jchen-utoronto / beps_hourly_site Goto Github PK

您好，模型的input数据 我看均是按照365天计算的，不知代码是否考虑了366天年份的情况

https://github.com/JChen-UToronto/BEPS_hourly_site/blob/master/soil_thermal_regime.c#L22-L31

L22 should be for (i = 1; i < p->n_layer; i++)

mass_snow_o_last is completely overwrited before pass value to others

I think those 3 lines should be removed

It will lead to the initial value Wcs_o[0]=var_o[16] never used.

hi，
i am a dummy of program,but i want to use your project to complete my study,so,if i use visual studio,how can i run this project.
many thx

BEPS script might have the problem of memory leak.
I am trying to interface BEPS with Rcpp. Most work has been done. But results returned by BEPS are different in different tries.

Any suggestion how to solve this problem?
https://github.com/rpkgs/rBEPS/blob/dev/Rcpp_debug.R

#! /usr/bin/Rscript
devtools::load_all()
# beps(system.file("examples/input", package = "rBEPS"))

library(data.table)
library(dplyr)

d_metro <- fread("inst/examples/input/p1_meteo.txt")
colnames(d_metro) <- c("day", "hour", "rad", "tem", "hum", "pre", "wind")

LAI <- read.table("inst/examples/input/p1_lai.txt") %>%
    unlist() %>%
    as.numeric()

opts <- c(
    long = 120.5, lat = 30.5,
    clumping = 0.85,
    LC = 25,
    soil_type = 8,
    soiltemp = 2.2,
    soilwater = 0.4115,
    snowdepth = 0, 
    j_start = 1, 
    j_end = 1
)

func <- function() {
    beps_main(indir, d_metro, LAI, opts) %>% mat2df() %>% 
        dplyr::select(-starts_with("npp"), -ends_with("resp"), -NEP) %>% 
        .[, -(1:4)]
}

indir <- "inst/examples/input"
res1 = func(); print(res1)
res2 = func(); print(res2)

diff = tibble::as_tibble(res1 - res2)
print(diff)

Results difference of 1st time and 2nd time

# A tibble: 24 × 6
         GPP  Net_Rad       SH          LH     Trans       Evap
       <dbl>    <dbl>    <dbl>       <dbl>     <dbl>      <dbl>
 1 -5.75e-10 -6.31e-8  1.36    -30.1       -8.15e-10 -4.41e-  2
 2 -7.23e-11  1.24e-9  0.841     0.340     -1.12e-10  4.96e-  4
 3  4.58e-10  2.77e-8 -0.00158   0.128      7.65e-10  1.87e-  4
 4  1.18e- 9  8.06e-8 -0.00120   0.128      2.12e- 9  1.87e-  4
 5  7.10e- 8 -1.93e-3 -0.00386   0.115      5.56e- 8  1.68e-  4
 6 -1.23e- 7  4.40e-3  0.00352   0.0917    -1.19e- 7  1.34e-  4
 7 -7.90e- 8  3.82e-3  0.00478   0.0679    -7.21e- 8  9.88e-  5
 8  4.58e- 9  2.09e-7  0.00588   0.0508     6.82e- 9  7.43e-  5
 9 -3.10e- 9 -1.41e-7 -0.0995    0.00759   -5.23e- 9  1.11e-  5
10 -1.10e-10  8.78e-4 -0.00736   0.0000175  2.55e- 8  1.78e-320

See details:

https://github.com/JChen-UToronto/BEPS_hourly_site/blob/master/aerodynamic_conductance.c

Julia is much easier to use, and more friendly to users.
It is amazing to find that BEPS in julia is about 2x faster than C.

The test case is at: https://github.com/CUG-hydro/BEPS.jl/blob/master/example_01.qmd.
Call for contributors for BEPS.jl.

In Line 422-424 https://github.com/JChen-UToronto/BEPS_hourly_site/blob/master/inter_prg.c#L422,

            Ci_u_sunlit_old=Ci_o_sunlit_new;
            Cs_u_sunlit_old=Cs_u_sunlit_new;
            Gs_u_sunlit_old=Gs_o_sunlit_new;   // m/s

might should be

            Ci_u_sunlit_old=Ci_u_sunlit_new;
            Cs_u_sunlit_old=Cs_u_sunlit_new;
            Gs_u_sunlit_old=Gs_u_sunlit_new;   // m/s