The algorithm estimates the capacity of ecosystems to provide opportunities for nature-based recreation and leisure (recreation opportunity spectrum). First, it bases upon look-up tables, to score access to or the quality of natural features (land suitability, protected areas, infrastructure, water resources) for their potential to support for outdoor recreation (potential recreation). Second, it implements a proximity-remoteness concept to integrate the recreation potential and the existing infrastructure.
The module offers two functionalities. One is the production of recreation related maps by using pre-processed maps that depict the quality of or the access to areas of recreational value. The other is to transform maps that depict natural features into scored maps that reflect the potential to support for outdoor recreational. Nevertheless, it is strongly advised to understand first the concepts and the terminology behind the algorithm, by reading the related sources.
An important category is the one with the Highest Recreation Spectrum. It includes areas of very high recreational value which, at the same time, are very near to access.
Potential | Opportunity | Near | Midrange | Far |
---|---|---|---|
Near | 1 | 2 | 3 |
Midrange | 4 | 5 | 6 |
Far | 7 | 8 | 9 |
The following equation represents the logic behind ESTIMAP:
Recreation Spectrum = Recreation Potential + Recreation Opportunity
( {Constant} + {Kappa} ) / ( {Kappa} + exp({alpha} * {Variable}) )
For the sake of demonstrating the usage of the module, we use the following component maps to derive a recreation potential map:
input_area_of_interest
input_land_suitability
input_water_resources
input_protected_areas
The maps shown above are available to download, among other sample maps, at: https://gitlab.com/natcapes/r.estimap.recreation.data.
Note, the prefix input_
in front of all maps is purposive in order to make the examples easier to understand. Similarly, all output maps and files will be prefixed with the string output_
.
Below, a table overviewing all input and output maps used or produced in the examples.
Input map name | Spatial Resolution | Remarks | |
---|---|---|---|
area_of_interest | 50 m | A map that can be used as a 'mask' | |
land_suitability | 50 m | A map scoring the potential for recreation over CORINE land classes | |
water_resources | 50 m | A map scoring access to water resources | |
protected_areas | 50 m | A map scoring the recreational value of natural protected areas | |
distance_to_infrastructure | 50 m | A map scoring access to infrastructure | |
population_2015 | 1000 m | The resolution of the raster map given to the 'populatio' input option will define the resolution of the output maps 'demand', 'unmet' and 'flow' | |
local_administrative_unit | 50 m | A rasterised version of Eurostat's Local Administrative Units map | |
Output map name | Spatial Resolution | Remarks | |
potential |
50 m | ||
potential_1 | 50 m | ||
potential_2 | 50 m | ||
potential_3 | 50 m | ||
potential_4 | 50 m | ||
spectrum | 50 m | ||
opportunity | 50 m | Requires to request for the 'spectrum' output | |
demand | 1000 m | Depends on the 'flow' map which, in turn, depends on the 'population' input map | |
unmet | 1000 m | Depends on the 'flow' map which, in turn, depends on the 'population' input map |
|
flow | 1000 m | Depends on the 'population' input map | |
Output table name | |||
supply | NA |
Before anything, we need to define the extent of interest by executing
g.region raster=input_area_of_interest -p
projection: 99 (ETRS89 / LAEA Europe) zone: 0 datum: etrs89 ellipsoid: grs80 north: 2879700 south: 2748850 west: 4735600 east: 4854650 nsres: 50 ewres: 50 rows: 2617 cols: 2381 cells: 6231077
The first four input options of the module, are designed to receive
pre-processed input maps that classify as either land
,
natural
, water
, and infrastructure
resources that add to the recreational value of the area.
Pro-processing means here to derive a map that scores the given
resources, in the context of recreation and the ESTIMAP algorithm.
To produce a recreation potential map, the simplest command requires
the user to define the input map option land
and name the output
map via the option potential
. Using a pre-processed map that
depicts the suitability of different land types to support for recreation
(here the map named land_suitability
), the command to execute
is:
r.estimap.recreation land=input_land_suitability potential=output_potential
Example of a recreation potential output map
Note, this will process the map input_land_suitability
over the
extent defined previously via g.region
, which is the standard
behaviour in GRASS GIS.
To exclude certain areas from the computations, we may use a raster map as a
mask and feed it to the input option mask
:
r.estimap.recreation land=input_land_suitability mask=input_area_of_interest potential=output_potential_1
Example of a recreation potential output map while using a MASK
The use of a mask (in GRASS GIS' terminology known as MASK)
will ignore areas of No Data (pixels in the
area_of_interest
map assigned the NULL value). Successively,
these areas will be empty in the output map output_potential_1
.
Actually, the same effect can be achieved by using GRASS GIS' native mask
creation module r.mask
and feed it with a raster map of
interest. The result will be a raster map named MASK whose
presence acts as a filter. In the following examples, it becomes obvious that
if a single input map features such No Data areas, they will
be propagated in the output map.
Nonetheless, it is good practice to use a MASK
when one needs to
ensure the exclusion of undesired areas from any computations. Note also the
--o
flag: it is required to overwrite the already existing map
named potential_1
.
Next, we add in the water component a map named water_resources
,
we modify the output map name to potential_2
and execute the new
command without a mask:
r.estimap.recreation land=input_land_suitability water=input_water_resources potential=output_potential_2
Example of a recreation potential output map while using a MASK, a land suitability map and a water resources map
At this point it becomes clear that all NULL
cells present in
the water map, are propagated in the output map
output_potential_2
.
Following, we provide a map of protected areas named
input_protected_areas
, we modify the output map name to
output_potential_3
and execute the updated command:
r.estimap.recreation land=input_land_suitability water=input_water_resources natural=input_protected_areas potential=output_potential_3
Example of a recreation potential output map while using a MASK, a land suitability map, a water resources map and a natural resources map
While the land
option accepts only one map as an input, both the
water
and the natural
options accept multiple maps
as inputs. For example, we add a second map named
input_bathing_water_quality
to the water component and
modify the output map name to output_potential_4
:
r.estimap.recreation land=input_land_suitability water=input_water_resources,input_bathing_water_quality natural=input_protected_areas potential=output_potential_4
In general, arbitrary number of maps, separated by comma, may be added to options that accept multiple inputs.
Example of a recreation potential output map while using a MASK, a land suitability map, two water resources maps and a natural resources map
This example, features also a title and a legend, so as to make sense of the map (however, we will skip for now important cartographic elements).
d.rast output_potential_4 d.legend -c -b output_potential_4 at=0,15,0,1 border_color=white d.text text="Potential" bgcolor=white
The different output map names are purposefully selected so as to enable a
visual comparison of the differences among the differenct examples. The
output maps output_potential_1
, output_potential_2
,
output_potential_3
and output_potential_4
, range
within [0,3]. Yet, they differ in the distribution of values due to the
different set of input maps.
All of the above examples base upon pre-processed maps that score the access to and quality of land, water and natural resources. For using raw, unprocessed maps, read section Using unprocessed maps.
We can remove all of the potential maps via
g.remove raster pattern=output_potential* -f
To derive a map with the recreation (opportunity) spectrum
, we
need in addition an infrastructure
component. In this example a
map that scores distance to infrastructure (such as the road network) named
input_distance_to_infrastructure
, is defined as an additional
input:
Example of an input map showing distances to infrastructure
Naturally, we need to define the output map option spectrum
too:
r.estimap.recreation \ land=input_land_suitability \ water=input_water_resources,input_bathing_water_quality \ natural=input_protected_areas \ infrastructure=input_distance_to_infrastructure \ spectrum=output_spectrum
or, the same command in a copy-paste friendly way for systems that won't
understand the special \
character:
r.estimap.recreation land=input_land_suitability water=input_water_resources,input_bathing_water_quality natural=input_protected_areas infrastructure=input_distance_to_infrastructure spectrum=output_spectrum
Example of a recreation spectrum output map while using a MASK, a land suitability map, a water resources map and a natural resources map
Missing to define an infrastructure
map, while asking for the
spectrum
output, the command will abort and inform about.
The image of the spectrum map was produced via the following native GRASS GIS commands
d.rast output_spectrum d.legend -c -b output_spectrum at=0,30,0,1 border_color=white d.text text="Spectrum" bgcolor=white
The opportunity
map is actually an intermediate step of the
algorithm. The option to output this map opportunity
is meant
for expert users who want to explore the fundamentals of the processing
steps. As such, and by design, it requires to also request for the output
option spectrum
. Be aware that this design choice is applied in
the case of the unmet
output map option too. Building upon the
previous command, we add the opportunity
output option:
r.estimap.recreation --o \ mask=input_area_of_interest \ land=input_land_suitability \ water=input_water_resources,input_bathing_water_quality \ natural=input_protected_areas \ infrastructure=input_distance_to_infrastructure \ opportunity=output_opportunity \ spectrum=output_spectrum
or, the same command in a copy-paste friendly way:
r.estimap.recreation --o mask=input_area_of_interest land=input_land_suitability water=input_water_resources,input_bathing_water_quality natural=input_protected_areas infrastructure=input_distance_to_infrastructure opportunity=output_opportunity spectrum=output_spectrum
We also add the --o
overwrite flag, because existing
output_spectrum
map will cause the module to abort.
Example of a recreation spectrum output map while using a MASK, a land suitability map, a water resources map and a natural resources map
The image of the opportunity map was produced via the following native GRASS GIS commands
d.rast output_opportunity d.legend -c -b output_opportunity at=0,20,0,1 border_color=white d.text text="Opportunity" bgcolor=white
To derive the outputs met demand
distribution,
unmet
demand distribution and the actual flow
,
additional requirements are a population
map and one of
boundaries, as an input to the option base
within which to
quantify the distribution of the population. Using a map of administrative
boundaries for the latter option, serves for deriving comparable figures
across these boundaries. The algorithm sets internally the spatial resolution
of all related output maps (demand
, unmet
and
flow
) to the spatial resolution of the population
input map.
Fragment of a population map (GHSL, 2015)
In this example, the population map named population_2015
is of
1000m^2.
Fragment of a local administrative units input map
The map named local_administrative_units
serves in the following
example as the base map for the zonal statistics to obtain the demand
map.
In this example command, we remove the previously added
opportunity
and spectrum
output options, and
logically add the demand
output option:
r.estimap.recreation --o \ mask=input_area_of_interest \ land=input_land_suitability \ water=input_water_resources,input_bathing_water_quality \ natural=input_protected_areas \ infrastructure=input_distance_to_infrastructure \ population=input_population_2015 \ base=input_local_administrative_units \ demand=output_demand
Of course, the maps output_opportunity
and output_spectrum
still exist in our data base,
unless explicitly removed.
Example of a demand distribution output map while using a MASK and inputs for land suitability, water resources, natural resources, infrastructure, population and base
In the following example, we add unmet
output map option. In
this case of the unmet distribution map too, by design the module
requires the user to define the demand
output map option.
r.estimap.recreation --o \ mask=input_area_of_interest \ land=input_land_suitability \ water=input_water_resources,input_bathing_water_quality \ natural=input_protected_areas \ infrastructure=input_distance_to_infrastructure \ population=input_population_2015 \ base=input_local_administrative_units \ demand=output_demand \ unmet=output_unmet_demand
Example of an 'unmet demand' output map while using a MASK and inputs for land suitability, water resources, natural resources, infrastructure, population and base
It is left as an exercise to the user to create screenshots of the
met, the unmet demand distribution and the flow
output maps. For example, is may be similar to the command examples that
demonstrate the use of the commands d.rast
,
d.legend
and d.text
, that draw the
potential, the spectrum and the opportunity
maps.
The flow bases upon the same function used to quantify the attractiveness of locations for their recreational value. It includes an extra score term.
The computation involves a distance map, reclassified in 5 categories as shown in the following table. For each distance category, a unique pair of coefficient values is assigned to the basic equation.
Distance | Kappa | Alpha |
---|---|---|
0 to 1 | 0.02350 | 0.00102 |
1 to 2 | 0.02651 | 0.00109 |
2 to 3 | 0.05120 | 0.00098 |
3 to 4 | 0.10700 | 0.00067 |
>4 | 0.06930 | 0.00057 |
Note, the last distance category is not considered in deriving the final "map of visits". The output is essentially a raster map with the distribution of the demand per distance category and within predefined geometric boundaries
r.estimap.recreation --o \ mask=input_area_of_interest \ land=input_land_suitability \ water=input_water_resources,input_bathing_water_quality \ natural=input_protected_areas \ infrastructure=input_distance_to_infrastructure \ population=input_population_2015 \ base=input_local_administrative_units \ flow=output_flow
Example of a flow output map while using a MASK and inputs for land suitability, water resources, natural resources, infrastructure, population and base
If we check the output values for the output_flow
map, they are
rounded by the module automatically to integers! Here the first few lines
reporting areal statistics on the output_flow
map:
r.stats output_flow -acpln --q |head
returns
52 125000000.000000 50000 1.72% 53 191000000.000000 76400 2.63% 54 303000000.000000 121200 4.17% 55 392000000.000000 156800 5.39% 56 196000000.000000 78400 2.69% 57 178000000.000000 71200 2.45% 58 286000000.000000 114400 3.93% 59 185000000.000000 74000 2.54% 60 207000000.000000 82800 2.85% 61 176000000.000000 70400 2.42%
If the user wants the real numbers, that derive from the mobility function,
the -r
flag comes in handy:
r.estimap.recreation --o -r \ mask=input_area_of_interest \ land=input_land_suitability \ water=input_water_resources,input_bathing_water_quality \ natural=input_protected_areas \ infrastructure=input_distance_to_infrastructure \ population=input_population_2015 \ base=input_local_administrative_units \ flow=output_flow
Querying again areal statistics via
r.stats output_flow -acpln --q |head
52-52.139117 from to 50000000.000000 20000 0.69% 52.139117-52.278233 from to 11000000.000000 4400 0.15% 52.278233-52.41735 from to 39000000.000000 15600 0.54% 52.41735-52.556467 from to 32000000.000000 12800 0.44% 52.556467-52.695583 from to 7000000.000000 2800 0.10% 52.695583-52.8347 from to 9000000.000000 3600 0.12% 52.8347-52.973817 from to 25000000.000000 10000 0.34% 52.973817-53.112933 from to 13000000.000000 5200 0.18% 53.112933-53.25205 from to 28000000.000000 11200 0.38% 53.25205-53.391167 from to 92000000.000000 36800 1.26%
The module outputs by request the supply and use tables in form of CSV files. Here is how:
r.estimap.recreation --o -r \ mask=input_area_of_interest \ land=input_land_suitability \ water=input_water_resources,input_bathing_water_quality \ natural=input_protected_areas \ infrastructure=input_distance_to_infrastructure \ population=input_population_2015 \ base=input_local_administrative_units \ supply=output_supply \ use=output_use
Not surprisingly, the above command fails!
It however informs that a land cover map
and corresponding reclassification rules,
for the classes of the landcover
map, are required.
Specifically, the algorithm's designers developed a MAES land classes scheme.
The "translation" of the CORINE land classes (left) into this scheme
(classes after the =
sign) are for example:
1 = 1 Urban 2 = 1 Urban 3 = 1 Urban 4 = 1 Urban 5 = 1 Urban 6 = 1 Urban 7 = 1 Urban 8 = 1 Urban 9 = 1 Urban 10 = 1 Urban 11 = 1 Urban 12 = 2 Cropland 13 = 2 Cropland 14 = 2 Cropland 15 = 2 Cropland 16 = 2 Cropland 17 = 2 Cropland 18 = 4 Grassland 19 = 2 Cropland 20 = 2 Cropland 21 = 2 Cropland 22 = 2 Cropland 23 = 3 Woodland and forest 24 = 3 Woodland and forest 25 = 3 Woodland and forest 26 = 4 Grassland 27 = 5 Heathland and shrub 28 = 5 Heathland and shrub 29 = 3 Woodland and forest 30 = 6 Sparsely vegetated land 31 = 6 Sparsely vegetated land 32 = 6 Sparsely vegetated land 33 = 6 Sparsely vegetated land 34 = 6 Sparsely vegetated land 35 = 7 Wetland 36 = 7 Wetland 37 = 8 Marine 38 = 8 Marine 39 = 8 Marine
We save this into a file named corine_to_maes_land_classes.rules
and feed it to the land_classes
option, then re-execute the
command:
r.estimap.recreation --o -r \ mask=input_area_of_interest \ land=input_land_suitability \ water=input_water_resources,input_bathing_water_quality \ natural=input_protected_areas \ infrastructure=input_distance_to_infrastructure \ population=input_population_2015 \ base=input_local_administrative_units \ landcover=input_corine_land_cover_2006 \ land_classes=corine_to_maes_land_classes.rules \ supply=output_supply \ use=output_use
This time it works. Here the first few lines from the output CSV files:
head -5 output_*.csv
==> output_supply.csv <== base,base_label,cover,cover_label,area,count,percents 3,,1,723.555560,9000000.000000,9,7.76% 3,,3,246142.186250,64000000.000000,64,55.17% 3,,2,21710.289271,47000000.000000,47,40.52% 1,,1,1235.207129,11000000.000000,11,7.97% ==> output_use.csv<== category,label,value 3,,268576.031081 4,,394828.563827 5,,173353.69508600002 1,,144486.484126
Using other land cover maps as input, would obviously require a similar set of land classes translation rules.
Of course it is possible to derive all output maps with one call:
r.estimap.recreation --o \ land=input_land_suitability \ natural=input_protected_areas,input_urban_green \ water=input_water_resources,input_bathing_water_quality \ infrastructure=input_distance_to_infrastructure \ landcover=input_corine_land_cover_2006 \ land_classes=corine_to_maes_land_classes.rules \ mask=input_area_of_interest \ potential=output_potential \ opportunity=output_opportunity \ spectrum=output_spectrum \ base=input_local_administrative_units \ aggregation=input_regions \ population=input_population_2015 \ demand=output_demand \ unmet=output_unmet_demand \ flow=output_flow \ supply=output_supply \ use=output_use \ timestamp='2015'
Note the use of the timestamp
parameter! This concerns the
spectrum
map. If plans include working with GRASS GIS' temporal
framework on time-series, maybe this will be useful.
A vector input map with the role of the base map, can be used too.
Example of a vector map showing local administrative units
r.estimap.recreation --o -r \ mask=input_area_of_interest \ land=input_land_suitability \ water=input_water_resources,input_bathing_water_quality \ natural=input_protected_areas \ infrastructure=input_distance_to_infrastructure \ population=input_population_2015 \ base=input_local_administrative_units \ base_vector=input_vector_local_administrative_units \ landcover=input_corine_land_cover_2006 \ land_classes=corine_to_maes_land_classes.rules \ supply=output_supply \ use=output_use
This command will also:
output_supply.csv
and output_use.csv
input_vector_local_administrative_units
vector map:spectrum_sum demand_sum unmet_sum flow_sum flow_1_sum flow_2_sum flow_3_sum flow_4_sum flow_5_sum flow_6_sum
all of which are of double precision.
For example, the
v.db.select input_vector_local_administrative_units columns=lau2_no_name,spectrum_sum,demand_sum,unmet_sum,flow_sum where="flow_sum IS NOT NULL" |head
following the analysis, returns
lau2_no_name|spectrum_sum|demand_sum|unmet_sum|flow_sum 801 Bad Erlach|22096|2810||700 841 Leopoldsdorf|8014|1800||426 630 Rabensburg|23358|8474||1546 468 Maissau|73168|6650||2580 10 Müllendorf|19419|1902||718 544 Straß im Straßertale|57314|4368||1471 67 Forchtenstein|53009|272||848 460 Guntersdorf|27408|12183||1955 103 Sankt Andrä am Zicksee|45130|3833||1926
Here the vector map used for administrative boundaries with the sum of flow for each unit:
Example of a vector map showing the flow per unit
and the corresponding unmet demand, based on the analysis
Example of a vector map showing the unmet demand in concerned units
In the latter screenshot, the units bearing the unmet demand results, are not
the same as the raster map previously shown. The different results are due to
the -r
flag used in this last analysis. The -r
flag
will round up floating point values during computations, thus the results
with or without it will differ. The reason to use, in this last example the
-r
flag, was to have short integer numbers to print as labels
inside the units (in the vector map).
The module offers a pre-processing functionality for all of the following input components:
A first look on how this works, is to experiment with the
landuse
and suitability_scores
input options.
Let's return to the first example, and use a fragment from the unprocessed
CORINE land data set, instead of the land_suitability
map. This
requires a set of "score" rules, that correspond to the CORINE nomenclature,
to translate the land cover types into recreation potential.
In this case, the rules are a simple ASCII file (for example named
corine_suitability.scores
) that contains the following:
1:1:0:0 2:2:0.1:0.1 3:9:0:0 10:10:1:1 11:11:0.1:0.1 12:13:0.3:0.3 14:14:0.4:0.4 15:17:0.5:0.5 18:18:0.6:0.6 19:20:0.3:0.3 21:22:0.6:0.6 23:23:1:1 24:24:0.8:0.8 25:25:1:1 26:29:0.8:0.8 30:30:1:1 31:31:0.8:0.8 32:32:0.7:0.7 33:33:0:0 34:34:0.8:0.8 35:35:1:1 36:36:0.8:0.8 37:37:1:1 38:38:0.8:0.8 39:39:1:1 40:42:1:1 43:43:0.8:0.8 44:44:1:1 45:45:0.3:0.3
This file is provided in the suitability_scores
option:
r.estimap.recreation landuse=input_corine_land_cover_2006 suitability_scores=corine_suitability.scores potential=output_potential_corine
Example of a recreation spectrum output map while using a MASK, based on a fragment from the CORINE land data base
The same can be achieved with a long one-line string too:
r.estimap.recreation \ landuse=input_corine_land_cover_2006 \ suitability_scores="1:1:0:0,2:2:0.1:0.1,3:9:0:0,10:10:1:1,11:11:0.1:0.1,12:13:0.3:0.3,14:14:0.4:0.4,15:17:0.5:0.5,18:18:0.6:0.6,19:20:0.3:0.3,21:22:0.6:0.6,23:23:1:1,24:24:0.8:0.8,25:25:1:1,26:29:0.8:0.8,30:30:1:1,31:31:0.8:0.8,32:32:0.7:0.7,33:33:0:0,34:34:0.8:0.8,35:35:1:1,36:36:0.8:0.8,37:37:1:1,38:38:0.8:0.8,39:39:1:1,40:42:1:1,43:43:0.8:0.8,44:44:1:1,45:45:0.3:0.3" \ potential=potential_1
In fact, this very scoring scheme, for CORINE land data sets, is integrated
in the module, so we obtain the same output even by discarding the
suitability_scores
option:
r.estimap.recreation \ landuse=input_corine_land_cover_2006 \ suitability_scores=suitability_of_corine_land_cover.scores \ potential=output_potential_1 --o
This is so because CORINE is a standard choice among existing land data bases that cover european territories. In case of a user requirement to provide an alternative scoring scheme, all what is required is either of
suitability_scores
optionNikos Alexandris
Copyright 2018 European Union
Licensed under the EUPL, Version 1.2 or -- as soon they will be approved by the European Commission -- subsequent versions of the EUPL (the "Licence");
You may not use this work except in compliance with the Licence. You may obtain a copy of the Licence at: https://joinup.ec.europa.eu/collection/eupl/eupl-text-11-12
Unless required by applicable law or agreed to in writing, software distributed under the Licence is distributed on an "AS IS" basis, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the Licence for the specific language governing permissions and limitations under the Licence.
Consult the LICENCE file for details.
Available at: r.estimap.recreation source code (history)
Latest change: Monday Jan 30 19:52:26 2023 in commit: cac8d9d848299297977d1315b7e90cc3f7698730
Main index | Raster index | Topics index | Keywords index | Graphical index | Full index
© 2003-2024 GRASS Development Team, GRASS GIS 8.4.1dev Reference Manual