- Based on stack bar charts, most municipalities have comparable ratios of different levels of LTS?
At the local and grid level, obviously much larger variations.
At the global level:
- 2
- 3
- 4
- 1
At the adm and socio level, the median levels ordered after share:
- 2
- 3
- 1
- 4
1 and 4 very close for global, adm, socio.
(ofc total car highest)
At grid level:
- 3
- 4
- 2
- 1
Highly uneven distribution of both network length AND density.
Density distributions are different at various aggregation levels. Very different distributions, also across space.
- At socio level - share of LTS 1 is not only an urban phenomenon - but absolute quantity is.
- Same for LTS 2
- Absolute values for LTS 3 a bit 'random' - but low in CPH. Share of LTS 3 low in larger cities.
- LTS 4 - not urban - neither based on share or absolute values - but a bit random - generally low values
Results for spatial autocorrelation:
- Significant clustering of network density - obviously
- Significant clustering of LTS 1 density and LTS relative length - but not the same places!
- Sig clustering of LTS 2 dens and rel. length - overlapping but not identical clusters.
- Sig clustering of LTS 3 dens - but small scattered clusters. Slightly bigger clusters for LTS 3 share.
- LTS 4 dens - sig but sparse clustering - this might change with new results!! LTS 4 share - sig clustering, especially of low LTS 4 share.
- Car dens/car share - sig. clustering - but reversed (i.e. high dens in cities, but low share).
- Same for LTS 1 - mostly not the same places with a high dens and high share, but both sig.
- LTS 2 - same
- LTS 3 - sig., overlapping but not identical clusters
- LTS 4 - stronger clustering tendency than hex? Overlapping but not identical clusters for dens and share
The number of components per level (steps):
- 1-2
- 1-3
- 1
- 1-4
- Car
- Total
Mean component size ranked smallest to biggest: (Median can be misleading - e.g. car have a few actual components and then tiny tracks and service)
- LTS 1-2
- LTS 1
- 1-3
- 1-4
- Car
- Total
Thus - LTS 1 is very fragmented - adding LTS 2 adds more network but does not result in less fragmentation! LTS 3 and 4 serve as connectors.
Based on rug plots, the pattern in distribution of comp counts is similar across aggregation levels.
Zipf plot confirms that LTS 4 works as connector (OBS - does this change after data update??)
Interpretation - for LTS 1, comp count increases with density/mixed picture - for LTS 1-2 and 1-3 it decreases? (see also ind. plots) (same at muni level - no pattern at hex level).
Some spatial clustering of comp per sqkm or km for lower LTS - less fragmentation around larger cities. (at socio level)
At hex grid:
- Longer largest components around CPH and across Sjælland
- Some clustering in component per length (lower in urban areas)
- But genereally very low Moran's I!
-
As expected based on density and fragmentation results?
-
Low reach for both LTS 1 and LTS 1-2
-
A bit better for 1-3
-
LTS 4 and car similar - and highest
-
There are a few locations with more reach for bikes than cars
-
Confirms picture from fragmentation:
-
Average reach is from smallest to biggest: 1, 1-2, 1-3, car, 1-4, (OBS on car) (OBS - might change with update??)
-
But for median reach is LTS 1-2 smaller than 1 because of smaller LTS 2 components
-
Because of fragmentation, the majority of LTS 1 and LTS 1-2 cells have no improvement in reach when increasing distance.
-
For LTS 1-3 it is around a third
-
Almost no LTS 1-4 cells have no improvements.
- Clear spatial patterns in reach - both for distance 5 reach and for where there is no reach increase when increasing distance!
- Sig. clustering in reach differences - high differences in 5-10/5-15 in urban areas - and some others! - low in non urban for LTS 1, 1-2. Opposite for LTS 4 and car.
Under 100k:
- Very low income positively correlated with other low income - with decreasing strength - up untill 500+ K, which is negatively correlated
- Very low income is negatively correlated with share of households with cars
- Positive with urban pct and pop density
100-150k:
- Similar for income 100-150 - but even stronger association with urban, pop, no car
Income 150-200 + 200-300:
- Similar, but positive corr with share of households with 1 car - but negative with 2!
- Positive but weak corr with pop and urban
Income 400 - 500k:
- positive corr with almot all income groups - but weak, except 200-300,400-400, and negative for 750+!
- Positive for share of households with 1 car, but negative for households with 2 cars --> results in negative in households w car and pos for no car
- Weak positive corr with urban, pop
Income 500-750k:
- Negativ corr with all income segments except 400-500 and 750+
- Strong positive with cars and 2 cars (!) - weaker positive corr with 1 car
- Negative corr with pop density and urban
Income 750+:
- Negative corr with all groups except 500-750
- Positive corr with car and 2 cars - but negative with 1 car!
- Negative corr with pop dens and urban (but only weak corr for urban)
- Fairly weak positive corr with total number of households with cars
1 car:
- Negatively correlated with two lowest income groups
- Positively corr with 3-6
- Negatively with two higest
- Quite weak corr between 1 car and 2 car
- Negative with urban and pop
2 cars:
- Negative corr with all income groups except two highest
- Strong positive corr with households with car
- Strong negative corr with pop and urban pct
POP & Urban pct:
- Pop and urban have similar corrs
- Positive with lower income groups - up to 200-300 k for pop and 300-400k for urban
- Negative with higer incomes
- Negative with car ownership
- Lower income groups live in areas with higher network density and higher density of low LTS network
- Medium income weak positive corr with high density/high lts 1 and 2 density
- High income live in low density areas, low dens of lts 1 and 2
- Negative corr with 1 and 2 dens and total network dens vs. cars
- Positve corr with 3 and 4 dens and cars (might change with updated data!)
-
Low income associated with lower fragmentation of LTS 1 and 2 networks
-
weak both pos and negative corr with middle income and lts 1 and 2 fragmentation
-
500-700 live in areas iwth high fragmentation of 1-3 lts but low fragmenation of lts 4
-
weaker pattern for 750+ (they live in slightly more urban areas?)
-
Car ownership positively correlated with fragmentation of low lts network - negatively correlated with fragmentation of car and lts 4 network
- High network reach possitive cor with income groups up to 300-400k - this group has slightly negative corr with network reach
- Slightly positive for 400-500
- Negative for high income - especially 500-700 (matches results for fragmentation and dens?)
-
Positive corr between LTS 1 and 2 density
-
Positive corr between 3 and 4
-
Negative corr between 1-2 and 3-4 (weaker for 3!)
-
Also holds for relative length!
-
Negative corr between lts 1 dens and lts comp count and comp per length - but positive with 4 and car comp count and comp per length
-
Similar with lts 2
-
On the other hand, high lts 3 and 4 positively correlated with many components of LTS 1 and 2 (lts 4 density also positively correlated with many lts 3 comps!)
-
BUT - high lts 3 and 4 density is negatively correlated with all levels of network reach! Only positively correlated with differences in 5-10 and 10-15 reach for own LTS level
-
High fragmentation results in low network reach
-
Much weaker pattern between components/fragmentation and density
-
Similar pattern as socio with areas with high 1-2 and low 3-4 and vice versa.
-
Different pattern for relative length - high relative LTS X is usually negatively associated with high relative lenght of other levels - because of the scale.
-
Reach corrs a bit different - still a stronger pos corr between LTS 1 reach and lts 4 dens, but also a weak positive corr between lts 4 dens and lts 4 reach (same for 3)
-
Reach comparison interesting - a high jump from lts 1 and lts 2 5 to 10 reach or 10 to 15 is associoated with low 3 and 4 dens. For LTS 3, associated with low lts 4 dens. For 4, associated with low lts 1-2 dens to, weaker assoc wiht low lts 3 --> seems like good network reach for low lts do not happen in areas with high lts 4, and vice versa.
Moran's I decrease for most LTS levels as K increase, clusters increase - but smaller differences and still positive
Same for relative length - mostly the same, but smaller variations
Usually higest Moran's I for smallest k - but only smaller differences.
Similar - does change number of areas in clusters, but does not change the general trend/global value a lot
Somewhat sensitive to spatial weights - but very low Moran's I, not significant
Not sensitive to changing weights.
-
Low density
-
low high stress
-
low-ish car pct
-
rel high share of 1
-
highest share of 2
-
low fragmentation
-
low reach
-
high local reach increase (1-5)
-
Both medium towns, outside of larger cities, and in some rural areas?
-
Second largest cluster
-
highest low stress dens
-
low high stress dens
-
highest car dens
-
highest total dens
-
low share of 3 and 4
-
low fragmentation
-
highest reach
-
higest reach increases for almost all metrics - except for car
-
In urban centers
-
fourth largest cluster (area wise)
-
Small cluster!
-
no/very low lts 1
-
high lts 4
-
almost only car
-
highly fragmented for lower stress
-
very low reach for lower stress
-
high increase for car and lts 4 at longer distances
-
low density
-
low low stress AND low 4
-
mostly lts 3
-
medium fragmentation for lts 2
-
very low reach
-
Only smaller islands?
-
Very low low stress
-
high high stress density
-
low total density
-
very high car pct
-
high fragmentation for 1 (a little) and 2 (medium)
-
low reach
-
low reach increases for low stress - but high for high stress and car
-
Rural areas?
-
Third biggest area-wise
-
low low stress
-
medium/high high stress density
-
low total density
-
medium lts 4 share
-
evenly split share between 2 and 3 (around 30)
-
medium/high car share
-
low fragmentation
-
low reach
-
medium/low reach increase for lower stress - some connectivity
-
higher increases for higher stress
-
by far the biggest cluster
-
rural
-
high low stress
-
low high stress density
-
high density
-
high share of low stress
-
low share of car
-
low fragmentation
-
high reach (second best)
-
high reach increases for all
-
Small cluster
-
Urban/suburban
-
medium low stress density
-
very low high stress dens
-
very high share of lts 1 and 2
-
very low share of lts 3 and esp 4
-
medium-low network density
-
high fragmentation for lts 3,4, car and all??
-
medium-high reach - but lower for car than lts 3 and 4!
-
high reach increases
-
Very small cluster