[gradsusr] trend analysis

Andrew Friedman andfried at gmail.com
Fri Jan 20 17:53:43 EST 2017


Hi Tereza,

Admittedly I haven't used GrADS for trend significance, but here are some thoughts that come to mind:

-ltrend.gs outputs the rmse, which can be squared to get the mse.
-For the trend r2, you could first use the built-in function tcorr with a linear trend time series to get the correlation, and then square this correlation coefficient to get the r2.
-I believe that the trend fit could be constructed as the slope x a linear trend time series, and that squaring and summing these would obtain the regression sum of squares (being careful about anomalies). Scaling this to obtain the mean regression sum of squares and dividing by the mse, you could get the F value.

Perhaps someone with more GrADS statistics experience can offer more concrete suggestions.

Best, Andrew

> On Jan 20, 2017, at 9:21 PM, Tereza Cavazos <tcavazos at cicese.mx> wrote:
> 
> Hi Andrew,
> 
> Is there a way to obtain the statistical significance of the trends in each point?
> 
> Cheers,
> 
> Tereza
>  
> On 20/01/2017 07:59 a.m., Andrew Friedman wrote:
>> Hi Bimal,
>> Yes - you can use ltrend.gs to obtain gridbox trends.
>> -Andrew
>> 
>> On Jan 20, 2017, at 3:19 PM, Bimal Gyawali <bml.gya at gmail.com> wrote:
>> 
>>> Hi Andrew, 
>>> Thank you very much. I think that script gives trend line. But I would like to find the trend on every grid point. Is that possible?
>>> 
>>> Bimal Gyawali
>>> Graduate Research Assistant
>>> Department of Physical and Environment Science
>>> Natural Resources Center
>>> Texas A&M University-Corpus Christi
>>> Corpus Christi, TX 78412
>>> Tel: 435-881-3603
>>> 
>>> 
>>> On Thu, Jan 19, 2017 at 5:14 PM, Andrew Friedman <andfried at gmail.com> wrote:
>>> Hi Bimal,
>>> 
>>> Check out Bin Guan’s script ltrend.gs: http://bguan.bol.ucla.edu/bGASL.html
>>> 
>>> Andrew
>>> 
>>> > On Jan 19, 2017, at 11:34 PM, Bimal Gyawali <bml.gya at gmail.com> wrote:
>>> >
>>> > Hi all,
>>> > Is it possible to calculate trend analysis in spatial domain using GrDAS? if is it possible how can it be done? Please help me. I have monthly data (3D variable, lat, lon and time)from 2000-2016. I wanna calculate the trend of the variable and plot the map. please help me with this if any of you have done it before.
>>> > THank you.
>>> > Bimal Gyawali
>>> >
>>> >
>>> >
>>> > _______________________________________________
>>> > gradsusr mailing list
>>> > gradsusr at gradsusr.org
>>> > http://gradsusr.org/mailman/listinfo/gradsusr
>>> 
>>> 
>>> _______________________________________________
>>> gradsusr mailing list
>>> gradsusr at gradsusr.org
>>> http://gradsusr.org/mailman/listinfo/gradsusr
>>> 
>>> _______________________________________________
>>> gradsusr mailing list
>>> gradsusr at gradsusr.org
>>> http://gradsusr.org/mailman/listinfo/gradsusr
>> 
>> 
>> _______________________________________________
>> gradsusr mailing list
>> 
>> gradsusr at gradsusr.org
>> http://gradsusr.org/mailman/listinfo/gradsusr
> 
> _______________________________________________
> gradsusr mailing list
> gradsusr at gradsusr.org
> http://gradsusr.org/mailman/listinfo/gradsusr




More information about the gradsusr mailing list