Assignment 3

Introduction:
Landslides are mass movements of material down a slope and are common in Oregon due to the wet steep conditions necessary for landslides. Earthquakes, extreme rainfall or volcanic eruptions, often trigger landslides; all of which are present in Western Oregon (Sassa, 2005; Sassa, 2013). Landslides are distributed across the globe butoften occur in moderate to steep sloped areas (Sassa, 2013). Landslides offer risk to surrounding infrastructure and can threaten people’s lives that live in the vicinity. Understanding the distribution of landslides in Oregon will provide useful in risk management.

Objective: The goal of this study is to try and understand the distribution of landslides in Oregon. Understanding whether landslides are clustered random or evenly distributed will help for preparedness. Additionally, it is important to understand the severity of the landslides, which are often determined by their length. This study aims to explain the probability of a landslide reaching critical length.

Methods:
The study site used for this analysis was the state of Oregon. I analyzed the statewide distribution of landslides across Oregon.
The data used for this analysis was acquired from the Statewide Landslide Information Database for Oregon. This dataset consists of records that are primarily from published geologic reports and geologic hazards studies carried out by the US Geologic Survey (USGS). This by now means claims to contain every landslide, but rather offers a sample of landslides for the state of Oregon.
The analytical methods I used in this study were to find basic descriptive statistics that would help explain the dataset. The next steps were to perform a quadrat analysis to further understand the distribution of landslides across the state. Additionally, I use a point pattern analysis to reinforce the distribution. The point pattern analysis finds the nearest neighbor of a landslide to another landslide. This is useful because the results will help describe the distribution of landslides. These analytical methods were performed in R, for more detail read the documentation at: https://www.r-project.org.

Results:
In Oregon landslides are highly clustered in the western portion of the state. A quadratic spatial analysis of Oregon’s landslides provided me with a VMR of 2.572189e+13, which indicated a highly clustered distribution of landslides across the state. With 95% confidence we can infer that the mean landslide length is between 281.22 and 310.78.

QuadratCounting
The mean nearest neighbor distance is 625.36 feet. The histogram below shows that the majority of landslides occur relatively close to another landslide and this is representative of a clustered distribution. The graph is skewed right.

Nearest Neighbor Distribution (title got cut off)
NND_Hist

After analyzing the landslide data, I found 1000 feet to be a fairly natural break in the data and therefore deemed it the critical length. Any landslide over 100 feet has reached critical length and is carries a larger risk. The probability of a landslide having critical length or longer is 7.08%. This was calculated by:
p(criticalLength) = sum(criticalLength)/allLandslides.

The more quadrats that I set the more clustered the data results became. This makes sense because a larger percentage of the quadrats had low numbers of landslides.

Discussion:
This analysis provided useful results. Because of the threat associated with landslides, it is important to understand their distribution. Understanding where landslides have occurred plays an important role in predicting where they will occur in the future. After this analysis, it is evident that the distribution of landslides in Oregon is highly clustered, with the majority of recorded landslides having occurred in the western portion of the state. Additionally, the majority of the landslides are relatively short in length. There is a probability of 7% that a landslide will reach critical length. When a landslide reaches critical length, it has a risk of destruction due to its long length and ability to take down infrastructure.
In future studies, it would be useful to perform quadrat analyses on different scaled study sites to understand a finer spatial resolution of landslide locations. From this analysis I am able to understand the spatial distribution at state-level scale. Now that I understand where the clusters are, it would be helpful to run this same analysis on subsets of the sample. For example, understanding the distribution of northwest portion of Oregon may provide useful information, as this is where a large portion of where landslides occur. This would also be important and effective when performing the nearest neighbor analysis. It would drastically change the results if the study site were a smaller portion of the state. If it was spatially constrained to a location where there was a high frequency of landslides, we may see a different distribution.

References:
Sassa, K. (2005). Landslides risk analysis and sustainable disaster management. Berlin: Springer.

Sassa, K. (2013). Landslides: Global risk preparedness. Berlin: Springer.