prep for DPS

NASA_CMS_2023/Mexico/DPS_algorithm/GMB_Mexico.R

+############# OPEN R, THEN RUN THE FOLLOWING COMMAND ###################
+
+options(repos=c(CRAN="https://cran.r-project.org"))
+install.packages("INLA",repos=c(getOption("repos"),INLA="https://inla.r-inla-download.org/R/stable"), dep=TRUE)
+
+############## LOAD PACKAGES ###########
+
+library("fmesher")
+library(MatrixModels)
+library(Matrix)
+library(INLA)
+library(inlabru)
+library(sf)
+library(terra)
+library(dplyr)
+library(spdep)
+library(exactextractr)
+library(sn)
+
+########################################
+args <- commandArgs(trailingOnly = TRUE)
+
+#### LOAD THE DATA USED TO MAKE THE MODEL ######
+mexico = st_read(args[2], quiet = TRUE) %>% st_union() %>% st_transform(crs = 6933) # 
+cci.rast = rast(args[3])
+hei.rast = rast(args[4])
+
+DATA <- read.csv(args[5]) #Read the saved data
+cci.plot <- DATA$cci.plot
+hei.plot <- DATA$hei.plot
+loc.plot <- data.frame(matrix(ncol = 2, nrow = length(hei.plot)))
+loc.plot[,1] <- DATA$X
+loc.plot[,2] <- DATA$X.1
+agbd.plot <- sqrt(DATA$agbd.plot)
+loc.plot <- data.matrix(loc.plot)
+
+###### REMAKE THE MESH IN CASE IT IS NOT STORED IN MEMORY ######
+max.edge = 10*10^3
+mexico.buffer = st_buffer(mexico, dist = max.edge*5) 
+mesh = inla.mesh.2d(boundary = list(as(mexico, "Spatial"), as(mexico.buffer, "Spatial")), max.edge = c(max.edge, 3*max.edge), cutoff = 2*max.edge/3, offset = c(max.edge, 5*max.edge)) 
+k = mesh$n 
+A.plot = inla.spde.make.A(mesh = mesh, loc = loc.plot)
+
+##### LOAD THE MODEL RESULTS ####
+load(args[6])
+samples = inla.posterior.sample(n = 250, result = model_fit_v2) 
+
+#### LOAD AN URBAN MASK AND PROBABILITY OF FOREST MASK FOR MEXICO ####
+URBAN = st_read(args[7],quiet=TRUE)
+FNF = rast(args[8])
+
+##### MAKE PREDICTIONS ###########
+
+PROJECTS = st_read(args[9], quiet = TRUE) %>% st_transform(crs = 6933)
+PROJECTS$ID <- seq.int(nrow(PROJECTS))
+PROJECTS$GMB_mean_AGBD_dense = NA
+PROJECTS$GMB_sd_AGBD_dense = NA
+
+for (f in (1:nrow(PROJECTS)))  { 
+    PROJECTS_ID = PROJECTS[f,]$ID
+    grid_to_predict = st_make_grid(PROJECTS[f,], cellsize = c(100,100), what = "centers") %>% st_as_sf() %>% st_filter(PROJECTS[f,])
+    grid_to_predict = grid_to_predict[URBAN, ,op=st_disjoint] 
+
+    if (nrow(grid_to_predict[URBAN, ,op=st_disjoint])>0){
+        grid = grid_to_predict %>% st_coordinates()
+        cci.pred = extract(cci.rast, grid)[,1]
+        hei.pred = extract(hei.rast, grid)[,1]
+
+        if (length(cci.pred) > 0 && length(hei.pred) > 0){
+            A.pred = inla.spde.make.A(mesh = mesh, loc = grid) 
+            pred_fun = function(...){
+              drop(intercept + 
+                cci.pred*cci + 
+                hei.pred*hei +
+                A.pred%*%alpha.spat[1:k] +   
+                Diagonal(x = cci.pred)%*%A.pred%*%beta.spat[1:k] +
+                Diagonal(x = hei.pred)%*%A.pred%*%eta.spat[1:k]) +
+                rnorm(nrow(A.pred), sd = sqrt(1/theta[1])) 
+            }
+
+            # Generate prediction samples
+            pred.samples = (inla.posterior.sample.eval(fun = pred_fun, samples = samples))
+            pred.samples[pred.samples < 0] = 0
+
+            # Model expectations and SD's at the grid locations
+            pred.mu_dense = Matrix::rowMeans(pred.samples^2, na.rm = TRUE) 
+            pred.sd_dense = apply(pred.samples^2, 1, sd, na.rm = TRUE) 
+            
+            # Enter information to CSV file
+            PROJECTS$GMB_mean_AGBD_dense[PROJECTS$ID == PROJECTS_ID] <- round(mean(pred.mu_dense,na.rm=TRUE),digits=2)
+            PROJECTS$GMB_sd_AGBD_dense[PROJECTS$ID == PROJECTS_ID] <- round(sd(colMeans(pred.samples^2, na.rm = T),na.rm=TRUE),digits=2)
+        }
+    }
+}
+
+#### SAVE OUR RESULTS 
+write.csv(st_drop_geometry(PROJECTS),paste0(args[1]))
\ No newline at end of file

NASA_CMS_2023/Mexico/DPS_algorithm/build_command_main_ADE.sh

+#!/bin/bash
+set -x
+basedir=$( cd "$(dirname "$0")" ; pwd -P )
+conda env update -n base --solver=libmamba -f ${basedir}/env_main_ADE.yaml
+
+conda install conda-forge::r-fmesher --yes
+conda install conda-forge::r-exactextractr --yes
+conda install conda-forge::r-sn --yes
+conda install conda-forge::r-MatrixModels --yes
+conda install conda-forge::r-inlabru --yes
+conda install conda-forge::r-terra --yes
+conda install conda-forge::r-dplyr --yes
+conda install conda-forge::r-spdep --yes
+conda install conda-forge::r-sf --yes
+conda install conda-forge::r-fields --yes
+conda install conda-forge::r-Matrix --yes
\ No newline at end of file

NASA_CMS_2023/Mexico/DPS_algorithm/env_main_ADE.yaml

+name: Packages_for_GMB
+channels:
+  - conda-forge
+  - defaults
+dependencies:
+    - r-fmesher
+    - r-exactextractr
+    - r-sn
+    - r-MatrixModels
+    - r-inlabru
+    - r-terra
+    - r-dplyr
+    - r-spdep
+    - r-sf
+    - r-fields
+    - r-Matrix
\ No newline at end of file

NASA_CMS_2023/Mexico/DPS_algorithm/run_GMBpredictions_MEX.sh

+#!/bin/bash
+
+basedir=$( cd "$(dirname "$0")" ; pwd -P )
+
+mkdir -p output
+
+output_csv=${1}
+mexico="input/ecort08gw_DESECON1_DISS.gpkg"
+ccirast="input/cci_mexico_6933.tif"
+heirast="input/GLAD_FH_mexico_UINT16_C_6399.tif"
+data="input/NFI_CCI_GEDIheights.csv"
+INLA_model_fit_v2="input/INLA_model_fit_v2.RData"
+MGN2020_INEGI_Urban_Mex_mask_DISS="input/MGN2020_INEGI_Urban_Mex_mask_DISS.gpkg"
+All_Products_Comp_Over30_Binary_6933="input/All_Products_Comp_Over30_Binary_6933.tif"
+Projects="input/Projects.gpkg" 
+
+OUTPUTDIR="${PWD}/output"
+
+Rscript ${basedir}/GMB_Mexico.R ${output_csv} ${mexico} ${ccirast} ${heirast} ${data} ${INLA_model_fit_v2} ${MGN2020_INEGI_Urban_Mex_mask_DISS} ${All_Products_Comp_Over30_Binary_6933} ${Projects} 
\ No newline at end of file

NASA_CMS_2023/Mexico/DPS_algorithm/run_GMBpredictions_MEX.yaml

+algorithm_description: Runs GMB predictions for Mexico
+algorithm_name: run_GMBpredictions_MEX
+algorithm_version: GMBpredictions_v1
+repository_url: https://repo.maap-project.org/lduncanson/biomass_harmonization
+docker_container_url: mas.maap-project.org/root/maap-workspaces/base_images/vanilla:v3.1.3
+queue: maap-dps-worker-16gb
+build_command: biomass_harmonization/NASA_CMS_2023/Mexico/DPS_algorithm/build_command_main_ADE.sh
+run_command: biomass_harmonization/NASA_CMS_2023/Mexico/DPS_algorithm/run_GMBpredictions_MEX.sh
+disk_space: 20GB 
+
+inputs:
+  config: []
+  file:
+  - default: 'mexico'
+    description: mexico
+    name: mexico
+    required: true
+  - default: 'ccirast'
+    description: ccirast
+    name: ccirast
+    required: true
+  - default: 'heirast'
+    description: heirast
+    name: heirast
+    required: true
+  - default: 'data'
+    description: data
+    name: data
+    required: true
+  - default: 'INLA_model_fit_v2'
+    description: INLA_model_fit_v2
+    name: INLA_model_fit_v2
+    required: true
+  - default: 'MGN2020_INEGI_Urban_Mex_mask_DISS'
+    description: MGN2020_INEGI_Urban_Mex_mask_DISS
+    name: MGN2020_INEGI_Urban_Mex_mask_DISS
+    required: true
+  - default: 'All_Products_Comp_Over30_Binary_6933'
+    description: All_Products_Comp_Over30_Binary_6933
+    name: All_Products_Comp_Over30_Binary_6933
+    required: true
+  - default: 'Projects'
+    description: Projects
+    name: Projects
+    required: true
+  positional:
+  - default: 'output_csv'
+    description: output_csv
+    name: output_csv
+    required: true
\ No newline at end of file