The Influence of Double ITCZ Phenomena on Equatorial Ecosystems Based on 32 Years of Remotely Sensed Data

Abstract:

Double ITCZ phenomena (dITCZ) occur in the east Pacific mainly during boreal spring. In our recent studies, 5 different states of the phenomenon could be distinguished from daily satellite observations during 1981-2012: 1) northern, 2) southern, 3) double, 4) equatorial and 5) non-present ITCZ. Analyses on a climatological time-scale suggest that interactions exist between the dITCZ and the large-scale tropical climate (e.g. Haffke et al, paperID A51E-3087, AGU 2014). In this work, we aim to quantify the influence of the dITCZ on vegetation in tropical regions.

We use 32 years of vegetation index data (GIMMS NDVI-3g) and 3 precipitation datasets derived from ERA-Interim reanalysis data, namely: a) the monthly average of ERA precipitation data, b) the monthly average over 32 years and c) the monthly average over 32 years weighted proportional to the occurrences of the 5 ITCZ states. For each dataset, we calculate the temporal correlation with the NDVI. As an indicator quantifying the influence of the dITCZ phenomena on the vegetation, we calculate an improvement factor in NDVI prediction capability when using ITCZ-weighted precipitation as compared to unweighted averaged precipitation (a factor of 1 indicates a prediction capability equivalent to the real precipitation; a factor of 0 indicates a prediction capability equivalent to the average precipitation, i.e. no improvement). This is done by calculating the ratio of the differences between the correlations of c) and b) with the NDVI and the correlations of a) and b) with the NDVI, respectively.

Our initial investigations show strong interactions of ITCZ-weighted precipitation and NDVI on the west coast of South America (i.e. Peru; Ecuador including Galapagos Islands), where 50% - 80% of the precipitation-induced vegetation variability can be explained by the dITCZ phenomena (see figure). In the rest of South America values reach up to 20%. In Africa, influences are rather small/noisy (minor improvements at the horn of Africa; however, degradations in Kenya, i.e. factor < 0). In Oceania/Australia, some regions suggest positive interactions of the dITCZ and the NDVI, most prominent on the east coast of Borneo with up to 30%. Finally, we present the large-scale dITCZ as a potential indicator for variations in vegetation patterns on a global scale in tropical ecosystems.