1.基本理论
由于地表质量发生变化导致地表的负荷变形,利用GRACE数据计算地表的垂直(2)、水平东向(3)和水平北向位移(4)方法如下:
我们之前已经利用冯伟老师提供的MATLAB工具包计算了EWH,下面我们具体看看两类计算公式的异同点:
我们可以发现,有一些部分是一致的,即
唯一不同的就是前面乘以的系数。我们看的,垂直位移的系数是
a*h/(1+k)
而等效水高形式的系数是
ap(2l+1)/3pw(k+1)
因此只要在冯伟老师的代码中修改这部分系数就可以利用GRACE球谐系数计算垂直位移。
然而对于水平位移涉及到勒让德函数的偏导数,下面的公式给出了计算勒让德偏一阶和二阶偏导数的递推公式:
2.计算勒让德函数的程序
我们先给出常用的勒让德函数的递推公式,然后再给出勒让德函数求偏导数的程序
勒让德函数递推公式:
function le=legendre_n(x,Nmax)
% 2024-3-20
le(1:Nmax+1,1:Nmax+1)=0;% l = m =0
le(1,1)=1.0;
% l = 1,m = 0
le(2,1)=sqrt(3.0)*x;
% l = m =1
le(2,2)=sqrt(3.0)*sqrt(1.0-x*x);
% l = m
for i=3:Nmax+1
le(i,i)=sqrt( (1.0-x*x) * (2*i-1) / (2*i-2) )*le(i-1,i-1);
end
% n-m = 1
for i=3:Nmax+1
le(i,i-1)=x*sqrt(2*i-1)*le(i-1,i-1);
end
%others
for j=1:Nmax-1
for i=j+2:Nmax+1
le(i,j)=x*sqrt((2.0*i-3)*(2.0*i-1)/((i-j)*(i+j-2)))*le(i-1,j) ...
-sqrt((2.0*i-1)*(i+j-3.0)*(i-j-1.0)/((2.0*i-5)*(i+j-2)*(i-j)))*le(i-2,j);
end
end
勒让德函数一阶偏导:
function le_dtheta=legendre_partial_n(x,Nmax)
% 2024-3-20
rP = legendre_n(x,Nmax);le_dtheta = zeros(Nmax+1);
for i = 0:Nmax
for j = 0:i
if i == j
le_dtheta(i+1,i+1) = i*x/sqrt(1-x^2) * rP(i+1,i+1);
else
le_dtheta(i+1,j+1) = i*x/sqrt(1-x^2) * rP(i+1,j+1)...
-1/sqrt(1-x^2)*sqrt( (i-j)*(i+j)*(2*i+1)/(2*i-1) ) * rP(i,j+1);
end
end
end
勒让德函数二阶偏导:
function le_dtheta=legendre_partial_nn(x,Nmax)
% 2024-3-20rP = legendre_n(x,Nmax);
le_dtheta = zeros(Nmax+1);
for i = 0:Nmax
for j = 0:i
if i == j
le_dtheta(i+1,i+1) = i*i-j-j*j*x/(1-x^2) * rP(i+1,i+1);
else
le_dtheta(i+1,j+1) = i*i-j-j*j*x/(1-x^2) * rP(i+1,j+1)...
+x/(1-x^2)*sqrt( (i-j)*(i+j)*(2*i+1)/(2*i-1) ) * rP(i,j+1);
end
end
end
实际运算例子:
lon = 0.5:1:359.5;lat = -89.5:1:89.5;
[lon,lat] = meshgrid(lon,lat);Nmax = 60; % max degree
Nlat = size(lon,1); % size of lat gridfor ilat = 1: Nlat
rCoLat= 90.0-lat(ilat,1);
rCoLat2= cos(deg2rad(rCoLat));
rP = legendre_n( rCoLat2, Nmax );
rP_d1 = legendre_partial_n( rCoLat2, Nmax );
rP_d2 = legendre_partial_nn( rCoLat2, Nmax );
end
可以得到61×61阶的结果,这与60阶GRACE球谐数据匹配。
3.计算GRACE数据得到的地表负荷位移
这次之前,我们需要弄清楚一个事实,即大地水准面系数和质量系数之间的关系。
回到Whar et al.(1998)年的论文,下式表达了两者之间的关系,即我们处理的GRACE数据是大地水准面系数,而利用GLDAS等水文模型导出的是质量球谐系数。
下面我们利用GLDAS数据计算地表位移。利用冯伟老师的工具箱,我们可以得到球谐展开的GLDAS系数,该系数是质量系数,即上式的右边,而如果要计算地表位移,需要首先将其转成大地水准面系数(左边),然后计算负荷位移。上一篇专栏我介绍了一个修改的,可以将GLDAS水文模型球形展开的例子,感兴趣的可以阅读一下。在冯伟老师的工具箱中,具体需要修改的函数是gmt_gc2mc。这里我编写一个gmt_mc2gc的函数,结合gmt_gc2lc即可得到负荷垂直位移。
function outcoeff = gmt_mc2gc(incoeff)
% Converts geoid coefficients (gc) to mass coefficients (mc)
% FENG Wei 19/04/2015
% State Key Laboratory of Geodesy and Earth's Dynamics
% Institute of Geodesy and Geophysics, Chinese Academy of Sciences
% [email protected]
% modified 20/3/2024 Chistrong Wen
[rows,cols] = size(incoeff);
if rows == cols % field is in CS-format
maxdeg = rows - 1;
% incoeff = cs2sc(incoeff);
elseif cols-2*rows == -1 % field is in SC-format already
maxdeg = rows - 1;
incoeff = gmt_sc2cs(incoeff); % convert to CS-format
else
error('Check format of field.')
endae = 6378136.3; % semi-major openaxis of ellipsoid [m]
rho_w = 1000; % density of water kg/m^3
rho_ave = 5517; % average density of the earth kg/m^3
%% mc2load
dens = (3.*rho_w)/(ae*rho_ave);
% Han and Wahr Love numbers
load LoveNumbers_lln.mat
% %Refer to Chamnbers deg1_coef.txt Fengwei
k(2) = 0.021;
sc = gmt_cs2sc(incoeff);
% loop over degrees to multiply each row of coefficients in sc-format
% with the factor from equation (13) of Wahr et al. 1998
for ll = 0:1:maxdeg
factor(ll+1) = dens*( 1+k(ll+1) )/( 2*ll+1 );
scnew(ll+1,:) = factor(ll+1)*sc(ll+1,:);
endoutcoeff = gmt_sc2cs(scnew);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% CCBY Chistrong Wen
% University of Chinese Academy of Sciences
% 2024-3-20
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% GLDAS INFO
O.lon = mascon_csr.lon;O.lat = mascon_csr.lat;O.rg = mascon_csr.rg(:,:,12);
figure
subplot(3,1,1),wzq_plot(O),title('surface mass'),caxis([-20,20]),colorbar
%% expand to 60 degree
grid = O.rg;
degree_max = 60;
[cs] = gmt_grid2cs(grid,degree_max);
%% applying filter method
destrip_method = 'NONE';
radius_filter = 000;
type = 1;gc_cs = gmt_mc2gc(cs);
load_cs = gmt_mc2load(gc_cs);outcoeff = gmt_gc2lc(gc_cs);
[grid_filter_load]=gmt_cs2grid(outcoeff,radius_filter,type,destrip_method);
[grid_filter_mass]=gmt_cs2grid(cs,radius_filter,type,destrip_method);lon = 0.5:1:359.5;
lat = -89.5:1:89.5;
[lon,lat] = meshgrid(lon,lat);
subplot(3,1,3)
out_soil_filter.lon = lon;
out_soil_filter.lat = lat;
out_soil_filter.rg = grid_filter_load;
wzq_plot(out_soil_filter)
title('degree 60 mass'),colorbarsubplot(3,1,2)
figure
out_soil_filter.lon = lon;
out_soil_filter.lat = lat;
out_soil_filter.rg = grid_filter_mass;
wzq_plot(out_soil_filter)
title('mass load'),colorbar
结果图:
代码包见资源。
♥欢迎点赞收藏♥
标签:Nmax,le,基本原理,lon,filter,lat,球谐,GRACE,gmt From: https://blog.csdn.net/weixin_43047707/article/details/136820666