 12 years between 1966 and 2017 and for natural forests in 2012 and 2017. Using these data, the areas of planted and natural forests in 13 forest age classes (i.e., 1–5, 6–10, 11–15, 16–20, 21–25, 26–30, 31–35, 36–40, 41–45, 46–50, 51–55, 56–60, and > 60 years) were estimated for every year during the analysis period. The detailed procedures for the calculation and the estimated area for each age class are available in Supplementary Information Text S1 and Fig. S1, respectively.
Using the areas of planted and natural forests (PFAi and NFAi, respectively) and the NLSI (NLSIi) for each age class i (with an interval of 5 years), we calculated the national-scale NLSI (NLSIJpn) for every year from 1966 to 2017.
where n is the number of the forest age classes (= 13). NLSIi, except NLSI13, is set as the average NLSI over the 5 years including in the interval. NLSI13 (i.e., NLSI for the age of > 60 years) is set as NLSI at the age of 60 years.
A rainfall index proposed by Shinohara and Komatsu (2016)—the total rainfall from May to October in northern, eastern, and western Japan—was used as the other explanatory variable. The rainfall index has a good correlation with the annual number of rainfall-triggered landslides in Japan (Shinohara and Komatsu 2016; Shinohara and Kume 2022). For the response variable, the number of rainfall-triggered landslides, reported by the Sabo and landslide Technical Center (STC), was used. The dataset includes landslides that affected houses or public facilities and is the most accurate among the four landslide datasets in Japan (Shinohara and Kume 2022). The dataset is available from 1982 to 2017 only. The number of landslides from 1966 to 1981 was estimated using the dataset of the National Police Agency (NPA) and the relation between the number by STC and that by NPA during 1982–2017 (r = 0.76, P < 0.05). The sources of datasets for forests, rainfall, and the number of landslides can be found in Table S1.
In the GLMs, a gamma distribution was assumed, and the logarithm link function was selected (Atkinson and Massari 1998). The explanatory variables (i.e., NLSIJpn and rainfall index) were selected using a stepwise method based on the Akaike Information Criterion (AIC). The models were developed using the four types of NLSI (see Sect. 2.1). The AIC, correlation coefficient (r), and root mean square error (RMSE) were used to evaluate the models. All analysis was calculated using the R program version 4.2.1 (R Core Team 2023).
2.3 Simulation of landslide frequency with different forest age distributions and rainfall amount scenarios
To quantify potential changes in the landslide frequency owing to changes in the forest age distribution, three scenarios of forest age distributions were prepared as inputs for the GLMs. In the first scenario, young-age-dominated forests were assumed using the forest age distribution in 1966 (Fyoung). In the second scenario, middle-age-dominated forests were assumed using the forest age distribution in 2002 (Fmiddle). In the third scenario, mature-age-dominated forests were assumed using the forest age distribution in 2017 (Fmature). Here, young-, middle-, and mature-ages corresponded to 1–20, 21–40, and > 40 years, respectively. Assuming a normal distribution with the mean and standard deviation (1046.2 ± 157.7 mm) for the rainfall index during 1966–2017 (Rcurrent), a datum was extracted using random numbers. The extraction was repeated 1,000 times and used as the input for the GLMs. Thus, the average and standard deviation of the number of landslides were obtained for the three forest distribution scenarios. In addition, four rainfall scenarios were prepared with an increase in the mean of 10% (R+ 10%) and 20% (R+ 20%) and a decrease in the mean of 10% (R–10%) and 20% (R–20%). For all scenarios, the standard deviation was assumed to be constant, and the extracted datum was repeated 1,000 times. The number of landslides was calculated for 15 scenarios with combinations of three forest age distribution (i.e., Fyoung, Fmiddle, and Fmature) and five rainfall (i.e., Rcurrent, R− 20%, R− 10%, R+ 10%, and R+ 20%) scenarios. The above simulation was conducted using the four types of NLSI.
3 Results
3.1 The relation between forest age and landslide susceptibility
We obtained 21 studies covering 11 sites that provided LSI in multiple forest age classes (Table 1). Among the 21 studies, 19 studies were conducted in South Korea, and the other two were in the United States and Japan. The 19 studies in South Korea reported FR and the two studies in the United States and Japan described the landslide density. Although not all studies indicated the types or species of the forests, these studies were conducted in both planted and natural forests and in both coniferous and broadleaved forests.
The LSI in all age classes for the 11 sites ranged from 0 to 18.00 (Fig. 1a), while the NLSI ranged from 0.13 to 4.93 (Fig. 1b). The NLSI for younger age classes tended to be larger than that for older age classes. For six sites, the highest NLSI was observed in the age class of 1–10 years. In 10 out of the 11 sites, the highest NLSI was observed in the age classes < 30 years. The relation between the forest age and NLSI was examined for the four types of regression (Fig. 2). The linear regression had the highest coefficient of determination (0.91), followed by the exponential (0.89), logarithmic (0.85), and power (0.74) regressions.
3.2 Development of models for estimating the annual number of rainfall-triggered landslides
The ratio of younger or older forests to the total changed substantially during the 51 years (Fig. 3a). The ratio of young forests decreased from 51% in 1966 to 3% in 2017. The ratio of mature forests increased from 29% in 1966 to 85% in 2017. Consequently, in all four types of NLSI, NLSIJpn decreased from 1966 to 2017. In 1966, NLSIJpn ranged from 2743.9 to 3209.7. In 2017, NLSIJpn ranged from 1288.1 to 1674.4. The maximum NLSIJpn was observed between 1966 and 1967. The average rainfall index during the analyzed period was 1046.2 mm with large year-to-year variations (Fig. 3b), whereas the trend of landslide numbers declined within the 51 years with an average of 1059 landslides per year. The average landslides were 1411 and 903 per year within 1966–1981 and 1982–2017, respectively (Fig. 3c). The number of landslides in 1972 was much larger than those in the years around 1972, probably because of the huge numbers of landslides over wide areas of Japan in that rainy season (Tani 1973), although the impact did not appear in the rainfall index.
The results from the GLMs using the four types of NLSIJpn are shown in Table 2. For all cases, both rainfall and NLSIJpn were selected as significant parameters (P < 0.01). The model using NLSIpower had the lowest AIC and RMSE and highest r values among the four cases.
3.3 Effect of differences in forest age distributions and rainfall on variability in landslide frequency
The areas of each forest age class for the three scenarios are shown in Fig. 4. Note that the area of each planted and natural forest is available in Fig. S1. In the current rainfall and forest age distribution scenarios (Rcurrent and Fmature), the number was not significantly different among the four types of NLSI (ANOVA, P = 0.54), as shown in Fig. S3. Therefore, the GLM using NLSI power, with the lowest AIC and RMSE and highest r values, was used in the following analysis. Results from the GLMs using NLSIlog, NLSIexp, and NLSIlinear, are shown in Table S2. Table 3; Fig. 5 indicate the number of landslides estimated in the combinations of five rainfall and three forest age distribution scenarios. The number of landslides based on the scenarios Fyoung and Fmiddle was estimated as ca. 2.4 and ca. 1.1 times that of the scenario Fmature. Under all forest age distribution scenarios, the number of landslides based on the scenarios R+ 20% and R+ 10% was estimated as ca. 1.7 and 1.3 times that of the scenario to Rcurrent. The ratio of Fmiddle to Fmature (ca. 1.1) was slightly smaller than that of Rcurrent to R+ 10% (ca. 1.3). The ratio of Fyoung to Fmature (ca. 2.4) was considerably larger than that of Rcurrent to R+ 20% (ca. 1.7).
4 Discussion
The relation was examined between the forest age and NLSI based on 21 previous studies in 11 sites. Using the relation and the area of each age class, we estimated changes in NLSIJpn between 1966 and 2017. The GLMs for estimating the number of landslides were developed from NLSIJpn and a rainfall index. On the basis of the GLMs, the authors found that the age distribution of forests potentially had a strong effect on landslide frequency. The number with Fmature, corresponding to the condition in 2017, was simulated to be 41% of that with Fyoung, corresponding to the condition in 1966. Thus, we succeeded in quantifying the reduction in landslide frequency through an increase in the maturity of forests at a national scale.
Shinohara and Kume (2022) speculated that the increase in forest maturity contributed to the decrease in the number of landslides in Japan between 1960 and 1989. Sato and Shuin (2022) concluded that there was a contribution from increasing forest carbon stocks to the reduction of landslides for the period before 1990. In our simulation (Table 3; Fig. 5), the number of landslides with Fyoung corresponding to the condition in 1966 was estimated at 2.2 times that of Fmiddle, corresponding to the condition in 2002. The result showed that the increase in forest maturity did contribute to the reduction in landslide frequency from 1966 to 2002, similar to the results in the two studies mentioned above. In addition, the number of landslides with Fmiddle was simulated to be 1.1 times that with Fmature, corresponding to the condition in 2017. The result suggests that increasing the maturity of forests potentially reduced landslide frequency even after 1990. The average value of rainfall index in the 2010s (= 1141.8 mm) was higher than that in the 1990s (= 1055.3 mm) and 2000s (= 1016.2 mm) (Fig. 3b). The higher amounts of rainfall might have offset the reduction in landslide numbers owing to increasing the maturity of forests after 1990.
When developing the relation between the forest age and NLSI, we collected data suitable for our criteria, that is, reporting the landslide density or FR with intervals of 5 or 10 years. Other studies that did not meet our criteria also reported decreases in landslide susceptibility owing to increasing forest age. For example, Gorsevski et al. (2006) examined landslide susceptibility for each class in north-central Idaho, USA, and found that the landslide susceptible area for the 0–5 and 6–10 year stand age categories was 2–3 times larger than that in older categories under rainstorm conditions. In addition, a study by the Oregon Board of Forestry (2001) showed that the landslide density in 0–9 years stand age forest was 2.5 times larger compared with a stand age of 31‒100 years. Thus, the relation we found may be similar in forests around the world.
The current study assumed a monotonic decrease in NLSI owing to the increase in forest age from 0 to 60 years (Fig. 2) and NLSI at ages of > 60 years was assumed to be the same as NLSI at an age of 60 years. However, because the forest development stage following “overmatured” is “decay” (Jacob et al. 2013), it is possible that NLSI at ages of > 60 years might have different trends. Furthermore, Cislaghi et al. (2021) suggested that root reinforcements could vary owing to the treatment of forests even at the same age. These factors were not considered in our study because of data limitations. This study also did not consider changes in root strength after harvesting and natural hazards (e.g., windstorms). When mature forests are harvested, the root reinforcement initially decreases owing to the root decay of the harvested trees and then increases owing to the root growth of the planted (or regenerated) trees. The rate of root reinforcement decay would be different depending on species, environmental conditions, and/or methodology for measuring root reinforcement (O’loughlin and Watson 1979; Cislaghi et al. 2021). In Japan, Kitamura and Namba (1981) examined the root pullout strength for living trees of various ages and stumps after various times in three coniferous and two broadleaved forest types and demonstrated that the total root reinforcement was at its minimum 5‒10 years after harvesting. Okada et al. (2023) reported that the minimum root reinforcement was approximately 9 years after harvesting in Japanese cedar trees. Imaizumi et al. (2008) described the landslide frequency in forests aged 26‒40 years as equal to control sites. This indicates that more than 20 years after forest harvesting, the root strength had almost completely returned to 0 years or the condition before harvesting (Kitamura and Namba 1981; Sidle and Ochiai 2006). In the current study, we calculated NLSIJpn simply considering the process of root regrowth and decay. In this calculation, we assumed that (1) NLSIs at 0 and 10 years were the same as NLSIlinear at 25 and 10 years, respectively; (2) NLSI increased linearly from 0 to 10 years; and (3) NLSI more than 10 years followed NLSIlinear. Although the recalculated NLSIJpn was different from the originals in the initial period of our analysis (Fig. S4), the estimated parameters of the GLMs (Table S3) and the simulated number of landslides (Table S4) were not altered. Thus, we verified that our conclusion remained the same, even if we used NLSIJpn on the assumption of root regrowth and decay.
In our simulation (Table 3; Fig. 5), the increase in the number of landslides from Fmature to Fmiddle under any rainfall scenario was slightly smaller than that from Rcurrent to R+ 10% (or from R–10% to Rcurrent) under any forest age distribution scenario. In contrast, the increase in the number of landslides from Fmature to Fyoung under any rainfall scenario was considerably larger than that from scenario Rcurrent to R+ 20% (or from R–20% to Rcurrent) under any forest age distributions. These results suggest that the change in maturity of forests potentially changes landslide frequency more than a change in rainfall. Furthermore, Sato et al. (2023) compared two landslide events that occurred in plantation forests of different ages in Japan and reported a 3.0-fold higher return period for the estimated groundwater level in mature forests than in young forests. Thus, the protection of mature forests is essential to avoid unnecessarily increasing landslide frequency in a changing climate. Japan has 10.3 million ha of planted forests; the ratio of planted forests to the total forested area is 42%, the second largest in the world (Yamaura et al. 2012). The increase in forest maturity results from low forestry activity because of increasing employer costs and decreasing timber prices (Komatsu et al. 2010). Recently, the loss of forestry workers has slowed, and the ratio of young workers has increased (Ministry of Agriculture Forestry and Fisheries 2021). In addition, large uncertainties exist in future rainfall prediction (Nikolopoulos et al. 2014). If both rainfall and the area of young forests increase in the future, the risk of landslides may substantially increase in Japan.
This study had several limitations. First, as we previously mentioned, the relation between the forest age and NLSI mostly came from South Korea. Second, the number of landslides was collected only for landslides that affected houses or public facilities. Shinohara and Kume (2022) expected that the number of people living in landslide-prone areas would increase during the analyzed period of this study. However, the construction of retaining walls, crib works, and ground anchors (Japan Landslide Society 2002; Junichi and Naoki 2020) will reduce the possibility of landslides on some slopes. These things might affect the number of landslides that we used. Third, we assumed that the relation between the total rainfall and rainfall intensity was constant for the simulation. The rainfall index that we used was highly correlated with the temporal changes in the number of landslides in recent years (Shinohara and Komatsu 2016; Shinohara and Kume 2022). The occurrence of landslides is affected by both the total (or long-term) rainfall and rainfall intensity (i.e., short-term rainfall) (Berti et al. 2012; Chen et al. 2015; Zhang et al. 2019). In Japan, a clear relation was observed between rainfall intensity and total rainfall (Shinohara et al. 2010), which is the probable reason why the simple index was well correlated with the number of landslides. Shinohara et al. (2010) reported that the relation between rainfall intensity and total rainfall had not changed from 1976 to 2007 in most stations of Japanese mountain regions. However, predictions of future trends in rainfall intensity are often different from that of total rainfall (Barbero et al. 2019; Lei et al. 2022). This may change the relation between the number of landslides and the rainfall index.
There are many types of landslides (Varnes 1978; Hungr et al. 2014). In Japan, landslides are classified into steep-slope failure, deep-seated landslide, and debris flow (Osanai et al. 2010). Shinohara and Watanabe (2023) indicated that the effect of forests on the number of rainfall-triggered landslides was clearly observed only for steep-slope failures among the three types of landslides. Some studies (e.g., Michelini et al. 2017; Wang et al. 2017) pointed out the contribution of forest roots to reducing the frequency of debris flows. In contrast, the slip surface of deep-seated landslides is often deeper than the depth of forest roots, suggesting that there would be no (or a weak) contribution of forests to reducing deep-seated landslide frequency. If we had used only the data for steep-slope failure (and debris flow), clearer effects might be observed. Furthermore, the authors combined planted and natural forests and did not consider the landslide susceptibility in each forest type (or species). The authors believe that this assumption is reasonable at the current stage. Nishioka et al. (2023) reported clear differences in landslide susceptibility between young and mature forests, but no differences in mature coniferous forests and broadleaved forests in Asakura, Japan. Note that, in Japan, most planted and natural forests are coniferous and broadleaved forests, respectively. If the relation between landslide susceptibility and the age is obtained for various types (or species) of forests, more detailed forest management (i.e., change in species) may be able to reduce landslide susceptibility. Furthermore, when rainfall exceeds the landslide resistance of mature forests, huge amounts of driftwood may cause substantial damage to residential areas (Sato et al. 2023). Thus, it may be important to consider both the positive and negative effects of mature forests in further studies.
5 Conclusion
The contribution of increased maturity of forests to landslide susceptibility was evaluated in Japan. The relationship between forest age and NLSI was developed based on 11 sites from three countries. The GLMs were created to estimate the annual number of rainfall-triggered landslides from a rainfall index and NLSIJpn, which was used to estimate the above relation and the area for each age class. Although NLSI was assumed to be a monotonic decreasing trend, the GLMs accurately represented the annual number of rainfall-triggered landslides. Using the GLMs and 15 scenarios with different rainfall amounts and forest age distributions, potential changes in the landslide frequency were simulated. The number of landslides with mature-age-dominated forests compared with middle- and young-age-dominated forests was estimated to increase by ca. 12% and 144%, respectively. Increases of 10% and 20% from current rainfall were estimated to increase the number of landslides by ca. 31% and 70%, respectively. Thus, the maturity of forests had potentially larger impacts on landslide frequency than changing rainfall. The authors conclude that the protection of forests is required to avoid unnecessarily increasing landslide frequency during climate change. In this study, the effects of forest characteristics except for age were not considered. If clear relations are obtained between landslide susceptibility and other forest characteristics, more detailed suggestions for forest management could possibly reduce landslide susceptibility.
References
Agrawal S, Gopalakrishnan T, Gorokhovich Y, Doocy S (2013) Risk factors for injuries in landslide- and flood-affected populations in Uganda. Prehosp Disaster Med 28:314–321. https://doi.org/10.1017/S1049023X13000356
Atkinson PM, Massari R (1998) Generalised linear modelling of susceptibility to landsliding in the central Apennines, Italy. Comput Geosci 24:373–385. https://doi.org/10.1016/S0098-3004(97)00117-9
Barbero R, Fowler HJ, Blenkinsop S et al (2019) A synthesis of hourly and daily precipitation extremes in different climatic regions. Weather Clim Extrem 26:100219. https://doi.org/10.1016/j.wace.2019.100219
Barik MG, Adam JC, Barber ME, Muhunthan B (2017) Improved landslide susceptibility prediction for sustainable forest management in an altered climate. Eng Geol 230:104–117. https://doi.org/10.1016/j.enggeo.2017.09.026
Berti M, Martina MLV, Franceschini S et al (2012) Probabilistic rainfall thresholds for landslide occurrence using a bayesian approach. J Geophys Res Earth Surf 117:F04006–F04020. https://doi.org/10.1029/2012JF002367
Brardinoni F, Hassan MA, Slaymaker HO (2002) Complex mass wasting response of drainage basins to forest management in coastal British Columbia. Geomorphology 49:109–124. https://doi.org/S0169-555X(02)00166-6
Chen CW, Saito H, Oguchi T (2015) Rainfall intensity–duration conditions for mass movements in Taiwan. Prog Earth Planet Sci 2:14. https://doi.org/10.1186/s40645-015-0049-2
Cislaghi A, Alterio E, Fogliata P et al (2021) Effects of tree spacing and thinning on root reinforcement in mountain forests of the European Southern Alps. Ecol Manage 482:118873. https://doi.org/10.1016/j.foreco.2020.118873
Dhakal AS, Sidle RC (2003) Long-term modelling of landslides for different forest management practices. Earth Surf Process Landf 28:853–868. https://doi.org/10.1002/esp.499
Douglas GB, McIvor IR, Manderson AK et al (2013) Reducing shallow landslide occurrence in pastoral hill country using wide-spaced trees. Land Degrad Dev 24:103–114. https://doi.org/10.1002/ldr.1106
Evans JR, Imaizumi F, Ohsaka O, Ogawa S (2020) Relationship between tree height and landslide characteristics obtained by GIS assessment. Earth Surf Process Landf 45:3280–3292. https://doi.org/10.1002/esp.4964
Fattet M, Fu Y, Ghestem M et al (2011) Effects of vegetation type on soil resistance to erosion: relationship between aggregate stability and shear strength. Catena (Amst) 87:60–69. https://doi.org/10.1016/j.catena.2011.05.006
Froude MJ, Petley DN (2018) Global fatal landslide occurrence from 2004 to 2016. Nat Hazards Earth Syst Sci 18:2161–2181. https://doi.org/10.5194/nhess-18-2161-2018
Gariano SL, Guzzetti F (2016) Landslides in a changing climate. Earth Sci Rev 162:227–252. https://doi.org/10.1016/j.earscirev.2016.08.011
Genet M, Kokutse N, Stokes A et al (2008) Root reinforcement in plantations of Cryptomeria japonica D. Don: effect of tree age and stand structure on slope stability. Ecol Manage 256:1517–1526. https://doi.org/10.1016/j.foreco.2008.05.050
Gorsevski PV, Gessler PE, Boll J et al (2006) Spatially and temporally distributed modeling of landslide susceptibility. Geomorphology 80:178–198. https://doi.org/10.1016/j.geomorph.2006.02.011
Guns M, Vanacker V (2013) Forest cover change trajectories and their impact on landslide occurrence in the tropical Andes. Environ Earth Sci 70:2941–2952. https://doi.org/10.1007/s12665-013-2352-9
Guzzetti F, Stark CP, Salvati P (2005) Evaluation of flood and landslide risk to the population of Italy. Environ Manage 36:15–36. https://doi.org/10.1007/s00267-003-0257-1
Hungr O, Leroueil S, Picarelli L (2014) The Varnes classification of landslide types, an update. Landslides 11:167–194. https://doi.org/10.1007/s10346-013-0436-y
Imaizumi F, Sidle RC, Kamei R (2008) Effects of forest harvesting on the occurrence of landslides and debris flows in steep terrain of central Japan. Earth Surf Process Landf 33:827–840. https://doi.org/10.1002/esp.1574
Jacob M, Bade C, Calvete H et al (2013) Significance of over-mature and decaying trees for carbon stocks in a central European Natural Spruce Forest. Ecosystems 16:336–346. https://doi.org/10.1007/s10021-012-9617-0
Japan Landslide Society (2002) Annual report on sediment-related disasters in Japan. https://www.stc.or.jp/jittai/jittai_2002.pdf. Accessed 1 Nov 2023
Ji J, Kokutse N, Genet M et al (2012) Effect of spatial variation of tree root characteristics on slope stability. A case study on Black Locust (Robinia pseudoacacia) and Arborvitae (Platycladus orientalis) stands on the Loess Plateau, China. Catena (Amst) 92:139–154. https://doi.org/10.1016/j.catena.2011.12.008
Jonkman SN, Kelman I (2005) An analysis of the causes and circumstances of flood disaster deaths. Disasters 29:75–97. https://doi.org/10.1111/j.0361-3666.2005.00275.x
Junichi K, Naoki I (2020) Outline of measures for sediment disaster by the Sabo department of MLIT, Japan. Landslides 17:2503–2513. https://doi.org/10.1007/s10346-020-01554-9
Kim J, Chululuong C, Soyoung P (2013) GIS-based landslide susceptibility analyses and cross-validation using a probabilistic model on two test areas in Korea. Disaster Adv 6:45–54
Kitamura Y, Namba S (1981) The function of tree roots upon landslide prevention presumed through the uprooting test. Bull Forestry for Prod Res Inst (Japan) 175–208. (in Japanese with English summary)
Komatsu H, Kume T, Shinohara Y et al (2010) Did annual run-off and low flow decrease with reduced forestry practices in Japan? Hydrol Process 24:2440–2451. https://doi.org/10.1002/hyp.7657
Lee S, Lee MJ (2006) Detecting landslide location using KOMPSAT 1 and its application to landslide-susceptibility mapping at the Gangneung area, Korea. Adv Space Res 38:2261–2271. https://doi.org/10.1016/j.asr.2006.03.036
Lee S, Min K (2001) Statistical analysis of landslide susceptibility at Yongin, Korea. Environ Geol 40:1095–1113. https://doi.org/10.1007/s002540100310
Lee S, Choi J, Min K (2002) Landslide susceptibility analysis and verification using the bayesian probability model. Environ Geol 43:120–131. https://doi.org/10.1007/s00254-002-0616-x
Lee S, Choi J, Min K (2004) Probabilistic landslide hazard mapping using GIS and remote sensing data at Boun, Korea. Int J Remote Sens 25:2037–2052. https://doi.org/10.1080/01431160310001618734
Lee S, Ryu JH, Lee MJ, Won JS (2006) The application of artificial neural networks to landslide susceptibility mapping at Janghung, Korea. Math Geol 38:199–220. https://doi.org/10.1007/s11004-005-9012-x
Lee S, Hwang J, Park I (2012) Application of data-driven evidential belief functions to landslide susceptibility mapping in Jinbu. Korea Catena (Amst) 100:15–30. https://doi.org/10.1016/j.catena.2012.07.014
Lee MJ, Park I, Lee S (2015) Forecasting and validation of landslide susceptibility using an integration of frequency ratio and neuro-fuzzy models: a case study of Seorak mountain area in Korea. Environ Earth Sci 74:413–429. https://doi.org/10.1007/s12665-015-4048-9
Lee S, Lee M-J, Lee S (2018) Spatial prediction of urban landslide susceptibility based on topographic factors using boosted trees. Environ Earth Sci 77:656. https://doi.org/10.1007/s12665-018-7778-7
Lei M, Li J, Zhang L et al (2022) Inconsistent frequency trends between hourly and daily precipitation during warm season in mainland of China. Geophys Res Lett 49. https://doi.org/10.1029/2022GL100277. :e2022GL100277
Lusiana N, Shinohara Y (2022) The role of citrus groves in rainfall-triggered landslide hazards in Uwajima. Japan Water (Switzerland) 14:2113. https://doi.org/10.3390/w14132113
Malek Ž, Boerboom L, Glade T (2015) Future forest cover change scenarios with implications for landslide risk: an example from Buzau subcarpathians, Romania. Environ Manage 56:1228–1243. https://doi.org/10.1007/s00267-015-0577-y
Malet J-P, Puissant A, Mathieu A et al (2013) Integrating spatial multi-criteria evaluation and expert knowledge for country-scale landslide susceptibility analysis: Application to France. Landslide Science and Practice: Volume 1: Landslide Inventory and Susceptibility and Hazard Zoning 303–311. https://doi.org/10.1007/978-3-642-31325-7_40
Michelini T, Bettella F, D’Agostino V (2017) Field investigations of the interaction between debris flows and forest vegetation in two Alpine fans. Geomorphology 279:150–164. https://doi.org/10.1016/j.geomorph.2016.09.029
Ministry of Agriculture Forestry and Fisheries 2021).Ministry of Agriculture Forestry and Fisheries (2021) Annual Report on Forest and Forestry in Japan. https://www.rinya.maff.go.jp/j/kikaku/hakusyo/r3hakusyo/attach/pdf/index-2.pdf. Accessed 15 Jul 2023
Moung-Jin L, Won-Kyong S, Joong-Sun W et al (2014) Spatial and temporal change in landslide hazard by future climate change scenarios using probabilistic-based frequency ratio model. Geocarto Int 29:639–662. https://doi.org/10.1080/10106049.2013.826739
Nikolopoulos EI, Crema S, Marchi L et al (2014) Impact of uncertainty in rainfall estimation on the identification of rainfall thresholds for debris flow occurrence. Geomorphology 221:286–297. https://doi.org/10.1016/j.geomorph.2014.06.015
Nishioka M, Inoue H, Ota T, Mizoue N (2023) Impact of forest type and age on shallow landslide susceptibility: a case study from the 2017 heavy rainfall in northern Kyushu, Japan. J for Res 1–8. https://doi.org/10.1080/13416979.2023.2227820
O’loughlin C, Watson A (1979) Root-wood strength deterioration in Radiata Pine after clearfelling. N Z J Sci 9:284–293
O’loughlin C, Ziemer RR (1982) The importance of root strength and deterioration rates upon edaphic stability in steepland forests. In: Proceedings of I.U.F.R.O. Workshop P.1.07-00 Ecology of Subalpine Ecosystems as a Key to Management. Oregon State University, Corvallis, Oregon, pp 70–78
Oh HJ, Kadavi PR, Lee CW, Lee S (2018) Evaluation of landslide susceptibility mapping by evidential belief function, logistic regression and support vector machine models. Geomatics Nat Hazards Risk 9:1053–1070. https://doi.org/10.1080/19475705.2018.1481147
Okada Y, Cai F, Kurokawa U (2023) Changes in Slope Stability over the growth and decay of Japanese Cedar Tree roots. Forests 14:256. https://doi.org/10.3390/f14020256
Oregon Board of Forestry (2001) Forestry, Landslides, and Public Safety. https://www.oregon.gov/odf/Documents/workingforests/landslidespublicsafety.pdf. Accessed 15 Jul 2023
Osanai N, Shimizu T, Kuramoto K et al (2010) Japanese early-warning for debris flows and slope failures using rainfall indices with radial basis function network. Landslides 7:325–338. https://doi.org/10.1007/s10346-010-0229-5
Park S, Choi C, Kim B, Kim J (2013) Landslide susceptibility mapping using frequency ratio, analytic hierarchy process, logistic regression, and artificial neural network methods at the Inje area, Korea. Environ Earth Sci 68:1443–1464. https://doi.org/10.1007/s12665-012-1842-5
Park S, Hamm SY, Kim J (2019) Performance evaluation of the GIS-based data-mining techniques decision tree, random forest, and rotation forest for landslide susceptibility modeling. Sustain (Switzerland) 11:5659. https://doi.org/10.3390/su11205659
Park S, Kim J (2019) Landslide susceptibility mapping based on random forest and boosted regression tree models, and a comparison of their performance. Appl Sci (Switzerland) 9:942. https://doi.org/10.3390/app9050942
Petley D (2012) Global patterns of loss of life from landslides. Geology 40:927–930. https://doi.org/10.1130/G33217.1
Pradhan AMS, Kim YT (2014) Relative effect method of landslide susceptibility zonation in weathered granite soil: a case study in Deokjeok-Ri Creek, South Korea. Nat Hazards 72:1189–1217. https://doi.org/10.1007/s11069-014-1065-z
Pradhan AMS, Kim YT (2017) Spatial data analysis and application of evidential belief functions to shallow landslide susceptibility mapping at Mt. Umyeon. Seoul Korea Bull Eng Geol Environ 76:1263–1279. https://doi.org/10.1007/s10064-016-0919-x
R Core Team (2023) R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/. Accessed 12 Aug 2023
Reichenbach P, Rossi M, Malamud BD et al (2018) A review of statistically-based landslide susceptibility models. Earth Sci Rev 180:60–91. https://doi.org/10.1016/j.earscirev.2018.03.001
Saro L, Woo JS, Kwan-Young O, Moung-Jin L (2016) The spatial prediction of landslide susceptibility applying artificial neural network and logistic regression models: a case study of Inje, Korea. Open Geosci 8:117–132. https://doi.org/10.1515/geo-2016-0010
Sato T, Shuin Y (2022) Impact of national-scale changes in forest cover on floods and rainfall-induced sediment-related disasters in Japan. J for Res. https://doi.org/10.1080/13416979.2022.2131221
Sato T, Katsuki Y, Shuin Y (2023) Evaluation of influences of forest cover change on landslides by comparing rainfall-induced landslides in Japanese artificial forests with different ages. Sci Rep 13:14258. https://doi.org/10.1038/s41598-023-41539-x
Shinohara Y, Komatsu H (2016) The recent trend in annual death toll by landslide disasters in Japan. J Japan Soc Eros Control Eng 68:3–9. https://doi.org/10.1002/hyp.7620
Shinohara Y, Kume T (2022) Changes in the factors contributing to the reduction of landslide fatalities between 1945 and 2019 in Japan. Sci Total Environ 827:154392. https://doi.org/10.1016/j.scitotenv.2022.154392
Shinohara Y, Watanabe Y (2023) Differences in factors determining landslide hazards among three types of landslides in Japan. Nat Hazards. https://doi.org/10.1007/s11069-023-06075-x
Shinohara Y, Kume T, Komatsu H, Otsuki K (2010) Spatial and temporal variations in summer precipitation in Japanese mountain areas. Hydrol Process 24:1844–1855. https://doi.org/10.1002/hyp.7620
Shu H, Hürlimann M, Molowny-Horas R et al (2019) Relation between land cover and landslide susceptibility in val d’Aran, Pyrenees (Spain): historical aspects, present situation and forward prediction. Sci Total Environ 693:1–14. https://doi.org/10.1016/j.scitotenv.2019.07.363
Sidle RC, Ochiai H (2006) Processes, prediction, and Land Use American Geophysical Union. American Geophysical Union, Washington DC, p 307
Sidle RC, Wu W (1999) Simulating effects of timber harvesting on the temporal and spatial distribution of shallow landslides. Z fur Geomorphologie 43:185–201. https://doi.org/10.1127/zfg/43/1999/185
Spiekermann RI, Smith HG, McColl S et al (2022) Quantifying effectiveness of trees for landslide erosion control. Geomorphology 396:107993. https://doi.org/10.1016/j.geomorph.2021.107993
Stokes A, Atger C, Bengough AG et al (2009) Desirable plant root traits for protecting natural and engineered slopes against landslides. Plant Soil 324:1–30. https://doi.org/10.1007/s11104-009-0159-y
Tani I (1973) Review of disasters by heavy rainfall in July 1972. J Japan Soc Eros Control Eng 25:3–9. https://doi.org/10.11475/sabo1948.25.3_3. (In Japanese)
Turner TR, Duke SD, Fransen BR et al (2010) Landslide densities associated with rainfall, stand age, and topography on forested landscapes, southwestern Washington, USA. Ecol Manage 259:2233–2247. https://doi.org/10.1016/j.foreco.2010.01.051
Varnes DJ, Krizek RJ (1978) In: Schuster RLkRJ (ed). Washington D.C, 11–33
Waldron LJ (1977) Shear resistance of root-permeated homogeneous and stratified soil. Soil Sci Soc Am J 41:843–849. https://doi.org/10.2136/sssaj1977.03615995004100050005x
Wang S, Meng X, Chen G et al (2017) Effects of vegetation on debris flow mitigation: a case study from Gansu province. China Geomorphology 282:64–73. https://doi.org/10.1016/j.geomorph.2016.12.024
Wu TH, McKinnell WP, Swanston DN (1979) Strength of tree roots and landslides on price of Wales island, Alaska. Can Geotech J 16:19–33. https://doi.org/10.1139/t79-003
Yamaura Y, Oka H, Taki H et al (2012) Sustainable management of planted landscapes: lessons from Japan. Biodivers Conserv 21:3107–3129. https://doi.org/10.1007/s10531-012-0357-4
Yeon YK, Han JG, Ryu KH (2010) Landslide susceptibility mapping in Injae, Korea, using a decision tree. Eng Geol 116:274–283. https://doi.org/10.1016/j.enggeo.2010.09.009
Yoo Y, Baek T, Kim J, Park S (2016) A comparative study of the frequency ratio and evidential belief function models for landslide susceptibility mapping. J Korean Soc Surveying Geodesy Photogrammetry Cartography 34:597–607. https://doi.org/10.7848/ksgpc.2016.34.6.597
Zhang B, Zhao QG, Horn R, Baumgartl T (2001) Shear strength of surface soil as affected by soil bulk density and soil water content. Soil Tillage Res 59:97–106. https://doi.org/10.1016/S0167-1987(01)00163-5
Zhang K, Wang S, Bao H, Zhao X (2019) Characteristics and influencing factors of rainfall-induced landslide and debris flow hazards in Shaanxi Province, China. Nat Hazards Earth Syst Sci 19:93–105. https://doi.org/10.5194/nhess-19-93-2019
Zhang Y, Shen C, Zhou S, Luo X (2022) Analysis of the influence of forests on landslides in the Bijie Area of Guizhou. Forests 13:1136. https://doi.org/10.3390/f13071136
Acknowledgements
We thank Leonie Seabrook, PhD, from Edanz (https://jp.edanz.com/ac) for editing a draft of this manuscript.
Funding
This work was supported by the Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Number JP21K04590, and the Japan Science and Technology Agency (JST) SPRING, Grant Number JPMJSP2105.
Open access funding provided by University of Miyazaki.
Author information
Authors and Affiliations
Contributions
Conceptualization: Y.S; Methodology, Formal analysis and investigation, Writing – original draft: N.L; Writing – review and editing: Y.S, F.I; Funding acquisition: Y.S, N.L.
Corresponding author
Ethics declarations
Competing interests
The authors declare that they have no known competing financial interests or personal relationships that could appear to influence the work reported in this paper.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Lusiana, N., Shinohara, Y. & Imaizumi, F. Quantifying effects of changes in forest age distribution on the landslide frequency in Japan. Nat Hazards (2024). https://doi.org/10.1007/s11069-024-06537-w
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s11069-024-06537-w