-
Notifications
You must be signed in to change notification settings - Fork 0
/
11-Compendium-of-code.Rmd
2128 lines (1745 loc) · 75.4 KB
/
11-Compendium-of-code.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# Annex I: Compendium of R scripts {-}
This chapter compiles the complete list of R scripts involved in the process of soil sampling design.
## Script 1: Introduction to R {-}
```{r intro R, eval=F}
# Introduction to R
# 0. Playground =================================================
# learn important keyboard shortcuts
# Ctrl + enter for running code
# tab after writing the first three
# characters of the function name
# F1 to access the help
# explore the use of <-, $, [], ==, !=, c(), :,
# data.frame(), list(), as.factor()
a <- 10:15
a[2]
a[2:3]
b <- c("1", "a", a )
length(b)
df <- data.frame(column_a = 1:8, column_b = b)
df[,1]
df$column_b
as.numeric(df$column_b)
plot(df)
df[1:3,]
df[,1]
as.factor(b)
d <- list(a, b, df)
d
names(d)
names(d) <- c("numeric_vector", "character_vector", "dataframe")
d
d[[1]]
d$numeric_vector
a == b
a != b
# 1. Set working directory ======================================
# Set working directory to a specific location
setwd("C:/GIT/Digital-Soil-Mapping/")
# Set working directory to source file location
setwd(dirname(rstudioapi::getActiveDocumentContext()$path))
# 2. Install and load packages ==================================
# readxl, tidyverse, and data.table packages using the functions
install.packages("tidyverse")
install.packages("readxl")
install.packages("data.table")
library(tidyverse)
library(readxl)
library(data.table)
# Install packages from other repository than CRAN
install.packages("remotes")
remotes::install_github("lemuscanovas/synoptReg")
# 3. Import an spreadsheet ======================================
## 3.1 Read the MS Excel file -----------------------------------
#Read the soil_data.xlsx file, spreadsheet 2, using read_excel
read_excel(path = "data/other/soil_data.xlsx", sheet = 2)
## 3.2 Read the csv file with the native function ---------------
# 01-Data/horizon.csv
read.csv("data/other/soil_profile_data.csv")
## 3.3 Read the csv file with the tidyverse function ------------
read_csv("data/other/soil_profile_data.csv")
## 3.4 Read the csv file with the data.table function -----------
fread("data/other/soil_profile_data.csv")
## 3.5 Assign the dataframe to an object called dat -------------
dat <- read_csv("data/other/soil_profile_data.csv")
# 4. Tidyverse functions ========================================
## 4.1 Select pid, hip, top, bottom, ph_h2o, cec from dat -------
dat_1 <- dat %>%
select(id_prof, id_hor, top, bottom, ph_h2o, cec)
## 4.2 Filter: pick observations by their values ----------------
# filter observations with cec > 50 cmolc/100g
dat_2 <- dat_1 %>%
filter(cec > 30)
dat_2
## 4.3 Mutate: create a new variable ----------------------------
# thickness = top - bottom
dat_3 <- dat_2 %>%
mutate(thickness = bottom - top)
## 4.4 Group_by and summarise -----------------------------------
# group by variable pid
# summarise taking the mean of pH and cec
dat_4 <- dat_3 %>%
group_by(id_prof) %>%
summarise(mean_ph = mean(ph_h2o),
mean_cec = mean(cec))
## 4.5 Reshape the table using pivot_longer ---------------------
# use dat_3
# put the names of the variables ph_h2o,
# cec and thickness in the column
# variable and keep the rest of the table. Save in dat_5
dat_5 <- dat_3 %>%
pivot_longer(ph_h2o:thickness, names_to = "soil_property",
values_to = "value")
## 4.6 Join the table sites.csv with dat_3 ----------------------
# Load soil_phys_data030.csv (in 01-Data folder)
# Join its columns with dat_3 keeping all the rows of dat_3
# save the result as dat_6
phys <- read_csv("data/other/soil_phys_data030_2.csv")
phys <- phys %>% rename(id_prof = "ProfID")
dat_6 <- dat_3 %>%
left_join(phys)
# or
dat_6 <- phys %>%
right_join(dat_3)
# 5. Data visualization with ggplot2 ============================
## 5.1 1D plot: histograms --------------------------------------
# histograms of cec and ph_h2o
ggplot(dat_3, aes(x=cec)) + geom_histogram()
## 5.2 2D plot: scatterplot -------------------------------------
# Scatterplot bottom vs. ph_h2o
ggplot(dat_3, aes(x = bottom, y = ph_h2o)) +
geom_point()
# add a fitting line
ggplot(dat_3, aes(x = bottom, y = ph_h2o)) +
geom_point() +
geom_smooth(method = "lm" )
## 5.3 3D plot: scatterplot -------------------------------------
# Scatterplot bottom vs. ph_h2o, add clay as color
# and size inside the
# function aes()
ggplot(dat_3,
aes(x = bottom, y = ph_h2o, color = cec, size = cec)) +
geom_point()
# 6. Geospatial data with terra =================================
## Load packages (install them if needed)
library(terra)
## 6.1 Load a raster and a vector layer -------------------------
# Load data/other/landcover2013.tif using rast() function, then plot it
# Load data/shapes/landcover.shp using vect() function and
# plot it
# explore the attributes of these layers
r <- rast("data/other/landcover2013.tif")
plot(r)
v <- vect("data/shapes/landcover.shp")
plot(v)
## 6.2 Load a raster and a vector layer -------------------------
# Check the current CRS (EPSG) of the raster and the vector.
# Find a *projected* CRS in http://epsg.io for Vietnam and
# copy the number
# Check the Arguments of function project (?project) that need to
# be defined
# Save the new object as r_proj and v_proj
# plot both objects
r_proj <- project(x = r, y = "epsg:3405", method = "near",
res = 250)
plot(r_proj)
v_proj <- project(x = v, y = "epsg:3405")
plot(v_proj, add = TRUE)
## 6.3 Cropping and masking a raster ----------------------------
# Compute the area of the polygons in v_proj
# (search for a function) and
# assign the values to a new column named area
# select the largest polygon using [], $, == and max() func.
# and save it as pol
# crop the raster with pol using the crop() function and save
#it as r_pol
# mask the raster r_pol with the polygon pol and save it
# with the same name
# plot each result
v_proj$area <- expanse(v_proj, unit = "ha")
pol <- v_proj[v_proj$area == max(v_proj$area)]
plot(pol)
r_pol <- crop(r_proj, pol)
plot(r_pol)
plot(pol, add = TRUE)
r_pol <- mask(r_pol, pol)
plot(r_pol)
## 6.4 Replace values in a raster by filtering their cells ------
# Explore the following link to understand how terra
#manage cell values
# https://rspatial.org/terra/pkg/4-algebra.html
# Replace values lower than 5 in r+pol by 0
r_pol[r_pol$landcover2013 < 5] <- 0
plot(r_pol)
## 6.5 Rasterize a vector layer ---------------------------------
# Use rasterize() function to convert v_proj to raster
# Use r_proj as reference raster
# Use field landcover to assign cell values, and plot the new map
v_class <- rasterize(x = v_proj, y = r_proj, field = "landcover" )
plot(v_class)
v_class
activeCat(v_class) <- 1
## 6.6 Extracting raster values using points --------------------
# Covert dat_6 to spatial points using vect() function
# (check help of vect())
# Note that the EPSG number is 3405
# Save the points as s
# Plot s and r_proj together in the same map (Argument add=TRUE)
# Extract the values of the raster using extract()
# function (check the help)
# Remove the ID column of the extracted values
# merge the extracted data with s using cbind() function
# Convert s as a dataframe
s <- vect(dat_6, geom=c("x", "y"), crs = "epsg:4326")
#s <- project(x = s, y = "epsg:3405")
plot(r_proj)
plot(s, add=TRUE)
x <- extract(r_proj,s, ID=FALSE)
s <- cbind(s,x)
d <- as.data.frame(s)
d
#GGally::ggscatmat(d)
## 6.7 Zonal statistics using polygons and rasters --------------
# Use the extract() func. to estimate the mean value of
# distance_proj at each polygon
# Use the fun= argument (check the help)
# Use the cbind() func. to merge v_proj and the extracted values
# convert v_proj to a dataframe
# Create a ggplot boxplot (geom_boxplot) with x=landcover
# and y=dist2access
distance <- rast("data/other/nghe_d2roads.tif")
plot(distance)
distance_proj <- project(x = distance, y = "epsg:3405", method = "bilinear",
res = 250)
plot(distance_proj)
x <- extract(distance_proj, v_proj, fun = mean, ID=FALSE)
v_proj <- cbind(v_proj, x)
d <- as_tibble(v_proj)
d %>%
ggplot(aes(x =landcover, y = dist2access, fill = landcover)) +
geom_boxplot() +
ylab("Distance to roads")
## END
```
## Script 2: Download environmental covariates {-}
```{js covs, echo=TRUE, eval=FALSE}
var assets =
["projects/digital-soil-mapping-gsp-fao/assets/CHELSA/bio1",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/bio12",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/bio13",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/bio14",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/bio16",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/bio17",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/bio5",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/bio6",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/ngd10",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/pet_penman_max",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/pet_penman_mean",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/pet_penman_min",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/pet_penman_range",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/sfcWind_max",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/sfcWind_mean",
"projects/digital-soil-mapping-gsp-fao/assets/CHELSA/sfcWind_range",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/fpar_030405_500m_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/fpar_030405_500m_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/fpar_060708_500m_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/fpar_060708_500m_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/fpar_091011_500m_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/fpar_091011_500m_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/fpar_120102_500m_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/fpar_120102_500m_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/lstd_030405_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/lstd_030405_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/lstd_060708_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/lstd_060708_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/lstd_091011_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/lstd_091011_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/lstd_120102_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/lstd_120102_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndlst_030405_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndlst_030405_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndlst_060708_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndlst_060708_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndlst_091011_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndlst_091011_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndlst_120102_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndlst_120102_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndvi_030405_250m_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndvi_030405_250m_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndvi_060708_250m_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndvi_060708_250m_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndvi_091011_250m_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndvi_091011_250m_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndvi_120102_250m_mean",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/ndvi_120102_250m_sd",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/snow_cover",
"projects/digital-soil-mapping-gsp-fao/assets/MODIS/swir_060708_500m_mean",
"projects/digital-soil-mapping-gsp-fao/assets/LANDCOVER/crops",
"projects/digital-soil-mapping-gsp-fao/assets/LANDCOVER/flooded_vegetation",
"projects/digital-soil-mapping-gsp-fao/assets/LANDCOVER/grass",
"projects/digital-soil-mapping-gsp-fao/assets/LANDCOVER/shrub_and_scrub",
"projects/digital-soil-mapping-gsp-fao/assets/LANDCOVER/trees",
"projects/digital-soil-mapping-gsp-fao/assets/OPENLANDMAP/dtm_curvature_250m",
"projects/digital-soil-mapping-gsp-fao/assets/OPENLANDMAP/dtm_downslopecurvature_250m",
"projects/digital-soil-mapping-gsp-fao/assets/OPENLANDMAP/dtm_dvm2_250m",
"projects/digital-soil-mapping-gsp-fao/assets/OPENLANDMAP/dtm_dvm_250m",
"projects/digital-soil-mapping-gsp-fao/assets/OPENLANDMAP/dtm_elevation_250m",
"projects/digital-soil-mapping-gsp-fao/assets/OPENLANDMAP/dtm_mrn_250m",
"projects/digital-soil-mapping-gsp-fao/assets/OPENLANDMAP/dtm_neg_openness_250m",
"projects/digital-soil-mapping-gsp-fao/assets/OPENLANDMAP/dtm_pos_openness_250m",
"projects/digital-soil-mapping-gsp-fao/assets/OPENLANDMAP/dtm_slope_250m",
"projects/digital-soil-mapping-gsp-fao/assets/OPENLANDMAP/dtm_tpi_250m",
"projects/digital-soil-mapping-gsp-fao/assets/OPENLANDMAP/dtm_twi_500m",
"projects/digital-soil-mapping-gsp-fao/assets/OPENLANDMAP/dtm_upslopecurvature_250m",
"projects/digital-soil-mapping-gsp-fao/assets/OPENLANDMAP/dtm_vbf_250m"];
// Load borders
/// Using UN 2020 (replace the countries to download)
/// var ISO = ['ITA'];
/// var aoi =
/// ee.FeatureCollection(
/// 'projects/digital-soil-mapping-gsp-fao/assets/UN_BORDERS/BNDA_CTY'
/// )
/// .filter(ee.Filter.inList('ISO3CD', ISO));
/// var region = aoi.geometry();
/// Using a shapefile
/// 1. Upload the borders of your countries as an asset
/// 2. Replace 'your_shapefile' with the path to your shapefile
var shapefile = ee.FeatureCollection('projects/digital-soil-mapping-gsp-fao/assets/Nghe_An');
var region = shapefile.geometry().bounds();
// Load assets as ImageCollection
var assetsCollection = ee.ImageCollection(assets);
// Clip each image in the collection to the region of interest
var clippedCollection = assetsCollection.map(function(img){
return img.clip(region).toFloat();
});
// Function to replace masked values with zeroes for fpar bands
function replaceMaskedFpar(img) {
var allBands = img.bandNames();
var fparBands =
allBands.filter(ee.Filter.stringStartsWith('item', 'fpar'));
var nonFparBands = allBands.removeAll(fparBands);
var fparImg = img.select(fparBands).unmask(0);
var nonFparImg = img.select(nonFparBands);
// If there are no fpar bands, return the original image
var result = ee.Algorithms.If(fparBands.length().eq(0),
img,
nonFparImg.addBands(fparImg));
return ee.Image(result);
}
// Clip each image in the collection to the region of
//interest and replace masked values for fpar bands
var clippedCollection = assetsCollection.map(function(img){
var clippedImg = img.clip(region).toFloat();
return replaceMaskedFpar(clippedImg);
});
// Stack the layers and maintain the
//layer names in the final file
var stacked = clippedCollection.toBands();
// Get the list of asset names
var assetNames = ee.List(assets).map(function(asset) {
return ee.String(asset).split('/').get(-1);
});
// Rename the bands with asset names
var renamed = stacked.rename(assetNames);
print(renamed, 'Covariates to be exported')
// Visualize the result
// Set a visualization parameter
// (you can adjust the colors as desired)
var visParams = {
bands: 'bio1',
min: 0,
max: 1,
palette: ['blue', 'green', 'yellow', 'red']
};
// Add the layer to the map
Map.centerObject(renamed, 6)
Map.addLayer(renamed, visParams, 'Covariates');
// Export the stacked image to Google Drive
Export.image.toDrive({
image: renamed,
description: 'covariates',
folder: 'GEE',
scale: 250,
maxPixels: 1e13,
region: region
});
/* Create mask for croplands ----------------------------*/
// Load the Copernicus Global Land Service image collection
var imageCollection =
ee.Image("COPERNICUS/Landcover/100m/Proba-V-C3/Global/2019")
.select("discrete_classification")
.clip(region)
var crs = 'EPSG:4326'; // WGS84
var res = 250; // Resolution in decimal degrees
// Default resampling is nearest neighbor
var image1 = imageCollection.resample()
.reproject({
crs: crs, // Add your desired CRS here
scale: res // Add your desired scale here
});
// Reclassify the land cover classes
var inList = [0, 20, 30, 40, 50, 60, 70, 80, 90, 100,
111, 112, 113, 114, 115, 116,
121, 122, 123, 124, 125, 126, 200];
var outList = [0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0];
var FAO_lu = image1.remap(inList, outList)
.toDouble()
.clip(region);
// print(FAO_lu)
// Convert 0 to NA
var mask = FAO_lu.neq(0);
print(mask)
FAO_lu = FAO_lu.updateMask(mask);
print(FAO_lu, "Mask")
var visParams = {
bands: 'remapped',
min: 0,
max: 1,
palette: ['green', 'yellow']
};
// Add the layer to the map
Map.addLayer(FAO_lu,visParams ,'Mask');
// Export the land cover image as a raster to Google Drive
Export.image.toDrive({
image: FAO_lu,
folder: 'GEE',
description: 'mask',
scale: res, // Add your desired scale here
region: region,
crs: crs, // Add your desired CRS here
maxPixels: 1e13 // Add a maximum number of pixels
//for export if needed
});
```
## Script 3: Evaluate Legacy Data {-}
```{r legacy-data, eval=FALSE}
#
# Digital Soil Mapping
# Soil Sampling Design
# Evaluation of Legacy Data
#
# GSP-Secretariat
# Contact: [email protected]
#________________________________________________________________
# Empty environment and cache
rm(list = ls())
gc()
# Content of this script ========================================
# Script for evaluation the degree of representativeness of a soil legacy dataset
# relative to the diversity of the environmental conditions described in a set
# of raster covariates.
#
# 0 - Set working directory and load packages
# 1 - User-defined variables
# 2 - Prepare data
# 3 - Extract environmental data from rasters at soil locations
# 4 - Compute variability matrix in covariates
# 5 - Calculate hypercube of "covariates" distribution (P)
# 6 - Calculate hypercube of "sample" distribution (Q)
# 7 - Calculate Representativeness of the Legacy Dataset
#________________________________________________________________
## 0 - Set working directory and load packages =================================
#remotes::install_github("lemuscanovas/synoptReg")
# Set working directory to source file location
setwd(dirname(rstudioapi::getActiveDocumentContext()$path))
setwd("../") # Move wd down to main folder
# List of packages
packages <- c("sp","terra","raster","sf","clhs", "sgsR","entropy", "tripack",
"manipulate","dplyr","synoptReg")
# Load packages
lapply(packages, require, character.only = TRUE)
# Remove object to save memory space
rm(packages)
## 1 - User-defined variables ==================================================
# Path to rasters
raster.path <- "data/rasters"
# Path to shapes
shp.path <- "data/shapes"
# Path to results
results.path <- "data/results/"
# Path to additional data
other.path <- "data/other/"
# Aggregation factor for up-scaling raster covariates (optional)
agg.factor = 10
## 2 - Prepare data ============================================================
## Load raster covariate data
# Read Spatial data covariates as rasters with terra
cov.dat <- list.files(raster.path, pattern = "tif$", recursive = TRUE, full.names = TRUE)
cov.dat <- terra::rast(cov.dat) # SpatRaster from terra
# Aggregate stack to simplify data rasters for calculations
cov.dat <- aggregate(cov.dat, fact=agg.factor, fun="mean")
# Load shape of district
nghe <- sf::st_read(file.path(paste0(shp.path,"/Nghe_An.shp")),quiet=TRUE)
# Crop covariates on administrative boundary
cov.dat <- crop(cov.dat, nghe, mask=TRUE)
# Transform raster information with PCA
pca <- raster_pca(cov.dat)
# Get SpatRaster layers
cov.dat <- pca$PCA
# Create a raster stack to be used as input in the clhs::clhs function
cov.dat.ras <- raster::stack(cov.dat)
# Subset rasters
cov.dat <- pca$PCA[[1:first(which(pca$summaryPCA[3,]>0.99))]]
cov.dat.ras <- cov.dat.ras[[1:first(which(pca$summaryPCA[3,]>0.99))]]
## 3 - Extract environmental data from rasters at soil locations ===============
# Load legacy soil data
p.dat <- terra::vect(file.path(paste0(shp.path,"/legacy_soils.shp")))
# Extract data
p.dat_I <- terra::extract(cov.dat, p.dat)
p.dat_I <- na.omit(p.dat_I) # Remove soil points outside study area
p.dat_I.df <- p.dat_I[,-1]
str(p.dat_I.df)
## 4 - Compute variability matrix in the covariates ====================================
# Define Number of bins
nb <- 25
# Quantile matrix of the covariate data
q.mat <- matrix(NA, nrow = (nb + 1), ncol = nlyr(cov.dat))
j = 1
for (i in 1:nlyr(cov.dat)) {
ran1 <- minmax(cov.dat[[i]])[2] - minmax(cov.dat[[i]])[1]
step1 <- ran1 / nb
q.mat[, j] <-
seq(minmax(cov.dat[[i]])[1],
to = minmax(cov.dat[[i]])[2],
by = step1)
j <- j + 1
}
q.mat
## 5 - Calculate hypercube of "covariates" distribution (P) ===================
# Convert SpatRaster to dataframe for calculations
cov.dat.df <- as.data.frame(cov.dat)
cov.mat <- matrix(1, nrow = nb, ncol = ncol(q.mat))
cov.dat.mx <- as.matrix(cov.dat.df)
for (i in 1:nrow(cov.dat.mx)) {
for (j in 1:ncol(cov.dat.mx)) {
dd <- cov.dat.mx[[i, j]]
if (!is.na(dd)) {
for (k in 1:nb) {
kl <- q.mat[k, j]
ku <- q.mat[k + 1, j]
if (dd >= kl && dd <= ku) {
cov.mat[k, j] <- cov.mat[k, j] + 1
}
}
}
}
}
cov.mat
## 6 - Calculate hypercube of "sample" distribution (Q) ========================
h.mat <- matrix(1, nrow = nb, ncol = ncol(q.mat))
for (i in 1:nrow(p.dat_I.df)) {
for (j in 1:ncol(p.dat_I.df)) {
dd <- p.dat_I.df[i, j]
if (!is.na(dd)) {
for (k in 1:nb) {
kl <- q.mat[k, j]
ku <- q.mat[k + 1, j]
if (dd >= kl && dd <= ku) {
h.mat[k, j] <- h.mat[k, j] + 1
}
}
}
}
}
h.mat
## 7 - Calculate Representativeness of the Legacy Dataset ==================
## Calculate the proportion of "variables" in the covariate spectra that fall within the convex hull of variables in the "environmental sample space"
# Principal component of the legacy data sample
pca.s = prcomp(p.dat_I[,2:(ncol(cov.dat.df)+1)],scale=TRUE, center=TRUE)
scores_pca1 = as.data.frame(pca.s$x)
# Plot the first 2 principal components and convex hull
rand.tr <- tri.mesh(scores_pca1[,1],scores_pca1[,2],"remove") # Delaunay triangulation
rand.ch <- convex.hull(rand.tr, plot.it=F) # convex hull
pr_poly = cbind(x=c(rand.ch$x),y=c(rand.ch$y)) # save the convex hull vertices
plot(scores_pca1[,1], scores_pca1[,2], xlab="PCA 1", ylab="PCA 2", xlim=c(min(scores_pca1[,1:2]), max(scores_pca1[,1:2])),ylim=c(min(scores_pca1[,1:2]), max(scores_pca1[,1:2])), main='Convex hull of soil legacy data')
lines(c(rand.ch$x,rand.ch$x[1]), c(rand.ch$y,rand.ch$y[1]),col="red",lwd=1) # draw the convex hull (domain of legacy data)
# PCA projection of study area population onto the principal components
PCA_projection <- predict(pca.s, cov.dat.df) # Project study area population onto sample PC
newScores = cbind(x=PCA_projection[,1],y=PCA_projection[,2]) # PC scores of projected population
# Plot the polygon and all points to be checked
plot(newScores, xlab="PCA 1", ylab="PCA 2", xlim=c(min(newScores[,1:2]), max(newScores[,1:2])), ylim=c(min(newScores[,1:2]), max(newScores[,1:2])), col='black', main='Environmental space plots over the convex hull of soil legacy data')
polygon(pr_poly,col='#99999990')
# Check which points fall within the polygon
pip <- point.in.polygon(newScores[,2], newScores[,1], pr_poly[,2],pr_poly[,1],mode.checked=FALSE)
newScores <- data.frame(cbind(newScores, pip))
# Plot points outside convex hull
points(newScores[which(newScores$pip==0),1:2],pch='X', col='red')
# Proportion of the conditions in the study area that fall within the convex hull of sample conditions
sum(nrow(newScores[newScores$pip>0,]))/nrow(newScores)*100
## END
```
## Script 4: Calculate Minimum and Optimal Sample Sizes {-}
```{r sample-size, eval=FALSE}
#
# Digital Soil Mapping
# Soil Sampling Design
# Optimizing Sample Size
#
# GSP-Secretariat
# Contact: [email protected]
#________________________________________________________________
#Empty environment and cache
rm(list = ls())
gc()
# Content of this script ========================================
# The goal of this script is to determine the minimum sample size required to describe an area
# while retaining for a 95% of coincidence in the environmental variability of covariates
# in the area
#
# 0 - Set working directory and load necessary packages
# 1 - User-defined variables
# 2 - Import national data
# 3 - Calculate the minimum sample size to describe the area
# 4 - Plot covariate diversity as PCA scores
# 5 - KL divergence and % similarity results for growing N samples
# 6 - Model KL divergence
# 7 - Determine the minimum sample size for 95% coincidence
# 8 - Determine the optimal iteration according to the minimum N size
# 9 - Plot minimum points from best iteration
#________________________________________________________________
## 0 - Set working directory and load necessary packages =======================
# Set working directory to source file location
setwd(dirname(rstudioapi::getActiveDocumentContext()$path))
setwd("../") # Move wd down to main folder
# List of packages
packages <- c("sp","terra","raster","sf","clhs",
"sgsR","entropy", "tripack",
"manipulate","dplyr","plotly","synoptReg")
# Load packages
lapply(packages, require, character.only = TRUE)
rm(packages) # Remove object to save memory space
## 1 - User-defined variables ==================================================
# Path to rasters
raster.path <- "data/rasters/"
# Path to shapes
shp.path <- "data/shapes/"
# Path to results
results.path <- "data/results/"
# Path to additional data
other.path <- "data/other/"
# Aggregation factor for up-scaling raster covariates (optional)
agg.factor = 10
# Define parameters to determine minimum sampling size
initial.n <- 60 # Initial sampling size to test
final.n <- 360 # Final sampling size to test
by.n <- 20 # Increment size
iters <- 5 # Number of trials on each size
## 2 - Import national data ====================================================
## Load raster covariate data
# Read Spatial data covariates as rasters with terra
cov.dat <- list.files(raster.path, pattern = "tif$", recursive = TRUE, full.names = TRUE)
cov.dat <- terra::rast(cov.dat) # SpatRaster from terra
# Aggregate stack to simplify data rasters for calculations
cov.dat <- aggregate(cov.dat, fact=agg.factor, fun="mean")
# Load shape of district
nghe <- sf::st_read(file.path(paste0(shp.path,"/Nghe_An.shp")),quiet=TRUE)
# Crop covariates on administrative boundary
cov.dat <- crop(cov.dat, nghe, mask=TRUE)
# Store elevation and slope separately
elevation <- cov.dat$dtm_elevation_250m
slope <- cov.dat$dtm_slope_250m
# Load roads
roads <- vect(file.path(paste0(shp.path,"/roads.shp")))
roads <- crop(roads, nghe)
# Simplify raster information with PCA
pca <- raster_pca(cov.dat)
# Get SpatRaster layers
cov.dat <- pca$PCA
# Subset rasters to main PC (var. explained <=0.99)
n_comps <- first(which(pca$summaryPCA[3,]>0.99))
cov.dat <- pca$PCA[[1:n_comps]]
# Plot covariates
plot(cov.dat)
# 3 - Calculate the minimum sample size to describe the area ===================
# Start computations ----
# Initialize empty vectors to store results
number_of_samples <- c()
prop_explained <- c()
klo_samples <- c()
samples_storage <- list()
# Convert SpatRaster to dataframe
cov.dat.df <- as.data.frame(cov.dat)
# Start evaluation with growing sample sizes
for (trial in seq(initial.n, final.n, by = by.n)) {
for (iteration in 1:iters) {
# Generate stratified clhs samples
p.dat_I <- sample_clhs(cov.dat,
nSamp = trial, iter = 10000,
progress = FALSE, simple = FALSE)
# Get covariate values as dataframe and delete NAs, avoid geometry
p.dat_I.df <- as.data.frame(p.dat_I) %>%
dplyr::select(!(geometry)) %>%
na.omit()
# Store samples as list
samples_storage[[paste0("N", trial, "_", iteration)]] <- p.dat_I
## Comparison of population and sample distributions - Kullback-Leibler (KL) divergence
# Define quantiles of the study area (number of bins)
nb <- 25
# Quantile matrix of the covariate data
q.mat <- matrix(NA, nrow = (nb + 1), ncol = nlyr(cov.dat))
j = 1
for (i in 1:nlyr(cov.dat)) {
ran1 <- minmax(cov.dat[[i]])[2] - minmax(cov.dat[[i]])[1]
step1 <- ran1 / nb
q.mat[, j] <-
seq(minmax(cov.dat[[i]])[1],
to = minmax(cov.dat[[i]])[2],
by = step1)
j <- j + 1
}
q.mat
# Hypercube of covariates in study area
# Initialize the covariate matrix
cov.mat <- matrix(1, nrow = nb, ncol = ncol(q.mat))
cov.dat.mx <- as.matrix(cov.dat.df)
for (i in 1:nrow(cov.dat.mx)) {
for (j in 1:ncol(cov.dat.mx)) {
dd <- cov.dat.mx[[i, j]]
if (!is.na(dd)) {
for (k in 1:nb) {
kl <- q.mat[k, j]
ku <- q.mat[k + 1, j]
if (dd >= kl && dd <= ku) {
cov.mat[k, j] <- cov.mat[k, j] + 1
}
}
}
}
}
cov.mat
# Compare whole study area covariate space with the selected sample
# Sample data hypercube (the same as for the raster data but on the sample data)
h.mat <- matrix(1, nrow = nb, ncol = ncol(q.mat))
for (i in 1:nrow(p.dat_I.df)) {
for (j in 1:ncol(p.dat_I.df)) {
dd <- p.dat_I.df[i, j]
if (!is.na(dd)) {
for (k in 1:nb) {
kl <- q.mat[k, j]
ku <- q.mat[k + 1, j]
if (dd >= kl && dd <= ku) {
h.mat[k, j] <- h.mat[k, j] + 1
}
}
}
}
}
h.mat
## Compute Kullback-Leibler (KL) divergence
kl.index <- c()
for (i in 1:ncol(cov.dat.df)) {
kl <- KL.empirical(c(cov.mat[, i]), c(h.mat[, i]))
kl.index <- c(kl.index, kl)
klo <- mean(kl.index)
}
## Calculate the proportion of "env. variables" in the covariate spectra that fall within the convex hull of variables in the "environmental sample space"
# Principal component of the data sample
pca.s = prcomp(p.dat_I.df, scale = TRUE, center = TRUE)
scores_pca1 = as.data.frame(pca.s$x)
# Plot the first 2 principal components and convex hull
rand.tr <-
tri.mesh(scores_pca1[, 1], scores_pca1[, 2], "remove") # Delaunay triangulation
rand.ch <- convex.hull(rand.tr, plot.it = F) # convex hull
pr_poly <-
cbind(x = c(rand.ch$x), y = c(rand.ch$y)) # save the convex hull vertices
# PCA projection of study area population onto the principal components
PCA_projection <-
predict(pca.s, cov.dat.df) # Project study area population onto sample PC
newScores = cbind(x = PCA_projection[, 1], y = PCA_projection[, 2]) # PC scores of projected population
# Check which points fall within the polygon
pip <-
point.in.polygon(newScores[, 2], newScores[, 1], pr_poly[, 2], pr_poly[, 1], mode.checked =
FALSE)
newScores <- data.frame(cbind(newScores, pip))
klo_samples <- c(klo_samples, klo)
prop_explained <-
c(prop_explained, sum(newScores$pip) / nrow(newScores) * 100)
number_of_samples <- c(number_of_samples, trial)
print(
paste(
"N samples = ",
trial,
" out of ",
final.n,
"; iteration = ",
iteration,
"; KL = ",
klo,
"; Proportion = ",
sum(newScores$pip) / nrow(newScores) * 100
)
)
}
}
# 4 - Plot covariate diversity as PCA scores ===================================
plot(newScores[,1:2],
xlab = "PCA 1",
ylab = "PCA 2",
xlim = c(min(newScores[,1:2], na.rm = T), max(newScores[,1:2], na.rm = T)),
ylim = c(min(newScores[,1:2], na.rm = T), max(newScores[,1:2], na.rm = T)),
col = 'black',
main = 'Environmental space plots on convex hull of soil samples')
polygon(pr_poly, col = '#99999990')
# # Plot points outside convex hull
points(newScores[which(newScores$pip == 0), 1:2],
col = 'red',
pch = 12,
cex = 1)
# 5 - KL divergence and % coincidence for growing N samples =============
# Merge data from number of samples, KL divergence and % coincidence
results <- data.frame(number_of_samples, klo_samples, prop_explained)
names(results) <- c("N", "KL", "Perc")
# Calculate mean results by N size
mean_result <- results %>%
group_by(N) %>%
summarize_all(mean)
mean_result
## Plot dispersion on KL and % by N
par(mar = c(5, 4, 5, 5))
boxplot(
Perc ~ N,
data = results,
col = rgb(1, 0.1, 0, alpha = 0.5),
ylab = "% coincidence"
)
mtext("KL divergence", side = 4, line = 3)
# Add new plot
par(new = TRUE, mar = c(5, 4, 5, 5))
# Box plot
boxplot(
KL ~ N,
data = results,
axes = FALSE,
outline = FALSE,
col = rgb(0, 0.8, 1, alpha = 0.5),
ylab = ""
)
axis(
4,
at = seq(0.02, 0.36, by = .06),
label = seq(0.02, 0.36, by = .06),
las = 3
)