spatiotemporal prediction of chlorophyll-a concentration in the caspian sea using logistic regression and markov chain

Primary production is the most important functional feature of terrestrial and aquatic ecosystems affecting many processes. in this study, we integrated logistic regression and markov chain to predict chlorophylla (chla) concentration as an index of primary production in the caspian sea. we categorized the continuous variable, chla, using quantile method for analysis and prediction. remotelysensed data of chla and nine environmental variables were downloaded from modis dataset for the years 2013 and 2016. the level of chla in 2019 was predicted across the caspian sea. chl‑a data was divided into three distinct levels (i.e. low, medium and high) based on 0.33 and 0.67 quantiles, and a logistic regression model was used based on transition between the levels of chla between 2013 and 2016, and between 2016 and 2019. the markov chain modelling indicated an increasing trend in chla levels (low to medium, low to high, medium to high) for some parts of the caspian sea, and also a stable condition for other parts including transition from medium to medium, high to high had the highest transition probabilities for both periods. from 2013 to 2019, the calculated areas of the pixels having low levels of chla decreased and there were considerable increases in the areas with medium and high chl‑a levels. accordingly, the chla level in the caspian sea at 2019 was predicted to be higher than those of the previous years, especially in the middle and southern parts of the sea.