I conducted a survey to assess farmers' perception on climate change and found out that their views on temperature were true on analysis of observed temperature data but views on rainfall were difficult to assess with observed rainfall data. While they think rainfall amount had decreases over the past study period, analysis of rainfall trend does not prove so. I therefore think that, apart from possible increasing trend of PET due to increasing temperature, possible increasing trend in rainfall variability over the period could account for their perception. Hence, I wish to assess rainfall variability trend over the study period.
Thanks in advance.
You may use the variation of Standarized Precipitation Index (SPI), to assess the variation of rainfall easily. The method is easy and often proven to be effective. Furthermore, it requires only precipitation data. In addition you could also use frequency analysis methods such as Gumbel frequency analysis or trend analysis techniques such as Mann-Kendalman analysis. Wish you would be able to utilized these tools to fulfill your cause.
I am grateful to you, Lahiru Adayanga.
I have actually used trend and break point detection analysis, specifically, Mann-Kendall, Pettitt test and Hubert Segmentation test. No trend and break points were detected. However, I want to believe those were not enough to detect variations since trend is different from variation, in my view. I shall try other ones (SPI & Gumbel) as you suggested. Once again, I am grateful.
Here you have two nice Investigators to ask, follow, and work...
https://www.researchgate.net/profile/Sara_Del_Rio2
https://www.researchgate.net/profile/A_Penas
an extremely simple approach to test for an increasing (or decreasing) variance trend, would be to divide your data arbitrarily into 2 parts and use an F-test (ratio of variances). The ratio follows a chi-squared distribution. This ignores the serial dependence in your data but would provide an initial indication as to whether the variance in rainfall has increased or decreased over time. More complicated (but more appropriate) methods to assess how rainfall variance may be changing over time include ARIMA and GARCH time-series modelling approaches (e.g.. see Harvey, A.C. 1993. Time Series Models, 2nd ed. MIT Press
I am most grateful to you all for your invaluable ideas and concern.
Dear Andrew,
very well placed by colleagues some methodologies, here we are working with the uncertainty of the observations, for this, we use neural networks as a tool, where we identified which the spatial and temporal variability of rain on a particular station compared to other observational stations along a spatial distribution. It could be an alternative.
best,
David
- Badges
- Science topic
- Similar topics
- Geoscience
- Atmospheric Sciences